Learning Disabilities Mortality Review Programme - ONS mortality data link

The overall purpose of the application is to link data about people with learning disabilities who have died, with mortality data, to identify the exact cause of death of people with learning disabilities in England.

Analysis of Medical Certificates of Cause of Death (MCCD) is important for monitoring the health of the population, designing and evaluating public health interventions, recognising priorities for medical research and health services, planning health services, and assessing the effectiveness of those services. However, at present, you cannot identify people with learning disabilities from analyses of MCCD or population level vital statistics, so there is little information about causes of death at population level in relation to people with learning disabilities, nor of ways in which services should be prioritised in relation to avoiding premature mortality in this population group.

Since the 1990s there have been a number of reports and case studies that have consistently highlighted that in England people with learning disabilities die younger than people without learning disabilities. Recent calculations by Public Health England using data drawn from the Clinical Practice Research Database indicates that the Standardised Mortality Ratio for people with learning disabilities in England as a whole is approximately three times that of people with no learning disabilities. Some research evidence about causes of deaths also exists (for example from the Confidential Inquiry into premature deaths of people with learning disabilities (2013)), but none drawn from national data.

Such is the concern about premature deaths of people with learning disabilities, and the lack of robust data about their patterns of mortality, NHS England has funded the national Learning Disabilities Mortality Review (LeDeR) programme until the end of May 2020. The University of Bristol submitted a tender for the work and was awarded the contract in May 2015.

The programme is commissioned by the Healthcare Quality Improvement Partnership (HQIP) on behalf of NHS England. The data controller is the Healthcare Quality Improvement Partnership (HQIP). A contract is in place between HQIP and the University of Bristol who process the data for the purpose of this study. No other organisations process the data for this purpose.

Processing of data referred to in this application relates to people who are already known to be deceased therefore GDPR does not apply.

A key part of the LeDeR Programme is to support local areas to review the deaths of people with learning disabilities. The programme is developing and rolling out a review process for the deaths of people with learning disabilities, helping to promote and implement the new review process, and providing support to local areas to take forward the lessons learned in the reviews in order to make improvements to service provision. The LeDeR Programme will also collate and share the pseudonymised information about the deaths of people with learning disabilities so that common themes, learning points and recommendations can be identified and taken forward into policy and practice improvements.

The programme team successfully delivered the Confidential Enquiry into premature deaths of people with learning disabilities (CIPOLD) from 2010-2013, funded by the Department of Health. During this time, the University of Bristol obtained mortality data including coding for all causes of death. It was found from undertaking CIPOLD and work to-date with the LeDeR programme that although local reviews identify the cause of the death of the individual, the ICD-10 coding (International Statistical Classification of Disease and Related Health Problems) for these causes of death as reported in mortality statistics is required to be sure of accurately reporting nationally and internationally comparable causes of death in this population. For example, some local reviews of deaths will simply report that a person died of pneumonia, but the ICD-10 code would be required for the pneumonia to be able to identify if this was an infectious or aspiration pneumonia.

The full coding for cause of deaths (rather than just the underlying cause of death) is required for processing because of two pressing reasons:

1. Possible inconsistencies in reporting causes of deaths for people with learning disabilities. As an example, during work on CIPOLD, it was found that some deaths of people with Down's Syndrome who had Alzheimers dementia to be coded as 1a. Dementia 1b. Down Syndrome (no Part 2 recorded). However, in other cases a similar set of circumstances leading to death was coded as 1a. Dementia Part 2 Down Syndrome. If the coding only was provided for underlying causes of death important evidence would be missed in the understanding of the extent of dementia in people with Down Syndrome. It is essential that data is linked to personal identifiers for this, so that the cause of death reported by reviewers can be assessed against the causes of death coded in ICD-10 descriptors.

2. A significant proportion of people with learning disabilities die with conditions that are not their underlying cause of death, for example pneumonia. Pathways to the immediate cause of death from the underlying cause of death are needed to assess whether there are any service improvements that can be made for this population that would affect this pathway. It is essential that data is linked to personal identifiers for this, so that the circumstances leading to death at individual as well as group levels can be assessed.

The number of people with learning disabilities who die each year is small, approximately 3,000 each year in England. When these deaths are sub-divided into ICD-10 chapters and sub-sections, the numbers become smaller still. Data is required over a number of years to have robust, large scale data to be able to draw accurate conclusions. The feasibility of selecting a reduced number of geographical areas to represent the nation rather than requesting national data was considered but was ruled out this approach because it would effect the validity of the findings. At present it is suspected that there may be pockets of particular causes of death (e.g. deaths related to epilepsy) in particular geographical regions according to the characteristics of those living there and the support and services available to them. It would be important to understand regional variation to guide local service improvement initiatives. Without using national data, it is not possible to guarantee that the reduced data select are accurately representative of England and without this certainty, the analysis would be weakened and the findings less credible. Given the potential impact of the findings on NHS policy and the provision of services to people with learning disabilities, it is essential to minimise uncertainty.

The applicant will not link the data in this application other than where specified within the scope of this application.

This is important, because if recommending service improvements (such as early diagnosis, or improvements in national cancer screening programmes) in relation to common causes of death of people with learning disabilities, there is a need to have a robust evidence base upon which to base recommendations.

The only way that the causes of death of people with learning disabilities can be accurately obtained, is to link the cause of death data with a list of known decedents with learning disabilities - as the LeDeR programme provides. There are no other alternative, less intrusive ways of achieving this purpose.

All efforts have been taken to minimise the data flowed to NHS Digital.

All organisations party to this agreement must comply with the Data Sharing Framework Contract requirements, including those regarding the use (and purposes of that use) by “Personnel” (as defined within the Data Sharing Framework Contract ie: employees, agents and contractors of the Data Recipient who may have access to that data).

Data is processed in accordance with the S251 application which covers all parts of the data flow process. No data (supplied by NHS Digital) is processed outside of the LeDeR programme (based at the University of Bristol), and all data is processed using the dedicated secure web-based platform, thus avoiding the need for hard copies of patient-identifiable information.

Data flows for the reviews of deaths (to place the agreement in context)

a) All relevant deaths are notified to the LeDeR programme (various senior level health and social care practitioners) via a central reporting point (a confidential 030 telephone line, or a dedicated online platform based at the University of Bristol).

b) The LeDeR team inform the Local Area Contact of the area in which the death occurred about the death and its need for review. The Local Area Contact allocates the case to a local reviewer via the secure web based platform.

c) Once the review of the death is completed, the final report is approved by the Local Area Contact and send to the LeDeR team via the secure web based platform for the programme.

d) Following a quality assurance process (to ensure that all reviews are of the required standard), the report is pseudonymised and returned to the Local Area Contact to be discussed at the next regional LeDeR Steering group meeting. The Steering Group is tasked with understanding patterns of mortality of people with learning disabilities in that geographic area, and taking forward any recommendations relating to individual deaths, or patterns arising from multiple deaths, into service improvement initiatives.

e) The LeDeR team uses the pseudonymised data for quantitative and qualitative analysis. Pseudononymised information about deaths of people with learning disabilities will be shared with local Steering groups, in the LeDeR programme annual report, and in presentations to academic, practitioner audiences, and those including people with learning disabilities and their families. Prior to use, the information will be checked to ensure that no potentially identifying information (e.g. unusual sequence of events leading to death) is being used.

Data flows for obtaining Civil Registration Mortality data (the subject of this agreement)

a) Each 6 months, (January and June) the LeDeR team will batch together the personal identifiers of the deaths notified to the LeDeR programme. Personal identifiers may include (depending on availability): NHS numbers (which is available for approximately 80% of deaths); date of birth; date of death; sex; name; postcode.

b) The LeDeR team will securely transfer a file of identifiers plus the Unique Study ID to NHS Digital.

c) NHS Digital returns linked pseudonymised mortality data including the Unique Study ID. Mortality data returned includes ICD-10 codes for causes of death listed and identified in Part 1a, 1b, 1c and 1d, and Part 2 of the MCCD.

d) The LeDeR team stores the data on a secure server in the University of Bristol. The data is stored in a separate location to the participant identifiers. The two datasets will not be re-linked.

e) Data will only be accessed by substantive employees of the University of Bristol who are members of the LeDeR team, and who have authorisation from the Programme Lead to access the data for the purpose(s) described.

f) The data will not be made available to any third parties except in the form of aggregated outputs with small numbers suppressed. Aggregate, anonymised data about ICD-10 coding for cause of death will be included in reports about the mortality of people with learning disabilities, including in the LeDeR programme annual report to the Secretary of State for Health.

The Data will only be used for the purposes described in this agreement.

Annual report for Secretary of State (December, annually)

The Secretary of State has requested that the LeDeR programme reports directly to him on an annual basis. The annual report will contain only aggregate level data with small numbers suppressed. This report is high-level and would only be available to the funders (NHS England) and government.

Public-facing annual report (to be submitted to NHSE in March 2020)

A public-facing annual report is produced each year, that is freely available via the website. The annual and any interim reports contain only aggregate level data with no possibility of identifying any individuals from the data. Copies of the public-facing reports are sent to the Study's contact list of interested practitioners, professionals, family networks, advocacy groups and providers of services. In addition they are send to all of those involved in undertaking reviews of deaths of people with learning disabilities and their relevant area-based Steering Groups.

Academic and practitioner journal articles

Population-level data about causes of deaths of people with learning disabilities will be of interest to both academics and practitioners, and the data will be published in both academic (Journal of Applied Research in Intellectual Disabilities; Health and Care in the Community; Journal of Epidemiology and Community Health) and practitioner-focused (BMJ Open, Nursing Times; Community Care) journals. It is unlikely that there will be sufficient data to publish before May 2020. Again only aggregate level data, with small numbers suppressed will be published, likely to be by ICD-10 chapter and main sub-sections of each chapter relating to causes of death at different ages, and by sex.

Accessible briefing papers

The LeDeR programme puts people with learning disabilities and their families at the centre of their work, and findings would be shared with networks of people with learning disabilities and family carers, advocacy and voluntary sector organisations. This is likely to be in summer/autumn 2019 onwards once sufficiently robust data is available. All accessible briefing papers will be freely available, in paper form or downloadable on the website. They are likely to present 'headline' data only, accompanied by fully anonymised case stories to illustrate the data in a more meaningful way. All case studies would have the consent of families prior to use.

Website/social media

The study website (www.bristol.ac.uk/sps/leder) is also used to aid dissemination. Additionally LeDeR has a Twitter and facebook presence to share information.} The website will provide a link to open access papers and include free downloads of accessible briefing papers, reports and summaries of findings.

Merely reviewing deaths and obtaining ICD-10 coding for causes of death is insufficient in itself. It is what happens to the learning and recommendations gained as a result of the reviews that is so important in being able to improve service provision for people with learning disabilities.

The benefits to health and social care will arise from

a) individual level reviews which will offer recommendations for service improvements as a result of that review of an individual's death, which may benefit others and address any potentially avoidable contributory causes to deaths in the future

b) collated information which may identify themes from across a number of reviews that local areas may wish to target (e.g. addressing rates of falls leading to deaths, or targeting deaths with ICD-10 codes for aspiration pneumonia).

The potential impact of service improvement initiatives for the 1.5 million people with learning disabilities is great. When CIPOLD was conducted a paper was published about the impact of the study (Heslop and Marriott 2015 British Journal of Learning Disabilities). Impacts were identified at individual, local, regional and national levels. At national level this work resulted in policy changes to the extension of Annual Health Checks for people with learning disabilities to those aged 14-17, an extension of GP held registers of adults with learning disabilities to all age registers, and guidance about treating people with learning disabilities as a priority group for seasonal flu vaccinations. At individual and local levels service providers focused on issues such as the early identification of illness in people with learning disabilities, and 'flagging' people with learning disabilities in healthcare systems.

In June and July 2017 a series of workshops were held across England to discuss with members of Steering Groups how they will translate the available information from mortality reviews into service improvements, how they would monitor and evaluate such initiatives, and the reporting mechanisms back to the LeDeR programme to demonstrate the effectiveness of mortality reviews in improving service provision. Benefits are already being felt in health and social care provision from the additional scrutiny that mortality reviews brings.

NHS England has now established a 'Learning into Action' group at national level to the address key findings from LeDeR. Separate workstreams are taking actions in relation to sepsis; adherence to the Mental Capacity Act; recognising deterioration in a person with learning disabilities, and constipation. A national conference was held in March 2019 to share the early work.

At local level, significant work is being undertaken to address problems with care locally. Many of these have been reported in the 2018 annual report so that they can be shared across the health and care sector. In addition, a 2-monthly bulletin is produced for health and social care providers that shares examples of positive practice, and offers a briefing paper to frontline carers about a specific topic.These are also available on the study website.