Understanding excess child and adolescent mortality in the UK

On 1 February 2023, NHS Digital merged with NHS England. NHS England has assumed responsibility for all activities previously undertaken by NHS Digital. The merger was completed by a statute change. Any reference made to NHS Digital within this Data Sharing Agreement is in reference to the merged organisation known as NHS England.

University College London (UCL) are the sole Data Controller who also process data for this project. There are no other organisations involved in the project.

This project was part of a successful application for a Medical Research Council (MRC) Clinical Research Training Fellowship. The project has now contributed to a successful PhD application since the first Data Access Request Service (DARS) application was submitted. The funder is not involved in any aspect of the analysis.

The objective of this project is to explore why the rate at which children and young people (CYP) die in the United Kingdom is higher than in many other developed countries.

In the 1970s United Kingdom (UK) child mortality rates (the number of child deaths per 100,00 population) were similar to those in comparable wealthy nations, and in many areas the UK performed well. Although UK child mortality has been falling since then, the rate of decline has been slower than in other countries, and the UK now has one of the highest child mortality rates in Europe. If the UK had a mortality rate similar to Sweden (one of the best performing countries), about 2000 fewer children would die each year, or 5 fewer a day. There is evidence to suggest health service factors and patterns of healthcare utilisation in the UK may contribute to these poor CYP mortality outcomes.

Aims of the study:

1) Identify causes of death in CYP 0-24 where the UK performs poorly compared to similar countries using publicly available data provided by the WHO World Mortality Database.

2) Analyse geographic and socioeconomic variability in mortality outcomes by cause for CYP 0-24 within England and Wales using death certification data provided by the Office for National Statistics (Office for National Statistics. (2017). Death Registrations in England and Wales, 1993-2017: Secure Access. [data collection] 2nd Edition. Accessed via UK Data Service).

3) Analyse the contribution of health service factors to excess mortality for CYP 0-24 in England for causes of death identified in aims 1 and 2 (i.e., causes of death where the UK performs poorly internationally and where there is wide geographic and socioeconomic variability in outcomes). This will require analysing data on health service use prior to death provided by Hospital Episode Statistics linked with Civil Registration (Deaths) data, requested from NHS Digital.

Analyses within aim 3 will be performed by age-group / sex as appropriate and will include:

a. A detailed analysis of national and regional trends in healthcare use in England over the study period (2007–latest data available) amongst children and young people. This will include analysing row level data on numbers and rate of hospital admissions by cause and type (emergency/planned/short stay), outpatient appointments by specialty, and accident and emergency attendances. Numbers and rates of health service use will be analysed by index of multiple deprivation, and regional variation described.

b. Variability in the contribution of health service factors to predominant mortality causes for CYP in England by NHS provider Trust in children and young people 0-24.

c. Variability in the contribution of health service factors to predominant mortality causes for CYP in England by demographic factors (e.g., socioeconomic status).

d. An analysis of how the contribution of health service factors to predominant mortality causes for CYP in England have changed over time 2007-latest data available.

For aim 3 processing, the data subjects are all children and young people aged 0-24 who have accessed secondary health services in England between 2007 and latest date available.

Initial analyses within aim 3 used data from 2007 – 2017, and contributed to the PhD thesis awarded in Dec 2021. Following this, under a previous iteration of this Agreement UCL requested additional years of the same data to latest completed year at the time (i.e., 2007 - to March 2022) to allow these analyses to be prepared for publication. The additional years of data will allow:

1) Analyses of trends during the COVID-19 pandemic, where hospital activity is known to have dramatically reduced for CYP for many causes. Understanding the impact on all causes of admission, and those related to conditions where the UK has higher mortality in CYP will mean this work is policy relevant in the post-Covid world.

2) This initial work using accident and emergency data was restricted to 2012 onwards, the year in which this data was nationally representative. The additional data request will allow a decade of trends in accident and emergency attendances to be described, through the combination of HES Accident and Emergency Data and the Emergency Care Dataset. This will allow greater understanding of how repeated attendances to A+E within CYP with chronic medical problems affects future mortality hazard.

3) The additional years of data requested would also allow for a specific focus on mental health admissions prior and during the COVID-19 pandemic, around which there is growing concern. Analyses undertaken within this project has identified concerning trends in mortality due to injury and self-harm prior to the pandemic, and that involvement with mental health services increases hazard for mortality for chronic conditions where there is excess mortality in the UK compared with other high-income countries. The additional years of data requested would build on this analysis and allow a detailed understanding of trends in admissions due to mental health concerns from CYP. It will also allow the analysis of how healthcare activity for mental health concerns is associated with mortality hazard for CYP who are admitted with Epilepsy, Diabetes and Asthma, conditions where the UK has high mortality outcomes compared with high income countries.

As described previously, UCL will identify cohorts of CYP who are admitted with predominant conditions where UK mortality is higher than in other wealthy countries (e.g., Asthma, Epilepsy and Diabetes), linked with mortality data. UCL will then describe the risk of death after the first (index) admission over follow up time (until age 24 or most recent data available within HES). UCL will then use cox regression analyses to stratify risk of death over follow up time by background characteristics (IMD, age, sex, English region), and time variant predictors of survival (health service use during follow up– number of A+E visits, number of additional admissions, number of missed out patient appointments, contact with mental health services). By doing this, UCL will be able to quantify the impact of different health service behaviours on mortality risk after controlling for variation by background characteristics.

The datasets requested are Hospital Episode Statistics (HES) Accident and Emergency, HES Outpatients, HES Admitted Patient Care, HES Critical Care, Civil Registration (Deaths) Secondary Care, Emergency Care Dataset (ECDS). The data are pseudonymised (no identifiable variables will be required).

The justification for the number of years now held will achieve the identified aims by first describing the overall impact of healthcare utilisation in CYP in England (inpatient, outpatient, accident and emergency and critical care) and how this has changed over time. Within this, trends in healthcare use for conditions identified in aims 1 and 2 as having higher mortality than other countries will be analysed, and their contribution to the overall impacts on hospital use described.

UCL have requested data on Accident and Emergency attendances within both HES Accident and Emergency and the Emergency Care Dataset, which replaced HES A+E in 2019/20. To ensure there is no loss of data between periods where the ECDS was introduced and the HES A+E dataset was discontinued, UCL have requested both datasets for the year 2019/20. UCL will use both datasets for this year, in order to cross reference Accident and Emergency attendances during 2019/20 which appear in both datasets, and intend include any data for CYP which appear in either dataset for this year in planned analyses.

Cohorts of individuals will then be identified of CYP who have accessed secondary health services with the predominant conditions where UK mortality is higher than in other wealthy countries within each age-group and by sex. This will allow a comparison of patterns of healthcare usage amongst CYP who died of these conditions with CYP who did not die but presented to health services with the same diagnosis. The predominant conditions will be determined from the analysis outcomes of aim 1 and 2 but are likely to include Epilepsy, Asthma and Diabetes.

The years of data that were requested are 2007 to latest date available. These dates were defined due to:

a) Constraints of data availability

In order to examine healthcare utilisation across all types of secondary care use, the study will require data for years where all datasets are available (HES Accident and Emergency, HES Outpatients, HES Admitted Patient Care 2007/2008; HES Critical Care 2008/2009; Emergency Care Dataset (2019/20-2022) to latest date available for completed financial year ( 2021/22). Note the accident and emergency dataset will only be available to 2019/20, after which this dataset was replaced by the ECDS.

b) Examine healthcare use prior to death

Multiple years of data are required to examine overall healthcare use in England (planned outpatient appointments / missed appointments / emergency admissions), and amongst CYP who died compared with those who did not die but attended secondary services with a diagnosis where the UK performs poorly. These patterns of healthcare use will be used as markers of severity, standard of care received, and predictors of mortality risk in the years prior to death.

c) The need to combine deaths over several years for some causes

It will be necessary to combine deaths/admission episodes over 3-5 year periods due to anticipated low numbers in some age / sex groups / regions of the country for some causes. UCL will do this to ensure outputs are in line with guidance on reporting small numbers within the HES analytic guide.

d) To examine trends over time

A key aim of this project is for a longitudinal analysis to examine trends in the overall changing impact on hospital use of healthcare utilization in England, and of any association between healthcare activity and CYP mortality over time, and how this has changed during the COVID-19 pandemic.

The geographic spread of the data (England) allows for an analysis of how patterns of overall healthcare utilisation, and specifically within conditions identified as having excess mortality, vary by NHS provider trust/geographic region in England.

The evidence for excess UK mortality extends throughout the early life course, with high total mortality amongst infants and 1-4 year olds, and high non-communicable diseases (NCD) mortality for all CYP age-groups, particularly adolescents and young people (10-24). In order to fully explore the contribution of health service factors to excess CYP mortality causes, and how this varies by age, the project will require data on secondary healthcare use and mortality within England for children and young people 0-24.

It was considered at length whether data could be filtered to specific conditions of relevance. This project requires data for all secondary care attendance in CYP 0-24, linked to mortality outcomes, for all causes over the study period (2007-2022). Causes of death will be mapped to the Global Burden of Disease mortality hierarchy across 4 levels. For example, acute lymphoblastic leukaemia (level 4) is classified within leukaemia (level 3), neoplasms (level 2) and noncommunicable diseases (NCD) (level 1). Due to the low number of deaths in CYP in each year/sex/age group, it will not always be possible to analyse mortality by level 4 cause, and causes may need to be aggregated by level 3, level 2, or even level 1 group. The level at which a cause of death can be analysed will only be determined after the number of deaths/attendances to secondary care within the dataset (by sex/age group/year) are known. This need to be flexible to allow grouping of causes over different levels depending on numbers of deaths and admissions will mean it will not be possible to perform the analysis if data are only requested on mortality and healthcare use for specific causes. Thus, it is not possible to limit the request to only specific conditions.

The study only used the minimum amount of personal data required to perform the analyses. The data requested was not identifying, and was pseudonymised. Data will then be aggregated by 5-year age group, cause of death group and 3-5 period (by year of death /admission) as required, to meet HES analytic guidelines for suppressing small numbers. UCL will apply the same methodology to the additional years of data requested here.

The research proposal and dissemination plan were presented to members of the National Children's Bureau Young Research Advisors (YRAs) group in March 2018, as part of a Patient and Public Involvement and Engagement initiative. The YRAs are a diverse group of CYP recruited from across the country who have received training in research methods and policy. A focus group of 25 young people aged 7-22 (and parents) was held to discuss the acceptability of the research methods (including the use of data without consent). The YRAs were supportive of the importance of the research and the necessity of analysing data without consent. Specific feedback regarding strategies to inform young people and their families of the research were incorporated in to the project proposal and transparency statement.

Data will only be accessed by individuals within UCL who have authorisation to access the data for the purpose described, all of whom are employees of UCL.

The wider study is the PhD project, which includes analysis where the study will use HES data, and the other analyses described in the application. The PhD project was successfully submitted in 2021. The PhD project has three aims: 1) Identify causes where the UK performs poorly compared with other wealthy nations 2) Analyse variation in cause specific mortality by region of the UK/England and socioeconomic status and then 3) Describe current impact and hospital use and trends in healthcare utilization amongst CYP in England, and compare health service use for predominant causes of child and young person mortality amongst children who die to those who did not die. The 3rd strand will be the strand of the PhD which will use the Data disseminated under this agreement.

In addition, the aggregated HES data used to describe the overall impact and trends of healthcare utilization in England amongst CYP has now contributed to Paediatrics 2040, a project led by the Royal College of Paediatrics and Child Health (RCPCH).

Paediatrics 2040 aimed to establish a credible vision for the future of paediatric services in the UK. A key strand of this work is to understand how patterns and trends in health service utilization may be contributing to the UK’s poor international performance for multiple health outcomes, (including CYP mortality).

The aggregated HES data used within the PhD project contributed to Paediatrics 2040 through the analysis of recent trends in health service activity. In addition to describing recent changes in activity, these descriptive aggregated data were used to predict the future healthcare use/impact in this age group, contribute to developing new models of paediatric care in the UK, and predict future demand for paediatric health services and personnel. This will inform efforts to reconfigure paediatric services in the UK, with the aim to improve healthcare outcomes including CYP mortality.

The lawful basis for processing

The lawful basis under which UCL are processing personal data is:

Article 6(1)(e) processing is necessary for the performance of a task carried out in the public interest or in the exercise of official authority vested the controller; As a public authority, most of UCL’s processing will be undertaken by Article 6(1)(e) using the ‘public task’ condition as the legal basis. This applies when the processing is necessary for UCL to perform a task in the public interest or in the exercise of official authority vested the controller.

The lawful basis under which UCL are processing a special category of data (i.e. health data) is:

Article 9(2)(j) processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) based on Union or Member State law which shall be proportionate to the aim pursued, respect the essence of the right to data protection and provide for suitable and specific measures to safeguard the fundamental rights and the interests of the data subject.

Whenever UCL uses personal data, it needs to establish a legal basis for doing so in line with the General Data Protection Regulation (GDPR). One of these legal bases is the 'public task' ground. This will apply where the processing of personal data is necessary to perform a task in the public interest, which is laid down by law, or in the exercise of official authority laid down by law. Current guidance issued by the UK Information Commissioner's Office (ICO) indicates that the 'public task' ground for processing is likely to apply to much of the processing of personal data carried out by universities, depending on the detail of their constitutions and legal powers.

This project is necessary for scientific research purposes:

The research could not be undertaken without processing this data. In order to establish how patterns of healthcare use differ throughout England among CYP prior to death it is necessary to use hospital episode statistics data linked with mortality outcomes as described in the study protocol.

This project is in the public interest:

As described above, there is a clear public interest in investigating why mortality rates amongst CYP in the UK are higher than in many other wealthy countries, so as to inform policies to improve outcomes. This includes investigating social and demographic factors which may influence higher mortality, understanding the causes of death where UK outcomes are poor, and exploring variability in health care use and how this impacts mortality. Further, exploring trends in healthcare use will allow health services to better meet the health needs of CYP in the UK, around which there is also a clear public interest.

Ethical or moral issues

The study was reviewed by the London-Brent Research Ethics Committee (NHS REC) on 23rd Jun 2018, and received a favourable opinion on 3rd August 2018. UCL will update Brent NHS REC regarding the ongoing research aims of this project, specifically around describing aggregate trends in total and cause specific admissions, including those due to mental health. UCL will submit an amendment to our ethics approval if this is required for this part of the analysis.

Potential harm to the public

Potential harm to the public resulting from this data request includes the identification of individuals within the requested datasets. However, the data processing activities detailed in this application will minimize this risk, the data requested includes no identifiable fields, and all data outputs will be aggregated in line with the HES analytic guide.

Completed outputs in this project have highlighted concerns regarding mortality outcomes for self-harm amongst children and young people in the UK. In view of this, and broader concerns regarding the increasing burden of mental health problems amongst young people, UCL will seek to specifically describe admission trends for mental problems amongst CYP in England as part of the analyses using aggregate data of health care utilisation described above. UCL will describe mental health admissions using the same methods as for overall healthcare utilisation by cause, using aggregate data, and illustrating trends over time by sex, age, index of multiple deprivation, ethnicity, region and (as data allow) NHS Trust. This work will contribute one work package of a separate mix-methods NIHR funded study exploring the impacts of mental health admissions on acute paediatric wards (The Mental Health Admissions to Paediatrics Wards Study NIHR135036).

The primary output will be the analysis of the impact and trends in overall secondary healthcare use in England, and patterns of activity amongst CYP prior to death for causes where UK mortality performance is poor. This will provide estimates of contributions of a range of health system and provider factors to excess CYP UK mortality.

The first stage of analysis within this project using HES data was completed in spring 2021, and contributed to the PhD thesis, which was awarded in Dec 2021.

The primary output will be the analysis of the impact and trends in overall secondary healthcare use in England, and patterns of activity amongst CYP prior to death for causes where UK mortality performance is poor. This will provide estimates of contributions of a range of health system and provider factors to excess CYP UK mortality.

The first stage of analysis within this project using HES data was completed in spring 2021, and contributed to the PhD thesis of one of the UCL researchers named on this study, which was awarded in Dec 2021.

Work completed within this project also contributed to the successful application for an NIHR funded Clinical Lecturer Post, which one of the researchers within this study has started in March 2022 at UCL. A key aim for work within this Clinical Lecturer post is to complete analysis described above for publication, using the additional years of data requested recently (this application has been approved and is being processed).

As described above, aggregated data used in this analysis contributed to the RCPCH Paediatrics 2040 project, and specifically the “data and evidence” and “future models of care” work streams. This report was published in Feb 2021 and can be found here: https://paediatrics2040.rcpch.ac.uk/ This focused on describing trends in admissions to hospital by cause, age group and sex, over time. Further analyses using this data set are ongoing. Data on admissions used within the Paediatrics 2040 report was also presented at a plenary session of the 2022 Royal College of Paediatrics and Child Health.

UCL aim to complete the updated analysis of work which contributed to the PhD thesis in Summer/Autumn 2023, and UCL aim to publish these results in Spring 2024. In this work, UCL will analyse health care utilisation and mortally hazard for conditions where there is concern regarding UK mortality outcomes compared to other high-income countries (including epilepsy, asthma, diabetes).

Building on this work, UCL will use the additional years of data (to latest date available) on to describe national trends and healthcare utilisation for mental health conditions amongst CYP, using aggregate data. The data application for these data has been approved in Autumn 2022 and is currently being processed. These analyses will use the same methodology to describe healthcare utilisation using aggregate data within the PhD thesis and Paediatrics 2040, but specifically explore mental health utilisation. UCL have successfully applied for an ethics amendment in order to focus on mental health care utilisation. This work will contribute to the separate NIHR funded study Mental Health Admissions to Paediatrics Wards Study (NIHR 135036). UCL expect preliminary results from this work in Spring 2023 and will aim for completion in Winter 2023 and publication in Spring 2024.

UCL will seek to present further findings from this project at national and international conferences such as the Royal College of Paediatrics and Child Health (RCPCH) and International Paediatric Association (IPA), and through public and media initiatives organised through UCL and Kings College London. Other professional bodies such as the Royal College of Nursing, Royal College of General Practitioners and the British Association for Child and Adolescent Public Health will provide further opportunities for knowledge exchange and communication to a range of interested parties. Charities focusing on CYP will also be potential partners for dissemination and will include the NSPCC and the Child Accident Prevention Trust, who actively campaign to reduce UK child mortality. All publications, conference presentations, media engagements and other dissemination activities are promoted on twitter, via institutional (UCL) accounts and the Principal Investigator’s (>3000 followers).

The aims, methods and ethical considerations of this project were presented to members of the National Children’s Bureau Young Research Advisors group in March 2018. As part of this process, the Young Research Advisors expressed interest in presenting the main research findings in an accessible way for young people, which will be facilitated by the National Children’s Bureau. This may include a written summary of the report, short videos, animations, or engaging with social media platforms. UCL will seek to collaborate with the National Children’s Bureau to help disseminate in this way, when the findings of the analysis using the updated data are finalised.

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide.

List of Specific Outputs Expected:

1) PhD Thesis and associated publications

Understanding Excess Child and Adolescent Mortality in the UK compared to the EU15+ countries, J Ward 2021

Ward JL, Azzopardi PS, Francis KL, et al. Global, regional, and national mortality among young people aged 10–24 years, 1950–2019: a systematic analysis for the Global Burden of Disease Study 2019. The Lancet 2021;398(10311):1593-618

Ward JL, Wolfe I, Viner RM Cause-specific child and adolescent mortality in the UK and EU15+ countries Arch Dis Child 2020

2) Paediatrics 2040 and associated analyses

Further details are available here https://paediatrics2040.rcpch.ac.uk/.

Yielded benefits of this project include a successful PHD submission in 2021 for one of the researchers named on this application. These analyses established causes of death where the UK performs poorly, and detailed geographic and socioeconomic variation in mortality. Within aim 1 of this project, analyses identified UK CYP mortality outcomes for common infections, neurological conditions (i.e., Epilepsy), chronic respiratory conditions (Asthma) and diabetes to be higher than comparable high-income countries. This analysis was published in Archives of Disease in Childhood in 2020, and was presented at a plenary session at the Royal College of Paediatrics and Child Health conference in 2019.

As part of aim 2, detailed analyses of variation in mortality over time, by age group, local authority and area level deprivation, both for causes where there is excess mortality for other leading contributors to total deaths in CYP have been completed.

Analyses as part of aim 3 have also been completed using Hospital Episode Statistics data outlined in this proposal, which also contributed to the PhD thesis. This has involved multiple steps to clean the data and prepare it for analysis. Analyses have included describing trends in emergency admissions for all causes and conditions where there is wide variation in outcomes within the UK, or excess mortality compared to other similar countries (identified in aims 1 and 2) over a ten-year period. These have included variation by English region and socioeconomic deprivation. These data also contributed to the Paediatrics 2040 report, which can be found here https://paediatrics2040.rcpch.ac.uk/

Also as part of aim 3 of the project, UCL have identified cohorts of children who have been admitted with conditions between 2007 – 2018 for causes of death where the UK has high mortality, and preliminary predictors of survival have been described. These include background characteristics of children and young people, but also time-varying predictors such as timing of transition from paediatric to adult services, subsequent emergency admissions (i.e., after the index admission), additional attendances to A+E, and contact with mental health services. This analysis also contributed to the PhD thesis

Analyses undertaken within this project, including those using Hospital Episode Statistics described here and already held by UCL, contributed to the successful application for an NIHR funded Clinical Lecturer post at UCL for one of the researchers named on this application. A key aims for work to be completed within this post include the analyses described here which require the updated years of data requested.

University College London (UCL) are the sole Data Controller who also process data for this project. There are no other organisations involved in the project.

This project was part of a successful application for a Medical Research Council (MRC) Clinical Research Training Fellowship. The project has now contributed to a successful PhD application since the first DARS application was submitted. The funder is not involved in any aspect of the analysis.

The objective of this project is to explore why the rate at which children and young people (CYP) die in the United Kingdom is higher than in many other developed countries.

In the 1970s UK child mortality rates (the number of child deaths per 100,00 population) were similar to those in comparable wealthy nations, and in many areas the UK performed well. Although UK child mortality has been falling since then, the rate of decline has been slower than in other countries, and the UK now has one of the highest child mortality rates in Europe. If the UK had a mortality rate similar to Sweden (one of the best performing countries), about 2000 fewer children would die each year, or 5 fewer a day. There is evidence to suggest health service factors and patterns of healthcare utilization in the UK may contribute to these poor CYP mortality outcomes.

Aims of the study:

1) Identify causes of death in CYP 0-24 where the UK performs poorly compared to similar countries using publicly available data provided by the WHO World Mortality Database.

2) Analyse geographic and socioeconomic variability in mortality outcomes by cause for CYP 0-24 within England and Wales using death certification data provided by the Office for National Statistics (Office for National Statistics. (2017). Death Registrations in England and Wales, 1993-2017: Secure Access. [data collection] 2nd Edition. Accessed via UK Data Service).

3) Analyse the contribution of health service factors to excess mortality for CYP 0-24 in England for causes of death identified in aims 1 and 2 (i.e., causes of death where the UK performs poorly internationally and where there is wide geographic and socioeconomic variability in outcomes). This will require analysing data on health service use prior to death provided by Hospital Episode Statistics linked with Civil Registration (Deaths) data, requested from NHS Digital.

Analyses within aim 3 will be performed by age-group / sex as appropriate and will include:

a. A detailed analysis of national and regional trends in healthcare use in England over the study period (2007–latest data available) amongst children and young people. This will include analysing row level data on numbers and rate of hospital admissions by cause and type (emergency/planned/short stay), outpatient appointments by specialty, and accident and emergency attendances. Numbers and rates of health service use will be analysed by index of multiple deprivation, and regional variation described.

b. Variability in the contribution of health service factors to predominant mortality causes for CYP in England by NHS provider Trust in children and young people 0-24.

c. Variability in the contribution of health service factors to predominant mortality causes for CYP in England by demographic factors (e.g., socioeconomic status).

d. An analysis of how the contribution of health service factors to predominant mortality causes for CYP in England have changed over time 2007-latest data available.

For aim 3 processing, the data subjects are all children and young people aged 0-24 who have accessed secondary health services in England between 2007 and latest date available.

Initial analyses within aim 3 used data from 2007 – 2017, and contributed to the PhD thesis awarded in Dec 2021. This amended application seeks to request additional years of the same data to latest completed year (i.e., 2007 - to March 2022) to allow these analyses to be prepared for publication. The additional years of data will allow:

1) Analyses of trends during the COVID-19 pandemic, where hospital activity is known to have dramatically reduced for CYP for many causes. Understanding the impact on all causes of admission, and those related to conditions where the UK has higher mortality in CYP will mean this work is policy relevant in the post-Covid world.

2) This initial work using accident and emergency data was restricted to 2012 onwards, the year in which this data was nationally representative. The additional data request will allow a decade of trends in accident and emergency attendances to be described, through the combination of HES Accident and Emergency Data and the Emergency Care Dataset. This will allow greater understanding of how repeated attendances to A+E within CYP with chronic medical problems affects future mortality hazard.

3) The additional years of data requested in this application will also allow for a specific focus on mental health admissions prior and during the COVID-19 pandemic, around which there is growing concern. Analyses undertaken within this project has identified concerning trends in mortality due to injury and self-harm prior to the pandemic, and that involvement with mental health services increases hazard for mortality for chronic conditions where there is excess mortality in the UK compared with other high-income countries. The additional years of data requested in this application will build on this analysis and allow a detailed understanding of trends in admissions due to mental health concerns from CYP. It will also allow the analysis of how healthcare activity for mental health concerns is associated with mortality hazard for CYP who are admitted with Epilepsy, Diabetes and Asthma, conditions where the UK has high mortality outcomes compared with high income countries.

As described previously, UCL will identify cohorts of CYP who are admitted with predominant conditions where UK mortality is higher than in other wealthy countries (e.g., Asthma, Epilepsy and Diabetes), linked with mortality data. UCL will then describe the risk of death after the first (index) admission over follow up time (until age 24 or most recent data available within HES). UCL will then use cox regression analyses to stratify risk of death over follow up time by background characteristics (IMD, age, sex, English region), and time variant predictors of survival (health service use during follow up– number of A+E visits, number of additional admissions, number of missed out patient appointments, contact with mental health services). By doing this, UCL will be able to quantify the impact of different health service behaviours on mortality risk after controlling for variation by background characteristics.

The datasets requested are Hospital Episode Statistics (HES) Accident and Emergency, HES Outpatients, HES Admitted Patient Care, HES Critical Care, Civil Registration (Deaths) Secondary Care, Emergency Care Dataset (ECDS). The data are pseudonymised (no identifiable variables will be required).

The additional years of data requested here, and the data that was previously requested (and subsequently now held) will achieve the identified aims by first describing the overall impact of healthcare utilization in CYP in England (inpatient, outpatient, accident and emergency and critical care) and how this has changed over time. Within this, trends in healthcare use for conditions identified in aims 1 and 2 as having higher mortality than other countries will be analysed, and their contribution to the overall impacts on hospital use described.

UCL have requested data on Accident and Emergency attendances within both HES Accident and Emergency and the Emergency Care Dataset, which replaced HES A+E in 2019/20. To ensure there is no loss of data between periods where the ECDS was introduced and the HES A+E dataset was discontinued, UCL have requested both datasets for the year 2019/20. UCL will use both datasets for this year, in order to cross reference Accident and Emergency attendances during 2019/20 which appear in both datasets, and intend include any data for CYP which appear in either dataset for this year in planned analyses.

Cohorts of individuals will then be identified of CYP who have accessed secondary health services with the predominant conditions where UK mortality is higher than in other wealthy countries within each age-group and by sex. This will allow a comparison of patterns of healthcare usage amongst CYP who died of these conditions with CYP who did not die but presented to health services with the same diagnosis. The predominant conditions will be determined from the analysis outcomes of aim 1 and 2 but are likely to include Epilepsy, Asthma and Diabetes.

The years of data that were requested are 2007 to latest date available. These dates were defined due to:

a) Constraints of data availability

In order to examine healthcare utilisation across all types of secondary care use, the study will require data for years where all datasets are available (HES Accident and Emergency, HES Outpatients, HES Admitted Patient Care 2007/2008; HES Critical Care 2008/2009; Emergency Care Dataset (2019/20-2022) to latest date available for completed financial year (i.e., 2021/22). Note the accident and emergency dataset will only be available to 2019/20, after which this dataset was replaced by the ECDS.

b) Examine healthcare use prior to death

Multiple years of data are required to examine overall healthcare use in England (planned outpatient appointments / missed appointments / emergency admissions), and amongst CYP who died compared with those who did not die but attended secondary services with a diagnosis where the UK performs poorly. These patterns of healthcare use will be used as markers of severity, standard of care received, and predictors of mortality risk in the years prior to death.

c) The need to combine deaths over several years for some causes

It will be necessary to combine deaths/admission episodes over 3-5 year periods due to anticipated low numbers in some age / sex groups / regions of the country for some causes. UCL will do this to ensure outputs are in line with guidance on reporting small numbers within the HES analytic guide.

d) To examine trends over time

A key aim of this project is for a longitudinal analysis to examine trends in the overall changing impact on hospital use of healthcare utilization in England, and of any association between healthcare activity and CYP mortality over time, and how this has changed during the COVID-19 pandemic.

The geographic spread of the data (England) allows for an analysis of how patterns of overall healthcare utilisation, and specifically within conditions identified as having excess mortality, vary by NHS provider trust/geographic region in England.

The evidence for excess UK mortality extends throughout the early life course, with high total mortality amongst infants and 1-4 year olds, and high non-communicable diseases (NCD) mortality for all CYP age-groups, particularly adolescents and young people (10-24). In order to fully explore the contribution of health service factors to excess CYP mortality causes, and how this varies by age, the project will require data on secondary healthcare use and mortality within England for children and young people 0-24.

It was considered at length whether data could be filtered to specific conditions of relevance. This project requires data for all secondary care attendance in CYP 0-24, linked to mortality outcomes, for all causes over the study period (2007-2022). Causes of death will be mapped to the Global Burden of Disease mortality hierarchy across 4 levels. For example, acute lymphoblastic leukaemia (level 4) is classified within leukaemia (level 3), neoplasms (level 2) and noncommunicable diseases (NCD) (level 1). Due to the low number of deaths in CYP in each year/sex/age group, it will not always be possible to analyse mortality by level 4 cause, and causes may need to be aggregated by level 3, level 2, or even level 1 group. The level at which a cause of death can be analysed will only be determined after the number of deaths/attendances to secondary care within the dataset (by sex/age group/year) are known. This need to be flexible to allow grouping of causes over different levels depending on numbers of deaths and admissions will mean it will not be possible to perform the analysis if data are only requested on mortality and healthcare use for specific causes. Thus, it is not possible to limit the request to only specific conditions.

The study only used the minimum amount of personal data required to perform the analyses. The data requested was not identifying, and was pseudonymised. Data will then be aggregated by 5-year age group, cause of death group and 3-5 period (by year of death /admission) as required, to meet HES analytic guidelines for suppressing small numbers. UCL will apply the same methodology to the additional years of data requested here.

The research proposal and dissemination plan were presented to members of the National Children's Bureau Young Research Advisors (YRAs) group in March 2018, as part of a Patient and Public Involvement and Engagement initiative. The YRAs are a diverse group of CYP recruited from across the country who have received training in research methods and policy. A focus group of 25 young people aged 7-22 (and parents) was held to discuss the acceptability of the research methods (including the use of data without consent). The YRAs were supportive of the importance of the research and the necessity of analysing data without consent. Specific feedback regarding strategies to inform young people and their families of the research were incorporated in to the project proposal and transparency statement.

Data will only be accessed by individuals within UCL who have authorisation to access the data for the purpose described, all of whom are employees of UCL.

The wider study is the PhD project, which includes analysis where the study will use HES data, and the other analyses described in the application. The PhD project was successfully submitted in 2021. The PhD project has three aims: 1) Identify causes where the UK performs poorly compared with other wealthy nations 2) Analyse variation in cause specific mortality by region of the UK/England and socioeconomic status and then 3) Describe current impact and hospital use and trends in healthcare utilization amongst CYP in England, and compare health service use for predominant causes of child and young person mortality amongst children who die to those who did not die. The 3rd strand will be the strand of the PhD which will use the Data disseminated under this agreement.

In addition, the aggregated HES data used to describe the overall impact and trends of healthcare utilization in England amongst CYP has now contributed to Paediatrics 2040, a project led by the Royal College of Paediatrics and Child Health (RCPCH).

Paediatrics 2040 aimed to establish a credible vision for the future of paediatric services in the UK. A key strand of this work is to understand how patterns and trends in health service utilization may be contributing to the UK’s poor international performance for multiple health outcomes, (including CYP mortality).

The aggregated HES data used within the PhD project contributed to Paediatrics 2040 through the analysis of recent trends in health service activity. In addition to describing recent changes in activity, these descriptive aggregated data were used to predict the future healthcare use/impact in this age group, contribute to developing new models of paediatric care in the UK, and predict future demand for paediatric health services and personnel. This will inform efforts to reconfigure paediatric services in the UK, with the aim to improve healthcare outcomes including CYP mortality.

The lawful basis for processing

The lawful basis under which UCL are processing personal data is:

Article 6(1)(e) processing is necessary for the performance of a task carried out in the public interest or in the exercise of official authority vested the controller; As a public authority, most of UCL’s processing will be undertaken by Article 6(1)(e) using the ‘public task’ condition as the legal basis. This applies when the processing is necessary for UCL to perform a task in the public interest or in the exercise of official authority vested the controller.

The lawful basis under which UCL are processing a special category of data (i.e. health data) is:

Article 9(2)(j) processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) based on Union or Member State law which shall be proportionate to the aim pursued, respect the essence of the right to data protection and provide for suitable and specific measures to safeguard the fundamental rights and the interests of the data subject.

Whenever UCL uses personal data, it needs to establish a legal basis for doing so in line with the General Data Protection Regulation (GDPR). One of these legal bases is the 'public task' ground. This will apply where the processing of personal data is necessary to perform a task in the public interest, which is laid down by law, or in the exercise of official authority laid down by law. Current guidance issued by the UK Information Commissioner's Office (ICO) indicates that the 'public task' ground for processing is likely to apply to much of the processing of personal data carried out by universities, depending on the detail of their constitutions and legal powers.

This project is necessary for scientific research purposes:

The research could not be undertaken without processing this data. In order to establish how patterns of healthcare use differ throughout England among CYP prior to death it is necessary to use hospital episode statistics data linked with mortality outcomes as described in the study protocol.

This project is in the public interest:

As described above, there is a clear public interest in investigating why mortality rates amongst CYP in the UK are higher than in many other wealthy countries, so as to inform policies to improve outcomes. This includes investigating social and demographic factors which may influence higher mortality, understanding the causes of death where UK outcomes are poor, and exploring variability in health care use and how this impacts mortality. Further, exploring trends in healthcare use will allow health services to better meet the health needs of CYP in the UK, around which there is also a clear public interest.

Ethical or moral issues

The study was reviewed by the London-Brent Research Ethics Committee (NHS REC) on 23rd Jun 2018, and received a favourable opinion on 3rd August 2018. UCL will update Brent NHS REC regarding the ongoing research aims of this project, specifically around describing aggregate trends in total and cause specific admissions, including those due to mental health. UCL will submit an amendment to our ethics approval if this is required for this part of the analysis.

Potential harm to the public

Potential harm to the public resulting from this data request includes the identification of individuals within the requested datasets. However, the data processing activities detailed in this application will minimize this risk, the data requested includes no identifiable fields, and all data outputs will be aggregated in line with the HES analytic guide.

Data refresh for 2022 and addition of Emergency Care Data Set

This amendment includes an application to update the data already held by UCL to include all years up to and including 2021/22, in order to prepare analyses already undertaken for publication. As the HES accident and emergency dataset has been replaced by the Emergency Care Data Set from 2020, the application also includes the addition of these data to analyses accident and emergency contacts in these years, using the same methods as University College London (UCL) are the sole Data Controller who also process data for this project. There are no other organisations involved in the project.

Completed outputs in this project have highlighted concerns regarding mortality outcomes for self-harm amongst children and young people in the UK. In view of this, and broader concerns regarding the increasing burden of mental health problems amongst young people, UCL will seek to specifically describe admission trends for mental problems amongst CYP in England following this data refresh, as part of the analyses using aggregate data of health care utilisation described above. UCL will describe mental health admissions using the same methods as for overall healthcare utilisation by cause, using aggregate data, and illustrating trends over time by sex, age, index of multiple deprivation, ethnicity, region and (as data allow) NHS Trust. This work will contribute one work package of a separate mix-methods NIHR funded study exploring the impacts of mental health admissions on acute paediatric wards (The Mental Health Admissions to Paediatrics Wards Study NIHR135036).

The primary output will be the analysis of the impact and trends in overall secondary healthcare use in England, and patterns of activity amongst CYP prior to death for causes where UK mortality performance is poor. This will provide estimates of contributions of a range of health system and provider factors to excess CYP UK mortality.

The first stage of analysis within this project using HES data was completed in spring 2021, and contributed to the PhD thesis, which was awarded in Dec 2021.

The primary output will be the analysis of the impact and trends in overall secondary healthcare use in England, and patterns of activity amongst CYP prior to death for causes where UK mortality performance is poor. This will provide estimates of contributions of a range of health system and provider factors to excess CYP UK mortality.

The first stage of analysis within this project using HES data was completed in spring 2021, and contributed to the PhD thesis of one of the researchers named on this application, which was awarded in Dec 2021.

Work completed within this project also contributed to the successful application for an NIHR funded Clinical Lecturer Post, which one of the researchers named on this application has started in March 2022 at UCL. A key aim for work within this Clinical Lecturer post is to complete analysis described above for publication, using the additional years of data requested here.

UCL aim to complete the updated analysis of work which contributed to the PhD thesis in Spring 2023, and UCL aim to publish these results in Summer/Autumn 2023. In this work, UCL will analyse health care utilisation and mortally hazard for conditions where there is concern regarding UK mortality outcomes compared to other high-income countries (including epilepsy, asthma, diabetes).

As described above, aggregated data used in this analysis contributed to the RCPCH Paediatrics 2040 project, and specifically the “data and evidence” and “future models of care” work streams. This report was published in Feb 2021 and can be found here: https://paediatrics2040.rcpch.ac.uk/ This focused on describing trends in admissions to hospital by cause, age group and sex, over time. Further analyses using this data set are ongoing. Data on admissions used within the Paediatrics 2040 report will also be presented at a plenary session of the 2022 Royal College of Paediatrics and Child Health and are being prepared for peer review publication.

Building on this work, UCL will use the additional years of data on to describe national trends and healthcare utilisation for mental health conditions amongst CYP, using aggregate data. These analyses will use the same mythology to describe healthcare utilisation using aggregate data within the PhD thesis and Paediatrics 2040, but specifically explore mental health utilisation. This work will contribute to the separate NIHR funded study Mental Health Admissions to Paediatrics Wars Study (NIHR 135036). UCL expect preliminary results from this work in Winter 2022 and will aim for publication in Summer 2023.

UCL will seek to present further findings from this project at national and international conferences such as the Royal College of Paediatrics and Child Health (RCPCH) and International Paediatric Association (IPA), and through public and media initiatives organised through UCL and Kings College London. Other professional bodies such as the Royal College of Nursing, Royal College of General Practitioners and the British Association for Child and Adolescent Public Health will provide further opportunities for knowledge exchange and communication to a range of interested parties. Charities focusing on CYP will also be potential partners for dissemination and will include the NSPCC and the Child Accident Prevention Trust, who actively campaign to reduce UK child mortality. All publications, conference presentations, media engagements and other dissemination activities are promoted on twitter, via institutional (UCL) accounts and the Principal Investigator’s (>3000 followers).

The aims, methods and ethical considerations of this project were presented to members of the National Children’s Bureau Young Research Advisors group in March 2018. As part of this process, the Young Research Advisors expressed interest in presenting the main research findings in an accessible way for young people, which will be facilitated by the National Children’s Bureau. This may include a written summary of the report, short videos, animations, or engaging with social media platforms. UCL will seek to collaborate with the National Children’s Bureau to help disseminate in this way, when the findings of the analysis using the updated data are finalised.

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide.

Yielded benefits of this project include a successful PHD submission in 2021 for one of the researchers named on this application. These analyses established causes of death where the UK performs poorly, and detailed geographic and socioeconomic variation in mortality. Within aim 1 of this project, analyses identified UK CYP mortality outcomes for common infections, neurological conditions (i.e., Epilepsy), chronic respiratory conditions (Asthma) and diabetes to be higher than comparable high-income countries. This analysis was published in Archives of Disease in Childhood in 2020, and was presented at a plenary session at the Royal College of Paediatrics and Child Health conference in 2019.

As part of aim 2, detailed analyses of variation in mortality over time, by age group, local authority and area level deprivation, both for causes where there is excess mortality for other leading contributors to total deaths in CYP have been completed.

Analyses as part of aim 3 have also been completed using Hospital Episode Statistics data outlined in this proposal, which also contributed to the PhD thesis. This has involved multiple steps to clean the data and prepare it for analysis. Analyses have included describing trends in emergency admissions for all causes and conditions where there is wide variation in outcomes within the UK, or excess mortality compared to other similar countries (identified in aims 1 and 2) over a ten-year period. These have included variation by English region and socioeconomic deprivation. These data also contributed to the Paediatrics 2040 report, which can be found here https://paediatrics2040.rcpch.ac.uk/

Also as part of aim 3 of the project, UCL have identified cohorts of children who have been admitted with conditions between 2007 – 2018 for causes of death where the UK has high mortality, and preliminary predictors of survival have been described. These include background characteristics of children and young people, but also time-varying predictors such as timing of transition from paediatric to adult services, subsequent emergency admissions (i.e., after the index admission), additional attendances to A+E, and contact with mental health services. This analysis also contributed to the PhD thesis

Analyses undertaken within this project, including those using Hospital Episode Statistics described here and already held by UCL, contributed to the successful application for an NIHR funded Clinical Lecturer post at UCL for one of the researchers named on this application. A key aims for work to be completed within this post include the analyses described here which require the updated years of data requested.

List of yielded benefits:

1) PhD Thesis and associated publications

Understanding Excess Child and Adolescent Mortality in the UK compared to the EU15+ countries, J Ward 2021

Ward JL, Azzopardi PS, Francis KL, et al. Global, regional, and national mortality among young people aged 10–24 years, 1950–2019: a systematic analysis for the Global Burden of Disease Study 2019. The Lancet 2021;398(10311):1593-618

Ward JL, Wolfe I, Viner RM Cause-specific child and adolescent mortality in the UK and EU15+ countries Arch Dis Child 2020

2) Paediatrics 2040 and associated analyses

Further details are available here https://paediatrics2040.rcpch.ac.uk/.

3) Appointment of NIHR Clinical Lecturer post for named DARS applicant.

University College London (UCL) are the sole Data Controller who also process data for this project. There are no other organisations involved in the project.

This project was part of a successful application for a Medical Research Council (MRC) Clinical Research Training Fellowship. The project will contribute to a PhD. The funder is not involved in any aspect of the analysis.

The objective of this project is to explore why the rate at which children and young people (CYP) die in the United Kingdom is higher than in many other developed countries.

In the 1970s UK child mortality rates (the number of child deaths per 100,00 population) were similar to those in comparable wealthy nations, and in many areas the UK performed well. Although UK child mortality has been falling since then, the rate of decline has been slower than in other countries, and the UK now has one of the highest child mortality rates in Europe. If the UK had a mortality rate similar to Sweden (one of the best performing countries), about 2000 fewer children would die each year, or 5 fewer a day. There is evidence to suggest health service factors and patterns of healthcare utilization in the UK may contribute to these poor CYP mortality outcomes.

Aims of the study:

1) Identify causes of death in CYP 0-24 where the UK performs poorly compared to similar countries using publicly available data provided by the WHO World Mortality Database.

2) Analyse geographic and socioeconomic variability in mortality outcomes by cause for CYP 0-24 within England and Wales using death certification data provided by the Office for National Statistics (Office for National Statistics. (2017). Death Registrations in England and Wales, 1993-2017: Secure Access. [data collection] 2nd Edition. Accessed via UK Data Service).

3) Analyse the contribution of health service factors to excess mortality for CYP 0-24 in England for causes of death identified in aims 1 and 2 (i.e. causes of death where the UK performs poorly internationally and where there is wide geographic and socioeconomic variability in outcomes). This will require analysing data on health service use prior to death provided by Hospital Episode Statistics linked with Civil Registration (Deaths) data, requested from NHS Digital.

Analyses within aim 3 will be performed by age-group / sex as appropriate and will include:

a. A detailed analysis of national and regional trends in healthcare use in England over the study period (2007 – 2017) amongst children and young people. This will include analysing row level data on numbers and rate of hospital admissions by cause and type (emergency/planned/short stay), outpatient appointments by specialty, and accident and emergency attendances. Numbers and rates of health service use will be analysed by index of multiple deprivation, and regional variation described.

b. Variability in the contribution of health service factors to predominant mortality causes for CYP in England by NHS provider Trust in children and young people 0-24.

c. Variability in the contribution of health service factors to predominant mortality causes for CYP in England by demographic factors (e.g. socioeconomic status).

d. An analysis of how the contribution of health service factors to predominant mortality causes for CYP in England have changed over time 2007-2017.

For aim 3 processing, the data subjects are all children and young people aged 0-24 who have accessed secondary health services in England between 2007 and latest date available.

Aged-Matched Controls information

As described previously, UCL will identify cohorts of CYP who are admitted with predominant conditions where UK mortality is higher than in other wealthy countries (e.g Asthma and Epilepsy), linked with mortality data. UCL will then describe the risk of death after the first (index) admission over follow up time (until age 24 or most recent data available within HES). UCL will then use cox regression analyses to stratify risk of death over follow up time by background characteristics (IMD, age, sex, English region), and time variant predictors of survival (health service use during follow up– number of A+E visits, number of additional admissions, number of missed out patient appointments). By doing this, UCL will be able to quantify the impact of different health service behaviours on mortality risk after controlling for variation by background characteristics.

The datasets requested are Hospital Episode Statistics (HES) Accident and Emergency, HES Outpatients, HES Admitted Patient Care, HES Critical Care, Civil Registration (Deaths) Secondary Care. The data are pseudonymised (no identifiable variables will be required).

The data that was requested (and subsequently now held) will achieve the identified aims by first describing the overall impact of healthcare utilization in CYP in England (inpatient, outpatient, accident and emergency and critical care) and how this has changed over time. Within this, trends in healthcare use for conditions identified in aims 1 and 2 as having higher mortality than other countries will be analysed, and their contribution to the overall impacts on hospital use described.

Cohorts of individuals will then be identified of CYP who have accessed secondary health services with the predominant conditions where UK mortality is higher than in other wealthy countries within each age-group and by sex. This will allow a comparison of patterns of healthcare usage amongst CYP who died of these conditions with age matched controls, who did not die but presented to health services with the same diagnosis. The predominant conditions will be determined from the analysis outcomes of aim 1 and 2.

The years of data that were requested are 2007 to latest date available. These dates were defined due to:

a) Constraints of data availability

In order to examine healthcare utilisation across all types of secondary care use, the study will require data for years where all datasets are available (HES Accident and Emergency, HES Outpatients, HES Admitted Patient Care 2007/2008; HES Critical Care 2008/2009

b) Examine healthcare use prior to death

Multiple years of data are required to examine overall healthcare use in England (planned outpatient appointments / missed appointments / emergency admissions), and amongst CYP who died compared with those who did not die but attended secondary services with a diagnosis where the UK performs poorly. These patterns of healthcare use will be used as markers of severity, standard of care received, and predictors of mortality risk in the years prior to death.

c) The need to combine deaths over several years for some causes

It will be necessary to combine deaths/admission episodes over 3-5 year periods due to anticipated low numbers in some age / sex groups / regions of the country for some causes.

d) To examine trends over time

A key aim of this project is for a longitudinal analysis to examine trends in the overall changing impact on hospital use of healthcare utilization in England, and of any association between healthcare activity and CYP mortality over time.

The geographic spread of the data (England) allows for an analysis of how patterns of overall healthcare utilisation, and specifically within conditions identified as having excess mortality, vary by NHS provider trust/geographic region in England.

The evidence for excess UK mortality extends throughout the early life course, with high total mortality amongst infants and 1-4 year olds, and high non-communicable diseases (NCD) mortality for all CYP age-groups, particularly adolescents and young people (10-24). In order to fully explore the contribution of health service factors to excess CYP mortality causes, and how this varies by age, the project will require data on secondary healthcare use and mortality within England for children and young people 0-24.

It was considered at length whether data could be filtered to specific conditions of relevance. This project requires data for all secondary care attendance in CYP 0-24, linked to mortality outcomes, for all causes over the study period (2007-2017). Causes of death will be mapped to the Global Burden of Disease mortality hierarchy across 4 levels. For example, acute lymphoblastic leukaemia (level 4) is classified within leukaemias (level 3), neoplasms (level 2) and noncommunicable diseases (NCD) (level 1). Due to the low number of deaths in CYP in each year/sex/age group, it will not always be possible to analyse mortality by level 4 cause, and causes may need to be aggregated by level 3, level 2, or even level 1 group. The level at which a cause of death can be analysed will only be determined after the number of deaths/attendances to secondary care within the dataset (by sex/age group/year) are known. This need to be flexible to allow grouping of causes over different levels depending on numbers of deaths and admissions will mean it will not be possible to perform the analysis if data are only requested on mortality and healthcare use for specific causes. Thus, it is not possible to limit the request to only specific conditions.

The study only used the minimum amount of personal data required to perform the analyses. The data requested was not be identifying, and was pseudonymised. Data will then be aggregated by 5-year age group, cause of death group and 3-5 period (by year of death /admission) as required, in order to meet HES analytic guidelines for suppressing small numbers.

The research proposal and dissemination plan were presented to members of the National Childrens Bureau Young Research Advisors (YRAs) group in March 2018, as part of a Patient and Public Involvement and Engagement initiative. The YRAs are a diverse group of CYP recruited from across the country who have received training in research methods and policy. A focus group of 25 young people aged 7-22 (and parents) was held to discuss the acceptability of the research methods (including the use of data without consent). The YRAs were supportive of the importance of the research and the necessity of analysing data without consent. Specific feedback regarding strategies to inform young people and their families of the research were incorporated in to the project proposal and transparency statement.

Data will only be accessed by individuals within UCL who have authorisation to access the data for the purpose described, all of whom are substantive employees of UCL and funded by an MRC Clinical Research Training Fellowship which is held by UCL.

The wider study is the PhD project, which includes analysis where the study will use HES data, and the other analyses described in the application. The PhD project has three aims: 1) Identify causes where the UK performs poorly compared with other wealthy nations 2) Analyse variation in cause specific mortality by region of the UK/England and socioeconomic status and then 3) Describe current impact and hospital use and trends in healthcare utilization amongst CYP in England, and compare health service use for predominant causes of child and young person mortality amongst children who die to those who did not die. The 3rd strand will be the strand of the PhD which will use the Data disseminated under this agreement.

In addition, the aggregated HES data used to describe the overall impact and trends of healthcare utilization in England amongst CYP will contribute to Paediatrics 2040, a project led by the Royal College of Paediatrics and Child Health (RCPCH).

Paediatrics 2040 seeks to establish a credible vision for the future of paediatric services in the UK. A key strand of this work is to understand how patterns and trends in health service utilization may be contributing to the UK’s poor international performance for multiple health outcomes, (including CYP mortality).

The aggregated HES data used within the PhD project will contribute to Paediatrics 2040 through the analysis of recent trends in health service activity. In addition to describing recent changes in activity, these descriptive aggregated data will be used to predict the future healthcare use/impact in this age group, contribute to developing new models of paediatric care in the UK, and predict future demand for paediatric health services and personnel. This will inform efforts to reconfigure paediatric services in the UK, with the aim to improve healthcare outcomes including CYP mortality.

The lawful basis for processing

The lawful basis under which UCL are processing personal data is:

Article 6(1)(e) processing is necessary for the performance of a task carried out in the public interest or in the exercise of official authority vested the controller;

The lawful basis under which UCL are processing a special category of data (i.e. health data) is:

Article 9(2)(j) processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) based on Union or Member State law which shall be proportionate to the aim pursued, respect the essence of the right to data protection and provide for suitable and specific measures to safeguard the fundamental rights and the interests of the data subject.

This project is necessary for scientific research purposes:

The research could not be undertaken without processing this data. In order to establish how patterns of healthcare use differ throughout England among CYP prior to death it is necessary to use hospital episode statistics data linked with mortality outcomes as described in the study protocol.

This project is in the public interest:

As described above, there is a clear public interest in investigating why mortality rates amongst CYP in the UK are higher than in many other wealthy countries, so as to inform policies to improve outcomes. This includes investigating social and demographic factors which may influence higher mortality, but also exploring variability in health care use.

Ethical or moral issues

The study was reviewed by the London-Brent Research Ethics Committee (NHS REC) on 23rd Jun 2018, and received a favourable opinion on 3rd August 2018.

Potential harm to the public

Potential harm to the public resulting from this data request includes the identification of individuals within the requested datasets. However, the data processing activities detailed in this application will minimize this risk, the data requested includes no identifiable fields, and all data outputs will be aggregated in line with the HES analytic guide.

The primary output will be the analysis of the impact and trends in overall secondary healthcare use in England, and patterns of activity amongst CYP prior to death for causes where UK mortality performance is poor. This will provide estimates of contributions of a range of health system and provider factors to excess CYP UK mortality.

The first stage of analysis will be completed within 6 months of gaining access to the data (Jun 2019) and the aim is to publish preliminary results by May 2020. The final analysis will be completed by August 2020. The wider project will contribute to a PhD thesis will be submitted in September 2020. As described above, this analysis will also contribute to the RCPCH Paediatrics 2040 work stream, which is due for completion in Spring 2021

Each sub-analysis within aim 3 will form a separate publication exploring the contribution of health service factors to mortality outcomes by NHS provider trust, socio-economic status and changes over time. The primary targets for publication will be peer-reviewed journals including the Lancet, British Medical Journal and Archives of Disease in Childhood. Estimated publication date for these analyses will be Jun 2020 - Sept 2020.

Findings will also be presented at national and international conferences such as the Royal College of Paediatrics and Child Health (RCPCH) and International Paediatric Association (IPA), and through public and media initiatives organised through UCL and Kings College London. Other professional bodies such as the Royal College of Nursing, Royal College of General Practitioners and the British Association for Child and Adolescent Public Health will provide further opportunities for knowledge exchange and communication to a range of interested parties. Charities focusing on CYP will also be potential partners for dissemination and will include the NSPCC and the Child Accident Prevention Trust, who actively campaign to reduce UK child mortality. All publications, conference presentations, media engagements and other dissemination activities are promoted on twitter, via institutional (UCL) accounts and the Principle Investigator’s (>3000 followers).

The aims, methods and ethical considerations of this project were presented to members of the National Children’s Bureau Young Research Advisors group in March 2018. As part of this process, the Young Research Advisors expressed interest in presenting the main research findings in an accessible way for young people, which will be facilitated by the National Children’s Bureau. This may include a written summary of the report, short videos, animations, or engaging with social media platforms.

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide.

Use of aggregated data from this analysis will also contribute to the RCPCH Paediatrics 2040 project. This project is due to be launched in Spring 2021, and the data described within this proposal will contribute to both the “data and evidence” and “future models of care” workstreams, and feature prominently within the final report. These analyses will also inform other elements of Paediatrics 2040 focusing on the UK paediatric workforce and the impact of innovation on future services. In addition to the Paediatrics 2040 report, these data will be presented at a Paediatrics 2040 launch event and at the RCPCH academic conference in 2021 (1500-2000 attendees). The results will be disseminated through both traditional media and social media platforms linked with the RCPCH (>20,000 followers), and partner child health organizations.

Analyses within this project to date have established causes of death where the UK performs poorly, and detailed geographic and socioeconomic variation in mortality. Within aim 1 of this project, analyses have identified UK CYP mortality outcomes for common infections, neurological conditions (ie Epilepsy), chronic respiratory conditions (Asthma) and diabetes to be higher than comparable high-income countries. This analysis is currently under final peer review for publication, and has been presented at a plenary session at the Royal College of Paediatrics and Child Health conference in 2019. RCPCH intends to use findings from the Paediatrics 2040 work stream to continue to influence the future shape of paediatric care, through ongoing engagement with government and national NHS bodies. As part of aim 2, detailed analyses of variation in mortality over time, by age group, local authority and area level deprivation, both for causes where there is excess mortality for other leading contributors to total deaths in CYP have been completed. This analysis is currently being prepared for submission for peer review publication. Aims 1 and 2 will constitute the first two chapters of the PhD thesis.

Preliminary analyses as part of aim 3 have also been completed using Hospital Episode Statistics data outlined in this proposal. This has involved multiple steps in order to clean the data and prepare it for analysis. Analyses have included describing trends in emergency admissions for conditions where there is wide variation in outcomes within the UK, or excess mortality compared to other similar countries (identified in aims 1 and 2) over a ten year period. These have included variation by English region and socioeconomic deprivation. This work is nearing completion and will constitute one chapter of the PhD thesis.

Cohorts of children who have been admitted with these conditions between 2007 – 2018 have also been identified, and preliminary predictors of survival have been described. These include background characteristics of children and young people, but also time-varying predictors such as timing of transition from paediatric to adult services, subsequent emergency admissions (i.e after the index admission), and additional attendances to A+E. This preliminary analysis is being prepared for publication. When complete this analysis will contribute to one chapter of the PhD thesis.

On reviewing the data, it has become apparent to that analyses by provider NHS trust in this study will not be feasible because of the low numbers of deaths. UCL will now analyse geographic variation in activity and mortality at the level of English government office region.

University College London (UCL) are the sole Data Controller who also process data for this project. There are no other organisations involved in the project.

This project was part of a successful application for a Medical Research Council (MRC) Clinical Research Training Fellowship. The project will contribute to a PhD. The funder is not involved in any aspect of the analysis.

The objective of this project is to explore why the rate at which children and young people (CYP) die in the United Kingdom is higher than in many other developed countries.

In the 1970s UK child mortality rates (the number of child deaths per 100,00 population) were similar to those in comparable wealthy nations, and in many areas the UK performed well. Although UK child mortality has been falling since then, the rate of decline has been slower than in other countries, and the UK now has one of the highest child mortality rates in Europe. If the UK had a mortality rate similar to Sweden (the best performing country), about 2000 fewer children would die each year, or 5 fewer a day.

Aims of the study:

1) Identify causes of death in CYP 0-24 where the UK performs poorly compared to similar countries using publicly available data provided by the WHO World Mortality Database.

2) Analyse geographic and socioeconomic variability in mortality outcomes by cause for CYP 0-24 within England and Wales using death certification data provided by the Office for National Statistics (Office for National Statistics. (2017). Death Registrations in England and Wales, 1993 – 2016: Secure Access. [data collection] 2nd Edition. Accessed via UK Data Service).

3) Analyse the contribution of health service factors to mortality for CYP 0-24 in England for causes of death identified in aims 1 and 2 (i.e causes of death where the UK performs poorly internationally and where there is wide geographic and socioeconomic variability in outcomes). This will require analysing data on health service use prior to death provided by Hospital Episode Statistics linked with Civil Registration (Deaths) data, requested from NHS Digital.

Analyses within aim 3 will be performed by age-group / sex as appropriate and will include:

a. Variability in the contribution of health service factors to predominant mortality causes for CYP in England by NHS provider Trust in children and young people 0-24.

b. Variability in the contribution of health service factors to predominant mortality causes for CYP in England by demographic factors (e.g socioeconomic status).

c. An analysis of how the contribution of health service factors to predominant mortality causes for CYP in England have changed over time 2007 – 2017.

For aim 3 processing, the data subjects are all children and young people aged 0-24 who have accessed secondary health services in England between 2007 and latest date available. The data requested is Civil Registration (Deaths) Secondary Care which is to be linked to Hospital Episode Statistics (HES) Accident and Emergency, HES Outpatients, HES Admitted Patient Care, HES Critical Care.

The datasets requested are Hospital Episode Statistics (HES) Accident and Emergency, HES Outpatients, HES Admitted Patient Care, HES Critical Care, Civil Registration (Deaths) Secondary Care. The data are pseudonymised (no identifiable variables will be required).

The data requested will achieve the identified aim by allowing cohorts to be identified of CYP who have accessed secondary health services with the predominant conditions where UK mortality is higher than in other wealthy countries within each age-group and by sex. This will allow a comparison of patterns of healthcare usage amongst CYP who died of these conditions with age matched controls, who did not die but presented to health services with the same diagnosis. The predominant conditions will be determined from the analysis outcomes of aim 1 and 2.

The years of data that will be requested are 2007 to latest date available. These dates were defined due to:

a) Constraints of data availability

In order to examine healthcare utilisation across all types of secondary care use, the study will require data for years where all datasets are available (HES Accident and Emergency, HES Outpatients, HES Admitted Patient Care 2007/2008; HES Critical Care 2008/2009).

b) Examine healthcare use prior to death

Multiple years of data are required to examine healthcare use (planned outpatient appointments / missed appointments / emergency admissions) over three year periods amongst CYP who died compared with those who did not die but attended secondary services with the same diagnosis. These patterns of healthcare use will be used as markers of severity, standard of care received, and predictors of mortality risk in the years prior to death.

c) The need to combine deaths over several years for some causes

It will be necessary to combine deaths/admission episodes over 3-5 year periods due to anticipated low numbers in some age / sex groups / regions of the country for some causes.

d) To examine trends over time

A key aim of this project is for a longitudinal analysis to examine trends in of any association between healthcare utilisation and mortality over time.

The geographic spread of the data (England) will allow for an analysis of how patterns of healthcare use prior to death vary by NHS provider trust/geographic region in England.

The evidence for excess UK mortality extends throughout the early life course, with high total mortality amongst infants and 1-4 year olds, and high non-communicable diseases (NCD) mortality for all CYP age-groups, particularly adolescents and young people (10-24). In order to fully explore the contribution of health service factors to excess CYP mortality causes, and how this varies by age, the project will require data on secondary healthcare use and mortality within England for children and young people 0-24.

It was considered at length whether data could be filtered to specific conditions of relevance. This project will require data for all secondary care attendance in CYP 0-24, linked to mortality outcomes, for all causes over the study period (2007 – 2016). Causes of death will be mapped to the Global Burden of Disease mortality hierarchy across 4 levels. For example, acute lymphoblastic leukaemia (level 4) is classified within leukaemias (level 3), neoplasms (level 2) and non-communicable diseases (NCD) (level 1). Due to the low number of deaths in CYP in each year/sex/age group, it will not always be possible to analyse mortality by level 4 cause, and causes may need to be aggregated by level 3, level 2, or even level 1 group. The level at which a cause of death can be analysed will only be determined after the number of deaths/attendances to secondary care within the dataset (by sex/age group/year) are known. This need to be flexible to allow grouping of causes over different levels depending on numbers of deaths and admissions will mean it will not be possible to perform the analysis if data are only requested on mortality and healthcare use for specific causes. Thus, it is not possible to limit the request to only specific conditions.

The study will only use the minimum amount of personal data required to perform the analyses. The data requested will not be identifying, and will be pseudonymised. Data will then be aggregated by 5-year age group, cause of death group and 3-5 period (by year of death /admission).

The research proposal and dissemination plan were presented to members of the National Children’s Bureau Young Research Advisors (YRAs) group in March 2018, as part of a Patient and Public Involvement and Engagement initiative. The YRAs are a diverse group of CYP recruited from across the country who have received training in research methods and policy. A focus group of 25 young people aged 7-22 (and parents) was held to discuss the acceptability of the research methods (including the use of data without consent). The YRAs were supportive of the importance of the research and the necessity of analysing data without consent. Specific feedback regarding strategies to inform young people and their families of the research were incorporated in to the project proposal and transparency statement.

The individual accessing the data under this agreement is a substantive employee of UCL and are funded by an MRC Clinical Research Training Fellowship which is held by UCL.

The wider study is the PhD project, which includes analysis where the study will use HES data, and the other analyses described in the application. The PhD project has three aims: 1) Identify causes where the UK performs poorly compared with other wealthy nations 2)Analyse variation in cause specific mortality by region of the UK/England and socioeconomic status and then 3)Compare health service use for predominant causes of child and young person mortality amongst children who die to those who did not die. the 3rd strand will be the strand of the PhD which will use the Data disseminated under this agreement.

The primary output will be the analysis of secondary healthcare usage amongst CYP prior to death in England for causes where UK mortality is poor, compared with age matched controls. This will provide estimates of contributions of a range of health system and provider factors to excess CYP UK mortality.

The first stage of analysis will be completed within 6 months of gaining access to the data (Jun 2019) and the aim is to publish preliminary results within 1 year (Dec 2019). The final analysis will be completed within 18 months (Jun 2020).

Each sub-analysis within aim 3 will form a separate publication exploring the contribution of health service factors to mortality outcomes by NHS provider trust, socio-economic status and changes over time. The primary targets for publication will be peer-reviewed journals including the Lancet, British Medical Journal and Archives of Disease in Childhood. Estimated publication date for these analyses will be Dec 2019 – Sept 2020.

The wider project will contribute to a PhD thesis which will be submitted to UCL in September 2020.

Findings will also be presented at national and international conferences such as the Royal College of Paediatrics and Child Health (RCPCH) and International Paediatric Association (IPA), and through public and media initiatives organised through UCL and Kings College London. Other professional bodies such as the Royal College of Nursing, Royal College of General Practitioners and the British Association for Child and Adolescent Public Health will provide further opportunities for knowledge exchange and communication to a range of interested parties. Charities focusing on CYP will also be potential partners for dissemination and will include the NSPCC and the Child Accident Prevention Trust, who actively campaign to reduce UK child mortality. All publications, conference presentations, media engagements and other dissemination activities are promoted on twitter, via institutional (UCL) accounts and the Principle Investigator’s (>1500 followers).

The aims, methods and ethical considerations of this project were presented to members of the National Children’s Bureau Young Research Advisors group in March 2018. As part of this process, the Young Research Advisors expressed interest in presenting the main research findings in an accessible way for young people, which will be facilitated by the National Children’s Bureau. This may include a written summary of the report, short videos, animations, or engaging with social media platforms.

All outputs will contain only data that is aggregated with small numbers supressed in line with the HES Analysis Guide.

Yielded Benefits is not a requirement for new applications.