NHS Digital Data Release Register - reformatted

University Of Hull projects

93 data files in total were disseminated unsafely (information about files used safely is missing for TRE/"system access" projects).


Is travel burden associated with differences in health outcomes for patients diagnosed with breast, lung, prostate, colorectal or oral cancers living in Yorkshire & Humberside and the North East Regions. — DARS-NIC-675446-G2S5Q

Type of data: information not disclosed for TRE projects

Opt outs honoured: Anonymised - ICO Code Compliant, No (Does not include the flow of confidential data)

Legal basis: Health and Social Care Act 2012 – s261(2)(a)

Purposes: No (Academic)

Sensitive: Sensitive

When:DSA runs 2023-07-01 — 2026-06-30 2023.10 — 2023.10.

Access method: One-Off

Data-controller type: UNIVERSITY OF HULL

Sublicensing allowed: No

Datasets:

  1. NDRS Cancer Registrations
  2. NDRS Linked HES APC
  3. NDRS National Cancer Patient Experience Survey (CPES)
  4. NDRS National Radiotherapy Dataset (RTDS)
  5. NDRS Systemic Anti-Cancer Therapy Dataset (SACT)

Objectives:

University of Hull requires access to NHS England data for the purpose of the following research project: "Is travel burden associated with differences in health outcomes for patients diagnosed with breast, lung, prostate, colorectal or oral cancers living in Yorkshire & Humberside and the North East Regions?"

The following is a summary of the aims of the research project provided by University of Hull:
"This study plans to examine the associations between travel times and travel distances and differences in outcomes for cancer patients. It also seeks to examine in direct response to the Chief Medical Officers call in 2021 to investigate potential differences in cancer outcomes for cancer patients living in coastal communities" www.gov.uk/government/publications/chief-medical-officers-annual-report-2021-health-in-coastal-communities.

The following NHS England data will be accessed:
• Cancer Registration – needed to compare the 5 different cancers.
• Hospital Episode Statistics Admitted Patient Care - needed to provide more information for the statistical models that will be focusing on treatment received for cancer (focusing on surgery and time spent in hospital).
• National Cancer Patient Experience Survey (CPES) - needed to provide additional information on the support that patients perceived that they received and how that is associated with the key outcomes. Waves 9 and 10 requested relate to years 2019 and 2020 and are the most recent questionnaires completed by cancer patients following treatment. The study wishes to assess for levels of support in the models and to use the most recent results.
• Radiotherapy dataset (RTDS) – needed to provide more information for the statistical models that will be focusing on treatment received for cancer (focusing on radiotherapy).
• Systemic Anti-Cancer Therapy Dataset (SACT) – needed to provide more information for the statistical models that will be focusing on treatment received for cancer (focusing on chemotherapy).

The level of the data will be pseudonymised.

The data will be minimised as follows:
• Limited to data for adults greater than 17 years old;
• Limited to data for cancers diagnosed between 01 January 2013 to 31 December 2020.
• Limited to the following data for patients residing in the following geographical boundary codes for the cancer alliances: Northern Cancer Alliance (E56000029), Humber Coast and Vale Cancer Alliance (E56000026), West Yorkshire and Harrogate Cancer Alliance (E56000030), South Yorkshire and Bassetlaw Cancer Alliance (E56000025)
• Limited to conditions relevant to the study identified by specific ICD or OPCS codes;
- Breast Cancer (ICD10 C50)
- Lung Cancer (ICD10 C33, C34 and C45)
- Colorectal Cancer (ICD10 C18, C19 and C20)
- Prostate Cancer (ICD10 C61)
- Oral Cancer (ICD10 C00 – C14)
• Only waves 9 and 10 are requested from the CPES dataset

University of Hull is the research sponsor and the controller as the organisation responsible for ensuring that the data will only be processed for the purpose described above.

The lawful basis for processing personal data under the UK GDPR 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 in the controller;

The lawful basis for processing special category data under the UK GDPR 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 processing is in the public interest on the basis of research and statistical purposes. Firstly, this research study is directly addressing a known deficiency in the evidence available on the cancer outcomes of communities living on the coast. This was highlighted by the Chief Medical Officers report on coastal communities published in 2021 (health in coastal communities). The hypothesis is that there are inequalities in cancer treatment and access to healthcare for those who live in coastal communities compared to those in non-coastal communities, which result in inequalities in outcomes. This data will test this hypothesis and provide evidence on cancer cases in coastal communities. Secondly, existing research has identified that travel burden (measured by distance/ travel times to healthcare) is associated with inequalities in access to treatment. The hypothesis is that living further from healthcare facilities (e.g. GP, radiotherapy centres) is associated with poorer healthcare outcome (e.g. survival) and inequalities in treatment received (e.g. not having palliative radiotherapy). University of Hull has worked closely with a patient public involvement (PPI) group to plan the proposed research and ensure that what is being proposed is in the public’s interest and will directly answer key questions that are relevant to patients diagnosed with cancer.

The funding for the data cost and staff time to undertake the analysis comes from the principal investigators (PI) fellowship funding (funded through a Yorkshire Cancer Research Career Development Fellowship). The funding includes the described study, as part of a programme of research. The funder will have no ability to suppress or otherwise limit the publication of findings. The Yorkshire Cancer Career Development Fellowship funding runs until May 2024 at which point the PI will transfer to being a Lecturer at the University of Hull.

AIMES Management Services Ltd are the data processors. They are responsible for the University of Hull Data Safe Haven.

There are no other organisations involved.

Data will be accessed by substantive employees of the university and by a PhD student enrolled at University of Hull. Individuals have completed mandatory data protection and confidentiality training and is subject to University of Hull’s policies on data protection and confidentiality. The PhD student accessing the data will do so under the supervision of a substantive employee of University of Hull. University of Hull would be responsible and liable for any work carried out by the individual. The PhD student would only work on the data for the purposes described in this Agreement.

A Public and Patient Involvement and Engagement group made up of four people diagnosed with cancer helped develop the programme of analysis that will produce findings that are in the public interest. Assurances were sought from the group members to ensure they were happy that this study is in the public interest.

The data requested in this agreement is not considered confidential under the Health and Social Care Act 2012 and therefore is not subject to a duty of confidence, as such the national data opt-out is not applied.

Where individuals have opted out of disease registration by the National Disease Registration Service (NDRS), their data has been permanently removed from the registry and therefore will not be disseminated under this Data Sharing Agreement (DSA). https://digital.nhs.uk/ndrs/patients/opting-out.

Expected Benefits:

This application directly supports the research that explores whether living in a coastal community is associated with differences in health outcomes from lung cancer. This directly meets the call from the Chief Medical Officers report in 2021 that evidence on cancer outcomes are lacking when looking specifically at coastal areas and there is a need for more evidence on coastal outcomes.

The potential impact of living further from a healthcare facility on a patient’s cancer outcomes and treatment choices will be disseminated based on the data is analysed.

Depending on the findings it is expected that this can be used as evidence to provide guidance on extra support to patients who may have to travel further or who have difficulties getting to the healthcare facilities for their treatment and diagnosis.

University of Hull work closely with the funders (Yorkshire Cancer Research) to communicate directly the findings. University of Hull work closely with the Hull Involve PPI group to disseminate the findings directly to the people who have helped shape and guide the research. University of Hull are also working closely with the local Cancer Alliances (e.g. Humber, Coast and Vale Cancer Alliance) to actively promote the research and findings for this region.

Outputs:

The expected outputs of the processing will be a submission to peer reviewed journals and publishing of academic papers.

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide (NHS Digital (2019) HES Analysis Guide).
All outputs from the logistic regression and Cox proportionate Hazard Models will be coefficients and 95% Confidence Intervals.

Outputs will be in the form of peer-review publications and conference presentations accessed by academics, policy makers, commissioners and clinicians, with lay summaries made available for service users and the public. Summaries of findings will be used to inform patient and public involvement (PPI) and clinical staff groups.
In March 2024 University of Hull will be holding a public exhibition of the research, of which this application is part, as part of the TRANSFORM team research at the University of Hull (https://www.hyms.ac.uk/research/transform).

The target date to publish the first paper in a health geography journal is 2023. Publishing academic papers that report on the findings from the association between living further from a healthcare facility and outcomes for cancer patients (stage at diagnosis, survival, and differences in treatments in academic journals) is targeted for between 2023 and 2026. University of Hull are planning a public dissemination exhibition in March 2024, with the target of disseminating the findings of this study. The PhD student will submit the PhD thesis in 2024.

Processing:

No data will flow to NHS England for the purposes of this Agreement.

Data will be transferred directly from NHS England to the University of Hull Data Safe Haven (DSH). Data related to this project will be stored and processed exclusively within the DSH with no touchpoints on wider University of Hull infrastructure.

NHS England data will provide the relevant records from the HESAPC, Cancer Registry, RTDS, CPES and SACT datasets to the University of Hull. The data will contain no direct identifying data items but will contain a unique person ID.

Data is stored within the University of Hull Data Safe Haven (DSH). This is a trusted research environment hosted by AIMES within their data centre in Liverpool. Data storage and analysis will be undertaken exclusively within the data safe haven with no touchpoints outside. Researchers will access the data via virtual machines, again hosted within the data centre

Data for each project is logically separated meaning there is no ability to access or link to datasets from other projects within the DSH. When working within the environment users have no way to move data from within the DSH to any device outside of it. All data enters and leaves through a digital airlock which is linked to the above authorisation workflow.

Data is backed up with the AIMES data centre using processes and systems exclusive to AIMES. There is no third party involved and no transfer of data outside the data centre.

Researchers access the environment via a secure VPN connection from University of Hull managed devices with 2-factor authentication. As such no data will leave the data centre and nothing will sit on the end users’ machine.

The data will not leave England/Wales at any time.

All processing / analysis is restricted to researchers who are employees of the University of Hull who have authorisation from the Principal Investigator and a PhD student enrolled at the university who holds an honorary contract.

AIMES Management Services Ltd is not permitted to access or process the data for purposes beyond the provisioning, securing and back-up of the project environment.
AIMES have system administrative access to the system but act on the instructions of University of Hull DSH admin. AIMES are processor for the purposes of provisioning, securing and back-up of the service.

Research staff complete NHS Data Security training along with DSH specific training prior to being granted access to the environment. Additionally, staff are trained on how to access and work within the DSH. This is all auditable. Staff also sign user terms and conditions prior to receiving access.

The datasets provided will not be linked to any other data outside of the scope of the agreement.

A database file has been created that has the travel times and distances for each LSOA (geographical variable) and OA (geographical variable) in the Yorkshire and Humber and North East Regions to each possible healthcare facility (GPs and hospitals). These have been calculated using the Visography TRACC software (https://basemap.co.uk/tracc), the UK road network, healthcare locations from NHS England and output areas and LSOA areas from the office for national statistics. In addition to the travel times and distances the LSOA (geographical variable) areas have been categorized as either coastal or non-coastal. This file will be added to the data safe haven at the University of Hull. This file will be merged with the data provided by NHS England in the data safe haven. It will be merged using the LSOA/OA geographical variable identifiers contained in both sets of data providing the travel times and travel distances needed and classification of coastal or non-coastal location. Once merged a working dataset will be created with the LSOA/OA geographical variables removed from it.

There will be no requirement and no attempt to reidentify individuals when using the data.

Researchers from the University of Hull will analyse the data for the purposes described above.


Examining the characteristics and predictors of alcohol withdrawal readmissions and emergency department attendances — DARS-NIC-226185-B6C2J

Type of data: information not disclosed for TRE projects

Opt outs honoured: No - data flow is not identifiable, Anonymised - ICO Code Compliant, No (Does not include the flow of confidential data)

Legal basis: Health and Social Care Act 2012 – s261(1) and s261(2)(b)(ii), Health and Social Care Act 2012 – s261(1) and s261(2)(b)(ii), Health and Social Care Act 2012 - s261 - 'Other dissemination of information', Health and Social Care Act 2012 – s261(2)(b)(ii)

Purposes: No (Academic)

Sensitive: Non Sensitive, and Non-Sensitive

When:DSA runs 2019-09-01 — 2020-12-01 2019.11 — 2019.11.

Access method: One-Off

Data-controller type: UNIVERSITY OF HULL

Sublicensing allowed: No

Datasets:

  1. Hospital Episode Statistics Accident and Emergency
  2. Hospital Episode Statistics Admitted Patient Care
  3. Hospital Episode Statistics Accident and Emergency (HES A and E)
  4. Hospital Episode Statistics Admitted Patient Care (HES APC)

Objectives:

With over 1 million alcohol-related hospital admissions the burden and unmet needs of excessive alcohol consumption and related conditions remain a priority under the NHS 10-year plan and for Public Health England (PHE, 2019).

In 20016/17 there were over 1 million alcohol-related admissions of which 300,000 hospital admissions were wholly attributable to alcohol in England, a rise of 29% over the last decade (PHE, 2018). With 65% of these admissions attributable to mental and behavioural disorders due to alcohol, the characteristics among those experiencing alcohol withdrawal and the relationship with readmission rates are not known.

Unplanned alcohol-related hospital admissions have been associated with physical multi-morbidity, coexisting mental health conditions and socioeconomic deprivation (Payne et al, 2013). Recent research in the US (Yedlapati and Stewart, 2018) has identified hospital readmission rates following alcohol withdrawal are linked to discharge against medical advice (AMA) and the complexity of the patients (i.e. co morbid mental health). Furthermore, being discharged against medical advice is associated with subsequent AMA events (Kraut et al, 2013). Whilst individual factors associated with re admissions rates for unplanned alcohol withdrawal can be identified, incomplete episodes of care for alcohol withdrawal may influence a return to excessive drinking on discharge and subsequent readmission.

The period required to complete a programme of alcohol withdrawal varies according to the needs of the individual although commonly require 5-7 days of clinical monitoring and treatment (NICE 2010,2011). HES-APC data contains primary and secondary diagnostic codes that allow for exploratory analysis of readmission rates amongst those with alcohol dependence including specifically alcohol withdrawal. Alcohol dependent patients (ICD-10 code: F10.2) who receive unplanned hospital care may experience alcohol withdrawal (F10.3/F10.4) due to the abrupt cessation or substantial reduction in alcohol consumption. Exploring the length of stay and characteristics for those experiencing unplanned alcohol withdrawal and the association with readmission will help to understand the impact of clinical practice and patient factors on their representation rates to A&E and re admissions. This study aims to examine routine hospital data to examine characteristics and predictors of alcohol withdrawal re admissions and ED attendances in England by linking Hospital Episode Statistics Admitted Patient Care (HES APC) data sets with HES Accident & Emergency (HES ED)

A previous study (Phillips et al., 2019) which was a PhD study considered the burden of alcohol disorders on ED and Inpatient Care and is published in Alcohol and Alcoholism (28 June, 2019). The study examined the characteristics of individuals with acute and chronic alcohol disorders using HES-APC and HES -ED data sets from 2009/10. This data that was used for this PhD is no longer accessible as it has been destroyed.

The study only used 2009/10 data. Since this time there have been changes in:
• quality of data reporting to HES
• increases in the recorded levels of alcohol withdrawal admissions from 21,590 in 2009/10 to 27,530 in 2017/18 despite no significant change in community prevalence of alcohol dependence
• significant changes in the commissioning of alcohol treatment which may have impacted on the care pathways for alcohol treatment
• Data published in the US indicating readmission rates for alcohol withdrawal are predicted by length of stay and discharges against medical advice. These areas have not previously been considered in UK populations.

The PhD study did not consider the entire inpatient dataset and differs in the following ways:

• The data requested in the PhD only included admission data following emergency presentations. This agreement includes admission data for those admitted via emergency presentations, booked and elective admissions. This is to include those cases where an elective admission is planned for and the patient experiences alcohol withdrawal.
• The data used in the PhD only used data where an individual case in the inpatient data could be matched to a coded ED presentation. As significant minority of ED presentations do not receive a coded ED ‘diagnosis’ (i.e. 36%). Hence, the PhD did not interrogate all possible cases if they meet the inclusion criteria set out in the protocol fr this study.
• The previous study did not examine predictor of readmission or re-attendance, the impact of length of stay nor discharges against medical advice. The study referenced in this agreement will examine alcohol withdrawal admissions and consider the characteristics that predict readmission. Principally, the hypothesis will be drawn from the US study that indicates shorter lengths of stay will predict greater likelihood of readmission and ED re-attendance.

To summarise; this study plans to examine the impact of length of stay and discharges against medical advice in predicting readmission and ED re-attendances within 30 days following alcohol withdrawal admissions.
This agreement will therefore build on the original work completed by Phillps et al (2019) and inform future research aimed at reducing the nature and burden of unplanned alcohol treatment within non-specialist care settings.

The lawful basis for undertaking this research under the General Data Protection Regulation (GDPR) articles are:
• Article 6 (1)(e): processing is necessary for the performance of a task carried out in the public interest, improving the care for people with alcohol dependency and related conditions.

• 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

The data requested will achieve the aim identified above by using encrypted HES ID to cases common to both the ED and APC data sets. No identifiable data will be shared to carry out this process. The ICD-10 codes will be searched to identify those who experienced alcohol diagnoses, including alcohol dependence, alcohol withdrawal and alcohol withdrawal with delirium. Previous research has identified an association between community prevalence and alcohol-related admissions (Brennan et al, 2019), furthermore clinical practice within hospital settings is also highly variable and therefore to obtain a robust understand of the impact of the length of stay for alcohol withdrawal data representing national coverage is required.

Socio-Demographic characteristics that influence clinical outcomes and patient engagement within these groups will be explored to characterise both cases (i.e. those who experience alcohol dependence and withdrawal) and controls (i.e. those without an in year history of alcohol dependence and withdrawal). Overall A&E attendances and admissions (i.e. hospital spells) will be identified. The length of hospital stay (LOS) will be identified for each spell where a patient experienced alcohol withdrawal (i.e. F10.3/4) statistical analysis will explore the association between LOS and subsequent re admissions and A&E attendances.

Common concomitant conditions will be examined and individuals with alcohol withdrawal and a common concomitant condition will be compared to controls without a recorded history of alcohol disorder (i.e. F10) but share the common concomitant condition. This analysis will help identify the relative burden of alcohol withdrawal accounting for the presence of co morbid health problems.

The hypothesis is that a shorter length of stay experienced by those with alcohol withdrawal during hospital admissions will be associated with greater ED re attendance and hospital readmission.

The research study requires demographic, and socioeconomic data to characterise cases and controls and to allow for matching cases using propensity score matching. This will allow the examination of the associations between individual characteristics and readmission. Additionally, data related to diagnosis will be essential in discriminating between cases and controls. Furthermore, being able to match cases and controls on age, gender and shared primary diagnosis (i.e. gastritis) will allow for the examination of the relative burden of alcohol on readmission. This is important, as gastritis may independently be associated with greater readmission rates. A series of analyses will be undertaken to examine characteristics between cases and controls (i.e. those without alcohol dependence/withdrawal). Demographic factors (age, gender, ethnicity, marital status, living circumstances, IMD, etc), clinical characteristics (length of stay, category of co-morbidity, A&E clinical diagnosis, emergency v elective admissions, previous discharges AMA, etc) and health service use (number of admissions, A&E re-attendances) will be explored against readmission.

Data Requested
i. Access to national data allows for examination of this hypothesis, which is unable to be achieved efficiently through prospective research. If an association is found between LOS and readmission and A&E attendance this will inform policy and future research applications through Research for Patient Benefit Grant or Programme Development grants applications. Therefore, the study request access to HES A&E data and APC data.

ii. The level of data: Pseudonymised is requested to allow for linkage between data sets at a ‘case’ level for the whole of England.

iii. One year of data would provide an adequate sample frame

iv. There is significant variation in the prevalence of alcohol disorders across the country and variation in admissions and practice – obtaining a countywide sample allows the study to establish national norms.

Recently published data for 2017/18 (PHE, 2019) identified that when both secondary and primary diagnosis were considered, alcohol withdrawal, and alcohol withdrawal with delirium accounted for over 28,000 alcohol-related hospital admissions. Previous research has identified an association between community prevalence of alcohol dependence and alcohol-related admissions (Brennan et al, 2019), furthermore, clinical practice within hospital settings is also highly variable and therefore to the most robust examination of the impact of LOS on hospital readmission and ED re-attendance a national spread of data is required.

v. This method of research is the least intrusive option to explore this question and is supported by the recently published Framework for Mental Health Research (DH, 2017), which supports the use of existing data sources to increase or understanding of mental health problems and how healthcare is provided. The fields requested have been kept to a minimum and do not include patient identifiable fields.

The University of Hull (UoH) is the Data Controller and the Data Processor. The Professor of Nursing (Addictions), University of Hull is the Chief Investigator for the study and is responsible for the overall design, and conduct of the study. The Epidemiologist, University of Hull will lead on the data management, and analysis. The Professor of Health Service Research, University of Kent is providing methodological support and is a collaborator on the study.

Data shared with the Collaborator (University of Kent) will be exclusively aggregate level summary table data with small numbers suppressed. No record level data will leave the UoH Data Safe Haven, within which only named UoH researchers will have access to it. The Collaborator will be given opportunity to make comment on the summary data. Based on those comments further analysis may be considered but the decision as to what analyses to perform will be made exclusively by the research team at UoH. Importantly the comments will not drive further data requests, change the variables requested or the means of storage, processing or data flow. The decision as to what variables are included in the DARs will be made exclusively by the research team at the UoH. Aggregate data will be shared using the Hull Health Trials Unit (HHTU) managed Box cloud storage. University of Kent (The Collaborator) will offer statistical support only and will have access to aggregate tables with small numbers suppressed in line with the HES analysis guide in read-only format only. In summary, the University of Kent has no role in determining the means by which the data are processed or influence over the outputs and dissemination.

Expected Benefits:

The outputs of this study will add to the growing literature and research related to the reduction of alcohol-related hospital burden which highlights the need to tackle unmet needs of patients with alcohol disorders to reduce the overall burden on health service provision. This study will identify the impact of length of stay on alcohol-related readmission rates.

The Lead Investigator retains membership of the Expert Group on Alcohol Treatment, Public Health England and has advised on the development of guidance, Developing pathways for referring patients from secondary care to specialist alcohol treatment care pathways (PHE, 2018). These analyses will further inform the development of improved care pathways for those experiencing unplanned alcohol withdrawal through the publication of peer-reviewed evidence and additional research projects designed to examine the service users experience and outcomes of specialist interventions (i.e. medically assisted alcohol withdrawal) in the non-specialist settings. The findings will be shared with Public Health Leads for alcohol and unplanned care.

The University of Hull works closely with commissioners of services within Yorkshire & The Humber and is also NIHR CRN Speciality Lead for Mental Health. Outcomes of this study will be shared with key stakeholders across the locality and region to inform service development. This will include working with the newly developed Yorkshire & Humber Applied Research Collaborative funded by the NIHR which is focused on reducing demands on Emergency Departments.

Outputs:

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

Outputs will be in the form of peer-review publications and conference presentations accessed by academics, commissioners and clinicians, with lay summaries made available for service users and the public. Summaries of findings will be used to inform patient and public involvement (PPI) and clinical staff focus groups supporting a programme grant application.

Peer-review publication: The University of Hull has identified the Journal Alcohol and Alcoholism as a key journal which is affiliated to the Medical Council on Alcohol, a national body which supports the translation of evidence to practice. All publications from the University of Hull are promoted through websites, blogs, and other social media (Twitter, LinkedIn)

The University of Hull aim to target National conferences (i.e. MCA Annual Conference, Society for the Study of Addiction (SSA) Annual Symposium) and international conferences (i.e. Research Society of Alcohol (RSA), USA). The latter publishes accepted abstracts in Alcohol: Clinical Experimental Research, a recognised peer-review journal.

The Lead Researcher is a member of a number of national expert groups, including the Alcohol Treatment Expert Group, Public Health England, which publishes guidance for providers and clinicians. The outcomes of this study will be reported to this group to support the recently published guidance by Public Health England, which considers the patient pathways between secondary and specialist care for alcohol dependent patients.

The Lead Researcher is also supporting the Humber, Coast and Vale Sustainability and Transformation Partnership (STP) in defining health priorities. The findings from this national study will also inform the work with a variety of NHS Trusts who are developing strategic responses to increasing alcohol-related admissions as part of the implementation of the NHS Long-Term Plan.

Analysis of data to inform publication is estimated to take 6 months with publications submitted in months 8-10 months. Conference presentations will commence end of 2019-2020. Specific dates will be set once data has been released to the applicant.

Additionally, the data will support the development of postgraduate, postdoctoral and research grant applications aimed at addressing the burden and unmet needs of alcohol patients receiving unplanned care.

Processing:

Data extracts will be stored and processed within the Data Safe Haven of the Hull Health Trials Unit (HHTU) at the new Institute for Clinical and Applied Health Research (ICAHR), University of Hull. The HHTU is among a few university departments across the country that has a dedicated Data Safe Haven (DSH) meeting the NHS Toolkit Approvals. Lead investigators for this project are co-located and will conduct data management, cleansing, and processing within the DSH suite in accordance with, and overseen by the Information Systems Manager at the HHTU. All processing activities will be undertaken in Stata 15SE statistical software package. Alcohol-attributable diagnosis will be considered as primary and secondary diagnoses. Hospital Admission will be calculated using Information Centre Method 4 and re admissions calculated initially using the NHS Digital IAS Ref Code: IAP00333. The study will consider the impact of length of stay on those admitted with alcohol dependence (including alcohol withdrawal) and A&E re attendance and readmission.

The project will not involve the flow of data into NHS Digital from the research team.

Once agreed the data will be received from the NHS Digital SFTP System directly into the Hull Health Trials Unit (HHTU), Date Safe Haven (DSH) by the Information Systems Manager. The storage, preservation, processing and statistical analysis will be conducted within the HHTU DSH suite with redacted tables made available for panel discussions and review informing and refining statistical analyses. The final report will only include aggregate data with small number suppression applied in line with the HES analysis guide.

Only those employed within the ICAHR, University of Hull, involved in this study who have been trained in data protection and confidentiality will be processing the data.

Existing data variables will be transformed to create dummy variables for use in the analysis which will include categorical variables relating to diagnosis, length of stay, etc. A&E Data will be explored to identify the frequency of A&E attendance and common presentations with linkage between inpatient spells to observe the frequency of A&E attendances in relation to admissions for alcohol withdrawal. Similarly, the frequency of inpatient spells will be calculated.

Data linkage will only occur between HES APC data and HES A&E data using the unique HES ID provided for the study. The use of the A&E attendance dates and admission dates together with the HES ID to identify A&E attendances leading to admission - parameters of age and gender will be used as secondary matching variables. No other data sets will be linked or matched to the HES data.

There will be no requirement or attempt to re-identify individuals

The HES data and all record level manipulations will be processed exclusively in the HHTU Data Safe Haven. This is a secure environment which is disconnected from the wider university network other than when importing and exporting datasets where a temporary connection is made to a white list of URLs. This service utilises dedicated DSH servers which are managed independently to the wider university network by specific DSH staff. The DSH has its own active directory where users and environments are provisioned specific to each research project. User access to the DSH is via dedicated thin clients machines housed in a secure specific DSH rooms which are managed by the HHTU. As part of gaining access to the environment staff will sign usage terms and conditions which will include a check on a member of staff’s IG training. Data entering and leaving the DSH is controlled by HHTU DSH admins and users have no ability to save or output data themselves. Data leaving the DSH will only be aggregate data and users will sign to confirm that their output meets the terms specified in their data sharing agreement.

For this project aggregate level derived data will be shared with collaborators throughout the analysis period. Based on feedback and comments an iterative analysis and review process will be followed. Aggregate data will be shared using the HHTU managed Box cloud storage instance. This instance is administered by HHTU admin staff, uses only EU data storage and has additional governance modules allowing specific data residency policies to be applied. All aggregate data with small number suppressed will be shared on a read-only basis.

Data shared with the Collaborator (University of Kent) will be exclusively aggregate level summary table data with small numbers suppressed. No record level data will leave the UoH Data Safe Haven, within which only named UoH researchers will have access to it. The Collaborator will be given opportunity to make comment on the summary data. Based on those comments further analysis may be considered but the decision as to what analyses to perform will be made exclusively by the research team at UoH. Importantly the comments will not drive further data requests, change the variables requested or the means of storage, processing or data flow. The decision as to what variables are included in the DARs will be made exclusively by the research team at the UoH. Aggregate data will be shared using the Hull Health Trials Unit (HHTU) managed Box cloud storage. University of Kent (The Collaborator) will offer statistical support only and will have access to aggregate tables with small numbers suppressed in line with the HES analysis guide in read-only format only. In summary, the University of Kent has no role in determining the means by which the data are processed or influence over the outputs and dissemination.

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).

There will be no data linkage undertaken with NHS Digital data provided under this agreement that is not already noted in the agreement.

Data will only be accessed and processed by substantive employees of the University of Hull and will not be accessed or processed by any other third parties not mentioned in this agreement.


An Impact Assessment of National Head Injury Guidelines Using an Interrupted Time Series — DARS-NIC-61042-K9Q3G

Type of data: information not disclosed for TRE projects

Opt outs honoured: N, Anonymised - ICO Code Compliant (Does not include the flow of confidential data)

Legal basis: Health and Social Care Act 2012, Section 42(4) of the Statistics and Registration Service Act (2007) as amended by section 287 of the Health and Social Care Act (2012), Health and Social Care Act 2012 – s261(1) and s261(2)(b)(ii), Health and Social Care Act 2012 – s261(2)(b)(ii)

Purposes: No (Academic)

Sensitive: Non Sensitive, and Sensitive, and Non-Sensitive

When:DSA runs 2019-03-10 — 2020-03-09 2018.03 — 2018.05.

Access method: One-Off, Ongoing

Data-controller type: UNIVERSITY OF HULL

Sublicensing allowed: No

Datasets:

  1. Hospital Episode Statistics Admitted Patient Care
  2. Office for National Statistics Mortality Data
  3. Bridge file: Hospital Episode Statistics to Mortality Data from the Office of National Statistics
  4. Civil Registration (Deaths) - Secondary Care Cut
  5. HES:Civil Registration (Deaths) bridge
  6. Civil Registrations of Death - Secondary Care Cut
  7. Hospital Episode Statistics Admitted Patient Care (HES APC)

Objectives:

Nature of Project:

This is a National Institute for Health Research funded PhD fellowship that will use hospital episode statistic data to assess the impact of the introduction of national head injury (NICE) guidelines on deaths from traumatic brain injury and hospital admissions for head injury. This forms part of a larger project that aims to evaluate and refine the current national clinical (NICE) guidelines used for the emergency management of head injured patients in the United Kingdom.

Background:

Head injury is very common. There are 1.4 million emergency department attendances annually in England and Wales following a head injury. The majority of these patients can be safely reassured and sent home. A small number of patients have life-threatening brain injuries that need to be immediately identified and treated. The difference between the two groups is not always initially clinically apparent.

National clinical (NICE) guidelines for the management of head injured patients were introduced in 2003, 2007 and 2014 in England and Wales. They recommended increased CT brain imaging of head injured patients and that patients with severe brain injuries should be managed in specialist centres. This aimed to ensure all patients with life-threatening injuries were identified and that patients with severe injuries had improved outcomes through specialist care. The increased costs of more imaging were planned to be offset by a reduction in hospital admissions of head-injured patients. It was thought that patients previously admitted for observation due to uncertainty about whether they had significant brain injuries would be discharged from the Emergency Department following normal CT imaging.

There has not previously been a comprehensive evaluation of the impact of the introduction of the NICE head injury guidelines. There is limited evidence that they have reduced deaths from traumatic brain injury. Hospital admissions for head injury have increased by over 50% since the introduction of the head injury guidelines. It has been hypothesised this may be due to increased CT imaging leading to the identification of small brain injuries of debatable clinical significance that would previously not have been identified. It important to assess whether the NICE head injury guidelines are clinically effective and whether their introduction has contributed to more hospital admissions and therefore increased costs for the NHS.

Aim:

This research will robustly assess the impact of the NICE head injury guidelines.

Objectives:

1) Assess the impact of the introduction of each iteration of the NICE guidelines on deaths from traumatic brain injury.

2) Assess the impact of the introduction of each iteration of the NICE guidelines on the number and rate of hospital admissions due to head injury.

3) Assess the size of any increase in hospital admissions for head injury due to the unintended identification of more traumatic brain injuries of lower clinical significance due to increased CT imaging.

Yielded Benefits:

Statistical analysis of the mortality linked HES data extract has provided evidence that the only NICE Head Injury guideline associated with a reduction in the mortality rate from traumatic brain injuries was the guideline iteration that recommended patients with severe injuries are managed in specialist centres. The effect was limited to the 16-65 age group. In patients over 65, the mortality rate from traumatic brain injury increased and was unaffected by the introduction of any guideline. This provides evidence to the NHS that although NICE guidelines recommending increased CT imaging may reduce hospital admissions and costs for the NHS, improved traumatic brain injury outcomes requires diagnosis to be coupled with specialist care. The increasing mortality rate in older patients from traumatic brain injury requires further investigation. It may mean either older patients have unequal access to specialist care or that specialist care is less effective in this group. This has implications for future research and the provision of NHS services treating the increasing number of older patients with traumatic brain injuries.

Expected Benefits:

The benefits noted below have been taken directly from the applicants NIHR Funding Application and is therefore written in the first person.

The nature of the proposed research involves the analysis of retrospective routinely collected data. Therefore the scope for public patient involvement in the execution phase of this research is limited. However, I have agreed to continue engaging with both stakeholder groups consulted in the preparatory phase of this research to explore the importance of the research to them, help with interpretation and in developing this research further.

My proposed research will explore whether the NICE guidelines have reduced head injury mortality and at what cost to the NHS. I will summarise and discuss these findings with the Trans-Humber Consumer Research Panel and a Headway Charity patient group to assess whether they feel that the NICE head injury guidelines represent effective care and good value for the NHS. Using these findings we will discuss how the NICE guidelines could be improved and what further research would be a priority.

In particular, the Trans-Humber research panel chairman and individual members were keen to be involved further in developing the decision rule aspect of the proposed research. I intend to form a long-term relationship with this group. I will return to them with a draft final protocol for this aspect of research for them to consider from a patient perspective. I will also share with them a draft risk-model for patients with abnormal CT head imaging.

We will discuss how such a model could be applied in practice to facilitate shared decision making between patients and clinicians. We will assess what further research may be necessary in order to inform the implementation of such a model into clinical practice. The Hull Headway patient group has also agreed to act as a patient advisory group for the proposed research. They have agreed to meet with me once a year. We will discuss how my research is progressing and whether it
could be refined to better encompass their priorities. At the end of this fellowship I will consult with the Headway patient group in developing a protocol for post-doctoral research aimed at assessing longer-term health care needs for patients with head injury (see future plans).

Benefit to patients and the NHS:

Implementing the NICE head injury guidelines represented a large expenditure of NHS resources. It is important to assess whether these guidelines have been effective, and potentially caused an increase in unnecessary
hospital admissions.

Developing a clinical risk assessment tool for mild traumatic brain injury patients that have minor CT head scan abnormalities will help refine the NICE head injury guidelines to allow better risk stratification of this group. Better risk stratification would inform shared decision-making and could reduce the rate and length of
admissions.


Outputs:


Aggregate level monthly data will be produced for: the non-adjusted and adjusted mortality rate of traumatic brain injury; emergency hospital admissions for head injury compared to emergency admissions for other conditions; proportion of patients with sociologically detected brain injuries that do not undergo neurosurgery. Aggregate data with small numbers suppressed in line with the HES analysis guide will be summarised in charts, tables and figures.

Aggregate level data with small number suppressed in line with the HES analysis guide will be: assessed by other researchers in the research team at meetings in Hull York Medical School.

Aggregate level data with small numbers suppressed in line with the HES analysis guide will be presented at relevant research conferences (this includes national Royal College of Emergency Medicine conferences, International Brain Injury Conferences and NIHR conferences of funded fellowships) ; disseminated in academic publications; and will form a final report for the National Institute of Health Research.

The projected timeline of specific formal outputs is detailed below, and data will be aggregated with small numbers suppressed in line with the HES analysis guidelines.

Study results will be compiled for academic publications and are projected to contribute to at least two peer-reviewed scientific articles published in relevant clinical journals such as the BMJ Emergency Medicine Journal and Journal of Neurotrauma by 2020.

Abstracts of study results will be submitted to 2 Annual Scientific Conferences in Emergency Medicine (September 2019, September 2020) and at the World Congress in Brain Injury 2018. Study results will also be submitted for presentation at NIHR conferences of funded research in 2019 and 2020.

An abstract of results of the study will also be submitted to the NICE annual conference in 2020 in order to help inform the development of a further iteration of national clinical head injury guidelines.

Results of the study will be disseminated directly in the form of a presentation to the local branch of the HEADWAY head injury charity by September 2019 and this group includes former brain injury patients, carers, and healthcare professionals. The results will be communicated to the HEADWAY charity at a national level in the form of a summary report to the charity’s Publication and Research Manager. The charity may choose to further disseminate the results of the research in national newsletters and on the charity’s website.

Results of the study will form a final report to the National Institute of Health Research in September 2019.

Processing:

The University of Hull is the data controller for this project the research contract for the NIHR Doctoral Fellowship for the lead investigator is between the University of Hull and Secretary of State for Health. However, no data will be: provided to; stored at; accessed in; or processed at; the University of Hull.

All data provided by NHS Digital will be stored, accessed and processed solely at the University of York Department of Health Sciences. This data will not be shared with any other parties and will be stored for the agreed duration of this project. The data set supplied will only be used for the purposes of this research project and not be used for any other purpose. All outputs will be at an aggregate level with small numbers suppressed in line with the HES analysis guidelines. Reccord level data will only be accessed by University of York researchers conducting analysis for this project. This includes: the lead investigator a Hull York Medical School NIHR Doctoral Research Fellow , a University of York Professor of Health Sciences and Hull York Medical School’s Senior Statistician.

Request Data:

An pseudonymised data set of record level data for all emergency admissions to hospital from 1998-2017 will be extracted by NHS Digital from the Inpatient Hospital Episode Statistic data set.

Linkage to mortality data is requested only for admission episodes with ICD 10 diagnostic codes related to head injury. These include S00-S09, T04.0 and T06.0.

HES data will be used to achieve the following objectives of the research project:
1) Assess the impact of the introduction of each iteration of the NICE guidelines on the number and rate of hospital admissions due to head injury.
2) Assess the size of any increase in hospital admissions for head injury due to the unintended identification of more traumatic brain injuries of lower clinical significance due to increased CT imaging.

In order to achieve the objectives with the minimum required HES data the Data Access Request Service technical team were involved in determining which data were required and the following measures have been taken:

1) Data fields have been selected to ensure the minimum required clinical data is provided for the planned analysis and duplicated forms of data are not requested.
2) Only the necessary pseudonymised HES specific episode information is requested to allow the planned analysis of changes in hospital admissions.
3) In total less than 50% of possible data fields are requested.
4) The focus of the project is the emergency management of head injury and therefore only data on emergency admissions is requested.
5) Data is requested for the period of 1998-2017 as the impact of 3 iterations of the NICE clinical guidelines is being assessed. These were introduced in 2003, 2007 and 2014 and therefore data from 1998-2017 is required in order to complete the planned analysis.
6) Data on all emergency admissions is requested for the period of 1998-2017. This is required to allow a comparison between trends in emergency admissions for head injured patients and patients with other conditions.

The objective of the part of the project that uses ONS mortality data is to assess whether the introduction of NICE clinical guidelines for head injury reduced deaths due to severe traumatic brain injury. The data requested is minimised in the following ways, whilst allowing this objective to be accomplished:

1) ONS mortality data is only requested for patients admitted to hospital with ICD 10 codes that relate to head injury (S00-S09, T04.0 and T06.0)
2) The period of the study is from April 1998 to December 2017, and so data is only requested for this period.
3) Patient identifiable data, including the date of death, is not requested as this is not required for the planned analysis.
4) Field selection has been minimised to that required to identify the cause of death and link patients to the HES inpatient data set.

Data Analysis:

1) Assess the impact of the introduction of each iteration of the NICE guidelines on deaths from traumatic brain injury.

Emergency admission episodes that are for traumatic brain injury will be identified by ICD 10 diagnostic coding (S00-S09, T04.0 and T06.0) for the period 1998-2017. These episodes will be linked to ONS mortality data for deaths as an inpatient and up to 30 days following discharge. Where a death has occurred data will be coded as either due to traumatic brain injury or due to another cause. Monthly aggregate totals of deaths due to traumatic brain injury for this population will be produced for this time period. These will be converted to a rate using estimates of the population of England. A monthly time series will be produced and segmented regression analysis will be used to assess whether national head injury guidelines introduced in 2003, 2007 and 2014 reduced deaths in this population. The analysis will be stratified into paediatric (aged under 16) and adult populations.

The monthly mortality rate will then be adjusted for age, sex, comorbidity and injury severity using multivariable logistic regression. Age at the beginning of each hospital admission episode for traumatic brain injury will be used. Gender recorded for the hospital admission episode for traumatic brain injury for each patient will be used and will be coded male, female or unknown when not recorded. Each ICD 10 head injury diagnostic code subtype will be coded separately and will be used to give an indication of the severity of injury. Co-morbidities will be grouped and coded based on subtype and significance. Regression analysis will be used to assess the impact of these factors on the likelihood of death for patients admitted with diagnostic codes relating to traumatic brain injury. An adjusted monthly mortality rate, taking into account these factors, will then be produced for deaths due to traumatic brain injury for the time period of interest. This will be used to create a monthly time series of adjusted mortality rates and segmented regression analysis will be used to assess the impact of the different iterations of the NICE head injury guidelines on deaths. This analysis of adjusted mortality will ensure that observed changes are not due to underlying changes in the population.

2) Assess the impact of the introduction of each iteration of the NICE guidelines on the number and rate of hospital admissions due to head injury.

Emergency hospital admission episodes for ICD10 codes related to head injury (ICD10 codes S00-S09, T04.0 and T06.0.) between 1998-2017 will be identified in the data extract of all emergency admissions for this time period of inpatient Hospital Episode Statistics. These will be aggregated on a monthly basis and converted into a rate of hospital admissions for head injury based on estimates of the population of England. A time series and segmented regression analysis will be used to assess for changes in the level and trend of admission rate after the introduction of each iteration of the head injury guidelines.

In order to ensure that increases are not due to underlying changes in the population analysis will be repeated with stratification by age grouping, significant comorbidity, gender and injury severity. Ages will be grouped into: 0-16; 16-30; 30-45; 45-65; 65-85; and 85+. The other variables will be separately coded for. Rates of admission for head injury will also be compared on a monthly basis for the time period of interest to the emergency admission rate for non-head injury trauma and medical emergency admissions. Analysis in these groups will be stratified in the same way into age grouping, comorbidity and gender. This will indicate whether there has been a disproportionate increase in head injury admissions attributable to the introduction of the NICE head injury guidelines or whether increases reflect general trends for increased hospital admissions due to other factors including the introduction of the 4-hour Emergency Department target that occurred contemporaneously to the introduction of the 2004 NICE head injury guideline.

3) Assess the size of any increase in hospital admissions for head injury due to the unintended identification of more traumatic brain injuries of lower clinical significance due to increased CT imaging.

ICD10 codes S02 and S06 will be used to identify emergency hospital admission episodes for patients with injuries that require CT imaging in the data extract of inpatient Hospital Episode Statistics of all emergency admissions between 1998-2017. OPSC-4 intervention codes for neurosurgical intervention (A05.2, A05.3, A05.4, A05.8, A05.9, A40.1, A40.8, A40.9, A41.1, A41.8, A41.9, V03.1, V03.2, V03.3, V03.4, V03.6, V03.7, V03.8, V03.9, V05.3 and V05.4) will be coded for to indicate a neurosurgical intervention has taken place. The proportion of these patients that do not undergo neurosurgery will be measured on a monthly basis. A monthly time series of the proportion of patients with injuries detected by CT imaging who do not undergo CT imaging will be plotted. Segmented regression analysis will be used to assess whether the introduction of the national head injury guidelines has led to an increase in admissions for patients that do not require neurosurgery.

An overall estimate of excess hospital admissions for patients that did not require neurosurgery will be estimated over the time-period from the introduction of the first NICE head injury guideline. This will give an indication of the total unexpected costs associated with the introduction of the guidelines and the potential impact a hospital admission risk tool for patients with small traumatic brain injuries being developing as part of this fellowship could have in reducing hospital admissions in this group.

A second parallel component of the NIHR doctoral fellowship involves the development of a risk stratification tool for the discharge of low-risk patients with traumatic brain injuries. This is separate to the analysis being completed with ONS linked data. However, the analysis being undertaken for objective 3 of the project using ONS linked HES data involves estimating the number of excess hospital admissions that resulted from increased CT imaging leading to the identification of injuries of uncertain clinical significance. The risk stratification tool being developed in parallel is aimed at identifying low-risk patients with brain injuries who could be safely discharged.

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).

ONS Terms and Conditions will be adhered to.