D5 - Office for National Statistics requirements for NHS-England (formerly NHS Digital) data, for the purposes of Statistics and Statistical Research, under section 45 of the Statistics and Registration Services Act 2007 as amended by the Digital Economy Act 2017

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS England, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://uksa.statisticsauthority.gov.uk/publication/data-access-policy-and-legislation/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this Agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this Data Sharing Agreement. This process involved working closely with NHS England’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS England will also publish the details of this Data Sharing Agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this Agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS England approval. It would also require an appropriate alternative legal gateway because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS England and enter into an amended Data Sharing Agreement before proceeding with that new purpose.

Dataset 1: Birth Notifications data

NHS England has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

There are a wide range of statistical uses to which the Office for National Statistics (ONS) intends to put Birth Notifications data. All use of Birth Notifications data by ONS will be to improve the availability and quality of statistics as part of ONS’s function to produce statistics for the public good.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required.

1.1 Birth and Child Mortality Statistics

The primary statistical purpose for which this information will be used is analysis of births, maternities, infant mortality and child health outcomes. Analyses are made publicly available as aggregate National Statistics.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics. However, there are some limitations with the GRO data, including a time lag, a lack of key information such as length of gestation and ethnicity of the baby, as well as some missing values in the fields that are available. To mitigate these limitations, the NHS England birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. The identifying information required from NHS England will enable and help quality assure this linkage, which in turn will enable ONS to produce more comprehensive and accurate statistics on births and child health outcomes.

As well as linking birth notifications data to birth registrations and deaths data to produce National Statistics these data will also be linked to other sources in order improve and development new statistics. For example, birth notifications data will be linked to:

• birth registration, deaths and census data to identify more detailed characteristics of the household and mother, such as ethnicity, and so better understand inequalities, risk factors and variation in child health outcomes

• deaths, to identify bereaved children, and to other NHS England data to identify subsequent inequalities and outcomes

• other ONS and NHS England data such as Hospital Episode Statistics to better determine underlying cause of death, understand inequalities, risk factors and variation in child outcomes

1.2 Improving data linkage methodology

ONS plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link information relating to siblings born at different times (i.e. not twins) using NHS number of the mother. This produces very accurate linkage of siblings.

ONS can then attempt to link siblings together using only the data available in the registration data – i.e. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the results of the latter linkage method matches those achieved when linking using mother’s NHS number. This will inform the best matching methodology to use when NHS number of the mother is not available (for example in pre-2005 birth registration data).

1.3 Improving population and migration statistics

ONS plans to use births notifications data to support development and improvements of population and migration statistics. This includes a range of work such as quality assurance of Census data, contribution to ONS’s population and migration statistics transformation programme, to put administrative data first and make recommendations on the future of the decennial Census. For example, birth notifications data will be:

• used to validate Census data, with a focus on population sub-groups, such as the under 1s, returned in Census data and subsequently supporting population outputs

• used to contribute to work to estimate or quality assure characteristics of the population, such as ethnicity or age, along with other data sources

• linked to other data sources in order to carry out such improvements to existing or develop new Official Statistics, and to quality assurance other data sources

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – e.g. additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – e.g. where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

• Research and development of methodology – e.g. improving linkage methodology for siblings

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS England data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Based Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Based Census Project (ABC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realised.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

As part of this work ONS will be producing a non-disclosive, de-identified Administrative Based Census (ABC) dataset for research use. This will be a heavily derived data product containing only the age, sex and Lower Layer Super Output Area (LSOA) for records included within the ABC. This ABC output dataset will be produced through amalgamating multiple administrative data sources (including NHS Personal Demographics Service, Hospital Episode Statistics, Emergency Care Dataset plus a multitude of data from other sources) to generate a de-identified record for every usual resident in England and Wales and help estimate the population at a local level. This de-identified, non-disclosive ONS product (the ABC) will be made available for approved / accredited research use within ONS operated Trusted Research Environments (TREs).

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity Data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity, but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data-based ethnicity estimates will be important when deciding whether to move to an admin data-based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS England data experts to understand the limitations of the HES ethnicity data and there are several; for example, coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about when, where, and why people have accessed hospital services, for example diagnosis and procedures codes, for a range of statistical purposes in line with ONS’s function to produce statistics for the public good. All use of the HES data for health analysis will be to improve the availability and quality of health statistics.

The diagnosis and procedures information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, for the first supply of HES data to ONS the volume of this information was minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services was shared with ONS, compared with the information about when and where people have accessed services. ONS has now used and published analysis using these HES data for several purposes described in this Agreement and DARS-NIC-400304-S1P1B (which allows reuse of the HES data ONS hold). Based on learning from this work ONS are now requesting further additional variables and historical times series updates where some variables were only requested for a subset of years to the current HES subset they receive.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

(i) Understanding healthy life expectancy.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

(ii) Understanding health inequalities.

Inclusion is one of the 4 pillars of the ONS strategy ‘Statistics for the Public Good’, as such exploring inequalities across outcomes is a priority for the ONS. Making use of linked morbidity, mortality, census and other data will allow ONS to produce statistics on health inequalities and provide these at a granular level. In particular, linking to these health data will help ONS to better understand if inequalities in outcomes persist when health status or pre-existing conditions are taken into account. For example, ONS have already used HES data to understand comorbidities of the population and to take this into account when modelling the risk of COVID-19 deaths by ethnicity (under DARS-NIC-400304-S1P1B). This allowed ONS to provide evidence that inequalities in outcomes still exist despite controlling for certain, but not all, health conditions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly.

It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible, to develop new measures of avoidable mortality in the elderly that could have been avoided.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality.

The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

d. ONS will use this data in conjunction with other health sources, and non-health sources such as Census, income and benefits data, and survey data, to produce a range of statistics on the interaction between health and the labour market. This is bi-directional, looking both at how health affects economic outcomes and vice versa. Work would include but not be limited to:

• Producing statistics on the prevalence of conditions in people who are inactive in the labour market due to ill health

• Modelling whether a change in the prevalence of certain conditions explain part of the increase in labour market inactivity observed since 2019.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation, and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, new-born and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS England and UK Health Security Agency (UKHSA) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability, and geography. In some cases, NHS England / UKHSA will not hold data that would enable this but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is usually released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released over the year in a series of themed packages. Child and infant mortality statistics and unexplained infant deaths are published annually. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

ONS are looking to improve and develop new statistics using newly linked data to explore inequalities, risk factors and variation in child outcomes. Outputs in the form of presentations, analytical articles and methodology reports will be created alongside data tables as appropriate. ONS publishes all its official statistics on its website.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Improving population and migration statistics: This will result in internal, and potentially external, ONS reports and presentations on how births notification data could be used to improve and develop official statistics, and potentially used directly in the development of new official statistics. ONS publishes all its official statistics on its website.

Dataset 2, 3 and 4: Hospital Episode Statistics, Improving Access to Psychological Therapies and Emergency Care Dataset data

The initial uses to which ONS will put HES, IAPT and ECDS data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Initial work has been carried out and new statistics published using HES data as part of the response to the coronavirus pandemic which has assured ONS of the quality and validity of using these data as initially intended such that additional data is being requested as part of this Agreement to address limitations in the initial supply of HES data. Research outputs have been published on the ONS website, for example: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES, IAPT or ECDS data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users and will seek to provide frequent updates on these projects during that first stage.

As part of ONS' work on producing more accurate statistics on the population at a local level through administrative data, a non-disclosive, de-identified Administrative Based Census (ABC) dataset will be created for research use. This will be a heavily derived data product containing only the age, sex and Lower Layer Super Output Area (LSOA) for records included within the ABC. This ABC output dataset will be produced through amalgamating multiple administrative data sources (including NHS Personal Demographics Service, Hospital Episode Statistics, Emergency Care Dataset plus a multitude of data from other sources) to generate a de-identified record for every usual resident in England and Wales and help estimate the population at a local level. This de-identified, non-disclosive ONS product (the ABC) will be made available for approved / accredited research use within ONS operated Trusted Research Environments (TREs).

Dataset 1: Benefits of the ONS births statistics which depend on the Birth Notifications data.

ONS has released research and statistics regarding Life expectancy, healthy life expectancy, disability-free life expectancy, slope index of inequality and range at birth and age 65 by national deprivation deciles across England from 2011 to 2019. Alongside our National Statistics 'Births' releases these have been used by various users across central government (Department of Health & Social Care, UK Health Security Agency, Department for Work & Pensions) local authorities, charities and international bodies (Eurostat and the UN).

Dataset 2: Predicted Benefits of the uses for Hospital Episode Statistics data

2.1. Benefits of the Admin Based Census Project and Improved Migration Statistics

ONS has used both HES and ECDS within construction of the Statistical Population Dataset (SPD), specifically they're used as a 'sign of life' or activity indicator to identify whether someone should be included in the usually resident population. This is a new methodology and still under development, most recently ONS published an update on this work in February 2023.

2.2. Benefits of the Health Analyses

ONS has used HES extensively for key health analysis:

Published April 2022, Suicides among people diagnosed with severe health conditions, England. This was based on mortality records linked to the 2011 Census and Hospital Episode Statistics.

Published September 2023, Climate-related mortality and hospital admissions, England. Counts and rates of deaths and hospital admissions associated with temperature for England from 2001 to 2020.

Published November 2023, Quality of ethnicity data in Hospital Episode Statistics, England, including APC, OP and A&E. Including ECDS.

Published April 2024, Health Outcomes of unpaid carers using, Hospital Episode Statistics, England.

2.3. Benefits of Improving the ONS Address Register

This work is still under development, however progress here has also helped in the construction of the Admin Based Census. Data on communal establishments in particular has been vital for improving ONS' Address Index.

2.4. Benefits of Improving UK household Expenditure

This work is still ongoing, ONS will provide an update on this in due course.

2.5. Benefits of Improving ONS Sustainable Development Indicators

ONS most recently published an article in April 2023 providing an update on our research to improve reporting of sex and gender within the context of the UN's Sustainable Development Goals, mentioning the use of NHS data to form part of the published indicators.

2.6 Benefits of supporting Covid-19 analysis

During the pandemic the ONS published extensive COVID-19 statistics, many of which used NHS data (provisioned under DARS-NIC-400304). Most recently ONS has published the following articles:

Published July 2022, COVID-19 hospital admissions by vaccination and pregnancy status, England.

Published March 2023, Risk of death following COVID-19 vaccination or positive SARS-CoV-2 test in young people, England.

More detailed benefits are set out within DARS-NIC-400304

Dataset 3: Benefits of the uses for IAPT data

3.1. Benefits of the Admin Data Census project and improved migration statistics

This data has been used under the same population statistics purposes as HES and ECDS, see 2.1 above.

3.2. Benefits of Statistical Research and Health Analyses using IAPT data

Most recently, ONS has published the following article:

Published June 2022, Socio-demographic differences in use of the Improving Access to Psychological Therapies services. IAPT services treatment rates, Common mental disorder rates and proportion of people with probable CMD who received treatment through IAPT services, for England, 1 April 2017 to 31 March 2018.

Dataset 4: Benefits of the uses for ECDS data

As ECDS is replacing the HES A&E data, the benefits described for HES data in section 2 will also apply to ECDS data.

In addition to the those described in section 2, ECDS data would also have an additional benefits as follows:

4.1. Benefits of using ECDS data for Crime Statistics (including knife and drug related crime)

Most recently, ONS has published the following article:

Published 16 November 2023, Offences involving the use of weapons, data relating to offences involving weapons as recorded by police within the ECDS dataset.

4.2 Waiting time on Mortality

This work is still ongoing, ONS will provide an update on this in due course.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS England, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://uksa.statisticsauthority.gov.uk/publication/data-access-policy-and-legislation/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this Agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this Data Sharing Agreement. This process involved working closely with NHS England’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS England will also publish the details of this Data Sharing Agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this Agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS England approval. It would also require an appropriate alternative legal gateway because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS England and enter into an amended Data Sharing Agreement before proceeding with that new purpose.

Dataset 1: Birth Notifications data

NHS England has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

There are a wide range of statistical uses to which the Office for National Statistics (ONS) intends to put Birth Notifications data. All use of Birth Notifications data by ONS will be to improve the availability and quality of statistics as part of ONS’s function to produce statistics for the public good.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required.

1.1 Birth and Child Mortality Statistics

The primary statistical purpose for which this information will be used is analysis of births, maternities, infant mortality and child health outcomes. Analyses are made publicly available as aggregate National Statistics.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics. However, there are some limitations with the GRO data, including a time lag, a lack of key information such as length of gestation and ethnicity of the baby, as well as some missing values in the fields that are available. To mitigate these limitations, the NHS England birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. The identifying information required from NHS England will enable and help quality assure this linkage, which in turn will enable ONS to produce more comprehensive and accurate statistics on births and child health outcomes.

As well as linking birth notifications data to birth registrations and deaths data to produce National Statistics these data will also be linked to other sources in order improve and development new statistics. For example, birth notifications data will be linked to:

• birth registration, deaths and census data to identify more detailed characteristics of the household and mother, such as ethnicity, and so better understand inequalities, risk factors and variation in child health outcomes

• deaths, to identify bereaved children, and to other NHS England data to identify subsequent inequalities and outcomes

• other ONS and NHS England data such as Hospital Episode Statistics to better determine underlying cause of death, understand inequalities, risk factors and variation in child outcomes

1.2 Improving data linkage methodology

ONS plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link information relating to siblings born at different times (i.e. not twins) using NHS number of the mother. This produces very accurate linkage of siblings.

ONS can then attempt to link siblings together using only the data available in the registration data – i.e. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the results of the latter linkage method matches those achieved when linking using mother’s NHS number. This will inform the best matching methodology to use when NHS number of the mother is not available (for example in pre-2005 birth registration data).

1.3 Improving population and migration statistics

ONS plans to use births notifications data to support development and improvements of population and migration statistics. This includes a range of work such as quality assurance of Census data, contribution to ONS’s population and migration statistics transformation programme, to put administrative data first and make recommendations on the future of the decennial Census. For example, birth notifications data will be:

• used to validate Census data, with a focus on population sub-groups, such as the under 1s, returned in Census data and subsequently supporting population outputs

• used to contribute to work to estimate or quality assure characteristics of the population, such as ethnicity or age, along with other data sources

• linked to other data sources in order to carry out such improvements to existing or develop new Official Statistics, and to quality assurance other data sources

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – e.g. additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – e.g. where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

• Research and development of methodology – e.g. improving linkage methodology for siblings

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS England data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Based Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Based Census Project (ABC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realised.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

As part of this work ONS will be producing a non-disclosive, de-identified Administrative Based Census (ABC) dataset for research use. This will be a heavily derived data product containing only the age, sex and Lower Layer Super Output Area (LSOA) for records included within the ABC. This ABC output dataset will be produced through amalgamating multiple administrative data sources (including NHS Personal Demographics Service, Hospital Episode Statistics, Emergency Care Dataset plus a multitude of data from other sources) to generate a de-identified record for every usual resident in England and Wales and help estimate the population at a local level. This de-identified, non-disclosive ONS product (the ABC) will be made available for approved / accredited research use within ONS operated Trusted Research Environments (TREs).

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity Data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity, but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data-based ethnicity estimates will be important when deciding whether to move to an admin data-based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS England data experts to understand the limitations of the HES ethnicity data and there are several; for example, coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about when, where, and why people have accessed hospital services, for example diagnosis and procedures codes, for a range of statistical purposes in line with ONS’s function to produce statistics for the public good. All use of the HES data for health analysis will be to improve the availability and quality of health statistics.

The diagnosis and procedures information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, for the first supply of HES data to ONS the volume of this information was minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services was shared with ONS, compared with the information about when and where people have accessed services. ONS has now used and published analysis using these HES data for several purposes described in this Agreement and DARS-NIC-400304-S1P1B (which allows reuse of the HES data ONS hold). Based on learning from this work ONS are now requesting further additional variables and historical times series updates where some variables were only requested for a subset of years to the current HES subset they receive.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

(i) Understanding healthy life expectancy.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

(ii) Understanding health inequalities.

Inclusion is one of the 4 pillars of the ONS strategy ‘Statistics for the Public Good’, as such exploring inequalities across outcomes is a priority for the ONS. Making use of linked morbidity, mortality, census and other data will allow ONS to produce statistics on health inequalities and provide these at a granular level. In particular, linking to these health data will help ONS to better understand if inequalities in outcomes persist when health status or pre-existing conditions are taken into account. For example, ONS have already used HES data to understand comorbidities of the population and to take this into account when modelling the risk of COVID-19 deaths by ethnicity (under DARS-NIC-400304-S1P1B). This allowed ONS to provide evidence that inequalities in outcomes still exist despite controlling for certain, but not all, health conditions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly.

It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible, to develop new measures of avoidable mortality in the elderly that could have been avoided.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality.

The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

d. ONS will use this data in conjunction with other health sources, and non-health sources such as Census, income and benefits data, and survey data, to produce a range of statistics on the interaction between health and the labour market. This is bi-directional, looking both at how health affects economic outcomes and vice versa. Work would include but not be limited to:

• Producing statistics on the prevalence of conditions in people who are inactive in the labour market due to ill health

• Modelling whether a change in the prevalence of certain conditions explain part of the increase in labour market inactivity observed since 2019.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation, and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, new-born and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS England and UK Health Security Agency (UKHSA) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability, and geography. In some cases, NHS England / UKHSA will not hold data that would enable this but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is usually released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released over the year in a series of themed packages. Child and infant mortality statistics and unexplained infant deaths are published annually. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

ONS are looking to improve and develop new statistics using newly linked data to explore inequalities, risk factors and variation in child outcomes. Outputs in the form of presentations, analytical articles and methodology reports will be created alongside data tables as appropriate. ONS publishes all its official statistics on its website.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Improving population and migration statistics: This will result in internal, and potentially external, ONS reports and presentations on how births notification data could be used to improve and develop official statistics, and potentially used directly in the development of new official statistics. ONS publishes all its official statistics on its website.

Dataset 2, 3 and 4: Hospital Episode Statistics, Improving Access to Psychological Therapies and Emergency Care Dataset data

The initial uses to which ONS will put HES, IAPT and ECDS data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Initial work has been carried out and new statistics published using HES data as part of the response to the coronavirus pandemic which has assured ONS of the quality and validity of using these data as initially intended such that additional data is being requested as part of this Agreement to address limitations in the initial supply of HES data. Research outputs have been published on the ONS website, for example: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES, IAPT or ECDS data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users and will seek to provide frequent updates on these projects during that first stage.

As part of ONS' work on producing more accurate statistics on the population at a local level through administrative data, a non-disclosive, de-identified Administrative Based Census (ABC) dataset will be created for research use. This will be a heavily derived data product containing only the age, sex and Lower Layer Super Output Area (LSOA) for records included within the ABC. This ABC output dataset will be produced through amalgamating multiple administrative data sources (including NHS Personal Demographics Service, Hospital Episode Statistics, Emergency Care Dataset plus a multitude of data from other sources) to generate a de-identified record for every usual resident in England and Wales and help estimate the population at a local level. This de-identified, non-disclosive ONS product (the ABC) will be made available for approved / accredited research use within ONS operated Trusted Research Environments (TREs).

Dataset 1: Benefits of the ONS births statistics which depend on the Birth Notifications data.

ONS has released research and statistics regarding Life expectancy, healthy life expectancy, disability-free life expectancy, slope index of inequality and range at birth and age 65 by national deprivation deciles across England from 2011 to 2019. Alongside our National Statistics 'Births' releases these have been used by various users across central government (Department of Health & Social Care, UK Health Security Agency, Department for Work & Pensions) local authorities, charities and international bodies (Eurostat and the UN).

Dataset 2: Predicted Benefits of the uses for Hospital Episode Statistics data

2.1. Benefits of the Admin Based Census Project and Improved Migration Statistics

ONS has used both HES and ECDS within construction of the Statistical Population Dataset (SPD), specifically they're used as a 'sign of life' or activity indicator to identify whether someone should be included in the usually resident population. This is a new methodology and still under development, most recently ONS published an update on this work in February 2023.

2.2. Benefits of the Health Analyses

ONS has used HES extensively for key health analysis:

Published April 2022, Suicides among people diagnosed with severe health conditions, England. This was based on mortality records linked to the 2011 Census and Hospital Episode Statistics.

Published September 2023, Climate-related mortality and hospital admissions, England. Counts and rates of deaths and hospital admissions associated with temperature for England from 2001 to 2020.

Published November 2023, Quality of ethnicity data in Hospital Episode Statistics, England, including APC, OP and A&E. Including ECDS.

Published April 2024, Health Outcomes of unpaid carers using, Hospital Episode Statistics, England.

2.3. Benefits of Improving the ONS Address Register

This work is still under development, however progress here has also helped in the construction of the Admin Based Census. Data on communal establishments in particular has been vital for improving ONS' Address Index.

2.4. Benefits of Improving UK household Expenditure

This work is still ongoing, ONS will provide an update on this in due course.

2.5. Benefits of Improving ONS Sustainable Development Indicators

ONS most recently published an article in April 2023 providing an update on our research to improve reporting of sex and gender within the context of the UN's Sustainable Development Goals, mentioning the use of NHS data to form part of the published indicators.

2.6 Benefits of supporting Covid-19 analysis

During the pandemic the ONS published extensive COVID-19 statistics, many of which used NHS data (provisioned under DARS-NIC-400304). Most recently ONS has published the following articles:

Published July 2022, COVID-19 hospital admissions by vaccination and pregnancy status, England.

Published March 2023, Risk of death following COVID-19 vaccination or positive SARS-CoV-2 test in young people, England.

More detailed benefits are set out within DARS-NIC-400304

Dataset 3: Benefits of the uses for IAPT data

3.1. Benefits of the Admin Data Census project and improved migration statistics

This data has been used under the same population statistics purposes as HES and ECDS, see 2.1 above.

3.2. Benefits of Statistical Research and Health Analyses using IAPT data

Most recently, ONS has published the following article:

Published June 2022, Socio-demographic differences in use of the Improving Access to Psychological Therapies services. IAPT services treatment rates, Common mental disorder rates and proportion of people with probable CMD who received treatment through IAPT services, for England, 1 April 2017 to 31 March 2018.

Dataset 4: Benefits of the uses for ECDS data

As ECDS is replacing the HES A&E data, the benefits described for HES data in section 2 will also apply to ECDS data.

In addition to the those described in section 2, ECDS data would also have an additional benefits as follows:

4.1. Benefits of using ECDS data for Crime Statistics (including knife and drug related crime)

Most recently, ONS has published the following article:

Published 16 November 2023, Offences involving the use of weapons, data relating to offences involving weapons as recorded by police within the ECDS dataset.

4.2 Waiting time on Mortality

This work is still ongoing, ONS will provide an update on this in due course.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS England, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://uksa.statisticsauthority.gov.uk/publication/data-access-policy-and-legislation/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this Agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this Data Sharing Agreement. This process involved working closely with NHS England’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS England will also publish the details of this Data Sharing Agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this Agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS England approval. It would also require an appropriate alternative legal gateway because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS England and enter into an amended Data Sharing Agreement before proceeding with that new purpose.

Dataset 1: Birth Notifications data

NHS England has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

There are a wide range of statistical uses to which the Office for National Statistics (ONS) intends to put Birth Notifications data. All use of Birth Notifications data by ONS will be to improve the availability and quality of statistics as part of ONS’s function to produce statistics for the public good.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required.

1.1 Birth and Child Mortality Statistics

The primary statistical purpose for which this information will be used is analysis of births, maternities, infant mortality and child health outcomes. Analyses are made publicly available as aggregate National Statistics.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics. However, there are some limitations with the GRO data, including a time lag, a lack of key information such as length of gestation and ethnicity of the baby, as well as some missing values in the fields that are available. To mitigate these limitations, the NHS England birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. The identifying information required from NHS England will enable and help quality assure this linkage, which in turn will enable ONS to produce more comprehensive and accurate statistics on births and child health outcomes.

As well as linking birth notifications data to birth registrations and deaths data to produce National Statistics these data will also be linked to other sources in order improve and development new statistics. For example, birth notifications data will be linked to:

• birth registration, deaths and census data to identify more detailed characteristics of the household and mother, such as ethnicity, and so better understand inequalities, risk factors and variation in child health outcomes

• deaths, to identify bereaved children, and to other NHS England data to identify subsequent inequalities and outcomes

• other ONS and NHS England data such as Hospital Episode Statistics to better determine underlying cause of death, understand inequalities, risk factors and variation in child outcomes

1.2 Improving data linkage methodology

ONS plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link information relating to siblings born at different times (i.e. not twins) using NHS number of the mother. This produces very accurate linkage of siblings.

ONS can then attempt to link siblings together using only the data available in the registration data – i.e. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the results of the latter linkage method matches those achieved when linking using mother’s NHS number. This will inform the best matching methodology to use when NHS number of the mother is not available (for example in pre-2005 birth registration data).

1.3 Improving population and migration statistics

ONS plans to use births notifications data to support development and improvements of population and migration statistics. This includes a range of work such as quality assurance of Census data, contribution to ONS’s population and migration statistics transformation programme, to put administrative data first and make recommendations on the future of the decennial Census. For example, birth notifications data will be:

• used to validate Census data, with a focus on population sub-groups, such as the under 1s, returned in Census data and subsequently supporting population outputs

• used to contribute to work to estimate or quality assure characteristics of the population, such as ethnicity or age, along with other data sources

• linked to other data sources in order to carry out such improvements to existing or develop new Official Statistics, and to quality assurance other data sources

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – e.g. additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – e.g. where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

• Research and development of methodology – e.g. improving linkage methodology for siblings

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS England data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realised.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity Data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity, but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data-based ethnicity estimates will be important when deciding whether to move to an admin data-based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS England data experts to understand the limitations of the HES ethnicity data and there are several; for example, coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about when, where, and why people have accessed hospital services, for example diagnosis and procedures codes, for a range of statistical purposes in line with ONS’s function to produce statistics for the public good. All use of the HES data for health analysis will be to improve the availability and quality of health statistics.

The diagnosis and procedures information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, for the first supply of HES data to ONS the volume of this information was minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services was shared with ONS, compared with the information about when and where people have accessed services. ONS has now used and published analysis using these HES data for several purposes described in this Agreement and DARS-NIC-400304-S1P1B (which allows reuse of the HES data ONS hold). Based on learning from this work ONS are now requesting further additional variables and historical times series updates where some variables were only requested for a subset of years to the current HES subset they receive.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

(i) Understanding healthy life expectancy.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

(ii) Understanding health inequalities.

Inclusion is one of the 4 pillars of the ONS strategy ‘Statistics for the Public Good’, as such exploring inequalities across outcomes is a priority for the ONS. Making use of linked morbidity, mortality, census and other data will allow ONS to produce statistics on health inequalities and provide these at a granular level. In particular, linking to these health data will help ONS to better understand if inequalities in outcomes persist when health status or pre-existing conditions are taken into account. For example, ONS have already used HES data to understand comorbidities of the population and to take this into account when modelling the risk of COVID-19 deaths by ethnicity (under DARS-NIC-400304-S1P1B). This allowed ONS to provide evidence that inequalities in outcomes still exist despite controlling for certain, but not all, health conditions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly.

It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible, to develop new measures of avoidable mortality in the elderly that could have been avoided.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality.

The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

d. ONS will use this data in conjunction with other health sources, and non-health sources such as Census, income and benefits data, and survey data, to produce a range of statistics on the interaction between health and the labour market. This is bi-directional, looking both at how health affects economic outcomes and vice versa. Work would include but not be limited to:

• Producing statistics on the prevalence of conditions in people who are inactive in the labour market due to ill health

• Modelling whether a change in the prevalence of certain conditions explain part of the increase in labour market inactivity observed since 2019.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation, and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, new-born and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS England and UK Health Security Agency (UKHSA) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability, and geography. In some cases, NHS England / UKHSA will not hold data that would enable this but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up-to-date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

A separate Agr

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is usually released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released over the year in a series of themed packages. Child and infant mortality statistics and unexplained infant deaths are published annually. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

ONS are looking to improve and develop new statistics using newly linked data to explore inequalities, risk factors and variation in child outcomes. Outputs in the form of presentations, analytical articles and methodology reports will be created alongside data tables as appropriate. ONS publishes all its official statistics on its website.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Improving population and migration statistics: This will result in internal, and potentially external, ONS reports and presentations on how births notification data could be used to improve and develop official statistics, and potentially used directly in the development of new official statistics. ONS publishes all its official statistics on its website.

Dataset 2, 3 and 4: Hospital Episode Statistics, Improving Access to Psychological Therapies and Emergency Care Dataset data

The initial uses to which ONS will put HES, IAPT and ECDS data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Initial work has been carried out and new statistics published using HES data as part of the response to the coronavirus pandemic which has assured ONS of the quality and validity of using these data as initially intended such that additional data is being requested as part of this Agreement to address limitations in the initial supply of HES data. Research outputs have been published on the ONS website, for example: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES, IAPT or ECDS data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS England, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://uksa.statisticsauthority.gov.uk/publication/data-access-policy-and-legislation/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this Agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this Data Sharing Agreement. This process involved working closely with NHS England’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS England will also publish the details of this Data Sharing Agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this Agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS England approval. It would also require an appropriate alternative legal gateway because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS England and enter into an amended Data Sharing Agreement before proceeding with that new purpose.

Dataset 1: Birth Notifications data

NHS England has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

There are a wide range of statistical uses to which the Office for National Statistics (ONS) intends to put Birth Notifications data. All use of Birth Notifications data by ONS will be to improve the availability and quality of statistics as part of ONS’s function to produce statistics for the public good.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required.

1.1 Birth and Child Mortality Statistics

The primary statistical purpose for which this information will be used is analysis of births, maternities, infant mortality and child health outcomes. Analyses are made publicly available as aggregate National Statistics.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics. However, there are some limitations with the GRO data, including a time lag, a lack of key information such as length of gestation and ethnicity of the baby, as well as some missing values in the fields that are available. To mitigate these limitations, the NHS England birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. The identifying information required from NHS England will enable and help quality assure this linkage, which in turn will enable ONS to produce more comprehensive and accurate statistics on births and child health outcomes.

As well as linking birth notifications data to birth registrations and deaths data to produce National Statistics these data will also be linked to other sources in order improve and development new statistics. For example, birth notifications data will be linked to:

• birth registration, deaths and census data to identify more detailed characteristics of the household and mother, such as ethnicity, and so better understand inequalities, risk factors and variation in child health outcomes

• deaths, to identify bereaved children, and to other NHS England data to identify subsequent inequalities and outcomes

• other ONS and NHS England data such as Hospital Episode Statistics to better determine underlying cause of death, understand inequalities, risk factors and variation in child outcomes

1.2 Improving data linkage methodology

ONS plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link information relating to siblings born at different times (i.e. not twins) using NHS number of the mother. This produces very accurate linkage of siblings.

ONS can then attempt to link siblings together using only the data available in the registration data – i.e. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the results of the latter linkage method matches those achieved when linking using mother’s NHS number. This will inform the best matching methodology to use when NHS number of the mother is not available (for example in pre-2005 birth registration data).

1.3 Improving population and migration statistics

ONS plans to use births notifications data to support development and improvements of population and migration statistics. This includes a range of work such as quality assurance of Census data, contribution to ONS’s population and migration statistics transformation programme, to put administrative data first and make recommendations on the future of the decennial Census. For example, birth notifications data will be:

• used to validate Census data, with a focus on population sub-groups, such as the under 1s, returned in Census data and subsequently supporting population outputs

• used to contribute to work to estimate or quality assure characteristics of the population, such as ethnicity or age, along with other data sources

• linked to other data sources in order to carry out such improvements to existing or develop new Official Statistics, and to quality assurance other data sources

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – e.g. additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – e.g. where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

• Research and development of methodology – e.g. improving linkage methodology for siblings

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS England data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realised.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity Data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity, but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data-based ethnicity estimates will be important when deciding whether to move to an admin data-based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS England data experts to understand the limitations of the HES ethnicity data and there are several; for example, coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about when, where, and why people have accessed hospital services, for example diagnosis and procedures codes, for a range of statistical purposes in line with ONS’s function to produce statistics for the public good. All use of the HES data for health analysis will be to improve the availability and quality of health statistics.

The diagnosis and procedures information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, for the first supply of HES data to ONS the volume of this information was minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services was shared with ONS, compared with the information about when and where people have accessed services. ONS has now used and published analysis using these HES data for several purposes described in this Agreement and DARS-NIC-400304-S1P1B (which allows reuse of the HES data ONS hold). Based on learning from this work ONS are now requesting further additional variables and historical times series updates where some variables were only requested for a subset of years to the current HES subset they receive.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

(i) Understanding healthy life expectancy.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

(ii) Understanding health inequalities.

Inclusion is one of the 4 pillars of the ONS strategy ‘Statistics for the Public Good’, as such exploring inequalities across outcomes is a priority for the ONS. Making use of linked morbidity, mortality, census and other data will allow ONS to produce statistics on health inequalities and provide these at a granular level. In particular, linking to these health data will help ONS to better understand if inequalities in outcomes persist when health status or pre-existing conditions are taken into account. For example, ONS have already used HES data to understand comorbidities of the population and to take this into account when modelling the risk of COVID-19 deaths by ethnicity (under DARS-NIC-400304-S1P1B). This allowed ONS to provide evidence that inequalities in outcomes still exist despite controlling for certain, but not all, health conditions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly.

It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible, to develop new measures of avoidable mortality in the elderly that could have been avoided.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality.

The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

d. ONS will use this data in conjunction with other health sources, and non-health sources such as Census, income and benefits data, and survey data, to produce a range of statistics on the interaction between health and the labour market. This is bi-directional, looking both at how health affects economic outcomes and vice versa. Work would include but not be limited to:

• Producing statistics on the prevalence of conditions in people who are inactive in the labour market due to ill health

• Modelling whether a change in the prevalence of certain conditions explain part of the increase in labour market inactivity observed since 2019.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation, and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, new-born and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS England and UK Health Security Agency (UKHSA) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability, and geography. In some cases, NHS England / UKHSA will not hold data that would enable this but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up-to-date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

A separate Agr

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is usually released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released over the year in a series of themed packages. Child and infant mortality statistics and unexplained infant deaths are published annually. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

ONS are looking to improve and develop new statistics using newly linked data to explore inequalities, risk factors and variation in child outcomes. Outputs in the form of presentations, analytical articles and methodology reports will be created alongside data tables as appropriate. ONS publishes all its official statistics on its website.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Improving population and migration statistics: This will result in internal, and potentially external, ONS reports and presentations on how births notification data could be used to improve and develop official statistics, and potentially used directly in the development of new official statistics. ONS publishes all its official statistics on its website.

Dataset 2, 3 and 4: Hospital Episode Statistics, Improving Access to Psychological Therapies and Emergency Care Dataset data

The initial uses to which ONS will put HES, IAPT and ECDS data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Initial work has been carried out and new statistics published using HES data as part of the response to the coronavirus pandemic which has assured ONS of the quality and validity of using these data as initially intended such that additional data is being requested as part of this Agreement to address limitations in the initial supply of HES data. Research outputs have been published on the ONS website, for example: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES, IAPT or ECDS data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS England, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this Agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this Data Sharing Agreement. This process involved working closely with NHS England’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS England will also publish the details of this Data Sharing Agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this Agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS England approval. It would also require an appropriate alternative legal gateway because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS England and enter into an amended Data Sharing Agreement before proceeding with that new purpose.

Dataset 1: Birth Notifications data

NHS England has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

There are a wide range of statistical uses to which the Office for National Statistics (ONS) intends to put Birth Notifications data. All use of Birth Notifications data by ONS will be to improve the availability and quality of statistics as part of ONS’s function to produce statistics for the public good.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required.

1.1 Birth and Child Mortality Statistics

The primary statistical purpose for which this information will be used is analysis of births, maternities, infant mortality and child health outcomes. Analyses are made publicly available as aggregate National Statistics.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics. However, there are some limitations with the GRO data, including a time lag, a lack of key information such as length of gestation and ethnicity of the baby, as well as some missing values in the fields that are available. To mitigate these limitations, the NHS England birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. The identifying information required from NHS England will enable and help quality assure this linkage, which in turn will enable ONS to produce more comprehensive and accurate statistics on births and child health outcomes.

As well as linking birth notifications data to birth registrations and deaths data to produce National Statistics these data will also be linked to other sources in order improve and development new statistics. For example, birth notifications data will be linked to:

• birth registration, deaths and census data to identify more detailed characteristics of the household and mother, such as ethnicity, and so better understand inequalities, risk factors and variation in child health outcomes

• deaths, to identify bereaved children, and to other NHS England data to identify subsequent inequalities and outcomes

• other ONS and NHS England data such as Hospital Episode Statistics to better determine underlying cause of death, understand inequalities, risk factors and variation in child outcomes

1.2 Improving data linkage methodology

ONS plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link information relating to siblings born at different times (i.e. not twins) using NHS number of the mother. This produces very accurate linkage of siblings.

ONS can then attempt to link siblings together using only the data available in the registration data – i.e. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the results of the latter linkage method matches those achieved when linking using mother’s NHS number. This will inform the best matching methodology to use when NHS number of the mother is not available (for example in pre-2005 birth registration data).

1.3 Improving population and migration statistics

ONS plans to use births notifications data to support development and improvements of population and migration statistics. This includes a range of work such as quality assurance of Census data, contribution to ONS’s population and migration statistics transformation programme, to put administrative data first and make recommendations on the future of the decennial Census. For example, birth notifications data will be:

• used to validate Census data, with a focus on population sub-groups, such as the under 1s, returned in Census data and subsequently supporting population outputs

• used to contribute to work to estimate or quality assure characteristics of the population, such as ethnicity or age, along with other data sources

• linked to other data sources in order to carry out such improvements to existing or develop new Official Statistics, and to quality assurance other data sources

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – e.g. additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – e.g. where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

• Research and development of methodology – e.g. improving linkage methodology for siblings

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS England data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realised.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity Data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity, but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data-based ethnicity estimates will be important when deciding whether to move to an admin data-based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS England data experts to understand the limitations of the HES ethnicity data and there are several; for example, coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about when, where, and why people have accessed hospital services, for example diagnosis and procedures codes, for a range of statistical purposes in line with ONS’s function to produce statistics for the public good. All use of the HES data for health analysis will be to improve the availability and quality of health statistics.

The diagnosis and procedures information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, for the first supply of HES data to ONS the volume of this information was minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services was shared with ONS, compared with the information about when and where people have accessed services. ONS has now used and published analysis using these HES data for several purposes described in this Agreement and DARS-NIC-400304-S1P1B (which allows reuse of the HES data ONS hold). Based on learning from this work ONS are now requesting further additional variables and historical times series updates where some variables were only requested for a subset of years to the current HES subset they receive.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

(i) Understanding healthy life expectancy.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

(ii) Understanding health inequalities.

Inclusion is one of the 4 pillars of the ONS strategy ‘Statistics for the Public Good’, as such exploring inequalities across outcomes is a priority for the ONS. Making use of linked morbidity, mortality, census and other data will allow ONS to produce statistics on health inequalities and provide these at a granular level. In particular, linking to these health data will help ONS to better understand if inequalities in outcomes persist when health status or pre-existing conditions are taken into account. For example, ONS have already used HES data to understand comorbidities of the population and to take this into account when modelling the risk of COVID-19 deaths by ethnicity (under DARS-NIC-400304-S1P1B). This allowed ONS to provide evidence that inequalities in outcomes still exist despite controlling for certain, but not all, health conditions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly.

It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible, to develop new measures of avoidable mortality in the elderly that could have been avoided.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality.

The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

d. ONS will use this data in conjunction with other health sources, and non-health sources such as Census, income and benefits data, and survey data, to produce a range of statistics on the interaction between health and the labour market. This is bi-directional, looking both at how health affects economic outcomes and vice versa. Work would include but not be limited to:

• Producing statistics on the prevalence of conditions in people who are inactive in the labour market due to ill health

• Modelling whether a change in the prevalence of certain conditions explain part of the increase in labour market inactivity observed since 2019.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation, and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, new-born and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS England and UK Health Security Agency (UKHSA) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability, and geography. In some cases, NHS England / UKHSA will not hold data that would enable this but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up-to-date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is usually released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released over the year in a series of themed packages. Child and infant mortality statistics and unexplained infant deaths are published annually. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

ONS are looking to improve and develop new statistics using newly linked data to explore inequalities, risk factors and variation in child outcomes. Outputs in the form of presentations, analytical articles and methodology reports will be created alongside data tables as appropriate. ONS publishes all its official statistics on its website.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Improving population and migration statistics: This will result in internal, and potentially external, ONS reports and presentations on how births notification data could be used to improve and develop official statistics, and potentially used directly in the development of new official statistics. ONS publishes all its official statistics on its website.

Dataset 2, 3 and 4: Hospital Episode Statistics, Improving Access to Psychological Therapies and Emergency Care Dataset data

The initial uses to which ONS will put HES, IAPT and ECDS data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Initial work has been carried out and new statistics published using HES data as part of the response to the coronavirus pandemic which has assured ONS of the quality and validity of using these data as initially intended such that additional data is being requested as part of this Agreement to address limitations in the initial supply of HES data. Research outputs have been published on the ONS website, for example: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES, IAPT or ECDS data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS England, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this Agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this Data Sharing Agreement. This process involved working closely with NHS England’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS England will also publish the details of this Data Sharing Agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this Agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS England approval. It would also require an appropriate alternative legal gateway because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS England and enter into an amended Data Sharing Agreement before proceeding with that new purpose.

Dataset 1: Birth Notifications data

NHS England has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

There are a wide range of statistical uses to which the Office for National Statistics (ONS) intends to put Birth Notifications data. All use of Birth Notifications data by ONS will be to improve the availability and quality of statistics as part of ONS’s function to produce statistics for the public good.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required.

1.1 Birth and Child Mortality Statistics

The primary statistical purpose for which this information will be used is analysis of births, maternities, infant mortality and child health outcomes. Analyses are made publicly available as aggregate National Statistics.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics. However, there are some limitations with the GRO data, including a time lag, a lack of key information such as length of gestation and ethnicity of the baby, as well as some missing values in the fields that are available. To mitigate these limitations, the NHS England birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. The identifying information required from NHS England will enable and help quality assure this linkage, which in turn will enable ONS to produce more comprehensive and accurate statistics on births and child health outcomes.

As well as linking birth notifications data to birth registrations and deaths data to produce National Statistics these data will also be linked to other sources in order improve and development new statistics. For example, birth notifications data will be linked to:

• birth registration, deaths and census data to identify more detailed characteristics of the household and mother, such as ethnicity, and so better understand inequalities, risk factors and variation in child health outcomes

• deaths, to identify bereaved children, and to other NHS England data to identify subsequent inequalities and outcomes

• other ONS and NHS England data such as Hospital Episode Statistics to better determine underlying cause of death, understand inequalities, risk factors and variation in child outcomes

1.2 Improving data linkage methodology

ONS plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link information relating to siblings born at different times (i.e. not twins) using NHS number of the mother. This produces very accurate linkage of siblings.

ONS can then attempt to link siblings together using only the data available in the registration data – i.e. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the results of the latter linkage method matches those achieved when linking using mother’s NHS number. This will inform the best matching methodology to use when NHS number of the mother is not available (for example in pre-2005 birth registration data).

1.3 Improving population and migration statistics

ONS plans to use births notifications data to support development and improvements of population and migration statistics. This includes a range of work such as quality assurance of Census data, contribution to ONS’s population and migration statistics transformation programme, to put administrative data first and make recommendations on the future of the decennial Census. For example, birth notifications data will be:

• used to validate Census data, with a focus on population sub-groups, such as the under 1s, returned in Census data and subsequently supporting population outputs

• used to contribute to work to estimate or quality assure characteristics of the population, such as ethnicity or age, along with other data sources

• linked to other data sources in order to carry out such improvements to existing or develop new Official Statistics, and to quality assurance other data sources

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – e.g. additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – e.g. where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

• Research and development of methodology – e.g. improving linkage methodology for siblings

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS England data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realised.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity Data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity, but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data-based ethnicity estimates will be important when deciding whether to move to an admin data-based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS England data experts to understand the limitations of the HES ethnicity data and there are several; for example, coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about when, where, and why people have accessed hospital services, for example diagnosis and procedures codes, for a range of statistical purposes in line with ONS’s function to produce statistics for the public good. All use of the HES data for health analysis will be to improve the availability and quality of health statistics.

The diagnosis and procedures information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, for the first supply of HES data to ONS the volume of this information was minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services was shared with ONS, compared with the information about when and where people have accessed services. ONS has now used and published analysis using these HES data for several purposes described in this Agreement and DARS-NIC-400304-S1P1B (which allows reuse of the HES data ONS hold). Based on learning from this work ONS are now requesting further additional variables and historical times series updates where some variables were only requested for a subset of years to the current HES subset they receive.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

(i) Understanding healthy life expectancy.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

(ii) Understanding health inequalities.

Inclusion is one of the 4 pillars of the ONS strategy ‘Statistics for the Public Good’, as such exploring inequalities across outcomes is a priority for the ONS. Making use of linked morbidity, mortality, census and other data will allow ONS to produce statistics on health inequalities and provide these at a granular level. In particular, linking to these health data will help ONS to better understand if inequalities in outcomes persist when health status or pre-existing conditions are taken into account. For example, ONS have already used HES data to understand comorbidities of the population and to take this into account when modelling the risk of COVID-19 deaths by ethnicity (under DARS-NIC-400304-S1P1B). This allowed ONS to provide evidence that inequalities in outcomes still exist despite controlling for certain, but not all, health conditions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly.

It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible, to develop new measures of avoidable mortality in the elderly that could have been avoided.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality.

The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation, and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, new-born and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS England and UK Health Security Agency (UKHSA) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability, and geography. In some cases, NHS England / UKHSA will not hold data that would enable this but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up-to-date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

A separate Agreement (DARS-NIC-400304-S1P1B) has now been set up to cover this area of work related to rapid response COVID-19 analysis and additional data beyond that covered in this Agreement such as primary care data and any additional statistical research projects will be updated through that Agreement. The Agreement, DARS-NIC-400304-S1P1B, includes permission to reuse some of the data acquired through this Agreement and to link this data to other sources as specified in that Agreement.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new Official Statistics.

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is usually released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released over the year in a series of themed packages. Child and infant mortality statistics and unexplained infant deaths are published annually. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

ONS are looking to improve and develop new statistics using newly linked data to explore inequalities, risk factors and variation in child outcomes. Outputs in the form of presentations, analytical articles and methodology reports will be created alongside data tables as appropriate. ONS publishes all its official statistics on its website.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Improving population and migration statistics: This will result in internal, and potentially external, ONS reports and presentations on how births notification data could be used to improve and develop official statistics, and potentially used directly in the development of new official statistics. ONS publishes all its official statistics on its website.

Dataset 2, 3 and 4: Hospital Episode Statistics, Improving Access to Psychological Therapies and Emergency Care Dataset data

The initial uses to which ONS will put HES, IAPT and ECDS data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Initial work has been carried out and new statistics published using HES data as part of the response to the coronavirus pandemic which has assured ONS of the quality and validity of using these data as initially intended such that additional data is being requested as part of this Agreement to address limitations in the initial supply of HES data. Research outputs have been published on the ONS website, for example: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES, IAPT or ECDS data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS England, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS England’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS England will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS England approval. It would also require an appropriate alternative legal gateway because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS England and enter into an amended Data Sharing Agreement before proceeding with that new purpose.

Dataset 1: Birth Notifications data

NHS England has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

There are a wide range of statistical uses to which the Office for National Statistics (ONS) intends to put Birth Notifications data. All use of Birth Notifications data by ONS will be to improve the availability and quality of statistics as part of ONS’s function to produce statistics for the public good.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required.

1.1 Birth and Child Mortality Statistics

The primary statistical purpose for which this information will be used is analysis of births, maternities, infant mortality and child health outcomes. Analyses are made publicly available as aggregate National Statistics.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics. However, there are some limitations with the GRO data, including a time lag, a lack of key information such as length of gestation and ethnicity of the baby, as well as some missing values in the fields that are available. To mitigate these limitations, the NHS England birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. The identifying information required from NHS England will enable and help quality assure this linkage, which in turn will enable ONS to produce more comprehensive and accurate statistics on births and child health outcomes.

As well as linking birth notifications data to birth registrations and deaths data to produce National Statistics these data will also be linked to other sources in order improve and development new statistics. For example, birth notifications data will be linked to:

• birth registration, deaths and census data to identify more detailed characteristics of the household and mother, such as ethnicity, and so better understand inequalities, risk factors and variation in child health outcomes

• deaths, to identify bereaved children, and to other NHS England data to identify subsequent inequalities and outcomes

• other ONS and NHS England data such as Hospital Episode Statistics to better determine underlying cause of death, understand inequalities, risk factors and variation in child outcomes

1.2 Improving data linkage methodology

ONS plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link information relating to siblings born at different times (i.e. not twins) using NHS number of the mother. This produces very accurate linkage of siblings.

ONS can then attempt to link siblings together using only the data available in the registration data – i.e. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the results of the latter linkage method matches those achieved when linking using mother’s NHS number. This will inform the best matching methodology to use when NHS number of the mother is not available (for example in pre-2005 birth registration data).

1.3 Improving population and migration statistics

ONS plans to use births notifications data to support development and improvements of population and migration statistics. This includes a range of work such as quality assurance of Census data, contribution to ONS’s population and migration statistics transformation programme, to put administrative data first and make recommendations on the future of the decennial Census. For example, birth notifications data will be:

• used to validate Census data, with a focus on population sub-groups, such as the under 1s, returned in Census data and subsequently supporting population outputs

• used to contribute to work to estimate or quality assure characteristics of the population, such as ethnicity or age, along with other data sources

• linked to other data sources in order to carry out such improvements to existing or develop new Official Statistics, and to quality assurance other data sources

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – e.g. additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – e.g. where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

• Research and development of methodology – e.g. improving linkage methodology for siblings

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS England data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realised.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity Data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity, but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data-based ethnicity estimates will be important when deciding whether to move to an admin data-based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS England data experts to understand the limitations of the HES ethnicity data and there are several; for example, coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about when, where, and why people have accessed hospital services, for example diagnosis and procedures codes, for a range of statistical purposes in line with ONS’s function to produce statistics for the public good. All use of the HES data for health analysis will be to improve the availability and quality of health statistics.

The diagnosis and procedures information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, for the first supply of HES data to ONS the volume of this information was minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services was shared with ONS, compared with the information about when and where people have accessed services. ONS has now used and published analysis using these HES data for several purposes described in this agreement and DARS-NIC-400304-S1P1B (which allows reuse of the HES data ONS hold). Based on learning from this work ONS are now requesting further additional variables and historical times series updates where some variables were only requested for a subset of years to the current HES subset they receive.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

(i) Understanding healthy life expectancy.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

(ii) Understanding health inequalities.

Inclusion is one of the 4 pillars of the ONS strategy ‘Statistics for the Public Good’, as such exploring inequalities across outcomes is a priority for the ONS. Making use of linked morbidity, mortality, census and other data will allow ONS to produce statistics on health inequalities and provide these at a granular level. In particular, linking to these health data will help ONS to better understand if inequalities in outcomes persist when health status or pre-existing conditions are taken into account. For example, ONS have already used HES data to understand comorbidities of the population and to take this into account when modelling the risk of COVID-19 deaths by ethnicity (under DARS-NIC-400304-S1P1B). This allowed ONS to provide evidence that inequalities in outcomes still exist despite controlling for certain, but not all, health conditions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly.

It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible, to develop new measures of avoidable mortality in the elderly that could have been avoided.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality.

The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation, and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, new-born and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS England and UK Health Security Agency (UKHSA) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability, and geography. In some cases, NHS England / UKHSA will not hold data that would enable this but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up-to-date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

A separate agreement (DARS-NIC-400304-S1P1B) has now been set up to cover this area of work related to rapid response COVID-19 analysis and additional data beyond that covered in this agreement such as primary care data and any additional statistical research projects will be updated through that agreement. The agreement, DARS-NIC-400304-S1P1B, includes permission to reuse some of the data acquired through this agreement and to link this data to other sources as specified in that agreement.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new Official Statistics.

2.7. Productivi

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is usually released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released over the year in a series of themed packages. Child and infant mortality statistics and unexplained infant deaths are published annually. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

ONS are looking to improve and develop new statistics using newly linked data to explore inequalities, risk factors and variation in child outcomes. Outputs in the form of presentations, analytical articles and methodology reports will be created alongside data tables as appropriate. ONS publishes all its official statistics on its website.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Improving population and migration statistics: This will result in internal, and potentially external, ONS reports and presentations on how births notification data could be used to improve and develop official statistics, and potentially used directly in the development of new official statistics. ONS publishes all its official statistics on its website.

Dataset 2, 3 and 4: Hospital Episode Statistics, Improving Access to Psychological Therapies and Emergency Care Dataset data

The initial uses to which ONS will put HES, IAPT and ECDS data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Initial work has been carried out and new statistics published using HES data as part of the response to the coronavirus pandemic which has assured ONS of the quality and validity of using these data as initially intended such that additional data is being requested as part of this agreement to address limitations in the initial supply of HES data. Research outputs have been published on the ONS website, for example: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES, IAPT or ECDS data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS Digital, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS Digital’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest, and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS Digital will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS Digital approval. It would also require an appropriate alternative legal gateway, because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS Digital and enter into an amended Data Sharing Agreement before proceeding with that new purpose .

Dataset 1: Birth Notifications data

NHS Digital has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

There are a wide range of statistical uses to which the Office for National Statistics (ONS) intends to put Birth Notifications data. All use of Birth Notifications data by ONS will be to improve the availability and quality of statistics as part of ONS’s function to produce statistics for the public good.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required.

1.1 Birth and Child Mortality Statistics

The primary statistical purpose for which this information will be used is analysis of births, maternities, infant mortality and child health outcomes. Analyses are made publicly available as aggregate National Statistics.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics. However, there are some limitations with the GRO data, including a time lag, a lack of key information such as length of gestation and ethnicity of the baby, as well as some missing values in the fields that are available. To mitigate these limitations, the NHS Digital birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. The identifying information required from NHS Digital will enable and help quality assure this linkage, which in turn will enable ONS to produce more comprehensive and accurate statistics on births and child health outcomes.

As well as linking birth notifications data to birth registrations and deaths data to produce National Statistics these data will also be linked to other sources in order improve and development new statistics. For example, birth notifications data will be linked to:

• birth registration, deaths and census data to identify more detailed characteristics of the household and mother, such as ethnicity, and so better understand inequalities, risk factors and variation in child health outcomes

• deaths, to identify bereaved children, and to other NHS digital data to identify subsequent inequalities and outcomes

• other ONS and NHS digital data such as Hospital Episode Statistics to better determine underlying cause of death, understand inequalities, risk factors and variation in child outcomes

1.2 Improving data linkage methodology

ONS plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link information relating to siblings born at different times (i.e. not twins) using NHS number of the mother. This produces very accurate linkage of siblings.

ONS can then attempt to link siblings together using only the data available in the registration data – i.e. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the results of the latter linkage method matches those achieved when linking using mother’s NHS number. This will inform the best matching methodology to use when NHS number of the mother is not available (for example in pre-2005 birth registration data).

1.3 Improving population and migration statistics

ONS plans to use births notifications data to support development and improvements of population and migration statistics. This includes a range of work such as quality assurance of Census data, contribution to ONS’s population and migration statistics transformation programme, to put administrative data first and make recommendations on the future of the decennial Census. For example, birth notifications data will be:

• used to validate Census data, with a focus on population sub-groups, such as the under 1s, returned in Census data and subsequently supporting population outputs

• used to contribute to work to estimate or quality assure characteristics of the population, such as ethnicity or age, along with other data sources

• linked to other data sources in order to carry out such improvements to existing or develop new Official Statistics, and to quality assurance other data sources

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – e.g. additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – e.g. where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

• Research and development of methodology – e.g. improving linkage methodology for siblings

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS Digital data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realized.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data based ethnicity estimates will be important when deciding whether to move to an admin data based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS Digital data experts to understand the limitations of the HES ethnicity data and there are several; for example coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘Activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about when, where and why people have accessed hospital services, for example diagnosis and procedures codes, for a range of statistical purposes in line with ONS’s function to produce statistics for the public good. All use of the HES data for health analysis will be to improve the availability and quality of health statistics.

The diagnosis and procedures information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, for the first supply of HES data to ONS the volume of this information was minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services was shared with ONS, compared with the information about when and where people have accessed services. ONS has now used and published analysis using these HES data for several purposes described in this agreement and DARS-NIC-400304-S1P1B (which allows reuse of the HES data ONS hold). Based on learning from this work ONS are now requesting further additional variables and historical times series updates where some variables were only requested for a subset of years to the current HES subset they receive.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

(i) Understanding healthy life expectancy.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

(ii) Understanding health inequalities.

Inclusion is one of the 4 pillars of the ONS strategy ‘Statistics for the Public Good’, as such exploring inequalities across outcomes is a priority for the ONS. Making use of linked morbidity, mortality, census and other data will allow ONS to produce statistics on health inequalities and provide these at a granular level. In particular, linking to these health data will help ONS to better understand if inequalities in outcomes persist when health status or pre-existing conditions are taken into account. For example, ONS have already used HES data to understand comorbidities of the population and to take this into account when modelling the risk of COVID-19 deaths by ethnicity (under DARS-NIC-400304-S1P1B). This allowed ONS to provide evidence that inequalities in outcomes still exist despite controlling for certain, but not all, health conditions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital, or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly.

It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible to develop new measures of avoidable mortality in the elderly that could have been avoided.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality.

The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments, and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, newborn and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS Digital and Public Health England (PHE) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability and geography. In some cases, NHS Digital / PHE will not hold data that would enable this, but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up to date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

A separate agreement (DARS-NIC-400304-S1P1B) has now been set up to cover this area of work related to rapid response COVID-19 analysis and additional data beyond that covered in this agreement such as primary care data and any additional statistical research projects will be updated through that agreement. The agreement, DARS-NIC-400304-S1P1B, includes permission to reuse some of the data acquired through this agreement and to link this data to other sources as specified in that agreement.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new Official Statistics.

2.7. Productivity

To f

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is usually released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released over the year in a series of themed packages. Child and infant mortality statistics and unexplained infant deaths are published annually. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

ONS are looking to improve and develop new statistics using newly linked data to explore inequalities, risk factors and variation in child outcomes. Outputs in the form of presentations, analytical articles and methodology reports will be created alongside data tables as appropriate. ONS publishes all its official statistics on its website.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Improving population and migration statistics: This will result in internal, and potentially external, ONS reports and presentations on how births notification data could be used to improve and develop official statistics, and potentially used directly in the development of new official statistics. ONS publishes all its official statistics on its website.

Dataset 2, 3 and 4: Hospital Episode Statistics, Improving Access to Psychological Therapies and Emergency Care Dataset data

The initial uses to which ONS will put HES,IAPT and ECDS data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Initial work has been carried out and new statistics published using HES data as part of the response to the coronavirus pandemic which has assured ONS of the quality and validity of using these data as initially intended such that additional data is being requested as part of this agreement to address limitations in the initial supply of HES data. Research outputs have been published on the ONS website, for example: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES, IAPT or ECDS data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users, and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS Digital, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS Digital’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest, and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS Digital will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS Digital approval. It would also require an appropriate alternative legal gateway, because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS Digital and enter into an amended Data Sharing Agreement before proceeding with that new purpose .

Dataset 1: Birth Notifications data

NHS Digital has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

There are a wide range of statistical uses to which the Office for National Statistics (ONS) intends to put Birth Notifications data. All use of Birth Notifications data by ONS will be to improve the availability and quality of statistics as part of ONS’s function to produce statistics for the public good.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required.

1.1 Birth and Child Mortality Statistics

The primary statistical purpose for which this information will be used is analysis of births, maternities, infant mortality and child health outcomes. Analyses are made publicly available as aggregate National Statistics.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics. However, there are some limitations with the GRO data, including a time lag, a lack of key information such as length of gestation and ethnicity of the baby, as well as some missing values in the fields that are available. To mitigate these limitations, the NHS Digital birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. The identifying information required from NHS Digital will enable and help quality assure this linkage, which in turn will enable ONS to produce more comprehensive and accurate statistics on births and child health outcomes.

As well as linking birth notifications data to birth registrations and deaths data to produce National Statistics these data will also be linked to other sources in order improve and development new statistics. For example, birth notifications data will be linked to:

• birth registration, deaths and census data to identify more detailed characteristics of the household and mother, such as ethnicity, and so better understand inequalities, risk factors and variation in child health outcomes

• deaths, to identify bereaved children, and to other NHS digital data to identify subsequent inequalities and outcomes

• other ONS and NHS digital data such as Hospital Episode Statistics to better determine underlying cause of death, understand inequalities, risk factors and variation in child outcomes

1.2 Improving data linkage methodology

ONS plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link information relating to siblings born at different times (i.e. not twins) using NHS number of the mother. This produces very accurate linkage of siblings.

ONS can then attempt to link siblings together using only the data available in the registration data – i.e. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the results of the latter linkage method matches those achieved when linking using mother’s NHS number. This will inform the best matching methodology to use when NHS number of the mother is not available (for example in pre-2005 birth registration data).

1.3 Improving population and migration statistics

ONS plans to use births notifications data to support development and improvements of population and migration statistics. This includes a range of work such as quality assurance of Census data, contribution to ONS’s population and migration statistics transformation programme, to put administrative data first and make recommendations on the future of the decennial Census. For example, birth notifications data will be:

• used to validate Census data, with a focus on population sub-groups, such as the under 1s, returned in Census data and subsequently supporting population outputs

• used to contribute to work to estimate or quality assure characteristics of the population, such as ethnicity or age, along with other data sources

• linked to other data sources in order to carry out such improvements to existing or develop new Official Statistics, and to quality assurance other data sources

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – e.g. additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – e.g. where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

• Research and development of methodology – e.g. improving linkage methodology for siblings

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS Digital data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realized.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data based ethnicity estimates will be important when deciding whether to move to an admin data based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS Digital data experts to understand the limitations of the HES ethnicity data and there are several; for example coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘Activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about when, where and why people have accessed hospital services, for example diagnosis and procedures codes, for a range of statistical purposes in line with ONS’s function to produce statistics for the public good. All use of the HES data for health analysis will be to improve the availability and quality of health statistics.

The diagnosis and procedures information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, for the first supply of HES data to ONS the volume of this information was minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services was shared with ONS, compared with the information about when and where people have accessed services. ONS has now used and published analysis using these HES data for several purposes described in this agreement and DARS-NIC-400304-S1P1B (which allows reuse of the HES data ONS hold). Based on learning from this work ONS are now requesting further additional variables and historical times series updates where some variables were only requested for a subset of years to the current HES subset they receive.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

(i) Understanding healthy life expectancy.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

(ii) Understanding health inequalities.

Inclusion is one of the 4 pillars of the ONS strategy ‘Statistics for the Public Good’, as such exploring inequalities across outcomes is a priority for the ONS. Making use of linked morbidity, mortality, census and other data will allow ONS to produce statistics on health inequalities and provide these at a granular level. In particular, linking to these health data will help ONS to better understand if inequalities in outcomes persist when health status or pre-existing conditions are taken into account. For example, ONS have already used HES data to understand comorbidities of the population and to take this into account when modelling the risk of COVID-19 deaths by ethnicity (under DARS-NIC-400304-S1P1B). This allowed ONS to provide evidence that inequalities in outcomes still exist despite controlling for certain, but not all, health conditions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital, or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly.

It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible to develop new measures of avoidable mortality in the elderly that could have been avoided.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality.

The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments, and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, newborn and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS Digital and Public Health England (PHE) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability and geography. In some cases, NHS Digital / PHE will not hold data that would enable this, but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up to date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

A separate agreement (DARS-NIC-400304-S1P1B) has now been set up to cover this area of work related to rapid response COVID-19 analysis and additional data beyond that covered in this agreement such as primary care data and any additional statistical research projects will be updated through that agreement. The agreement, DARS-NIC-400304-S1P1B, includes permission to reuse some of the data acquired through this agreement and to link this data to other sources as specified in that agreement.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new Official Statistics.

2.7. Productivity

To f

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is usually released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released over the year in a series of themed packages. Child and infant mortality statistics and unexplained infant deaths are published annually. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

ONS are looking to improve and develop new statistics using newly linked data to explore inequalities, risk factors and variation in child outcomes. Outputs in the form of presentations, analytical articles and methodology reports will be created alongside data tables as appropriate. ONS publishes all its official statistics on its website.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Improving population and migration statistics: This will result in internal, and potentially external, ONS reports and presentations on how births notification data could be used to improve and develop official statistics, and potentially used directly in the development of new official statistics. ONS publishes all its official statistics on its website.

Dataset 2, 3 and 4: Hospital Episode Statistics, Improving Access to Psychological Therapies and Emergency Care Dataset data

The initial uses to which ONS will put HES,IAPT and ECDS data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Initial work has been carried out and new statistics published using HES data as part of the response to the coronavirus pandemic which has assured ONS of the quality and validity of using these data as initially intended such that additional data is being requested as part of this agreement to address limitations in the initial supply of HES data. Research outputs have been published on the ONS website, for example: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES, IAPT or ECDS data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users, and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS Digital, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS Digital’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest, and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS Digital will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS Digital approval. It would also require an appropriate alternative legal gateway, because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS Digital and enter into an amended Data Sharing Agreement before proceeding with that new purpose .

Dataset 1: Birth Notifications data

NHS Digital has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

There are a wide range of statistical uses to which the Office for National Statistics (ONS) intends to put Birth Notifications data. All use of Birth Notifications data by ONS will be to improve the availability and quality of statistics as part of ONS’s function to produce statistics for the public good.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required.

1.1 Birth and Child Mortality Statistics

The primary statistical purpose for which this information will be used is analysis of births, maternities, infant mortality and child health outcomes. Analyses are made publicly available as aggregate National Statistics.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics. However, there are some limitations with the GRO data, including a time lag, a lack of key information such as length of gestation and ethnicity of the baby, as well as some missing values in the fields that are available. To mitigate these limitations, the NHS Digital birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. The identifying information required from NHS Digital will enable and help quality assure this linkage, which in turn will enable ONS to produce more comprehensive and accurate statistics on births and child health outcomes.

As well as linking birth notifications data to birth registrations and deaths data to produce National Statistics these data will also be linked to other sources in order improve and development new statistics. For example, birth notifications data will be linked to:

• birth registration, deaths and census data to identify more detailed characteristics of the household and mother, such as ethnicity, and so better understand inequalities, risk factors and variation in child health outcomes

• deaths, to identify bereaved children, and to other NHS digital data to identify subsequent inequalities and outcomes

• other ONS and NHS digital data such as Hospital Episode Statistics to better determine underlying cause of death, understand inequalities, risk factors and variation in child outcomes

1.2 Improving data linkage methodology

ONS plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link information relating to siblings born at different times (i.e. not twins) using NHS number of the mother. This produces very accurate linkage of siblings.

ONS can then attempt to link siblings together using only the data available in the registration data – i.e. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the results of the latter linkage method matches those achieved when linking using mother’s NHS number. This will inform the best matching methodology to use when NHS number of the mother is not available (for example in pre-2005 birth registration data).

1.3 Improving population and migration statistics

ONS plans to use births notifications data to support development and improvements of population and migration statistics. This includes a range of work such as quality assurance of Census data, contribution to ONS’s population and migration statistics transformation programme, to put administrative data first and make recommendations on the future of the decennial Census. For example, birth notifications data will be:

• used to validate Census data, with a focus on population sub-groups, such as the under 1s, returned in Census data and subsequently supporting population outputs

• used to contribute to work to estimate or quality assure characteristics of the population, such as ethnicity or age, along with other data sources

• linked to other data sources in order to carry out such improvements to existing or develop new Official Statistics, and to quality assurance other data sources

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – e.g. additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – e.g. where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

• Research and development of methodology – e.g. improving linkage methodology for siblings

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS Digital data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realized.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data based ethnicity estimates will be important when deciding whether to move to an admin data based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS Digital data experts to understand the limitations of the HES ethnicity data and there are several; for example coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘Activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about when, where and why people have accessed hospital services, for example diagnosis and procedures codes, for a range of statistical purposes in line with ONS’s function to produce statistics for the public good. All use of the HES data for health analysis will be to improve the availability and quality of health statistics.

The diagnosis and procedures information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, for the first supply of HES data to ONS the volume of this information was minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services was shared with ONS, compared with the information about when and where people have accessed services. ONS has now used and published analysis using these HES data for several purposes described in this agreement and DARS-NIC-400304-S1P1B (which allows reuse of the HES data ONS hold). Based on learning from this work ONS are now requesting further additional variables and historical times series updates where some variables were only requested for a subset of years to the current HES subset they receive.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

(i) Understanding healthy life expectancy.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

(ii) Understanding health inequalities.

Inclusion is one of the 4 pillars of the ONS strategy ‘Statistics for the Public Good’, as such exploring inequalities across outcomes is a priority for the ONS. Making use of linked morbidity, mortality, census and other data will allow ONS to produce statistics on health inequalities and provide these at a granular level. In particular, linking to these health data will help ONS to better understand if inequalities in outcomes persist when health status or pre-existing conditions are taken into account. For example, ONS have already used HES data to understand comorbidities of the population and to take this into account when modelling the risk of COVID-19 deaths by ethnicity (under DARS-NIC-400304-S1P1B). This allowed ONS to provide evidence that inequalities in outcomes still exist despite controlling for certain, but not all, health conditions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital, or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly.

It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible to develop new measures of avoidable mortality in the elderly that could have been avoided.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality.

The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments, and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, newborn and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS Digital and Public Health England (PHE) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability and geography. In some cases, NHS Digital / PHE will not hold data that would enable this, but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up to date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

A separate agreement (DARS-NIC-400304-S1P1B) has now been set up to cover this area of work related to rapid response COVID-19 analysis and additional data beyond that covered in this agreement such as primary care data and any additional statistical research projects will be updated through that agreement. The agreement, DARS-NIC-400304-S1P1B, includes permission to reuse some of the data acquired through this agreement and to link this data to other sources as specified in that agreement.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new Official Statistics.

2.7. Productivity

To f

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is usually released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released over the year in a series of themed packages. Child and infant mortality statistics and unexplained infant deaths are published annually. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

ONS are looking to improve and develop new statistics using newly linked data to explore inequalities, risk factors and variation in child outcomes. Outputs in the form of presentations, analytical articles and methodology reports will be created alongside data tables as appropriate. ONS publishes all its official statistics on its website.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Improving population and migration statistics: This will result in internal, and potentially external, ONS reports and presentations on how births notification data could be used to improve and develop official statistics, and potentially used directly in the development of new official statistics. ONS publishes all its official statistics on its website.

Dataset 2, 3 and 4: Hospital Episode Statistics, Improving Access to Psychological Therapies and Emergency Care Dataset data

The initial uses to which ONS will put HES,IAPT and ECDS data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Initial work has been carried out and new statistics published using HES data as part of the response to the coronavirus pandemic which has assured ONS of the quality and validity of using these data as initially intended such that additional data is being requested as part of this agreement to address limitations in the initial supply of HES data. Research outputs have been published on the ONS website, for example: https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES, IAPT or ECDS data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users, and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS Digital, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS Digital’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest, and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS Digital will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS Digital approval . It would also require an appropriate alternative legal gateway, because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS Digital and enter into an amended Data Sharing Agreement before proceeding with that new purpose .

Dataset 1: Birth Notifications data

NHS Digital has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

The birth notifications data contribute to ONS statistical analyses of births, maternities, and infant mortality outcomes. Analyses are made publicly available as aggregate National Statistics. These statistics help a range of public and other bodies make better decisions (see section 5d). They also feed into the Department of Health's NHS Outcomes Framework for monitoring low birthweight of term babies.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics, and ONS become controllers of that data under Section 42 of the 2007 Statistics and Registration Services Act. However, there are some limitations with the GRO data, including a lack of medical information such as length of gestation, as well as some missing and implausible values in the fields that are available.

To mitigate these limitations, the NHS Digital birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. Several identifying variables such as NHS number are received to enable this linkage.

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

ONS also plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link siblings born at different times (i.e. not twins) using the NHS number of the mother which is only available on the notification data. This provides a ‘gold standard’ linkage method.

ONS can then then attempt to link siblings together using only the data available in the registration data – e.g. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the latter linkage method matches the gold standard. This will inform the best matching methodology to use when seeking to link siblings if NHS number of mother is not available. This is needed to link pre-2005 birth registration data, a time when the birth notification data is not available to ONS. This purpose would fall under the Research and development of methodology category in the uses framework above.

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS Digital data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realized.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data based ethnicity estimates will be important when deciding whether to move to an admin data based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS Digital data experts to understand the limitations of the HES ethnicity data and there are several; for example coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘Activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about why people have accessed hospital services, for example diagnosis, for a range of statistical purposes.

This information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, the volume of this information is being minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services will be shared with ONS, compared with the information about when and where people have accessed services.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital, or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly. It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible to develop new measures of avoidable mortality in the elderly.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality. The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments, and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, newborn and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS Digital and Public Health England (PHE) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability and geography. In some cases, NHS Digital / PHE will not hold data that would enable this, but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (2011 Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up to date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

Dataset 3: Improving Access to Psychological Therapies (IAPT) Dataset

3.1. To enable research being conducted by ONS’ Administrative Data Census and Migration Statistics improvement projects using ‘activity’ and characteristics data from IAPT.

This first use is essentially the same as described for the uses of HES data within these projects: The IAPT data provides evidence of presence at a particular address, and it also includes information on characteristics including ethnicity. See section 2.1 above (within the HES section) for the full rationale for why this information is needed.

3.2. To conduct a range of Statistical Research and Health Analyses using IAPT data

a. Statistical Research to inform Primary Mental Health Service Policy Making

This project will focus on common mental health disorders (CMDs) such as anxiety and depression. Using a phased approach, ONS will look first at the mortality risk of people with CMDs, co-morbidities between mental and physical health problems, and investigate inequalities around mental health. In the second phase, ONS will investigate income and employment transitions for patients who have been through mental health treatment.

The first phase will address existing evidence gaps on co-morbidities between mental and physical health, improve understanding on the demographics of people with CMDs, and investigate whether some or all people with CMDs are more at risk of death than the general population.

The IAPT data for 2012 to 2017 will be linked to the 2011 Census to provide detailed demographic background, and to death registrations from 2012 to 2018. The mortality analysis will focus on specific causes of death which may be connected to mental health (suicide, alcohol and drug abuse) as well as overall risk. In addition, the causes of death will be compared to the distribution of causes in the general population to identify any common co-morbidity with life-threatening illnesses. This goes some way to provide insight into important issues raised by the NHS England Five Year Forward View on mental health:

“An important barrier to good care is the lack of appropriate data sharing to enable organisations to identify co-morbidities…People with poor mental health may require primary care, secondary physical care and social care, as well as mental health services, but the lack of linked datasets hinders effective provision.”

IAPT data is estimated to cover over 15% of people with CMDs in England. Because of the service’s large, national scale and focus on people with mild and moderate mental health conditions, it provides a reasonable proxy for patterns and trends in the population of people with diagnosable CMDs.

The IAPT data will be compared with the findings of the Adult Psychiatric Morbidity Survey (2007 and 2014) to assess likely issues of representativeness, such as the under-representation of specific population groups in the treatment cohort. People with severe mental health conditions are not typically treated in the IAPT programme.

The three-way linkage will provide an independent and more detailed demographic baseline than the IAPT data could do alone, and allow ONS to investigate if there have been changes in peoples’ circumstances between the Census and treatment in IAPT (e.g. becoming disabled or living alone). Having the mortality data linked as well allows ONS to see the overall trends in mortality, plus to see if there is any relationship between changes in demographics and the cause of death outcomes.

The research is not aiming to look at individual level outcomes or to evaluate the IAPT treatment, but to look for trends in the aggregate data after linkage, to provide population level analysis to inform policy.

Entry into IAPT treatment will be used as the main indicator of having a diagnosable CMD. The clinical data will not be analysed except to:

• Group the cohort into broad types of CMD

• Potentially, link successful/unsuccessful treatment outcome to risk of subsequent death.

b. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released between August and December in a series of themed packages. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison, sometimes back to 1837. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

There are several published packages:

Birth summary tables: includes the number of live births and stillbirths, fertility rates, percentage of live births outside marriage and civil partnership, mean age of mother and percentage of live births to non-UK born mothers for England and Wales as a whole. Live births (number and rate) and the number of stillbirths are also provided down to local authority level. To aid with user interpretation, ONS also publishes an interactive fertility mapping tool, which enables users to analyse trends in fertility by county district and unitary authority; this is contained within the statistical bulletin.

Parents’ country of birth: includes births by country of birth of mother and total fertility rates for UK born and non-UK born women for England and Wales as a whole. Summary figures are also available down to local authority level. ONS publishes detailed analysis on parents’ country of birth because this information is collected at birth registration and does not change over time, while their nationality or ethnicity may change.

Birth characteristics and by area of usual residence: contains statistics on stillbirths and maternities for England and Wales, birthweight data for live and stillbirths by mother's region of usual residence, and live births and stillbirths in hospitals and communal establishments by region of occurrence. These tables also provide figures on month and quarter of occurrence, place of birth, ethnicity and gestational age and multiple births for England and Wales as a whole. Also provides summary data for live births down to local authority level including figures by age of mother figures are published using boundaries in place during the year the birth occurred.

Births by parents’ characteristics: provides live birth, stillbirth and maternity statistics by age of mother and type of registration (within marriage and civil partnership, joint, sole). It also provides data on previous live-born children, National Statistics Socio-economic Classification (NS-SEC), median birth intervals, age-specific fertility rates for men and mean age of fathers. All tables are for England and Wales as a whole with no sub-national breakdown.

Childbearing for women born in different years (formerly known as Cohort fertility): presents data on fertility by year of birth of mother rather than the year of birth of child for England and Wales as a whole this package includes the average number of live-born children and the proportion of women remaining childless for women born in different years.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Dataset 2 and dataset 3: Hospital Episode Statistics and Improving Access to Psychological Therapies data

The initial uses to which ONS will put HES and IAPT data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES or IAPT data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users, and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS Digital, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS Digital’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest, and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS Digital will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS Digital approval . It would also require an appropriate alternative legal gateway, because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS Digital and enter into an amended Data Sharing Agreement before proceeding with that new purpose .

Dataset 1: Birth Notifications data

NHS Digital has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

The birth notifications data contribute to ONS statistical analyses of births, maternities, and infant mortality outcomes. Analyses are made publicly available as aggregate National Statistics. These statistics help a range of public and other bodies make better decisions (see section 5d). They also feed into the Department of Health's NHS Outcomes Framework for monitoring low birthweight of term babies.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics, and ONS become controllers of that data under Section 42 of the 2007 Statistics and Registration Services Act. However, there are some limitations with the GRO data, including a lack of medical information such as length of gestation, as well as some missing and implausible values in the fields that are available.

To mitigate these limitations, the NHS Digital birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. Several identifying variables such as NHS number are received to enable this linkage.

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

ONS also plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link siblings born at different times (i.e. not twins) using the NHS number of the mother which is only available on the notification data. This provides a ‘gold standard’ linkage method.

ONS can then then attempt to link siblings together using only the data available in the registration data – e.g. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the latter linkage method matches the gold standard. This will inform the best matching methodology to use when seeking to link siblings if NHS number of mother is not available. This is needed to link pre-2005 birth registration data, a time when the birth notification data is not available to ONS. This purpose would fall under the Research and development of methodology category in the uses framework above.

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS Digital data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realized.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data based ethnicity estimates will be important when deciding whether to move to an admin data based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS Digital data experts to understand the limitations of the HES ethnicity data and there are several; for example coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘Activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about why people have accessed hospital services, for example diagnosis, for a range of statistical purposes.

This information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, the volume of this information is being minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services will be shared with ONS, compared with the information about when and where people have accessed services.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital, or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly. It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible to develop new measures of avoidable mortality in the elderly.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality. The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments, and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, newborn and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS Digital and Public Health England (PHE) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability and geography. In some cases, NHS Digital / PHE will not hold data that would enable this, but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (2011 Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up to date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

Dataset 3: Improving Access to Psychological Therapies (IAPT) Dataset

3.1. To enable research being conducted by ONS’ Administrative Data Census and Migration Statistics improvement projects using ‘activity’ and characteristics data from IAPT.

This first use is essentially the same as described for the uses of HES data within these projects: The IAPT data provides evidence of presence at a particular address, and it also includes information on characteristics including ethnicity. See section 2.1 above (within the HES section) for the full rationale for why this information is needed.

3.2. To conduct a range of Statistical Research and Health Analyses using IAPT data

a. Statistical Research to inform Primary Mental Health Service Policy Making

This project will focus on common mental health disorders (CMDs) such as anxiety and depression. Using a phased approach, ONS will look first at the mortality risk of people with CMDs, co-morbidities between mental and physical health problems, and investigate inequalities around mental health. In the second phase, ONS will investigate income and employment transitions for patients who have been through mental health treatment.

The first phase will address existing evidence gaps on co-morbidities between mental and physical health, improve understanding on the demographics of people with CMDs, and investigate whether some or all people with CMDs are more at risk of death than the general population.

The IAPT data for 2012 to 2017 will be linked to the 2011 Census to provide detailed demographic background, and to death registrations from 2012 to 2018. The mortality analysis will focus on specific causes of death which may be connected to mental health (suicide, alcohol and drug abuse) as well as overall risk. In addition, the causes of death will be compared to the distribution of causes in the general population to identify any common co-morbidity with life-threatening illnesses. This goes some way to provide insight into important issues raised by the NHS England Five Year Forward View on mental health:

“An important barrier to good care is the lack of appropriate data sharing to enable organisations to identify co-morbidities…People with poor mental health may require primary care, secondary physical care and social care, as well as mental health services, but the lack of linked datasets hinders effective provision.”

IAPT data is estimated to cover over 15% of people with CMDs in England. Because of the service’s large, national scale and focus on people with mild and moderate mental health conditions, it provides a reasonable proxy for patterns and trends in the population of people with diagnosable CMDs.

The IAPT data will be compared with the findings of the Adult Psychiatric Morbidity Survey (2007 and 2014) to assess likely issues of representativeness, such as the under-representation of specific population groups in the treatment cohort. People with severe mental health conditions are not typically treated in the IAPT programme.

The three-way linkage will provide an independent and more detailed demographic baseline than the IAPT data could do alone, and allow ONS to investigate if there have been changes in peoples’ circumstances between the Census and treatment in IAPT (e.g. becoming disabled or living alone). Having the mortality data linked as well allows ONS to see the overall trends in mortality, plus to see if there is any relationship between changes in demographics and the cause of death outcomes.

The research is not aiming to look at individual level outcomes or to evaluate the IAPT treatment, but to look for trends in the aggregate data after linkage, to provide population level analysis to inform policy.

Entry into IAPT treatment will be used as the main indicator of having a diagnosable CMD. The clinical data will not be analysed except to:

• Group the cohort into broad types of CMD

• Potentially, link successful/unsuccessful treatment outcome to risk of subsequent death.

b. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released between August and December in a series of themed packages. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison, sometimes back to 1837. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

There are several published packages:

Birth summary tables: includes the number of live births and stillbirths, fertility rates, percentage of live births outside marriage and civil partnership, mean age of mother and percentage of live births to non-UK born mothers for England and Wales as a whole. Live births (number and rate) and the number of stillbirths are also provided down to local authority level. To aid with user interpretation, ONS also publishes an interactive fertility mapping tool, which enables users to analyse trends in fertility by county district and unitary authority; this is contained within the statistical bulletin.

Parents’ country of birth: includes births by country of birth of mother and total fertility rates for UK born and non-UK born women for England and Wales as a whole. Summary figures are also available down to local authority level. ONS publishes detailed analysis on parents’ country of birth because this information is collected at birth registration and does not change over time, while their nationality or ethnicity may change.

Birth characteristics and by area of usual residence: contains statistics on stillbirths and maternities for England and Wales, birthweight data for live and stillbirths by mother's region of usual residence, and live births and stillbirths in hospitals and communal establishments by region of occurrence. These tables also provide figures on month and quarter of occurrence, place of birth, ethnicity and gestational age and multiple births for England and Wales as a whole. Also provides summary data for live births down to local authority level including figures by age of mother figures are published using boundaries in place during the year the birth occurred.

Births by parents’ characteristics: provides live birth, stillbirth and maternity statistics by age of mother and type of registration (within marriage and civil partnership, joint, sole). It also provides data on previous live-born children, National Statistics Socio-economic Classification (NS-SEC), median birth intervals, age-specific fertility rates for men and mean age of fathers. All tables are for England and Wales as a whole with no sub-national breakdown.

Childbearing for women born in different years (formerly known as Cohort fertility): presents data on fertility by year of birth of mother rather than the year of birth of child for England and Wales as a whole this package includes the average number of live-born children and the proportion of women remaining childless for women born in different years.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Dataset 2 and dataset 3: Hospital Episode Statistics and Improving Access to Psychological Therapies data

The initial uses to which ONS will put HES and IAPT data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES or IAPT data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users, and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS Digital, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS Digital’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest, and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS Digital will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS Digital approval . It would also require an appropriate alternative legal gateway, because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS Digital and enter into an amended Data Sharing Agreement before proceeding with that new purpose .

Dataset 1: Birth Notifications data

NHS Digital has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

The birth notifications data contribute to ONS statistical analyses of births, maternities, and infant mortality outcomes. Analyses are made publicly available as aggregate National Statistics. These statistics help a range of public and other bodies make better decisions (see section 5d). They also feed into the Department of Health's NHS Outcomes Framework for monitoring low birthweight of term babies.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics, and ONS become controllers of that data under Section 42 of the 2007 Statistics and Registration Services Act. However, there are some limitations with the GRO data, including a lack of medical information such as length of gestation, as well as some missing and implausible values in the fields that are available.

To mitigate these limitations, the NHS Digital birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. Several identifying variables such as NHS number are received to enable this linkage.

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

ONS also plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link siblings born at different times (i.e. not twins) using the NHS number of the mother which is only available on the notification data. This provides a ‘gold standard’ linkage method.

ONS can then then attempt to link siblings together using only the data available in the registration data – e.g. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the latter linkage method matches the gold standard. This will inform the best matching methodology to use when seeking to link siblings if NHS number of mother is not available. This is needed to link pre-2005 birth registration data, a time when the birth notification data is not available to ONS. This purpose would fall under the Research and development of methodology category in the uses framework above.

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS Digital data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realized.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data based ethnicity estimates will be important when deciding whether to move to an admin data based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS Digital data experts to understand the limitations of the HES ethnicity data and there are several; for example coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘Activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about why people have accessed hospital services, for example diagnosis, for a range of statistical purposes.

This information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, the volume of this information is being minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services will be shared with ONS, compared with the information about when and where people have accessed services.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital, or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly. It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible to develop new measures of avoidable mortality in the elderly.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality. The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments, and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, newborn and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS Digital and Public Health England (PHE) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability and geography. In some cases, NHS Digital / PHE will not hold data that would enable this, but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (2011 Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up to date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

Dataset 3: Improving Access to Psychological Therapies (IAPT) Dataset

3.1. To enable research being conducted by ONS’ Administrative Data Census and Migration Statistics improvement projects using ‘activity’ and characteristics data from IAPT.

This first use is essentially the same as described for the uses of HES data within these projects: The IAPT data provides evidence of presence at a particular address, and it also includes information on characteristics including ethnicity. See section 2.1 above (within the HES section) for the full rationale for why this information is needed.

3.2. To conduct a range of Statistical Research and Health Analyses using IAPT data

a. Statistical Research to inform Primary Mental Health Service Policy Making

This project will focus on common mental health disorders (CMDs) such as anxiety and depression. Using a phased approach, ONS will look first at the mortality risk of people with CMDs, co-morbidities between mental and physical health problems, and investigate inequalities around mental health. In the second phase, ONS will investigate income and employment transitions for patients who have been through mental health treatment.

The first phase will address existing evidence gaps on co-morbidities between mental and physical health, improve understanding on the demographics of people with CMDs, and investigate whether some or all people with CMDs are more at risk of death than the general population.

The IAPT data for 2012 to 2017 will be linked to the 2011 Census to provide detailed demographic background, and to death registrations from 2012 to 2018. The mortality analysis will focus on specific causes of death which may be connected to mental health (suicide, alcohol and drug abuse) as well as overall risk. In addition, the causes of death will be compared to the distribution of causes in the general population to identify any common co-morbidity with life-threatening illnesses. This goes some way to provide insight into important issues raised by the NHS England Five Year Forward View on mental health:

“An important barrier to good care is the lack of appropriate data sharing to enable organisations to identify co-morbidities…People with poor mental health may require primary care, secondary physical care and social care, as well as mental health services, but the lack of linked datasets hinders effective provision.”

IAPT data is estimated to cover over 15% of people with CMDs in England. Because of the service’s large, national scale and focus on people with mild and moderate mental health conditions, it provides a reasonable proxy for patterns and trends in the population of people with diagnosable CMDs.

The IAPT data will be compared with the findings of the Adult Psychiatric Morbidity Survey (2007 and 2014) to assess likely issues of representativeness, such as the under-representation of specific population groups in the treatment cohort. People with severe mental health conditions are not typically treated in the IAPT programme.

The three-way linkage will provide an independent and more detailed demographic baseline than the IAPT data could do alone, and allow ONS to investigate if there have been changes in peoples’ circumstances between the Census and treatment in IAPT (e.g. becoming disabled or living alone). Having the mortality data linked as well allows ONS to see the overall trends in mortality, plus to see if there is any relationship between changes in demographics and the cause of death outcomes.

The research is not aiming to look at individual level outcomes or to evaluate the IAPT treatment, but to look for trends in the aggregate data after linkage, to provide population level analysis to inform policy.

Entry into IAPT treatment will be used as the main indicator of having a diagnosable CMD. The clinical data will not be analysed except to:

• Group the cohort into broad types of CMD

• Potentially, link successful/unsuccessful treatment outcome to risk of subsequent death.

b. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released between August and December in a series of themed packages. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison, sometimes back to 1837. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

There are several published packages:

Birth summary tables: includes the number of live births and stillbirths, fertility rates, percentage of live births outside marriage and civil partnership, mean age of mother and percentage of live births to non-UK born mothers for England and Wales as a whole. Live births (number and rate) and the number of stillbirths are also provided down to local authority level. To aid with user interpretation, ONS also publishes an interactive fertility mapping tool, which enables users to analyse trends in fertility by county district and unitary authority; this is contained within the statistical bulletin.

Parents’ country of birth: includes births by country of birth of mother and total fertility rates for UK born and non-UK born women for England and Wales as a whole. Summary figures are also available down to local authority level. ONS publishes detailed analysis on parents’ country of birth because this information is collected at birth registration and does not change over time, while their nationality or ethnicity may change.

Birth characteristics and by area of usual residence: contains statistics on stillbirths and maternities for England and Wales, birthweight data for live and stillbirths by mother's region of usual residence, and live births and stillbirths in hospitals and communal establishments by region of occurrence. These tables also provide figures on month and quarter of occurrence, place of birth, ethnicity and gestational age and multiple births for England and Wales as a whole. Also provides summary data for live births down to local authority level including figures by age of mother figures are published using boundaries in place during the year the birth occurred.

Births by parents’ characteristics: provides live birth, stillbirth and maternity statistics by age of mother and type of registration (within marriage and civil partnership, joint, sole). It also provides data on previous live-born children, National Statistics Socio-economic Classification (NS-SEC), median birth intervals, age-specific fertility rates for men and mean age of fathers. All tables are for England and Wales as a whole with no sub-national breakdown.

Childbearing for women born in different years (formerly known as Cohort fertility): presents data on fertility by year of birth of mother rather than the year of birth of child for England and Wales as a whole this package includes the average number of live-born children and the proportion of women remaining childless for women born in different years.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Dataset 2 and dataset 3: Hospital Episode Statistics and Improving Access to Psychological Therapies data

The initial uses to which ONS will put HES and IAPT data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES or IAPT data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users, and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS Digital, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS Digital’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest, and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS Digital will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS Digital approval . It would also require an appropriate alternative legal gateway, because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS Digital and enter into an amended Data Sharing Agreement before proceeding with that new purpose .

Dataset 1: Birth Notifications data

NHS Digital has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

The birth notifications data contribute to ONS statistical analyses of births, maternities, and infant mortality outcomes. Analyses are made publicly available as aggregate National Statistics. These statistics help a range of public and other bodies make better decisions (see section 5d). They also feed into the Department of Health's NHS Outcomes Framework for monitoring low birthweight of term babies.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics, and ONS become controllers of that data under Section 42 of the 2007 Statistics and Registration Services Act. However, there are some limitations with the GRO data, including a lack of medical information such as length of gestation, as well as some missing and implausible values in the fields that are available.

To mitigate these limitations, the NHS Digital birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. Several identifying variables such as NHS number are received to enable this linkage.

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

ONS also plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link siblings born at different times (i.e. not twins) using the NHS number of the mother which is only available on the notification data. This provides a ‘gold standard’ linkage method.

ONS can then then attempt to link siblings together using only the data available in the registration data – e.g. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the latter linkage method matches the gold standard. This will inform the best matching methodology to use when seeking to link siblings if NHS number of mother is not available. This is needed to link pre-2005 birth registration data, a time when the birth notification data is not available to ONS. This purpose would fall under the Research and development of methodology category in the uses framework above.

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS Digital data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realized.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data based ethnicity estimates will be important when deciding whether to move to an admin data based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS Digital data experts to understand the limitations of the HES ethnicity data and there are several; for example coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘Activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about why people have accessed hospital services, for example diagnosis, for a range of statistical purposes.

This information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, the volume of this information is being minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services will be shared with ONS, compared with the information about when and where people have accessed services.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital, or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly. It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible to develop new measures of avoidable mortality in the elderly.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality. The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments, and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, newborn and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS Digital and Public Health England (PHE) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability and geography. In some cases, NHS Digital / PHE will not hold data that would enable this, but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

2.6. Rapid response project to investigate the socio-economic factors and underlying health conditions associated with worse outcomes from contracting the COVID-19 virus

Hospital episodes linked to COVID-19 allows ONS to identify incidences where people are hospitalised but recover, filling a key gap in visibility of cases. This will be used as part of a large scale COVID-19 linkage project. Data will be linked to data on Deaths, demographics (2011 Census) and primary care data to establish and assess commodities and risk factors associated with COVID-19.

This is of critical priority across government as part of the UK’s response to the COVID-19 pandemic. This will contribute to the wider understanding of the virus, helping to inform a range of policy decisions taken by central government, health services and others.

ONS previously held HES data covering up until March 2019 and was to receive annual updates. This is not timely enough to enable all of the aims of the project. The gap in HES data between April 2019 to February 2020 (from before the UK epidemic) means ONS did not have an up to date picture of underlying health conditions. Getting retrospective monthly HES data for this period will fill this gap sooner than waiting for the final annual 2019/20 extract which will be supplied by NHSD in October 2020.

ONS also needs hospital data from during the UK epidemic to have visibility of those who are hospitalised from COVID-19 but then recover. An ongoing monthly supply of HES data will provide ONS with this information much sooner than an annual supply would provide.

Dataset 3: Improving Access to Psychological Therapies (IAPT) Dataset

3.1. To enable research being conducted by ONS’ Administrative Data Census and Migration Statistics improvement projects using ‘activity’ and characteristics data from IAPT.

This first use is essentially the same as described for the uses of HES data within these projects: The IAPT data provides evidence of presence at a particular address, and it also includes information on characteristics including ethnicity. See section 2.1 above (within the HES section) for the full rationale for why this information is needed.

3.2. To conduct a range of Statistical Research and Health Analyses using IAPT data

a. Statistical Research to inform Primary Mental Health Service Policy Making

This project will focus on common mental health disorders (CMDs) such as anxiety and depression. Using a phased approach, ONS will look first at the mortality risk of people with CMDs, co-morbidities between mental and physical health problems, and investigate inequalities around mental health. In the second phase, ONS will investigate income and employment transitions for patients who have been through mental health treatment.

The first phase will address existing evidence gaps on co-morbidities between mental and physical health, improve understanding on the demographics of people with CMDs, and investigate whether some or all people with CMDs are more at risk of death than the general population.

The IAPT data for 2012 to 2017 will be linked to the 2011 Census to provide detailed demographic background, and to death registrations from 2012 to 2018. The mortality analysis will focus on specific causes of death which may be connected to mental health (suicide, alcohol and drug abuse) as well as overall risk. In addition, the causes of death will be compared to the distribution of causes in the general population to identify any common co-morbidity with life-threatening illnesses. This goes some way to provide insight into important issues raised by the NHS England Five Year Forward View on mental health:

“An important barrier to good care is the lack of appropriate data sharing to enable organisations to identify co-morbidities…People with poor mental health may require primary care, secondary physical care and social care, as well as mental health services, but the lack of linked datasets hinders effective provision.”

IAPT data is estimated to cover over 15% of people with CMDs in England. Because of the service’s large, national scale and focus on people with mild and moderate mental health conditions, it provides a reasonable proxy for patterns and trends in the population of people with diagnosable CMDs.

The IAPT data will be compared with the findings of the Adult Psychiatric Morbidity Survey (2007 and 2014) to assess likely issues of representativeness, such as the under-representation of specific population groups in the treatment cohort. People with severe mental health conditions are not typically treated in the IAPT programme.

The three-way linkage will provide an independent and more detailed demographic baseline than the IAPT data could do alone, and allow ONS to investigate if there have been changes in peoples’ circumstances between the Census and treatment in IAPT (e.g. becoming disabled or living alone). Having the mortality data linked as well allows ONS to see the overall trends in mortality, plus to see if there is any relationship between changes in demographics and the cause of death outcomes.

The research is not aiming to look at individual level outcomes or to evaluate the IAPT treatment, but to look for trends in the aggregate data after linkage, to provide population level analysis to inform policy.

Entry into IAPT treatment will be used as the main indicator of having a diagnosable CMD. The clinical data will not be analysed except to:

• Group the cohort into broad types of CMD

• Potentially, link successful/unsuccessful treatment outcome to risk of subsequent death.

b. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released between August and December in a series of themed packages. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison, sometimes back to 1837. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

There are several published packages:

Birth summary tables: includes the number of live births and stillbirths, fertility rates, percentage of live births outside marriage and civil partnership, mean age of mother and percentage of live births to non-UK born mothers for England and Wales as a whole. Live births (number and rate) and the number of stillbirths are also provided down to local authority level. To aid with user interpretation, ONS also publishes an interactive fertility mapping tool, which enables users to analyse trends in fertility by county district and unitary authority; this is contained within the statistical bulletin.

Parents’ country of birth: includes births by country of birth of mother and total fertility rates for UK born and non-UK born women for England and Wales as a whole. Summary figures are also available down to local authority level. ONS publishes detailed analysis on parents’ country of birth because this information is collected at birth registration and does not change over time, while their nationality or ethnicity may change.

Birth characteristics and by area of usual residence: contains statistics on stillbirths and maternities for England and Wales, birthweight data for live and stillbirths by mother's region of usual residence, and live births and stillbirths in hospitals and communal establishments by region of occurrence. These tables also provide figures on month and quarter of occurrence, place of birth, ethnicity and gestational age and multiple births for England and Wales as a whole. Also provides summary data for live births down to local authority level including figures by age of mother figures are published using boundaries in place during the year the birth occurred.

Births by parents’ characteristics: provides live birth, stillbirth and maternity statistics by age of mother and type of registration (within marriage and civil partnership, joint, sole). It also provides data on previous live-born children, National Statistics Socio-economic Classification (NS-SEC), median birth intervals, age-specific fertility rates for men and mean age of fathers. All tables are for England and Wales as a whole with no sub-national breakdown.

Childbearing for women born in different years (formerly known as Cohort fertility): presents data on fertility by year of birth of mother rather than the year of birth of child for England and Wales as a whole this package includes the average number of live-born children and the proportion of women remaining childless for women born in different years.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Dataset 2 and dataset 3: Hospital Episode Statistics and Improving Access to Psychological Therapies data

The initial uses to which ONS will put HES and IAPT data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES or IAPT data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users, and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS Digital, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS Digital’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest, and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS Digital will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS Digital approval . It would also require an appropriate alternative legal gateway, because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS Digital and enter into an amended Data Sharing Agreement before proceeding with that new purpose .

Dataset 1: Birth Notifications data

NHS Digital has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

The birth notifications data contribute to ONS statistical analyses of births, maternities, and infant mortality outcomes. Analyses are made publicly available as aggregate National Statistics. These statistics help a range of public and other bodies make better decisions (see section 5d). They also feed into the Department of Health's NHS Outcomes Framework for monitoring low birthweight of term babies.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics, and ONS become controllers of that data under Section 42 of the 2007 Statistics and Registration Services Act. However, there are some limitations with the GRO data, including a lack of medical information such as length of gestation, as well as some missing and implausible values in the fields that are available.

To mitigate these limitations, the NHS Digital birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. Several identifying variables such as NHS number are received to enable this linkage.

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

ONS also plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link siblings born at different times (i.e. not twins) using the NHS number of the mother which is only available on the notification data. This provides a ‘gold standard’ linkage method.

ONS can then then attempt to link siblings together using only the data available in the registration data – e.g. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the latter linkage method matches the gold standard. This will inform the best matching methodology to use when seeking to link siblings if NHS number of mother is not available. This is needed to link pre-2005 birth registration data, a time when the birth notification data is not available to ONS. This purpose would fall under the Research and development of methodology category in the uses framework above.

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS Digital data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realized.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data based ethnicity estimates will be important when deciding whether to move to an admin data based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS Digital data experts to understand the limitations of the HES ethnicity data and there are several; for example coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘Activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about why people have accessed hospital services, for example diagnosis, for a range of statistical purposes.

This information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, the volume of this information is being minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services will be shared with ONS, compared with the information about when and where people have accessed services.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital, or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly. It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible to develop new measures of avoidable mortality in the elderly.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality. The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments, and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, newborn and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS Digital and Public Health England (PHE) to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability and geography. In some cases, NHS Digital / PHE will not hold data that would enable this, but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

Dataset 3: Improving Access to Psychological Therapies (IAPT) Dataset

3.1. To enable research being conducted by ONS’ Administrative Data Census and Migration Statistics improvement projects using ‘activity’ and characteristics data from IAPT.

This first use is essentially the same as described for the uses of HES data within these projects: The IAPT data provides evidence of presence at a particular address, and it also includes information on characteristics including ethnicity. See section 2.1 above (within the HES section) for the full rationale for why this information is needed.

3.2. To conduct a range of Statistical Research and Health Analyses using IAPT data

a. Statistical Research to inform Primary Mental Health Service Policy Making

This project will focus on common mental health disorders (CMDs) such as anxiety and depression. Using a phased approach, ONS will look first at the mortality risk of people with CMDs, co-morbidities between mental and physical health problems, and investigate inequalities around mental health. In the second phase, ONS will investigate income and employment transitions for patients who have been through mental health treatment.

The first phase will address existing evidence gaps on co-morbidities between mental and physical health, improve understanding on the demographics of people with CMDs, and investigate whether some or all people with CMDs are more at risk of death than the general population.

The IAPT data for 2012 to 2017 will be linked to the 2011 Census to provide detailed demographic background, and to death registrations from 2012 to 2018. The mortality analysis will focus on specific causes of death which may be connected to mental health (suicide, alcohol and drug abuse) as well as overall risk. In addition, the causes of death will be compared to the distribution of causes in the general population to identify any common co-morbidity with life-threatening illnesses. This goes some way to provide insight into important issues raised by the NHS England Five Year Forward View on mental health:

“An important barrier to good care is the lack of appropriate data sharing to enable organisations to identify co-morbidities…People with poor mental health may require primary care, secondary physical care and social care, as well as mental health services, but the lack of linked datasets hinders effective provision.”

IAPT data is estimated to cover over 15% of people with CMDs in England. Because of the service’s large, national scale and focus on people with mild and moderate mental health conditions, it provides a reasonable proxy for patterns and trends in the population of people with diagnosable CMDs.

The IAPT data will be compared with the findings of the Adult Psychiatric Morbidity Survey (2007 and 2014) to assess likely issues of representativeness, such as the under-representation of specific population groups in the treatment cohort. People with severe mental health conditions are not typically treated in the IAPT programme.

The three-way linkage will provide an independent and more detailed demographic baseline than the IAPT data could do alone, and allow ONS to investigate if there have been changes in peoples’ circumstances between the Census and treatment in IAPT (e.g. becoming disabled or living alone). Having the mortality data linked as well allows ONS to see the overall trends in mortality, plus to see if there is any relationship between changes in demographics and the cause of death outcomes.

The research is not aiming to look at individual level outcomes or to evaluate the IAPT treatment, but to look for trends in the aggregate data after linkage, to provide population level analysis to inform policy.

Entry into IAPT treatment will be used as the main indicator of having a diagnosable CMD. The clinical data will not be analysed except to:

• Group the cohort into broad types of CMD

• Potentially, link successful/unsuccessful treatment outcome to risk of subsequent death.

b. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and IAPT data could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data to make inferences about the health of the population. However, ONS will investigate the possibilities of IAPT data contributing to more complete estimation of morbidity, to then use alongside mortality data and other relevant sources in health projection modelling. This assessment of morbidity would be in conjunction with other sources such as NHS Digital’s Hospital Episode Statistics (HES) that the Board has already required be shared with ONS.

c. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions, would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking IAPT data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released between August and December in a series of themed packages. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison, sometimes back to 1837. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

There are several published packages:

Birth summary tables: includes the number of live births and stillbirths, fertility rates, percentage of live births outside marriage and civil partnership, mean age of mother and percentage of live births to non-UK born mothers for England and Wales as a whole. Live births (number and rate) and the number of stillbirths are also provided down to local authority level. To aid with user interpretation, ONS also publishes an interactive fertility mapping tool, which enables users to analyse trends in fertility by county district and unitary authority; this is contained within the statistical bulletin.

Parents’ country of birth: includes births by country of birth of mother and total fertility rates for UK born and non-UK born women for England and Wales as a whole. Summary figures are also available down to local authority level. ONS publishes detailed analysis on parents’ country of birth because this information is collected at birth registration and does not change over time, while their nationality or ethnicity may change.

Birth characteristics and by area of usual residence: contains statistics on stillbirths and maternities for England and Wales, birthweight data for live and stillbirths by mother's region of usual residence, and live births and stillbirths in hospitals and communal establishments by region of occurrence. These tables also provide figures on month and quarter of occurrence, place of birth, ethnicity and gestational age and multiple births for England and Wales as a whole. Also provides summary data for live births down to local authority level including figures by age of mother figures are published using boundaries in place during the year the birth occurred.

Births by parents’ characteristics: provides live birth, stillbirth and maternity statistics by age of mother and type of registration (within marriage and civil partnership, joint, sole). It also provides data on previous live-born children, National Statistics Socio-economic Classification (NS-SEC), median birth intervals, age-specific fertility rates for men and mean age of fathers. All tables are for England and Wales as a whole with no sub-national breakdown.

Childbearing for women born in different years (formerly known as Cohort fertility): presents data on fertility by year of birth of mother rather than the year of birth of child for England and Wales as a whole this package includes the average number of live-born children and the proportion of women remaining childless for women born in different years.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Dataset 2 and dataset 3: Hospital Episode Statistics and Improving Access to Psychological Therapies data

The initial uses to which ONS will put HES and IAPT data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES or IAPT data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users, and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS Digital, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS Digital’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest, and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS Digital will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS Digital approval . It would also require an appropriate alternative legal gateway, because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS Digital and enter into an amended Data Sharing Agreement before proceeding with that new purpose .

Dataset 1: Birth Notifications data

NHS Digital has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

The birth notifications data contribute to ONS statistical analyses of births, maternities, and infant mortality outcomes. Analyses are made publicly available as aggregate National Statistics. These statistics help a range of public and other bodies make better decisions (see section 5d). They also feed into the Department of Health's NHS Outcomes Framework for monitoring low birthweight of term babies.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics, and ONS become controllers of that data under Section 42 of the 2007 Statistics and Registration Services Act. However, there are some limitations with the GRO data, including a lack of medical information such as length of gestation, as well as some missing and implausible values in the fields that are available.

To mitigate these limitations, the NHS Digital birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. Several identifying variables such as NHS number are received to enable this linkage.

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

ONS also plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link siblings born at different times (i.e. not twins) using the NHS number of the mother which is only available on the notification data. This provides a ‘gold standard’ linkage method.

ONS can then then attempt to link siblings together using only the data available in the registration data – e.g. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the latter linkage method matches the gold standard. This will inform the best matching methodology to use when seeking to link siblings if NHS number of mother is not available. This is needed to link pre-2005 birth registration data, a time when the birth notification data is not available to ONS. This purpose would fall under the Research and development of methodology category in the uses framework above.

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS Digital data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realized.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data based ethnicity estimates will be important when deciding whether to move to an admin data based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS Digital data experts to understand the limitations of the HES ethnicity data and there are several; for example coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘Activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about why people have accessed hospital services, for example diagnosis, for a range of statistical purposes.

This information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, the volume of this information is being minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services will be shared with ONS, compared with the information about when and where people have accessed services.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital, or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly. It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible to develop new measures of avoidable mortality in the elderly.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality. The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments, and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be used to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, newborn and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS Digital and Public Health England (PHE)to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability and geography. In some cases, NHS Digital / PHE will not hold data that would enable this, but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released between August and December in a series of themed packages. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison, sometimes back to 1837. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

There are several published packages:

Birth summary tables: includes the number of live births and stillbirths, fertility rates, percentage of live births outside marriage and civil partnership, mean age of mother and percentage of live births to non-UK born mothers for England and Wales as a whole. Live births (number and rate) and the number of stillbirths are also provided down to local authority level. To aid with user interpretation, ONS also publishes an interactive fertility mapping tool, which enables users to analyse trends in fertility by county district and unitary authority; this is contained within the statistical bulletin.

Parents’ country of birth: includes births by country of birth of mother and total fertility rates for UK born and non-UK born women for England and Wales as a whole. Summary figures are also available down to local authority level. ONS publishes detailed analysis on parents’ country of birth because this information is collected at birth registration and does not change over time, while their nationality or ethnicity may change.

Birth characteristics and by area of usual residence: contains statistics on stillbirths and maternities for England and Wales, birthweight data for live and stillbirths by mother's region of usual residence, and live births and stillbirths in hospitals and communal establishments by region of occurrence. These tables also provide figures on month and quarter of occurrence, place of birth, ethnicity and gestational age and multiple births for England and Wales as a whole. Also provides summary data for live births down to local authority level including figures by age of mother figures are published using boundaries in place during the year the birth occurred.

Births by parents’ characteristics: provides live birth, stillbirth and maternity statistics by age of mother and type of registration (within marriage and civil partnership, joint, sole). It also provides data on previous live-born children, National Statistics Socio-economic Classification (NS-SEC), median birth intervals, age-specific fertility rates for men and mean age of fathers. All tables are for England and Wales as a whole with no sub-national breakdown.

Childbearing for women born in different years (formerly known as Cohort fertility): presents data on fertility by year of birth of mother rather than the year of birth of child for England and Wales as a whole this package includes the average number of live-born children and the proportion of women remaining childless for women born in different years.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Dataset 2: Hospital Episode Statistics

The initial uses to which ONS will put HES data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users, and will seek to provide frequent updates on these projects during that first stage.

The Office for National Statistics (ONS), as the executive arm of the UK Statistics Authority (UKSA) requires access to administrative data held by NHS Digital, for the production of official statistics.

In the past it has been difficult for ONS to access administrative data controlled by other Government departments, information that could potentially transform official statistics and the impact they have on decision making for the better. Often, this has been caused by the lack of a clear legal basis under which the data can be shared with ONS. As a result, in 2016, ONS set out why legislation was needed for better access to data:

https://www.statisticsauthority.gov.uk/publication/delivering-better-statistics-for-better-decisions-data-access-legislation-march-2016/

As a result, the Digital Economy Act in April 2017 amended the Statistics and Registration Services Act (2007) (SRSA) such that ONS can require public authorities to share data with it. See the Digital Economy Act (chapter 7 of part 5):

http://www.legislation.gov.uk/ukpga/2017/30/part/5/chapter/7/enacted

More specifically, section 45c of the SRSA 2007 (as inserted by section 80 of the Digital Economy Act 2017) permits the Statistics Board (of which ONS is part) to serve a Notice on a public authority requiring it to disclose information it holds in connection with its functions:

http://www.legislation.gov.uk/ukpga/2007/18/section/45C

To do so, the information so disclosed must be required by the Statistics Board for one or more of its functions as set out in the SRSA 2007 and the Census Act 1920.

The SRSA (2007) states that the ONS’s objectives include ‘promoting and safeguarding the production and publication of official statistics that serve the public good, where serving public good includes informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. It also sets out the Board’s functions, which are the specifically referred to in section 45c of the amended SRSA. Notably they include, under section 20, that ONS ‘may produce and publish statistics relating to any matter relating to the United Kingdom or any part of it’.

Requirements made under section 45 must also be in line with a statistical statement of principles that has been approved by parliament:

https://www.gov.uk/government/publications/digital-economy-act-2017-part-5-codes-of-practice/statistics-statement-of-principles-and-code-of-practice-on-changes-to-data-systems

This states that ‘We will only seek access to data for the purposes of fulfilling one or more of our statutory functions, including to produce official statistics and undertake statistical research that meets identifiable user needs for the public good.’

The statement also sets out six principles to which ONS will adhere when requiring information under section 45; they state that ONS will:

• safeguard confidentiality

• be transparent about what data it is accessing and why

• ensure accessing the data is lawful and meet strict ethical standards

• ensure that accessing the data is in the public interest - for example that the data are fit for purpose for the statistical use which ONS intends

• ensure requiring that the data be supplied is proportionate – for example, ONS will have exhausted possible alternatives

• seek to collaborate with suppliers at all times

In addition, the following is a useful framework for categorizing ONS’s statistical uses for information such as that covered under this agreement. They are all ultimately all related to ONS’s functions of producing Official Statistics mentioned earlier:

• Improvements to existing Official Statistics

• Development of new Official Statistics – this may involve testing to investigate whether statistics of sufficient quality can be produced, and may also involve the production of statistics badged as ‘experimental’ while further work is done to improve quality aspects such as accuracy

• Quality assurance of Official Statistics

• Development of commentary around Official Statistics

• Replacement of current survey questions – developing statistics from available data to directly replace the need to collect the information through survey questions

• Improving efficiency or accuracy of sampling – for example, ensuring that a representative sample of the target population is taken when conducting a survey of the public, such that the statistics produced from the survey are the best possible reflection of reality

• Research and development of methodology – for example, using data to develop and test linkage methodology that is ultimately used to help produce statistics based on other data rather than the original data source

Using robust information governance processes, ONS has determined that the conditions associated with requiring data under section 45c of the amended SRSA have been met for the information in this data sharing agreement. This process involved working closely with NHS Digital’s experts to help determine that the data would likely be of good enough quality to meet the proposed statistical purposes. This work guided ONS’s assessment against some of the principles underpinning its legal powers – for example whether sharing the data is in the public interest, and proportionate in terms of burden on the supplier. In addition, as part of its commitment to transparency, ONS will publish full details of the reasons for acquiring the information, and ONS notes that NHS Digital will also publish this data sharing agreement.

In terms of public interest, it is worth noting that the benefits gained from the statistics enabled by this data share do not need to be specific to health and social care when data are flowing under section 45 of the SRSA. For example, some of the data being required will help improve ONS’s population and economic statistics, and in these cases, the improved statistics may not benefit health and social care directly.

The data shared with ONS under this agreement will not be onwardly disseminated or shared, except as disclosure controlled aggregate statistics and/or analysis as aggregated data with small numbers suppressed, in line with the Hospital Episode Statistics Analysis Guide. Any exceptions to this would require additional NHS Digital approval. It would also require an appropriate alternative legal gateway, because section 45c of the SRSA as amended by the Digital Economy Act only enables data to be shared with ONS (not for example, other Government departments or academic researchers).

The rest of this section will set out the specific purposes for which ONS requires each dataset. Each purpose will be linked to the framework of statistical uses set out above.

In future, ONS may decide to put a dataset to new uses not explained below. In these cases, the new use will be in line with ONS’s legally defined functions. ONS will inform NHS Digital and enter into an amended Data Sharing Agreement before proceeding with that new purpose.

Dataset 1: Birth Notifications data

NHS Digital has disseminated birth notifications data to ONS since 2005. Support under section 251 of the NHS Act 2006 (reference PIAG 4-05(d)/2005) permitted this sharing but the legal gateway under which the data will continue to flow will change to section 45c of the amended SRSA 2007.

The birth notifications data contribute to ONS statistical analyses of births, maternities, and infant mortality outcomes. Analyses are made publicly available as aggregate National Statistics. These statistics help a range of public and other bodies make better decisions (see section 5d). They also feed into the Department of Health's NHS Outcomes Framework for monitoring low birthweight of term babies.

Birth registration data that ONS receives from the General Register Office (GRO) is the primary source for producing these statistics, and ONS become controllers of that data under Section 42 of the 2007 Statistics and Registration Services Act. However, there are some limitations with the GRO data, including a lack of medical information such as length of gestation, as well as some missing and implausible values in the fields that are available.

To mitigate these limitations, the NHS Digital birth notifications data are used to improve and validate the registration data. Before this can be done, the two datasets must be linked at an individual level. Several identifying variables such as NHS number are received to enable this linkage.

In terms of the statistical uses framework set out earlier, then the data are used for:

• Improving official statistics – additional information not on the birth registrations data can be added at the record level once the two sources have been linked

• Quality assurance of official statistics – where information is on both sources, the birth notifications data can be used to validate the values contained in the birth registration data, and potentially edit (overwrite) the birth registrations data where that value is missing or implausible

ONS also plans to use birth notifications data to help develop and improve its data linkage methodology. For example, the birth notifications data allows ONS to link siblings born at different times (i.e. not twins) using the NHS number of the mother which is only available on the notification data. This provides a ‘gold standard’ linkage method.

ONS can then then attempt to link siblings together using only the data available in the registration data – e.g. mother’s name and date of birth, but not NHS number. ONS can then assess how closely the latter linkage method matches the gold standard. This will inform the best matching methodology to use when seeking to link siblings if NHS number of mother is not available. This is needed to link pre-2005 birth registration data, a time when the birth notification data is not available to ONS. This purpose would fall under the Research and development of methodology category in the uses framework above.

Dataset 2: Hospital Episode Statistics

There are a range of initial statistical uses to which ONS intends to put Hospital Episodes Statistics (HES) data.

Generally, linkage to other sources at a record level is a prerequisite to success for all proposed uses, and therefore identifiers including postcode, date of birth, sex and NHS number are required. The other HES information required varies by purpose, broken down below.

The specification of the variables being required has been developed in collaboration with NHS Digital data experts to ensure the data being shared are of sufficient quality (e.g. coverage, accuracy, relevance) to be likely to support the statistical purpose intended. The proposed uses of the HES data are as follows.

2.1. To enable ONS’s Administrative Data Census Project, including placing administrative data at the core of migration statistics, using ‘activity’ and characteristics data from HES

ONS’s Administrative Data Census Project (ADC) is assessing whether the Government’s ambition that ‘censuses after 2021 be conducted using other sources of data’ can be realized.

ONS aims to replicate the type of information collected through a census by using administrative data already held by government, supplemented by surveys. This can then be compared with the data collected by the 2021 census itself. This will allow ONS to determine whether this alternative approach can meet users’ needs.

In addition, ONS set out a cross-Government Statistical Service (GSS) programme working with the Home Office (the lead policy department), the devolved administrations and other government departments who have a strong interest in improving the migration evidence base. ONS aims to deliver improvements in migration statistics by putting administrative data at the core of migration statistics as part of the wider transformation to an administrative data-based population statistics system. The programme also recognises the changing demand from users of migration statistics and the need for more information on the impact migrants have while they are in the UK:

https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/articles/migrationstatisticstransformationupdate/2018-05-24

There are two main types of information from the Hospital Episodes Statistics dataset that are needed for these projects; so called ‘activity data’, and characteristics data.

a. Activity data

ONS has access to administrative sources that include a large proportion of the population such as GP patient registration information and tax records. These provide evidence of how many people live in each area of the country. However, these sources often suffer from over coverage. This is because people may have left the country but still appear in the data, creating the risk that the size of the national population is overestimated. Even when someone is still in the country, they may move without updating their address information with relevant services – for example, they may not register with a new GP at their new location until they need to see a doctor. In this case, there is a risk of ONS including them as contributing to the resident population in the wrong part of the country.

ONS can mitigate these limitations using other sources such as HES. For example, where these other sources show that an individual is interacting with a service, it provides evidence that they are in the country, and indeed which address information is correct (if the main sources mentioned earlier do not agree on this). For this particular use, ONS only requires information about where and when individuals are interacting with hospital services, not why.

b. Characteristics data

Ethnicity and national identity received one of the highest user needs scores from the 2015 census topic consultation, and the census ethnicity information is used by national and local decision makers; for example, in equality impact assessments when local authorities make changes to service delivery. The traditional census includes questions on ethnicity but it is currently very difficult to estimate ethnicity at a local level between censuses. The feasibility of producing admin data based ethnicity estimates will be important when deciding whether to move to an admin data based census after 2021.

Very few administrative sources capture ethnicity at all, so including ethnicity on an administrative data census is challenging. However, HES is one of the few sources where ethnicity is captured. ONS has worked with NHS Digital data experts to understand the limitations of the HES ethnicity data and there are several; for example coverage and differences between the ethnicity categories used on HES vs on the Census. However, there are methodological approaches that can be used to mitigate these, and ONS is of the view that it is in the public interest this ethnicity information is acquired from HES.

In terms of the framework of statistical uses presented earlier in this section, then the Administrative Data Census project work described (both a and b) falls into multiple categories:

• Improvements to existing Official Statistics - If an Administrative Data Census proves feasible, ONS will be able to produce census-type population and other statistics more often, in more granular detail, produce new analyses not possible using traditional census data, and reduce the cost and burden on the public by avoiding a traditional decennial census

• Development of new Official Statistics - In the short term, ‘activity data’ from HES may contribute to new admin data-based migration statistics

• Quality assurance of Official Statistics - ‘Activity data’ will help ONS quality assure presence and address information from other sources

• Development of commentary around Official Statistics - Identification of interaction by migrants with secondary care will allow ONS to expand on and increase the frequency of commentary on population changes and impacts, meeting user demand and providing better evidence to better inform policy-makers; for example, impact of migrants on public service demand

• Research and development of methodology - Estimating ethnicity at a population level by local area using an Administrative Data Census approach will be challenging. Using HES ethnicity data, methodological teams will gain experience of developing methods to mitigate the statistical weaknesses often found in administrative data. For example, how to adjust for bias in coverage, and also data being collected on a different statistical definition compared to the desired definition

2.2. To conduct a range of Statistical Research and Health Analyses using clinical data from HES

ONS’s health analysts will use information about why people have accessed hospital services, for example diagnosis, for a range of statistical purposes.

This information is clearly more sensitive, and the intended statistical uses will require testing to determine whether official statistics of sufficient quality can be produced using HES data. As such, the volume of this information is being minimised to that absolutely necessary to do this. In practice, this means fewer years’ worth of information about why people have accessed hospital services will be shared with ONS, compared with the information about when and where people have accessed services.

a. Exploring the feasibility of producing robust projections of the future health state of the nation.

The State pension age review, 2017, called for more work on healthy life expectancy projections to better inform future decisions about the state pension age. The review also noted their potential value in informing planning future health and social care provision at a local and national level.

These projections would need to take into account population projections, morbidity and mortality trends, and other characteristics, and HES could provide some of the information required. ONS recognises that there are serious limitations when using healthcare activity data, particularly hospital episodes, to make inferences about the health of the population. However, using the HES data experimentally will allow ONS to investigate the possibilities of this dataset contributing to more complete estimation of selected serious and acute illnesses, in combination with mortality data and other relevant sources.

It will be necessary to link the HES data with other data sources to prevent double counting of cases and understand the relative completeness, coverage and quality of each data source, and to enable additional demographic variables to be applied to the HES data, therefore record level identifiable data is required.

In terms of the framework of statistical uses, this would be Research and Development of Methodology in the first instance, with the ultimate goal of Developing New National Statistics.

b. Exploring the use of linked morbidity, mortality, census, benefits and other data to produce more granular statistics on health inequalities and health state life expectancies.

ONS healthy life expectancy statistics are central amongst the public health indicators that help guide decisions by Local Authorities (LAs) about the distribution and prioritisation of services. More local level health expectancy statistics, and more breakdowns such as ethnicity, educational attainment and occupation based socioeconomic position to examine interactions would provide insight allowing LAs to better target interventions to reduce health inequalities.

Researching the feasibility of meeting this need will involve linking the HES data to individuals’ self-assessments of their health and disability status as collected by the 2011 Census, the ONS annual population survey since 2011 (for those surveyed), and ultimately the 2021 Census once collected in due course. ONS will explore the relationship between hospital admissions and self-reported health status at both individual and small area levels, and with reference to potentially mediating or confounding demographic and geographic variables. Therefore, identifiable record level data is required, including postcodes.

Research will include exploring the feasibility of using actual morbidity data such as HES to supplement or even replace survey data to produce healthy life expectancy estimates, potentially allowing more granular statistics.

In terms of the framework of statistical uses, this would be this would be Developing New National Statistics and potentially Replacing current survey questions.

c. Exploring the completeness of death certification and patterns of comorbidities in specific population groups

ONS holds data from the compulsory registration of all deaths in England and Wales. The information recorded about causes of death is sometimes unclear or inadequate for the range of public health, monitoring and research purposes to which the data can be put. The majority of deaths occur in hospital, or following an illness for which the deceased had hospital treatment. Linking the diagnosis data in HES with the registered causes of death will allow exploration of the relationships between them, including:

(i) Understanding multi-morbidity and vulnerability in the elderly. It is well-known that deaths of elderly people tend to mention more health conditions, but also to be less specific in a way which makes identifying the factor(s) which contributed most to death difficult. Terms such as ‘old age’ and ‘frailty’ are often used on death certificates with no specific clinical cause of death. By examining the HES diagnoses and registered causes of death together, ONS will aim to throw more light on the combinations of health conditions in elderly people (multimorbidity), the role and frequency of key conditions such as pneumonia and sepsis in the causal pathways leading to death, and if possible to develop new measures of avoidable mortality in the elderly.

This use would require the linkage of HES to deaths at the individual record level. ONS would also link the data to the Census and/or survey data, so as to explore the role of social factors such as living alone in deaths of the elderly along with clinical factors, with the potential to identify at-risk groups and improve targeting of preventive interventions.

(ii) Understanding infant mortality. The causes of death recorded at registration of perinatal deaths in particular are often very broad and not clinically meaningful. ONS is discussing with clinical and scientific experts ways to improve this information and to determine the underlying cause of death. Linkage of the HES data to registered deaths will provide extra information on the factors underlying the recorded causes of death. ONS will aim to improve the accuracy and completeness of infant mortality statistics, potentially contributing to the government ambition to halve infant mortality by 2025.

In terms of the framework of statistical uses, these projects would contribute to Improvements to existing Official Statistics, Quality Assurance of Official Statistics and Developing New National Statistics.

2.3. improving ONS’ Address Register

This project will investigate using HES data to identify and/or validate the addresses of communal establishments, and would require information including where individuals were admitted from and discharged to. Also:

• Length of stay information will provide evidence of how many people ONS would expect to be classed as usually resident (> 6 months stay) in hospital at any given time

• Sex information may assist with identifying communal establishments that are male or female only.

In terms of the framework of statistical uses, this research, if successful, would enable Quality Assurance of Official Statistics and Improved efficiency / accuracy of sampling.

2.4. Creating a better estimate of the UK household expenditure on hospital services (inpatient only) and medical and paramedical services (outpatient)

The ONS national accounts framework provides a simple and understandable description of national production, income, consumption, accumulation and wealth.

The national accounts research team will investigate whether HES data can improve estimates of revenue paid by patients, split into outpatient and inpatient activity, private patient episodes split by outpatient and inpatient activity, and outpatient activity split between medical services and paramedical services.

The data may also be sued to improve the figures on UK healthcare resources, activity and expenditure which are provided regularly to the international institutions (Eurostat, OECD and WHO) for comparative purposes.

In terms of the framework of statistical uses, the ultimate aim would be to Improve an existing National Statistic – i.e. UK national accounts.

2.5. Enabling the UK to report data or proxy indicator data to measure its progress against the United Nation's Sustainable Development Goals (SDGs)

The UK is committed to reporting progress against all of the internationally agreed Sustainable Development Goals (SDGs), and ONS will lead on delivering this. In some cases, new indicators will need to be developed, and/or new uses made of existing data. Interest in HES is specifically around the feasibility of providing data for the following Sustainable Development indicators:

• Maternal mortality ratio

• Proportion of births attended by skilled health personnel

• Number of people requiring interventions against neglected tropical diseases

• Coverage of treatment interventions (pharmacological, psychosocial and rehabilitation and aftercare services) for substance use disorders

• Proportion of women of reproductive age (aged 15-49 years) who have their need for family planning satisfied with modern methods

• Coverage of essential health services (defined as the average coverage of essential services based on tracer interventions that include reproductive, maternal, newborn and child health, infectious diseases, non-communicable diseases and service capacity and access, among the general and the most disadvantaged population)

ONS’s SDGs team are working with NHS Digital and Public Health England (PHE)to produce these indicators without the need for data sharing. However, ONS also needs to disaggregate these headline indicators by ethnicity, age, sex, disability and geography. In some cases, NHS Digital / PHE will not hold data that would enable this, but linking HES data to ONS held data such as from Census 2011 at an individual level may fill this gap.

In terms of the framework of statistical uses, the ultimate aim would be to Develop a new National Statistic.

Dataset 1: Birth Notifications

Official Birth Statistics

Annual birth outputs represent births occurring in England and Wales in a given year. A package containing summary tables for the previous calendar year is released in July, with supporting commentary in a statistical bulletin. More detailed figures are then released between August and December in a series of themed packages. Each package consists of a number of data tables; these are generally accompanied by a statistical bulletin. ONS’ tables provide the latest year’s figures with some also showing historical data for comparison, sometimes back to 1837. ONS publishes all its statistics on its website, and also extends its reach through social media, for example its twitter feed.

There are several published packages:

Birth summary tables: includes the number of live births and stillbirths, fertility rates, percentage of live births outside marriage and civil partnership, mean age of mother and percentage of live births to non-UK born mothers for England and Wales as a whole. Live births (number and rate) and the number of stillbirths are also provided down to local authority level. To aid with user interpretation, ONS also publishes an interactive fertility mapping tool, which enables users to analyse trends in fertility by county district and unitary authority; this is contained within the statistical bulletin.

Parents’ country of birth: includes births by country of birth of mother and total fertility rates for UK born and non-UK born women for England and Wales as a whole. Summary figures are also available down to local authority level. ONS publishes detailed analysis on parents’ country of birth because this information is collected at birth registration and does not change over time, while their nationality or ethnicity may change.

Birth characteristics and by area of usual residence: contains statistics on stillbirths and maternities for England and Wales, birthweight data for live and stillbirths by mother's region of usual residence, and live births and stillbirths in hospitals and communal establishments by region of occurrence. These tables also provide figures on month and quarter of occurrence, place of birth, ethnicity and gestational age and multiple births for England and Wales as a whole. Also provides summary data for live births down to local authority level including figures by age of mother figures are published using boundaries in place during the year the birth occurred.

Births by parents’ characteristics: provides live birth, stillbirth and maternity statistics by age of mother and type of registration (within marriage and civil partnership, joint, sole). It also provides data on previous live-born children, National Statistics Socio-economic Classification (NS-SEC), median birth intervals, age-specific fertility rates for men and mean age of fathers. All tables are for England and Wales as a whole with no sub-national breakdown.

Childbearing for women born in different years (formerly known as Cohort fertility): presents data on fertility by year of birth of mother rather than the year of birth of child for England and Wales as a whole this package includes the average number of live-born children and the proportion of women remaining childless for women born in different years.

Data Linkage Methodology Research: This will result in internal, and potentially external, ONS reports and presentations on how best to link siblings / family units together when linkage based on NHS number is not possible. Any reports or presentations would not include statistics derived from the birth notifications data. They would only include figures comparing the success of various matching strategies compared to one based on linking using mother’s NHS number.

Dataset 2: Hospital Episode Statistics

The initial uses to which ONS will put HES data are most commonly new or improved official statistics that will enable better decision making (see sections 5a and 5d). To reach this goal, a lot of development work, testing, and quality assurance will be required to determine whether official statistics of sufficient quality can be produced in each case.

Generally, this initial work will be disseminated through a range of products and channels, in particular research updates and research outputs. For example, the Admin Data Census project already publishes its research outputs and work involving HES will be reported in similar fashion on this section of the ONS website:

https://www.ons.gov.uk/census/censustransformationprogramme/administrativedatacensusproject/administrativedatacensusresearchoutputs

Subsequently, projects will move on to the production of experimental statistics and potentially in due course, National Statistics (a status that can only be gained once certain quality standards are met). Both types are released via the ONS website.

By way of illustration, a good example of an experimental statistic is here:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/estimatingsuicideamonghighereducationstudentsenglandandwalesexperimentalstatistics/2018-06-25

This release is based on a project linking information about suicides with information on higher education students to increase the evidence base on suicides by those in higher education.

No targets can be given as to if and when experimental or National Statistics will be produced using HES data until the initial stage of any given project is complete. All ONS statistical teams engage regularly with users, and will seek to provide frequent updates on these projects during that first stage.

Yielded Benefits is not a requirement for new applications.