NHS Digital Data Release Register - reformatted

Imperial College London

Project 1 — DARS-NIC-02077-R7M9C

Opt outs honoured: Y, N

Sensitive: Sensitive

When: 2016/09 — 2016/11.

Repeats: Ongoing

Legal basis: Informed Patient consent to permit the receipt, processing and release of data by the HSCIC

Categories: Identifiable

Datasets:

  • MRIS - Cause of Death Report

Benefits:

The results will help physicians to estimate a COPD patient’s life-span. This will be useful in guiding therapeutic intervention and explaining to patients the need to modify behaviour and life-style. The study is academic research only and is in no way commercial.

Outputs:

The results of this study will be published in general and respiratory journals and they will be presented and discussed at national and international meetings such as the British Thoracic Society, European Respiratory Society and American Thoracic Society. The results will be also disseminated at other educational meetings on COPD and respiratory medicine. The results may also be submitted to NICE for the next COPD guideline update

Processing:

Cox proportional hazard models would be used to assess whether bronchiectatic scores of 2 or less were associated with less mortality than scores of 3 or more. Similarly, for emphysema, Imperial College would test whether scores above a median of 15.6% were associated with increased mortality. Data will be visualized with Kaplan-Meier plots; and co-variants included in the cox proportional hazard model might include disease severity and smoking history. Imperial College also would like to know whether a rising trends in airway inflammation is associated with early mortality. The inflammatory markers Imperial College wish to examine are fibrinogen, sputum interleukin-8 and interleukin-6 and C-reactive protein. Imperial College also has extensive historical data on FEV1 and exacerbations, and wish to find out whether lung function decline or exacerbation frequency predicts mortality. Imperial College will examine whether these longitudinal trends are related to survival using cox-proportional hazards and joint models. Joint models investigate how a marker that is repeatedly measure in time is associated with a time to an event of interest, such as death. Imperial College London will submit their Cohort to HSCIC containing full names, date of birth, address, gender and postcode together with a study ID against each patient. The HSCIC provide ONS Mortality Data back to Imperial College London containing study ID, cause of death and date of death.

Objectives:

The aim is: (a) to examine whether measures of emphysema/bronchiestasis collected in 2003/2004 predict mortality (b) to see whether longitudinal changes in airway inflammation, lung function decline and exacerbation frequency predict mortality Previously, it has been shown that frequent exacerbations of COPD are associated with increased mortality but the mechanisms remain to be elucidated. Frequent exacerbations are associated with greater emphysema and bronchiectasis (airway wall thickening). Imperial College’s initial objective is to examine whether indices of emphysema and bronchiectasis previous acquired by CT scanning predict mortality. The data was collected in 2003/2004 on 54 patients. Imperial Colleges were one of the first groups to CT scan COPD patients and therefore this would have one of the longest follow-up periods available.


Project 2 — DARS-NIC-12828-M0K2D

Opt outs honoured: N, Y

Sensitive: Sensitive, and Non Sensitive

When: 2016/09 — 2018/05.

Repeats: Ongoing

Legal basis: Health and Social Care Act 2012, Section 251 approval is in place for the flow of identifiable data

Categories: Anonymised - ICO code compliant, Identifiable

Datasets:

  • Hospital Episode Statistics Outpatients
  • Hospital Episode Statistics Accident and Emergency
  • Hospital Episode Statistics Critical Care
  • Hospital Episode Statistics Admitted Patient Care

Benefits:

Imperial College London Dr Foster Unit (ICL DFU) works with the Care Quality Commission (CQC), contributing to its surveillance remit using the same methods and data. The unit generates monthly mortality alerts since 2007, based on high thresholds [1]. This was pivotal in alerting the then Healthcare Commission (HCC) to problems at the Mid Staffordshire NHS Foundation Trust between July and November 2007[2]. The resulting Public Inquiry recognised the role that the unit’s surveillance system of mortality alerts had to play in identifying Mid Staffs as an outlier [3]. Key recommendations, [4] reflecting the unit’s work, are that all healthcare provider organisations should develop and maintain systems which give effective real-time information on the performance of each of their services, specialist teams and consultants in relation to mortality, patient safety and minimum quality standards. A further recommendation is that summary hospital-level mortality indicators should be recognised as official statistics [5]. If ICL DFU is given continued access to the data, this monitoring tool that detected Mid Staffs will continue to monitor patient outcomes at acute hospitals and be ready to detect any future outliers. The unit will be able to assist the investigation of variations in outcomes at a local level by providing Local Patient ID, NHS Number and Consultant Code from the unit’s analyses to authorised users within trusts to enable reconciliation with local information systems and the instigation of clinical audits and case note reviews. ICL DFU mortality outlier outputs are used by CQC within their Hospital Inspection framework.(on-going) As a result of the unit’s leading role in the development of hospital mortality measures, in 2010 ICL DFU was invited to contribute to a DoH Commissioned expert panel (Steering Group for the National Review of the Hospital Standardised Mortality Ratio) [6] to develop a national indicator of hospital mortality. The resultant Summary-level Hospital Mortality Indicator (based in part on their HSMR methods) is now a public indicator used by all acute trusts. [7] Professor Sir Bruce Keogh suggests that a relatively “poor” SHMI should trigger further analysis or investigation by the hospital Board. The recent review (published in July 2013) into the quality of care and treatment provided by 14 hospital trusts with consistently high mortality in either measure led to 11 out of the 14 trusts identified being immediately placed on special measures. The review also informs the way in which hospital reviews and inspections are to be carried out with the recommendation that mortality is used as part of a broad set of triggers for conducting future inspections [8]. ICL DFU continues to advise the HSCIC on methodological issues around the Summary level Hospital Mortality Index (SHMI), and carry out analyses relating to this measure to assist in its development. (ongoing) The unit’s research on specific aspects of care has received a high media profile and has been highly cited. Their research on weekend mortality in emergency care, analysis of mortality associated with the junior doctor changeover and work on elective procedures and mortality by day of the week resulted in front page broad sheet coverage, and radio and TV interviews. (ongoing) https://www1.imperial.ac.uk/publichealth/departments/pcph/research/drfosters/inthemedia/ The unit’s “Out of hours” work has been a key driver in moving NHS towards 7/7 care. Headlines include, “NHS Services – open seven days a week: every day counts” and, “Sunday Times Safe Weekend Care”. As a result of the unit’s published research into the junior doctor changeover, Bruce Keogh introduced a week's shadowing where newly qualified doctors worked alongside more senior ones for a week before they start work in August. The Academy of Medical Royal Colleges published proposals (16th April 2014) suggesting all Foundation Year 1 posts should begin on the first Wednesday in August as has always been the case, but other training posts should begin in September.(on-going) As part of the ‘biggest bang per buck’ analysis, econometric modelling will suggest which elements of the patient pathway are the most costly. Combining this with modelling of variation by unit will suggest priorities for improvement. Outputs will benefit managers, commissioners and patients. (Dec 2017) Analyses of return to theatre and joint revision for elective hip and knee surgery will help orthopaedic surgeons, commissioners and patients understand these key quality markers for this specialty and devise appropriate improvement projects, for instance by determining which patients are at the highest risk and therefore need more rigorous follow-up. (on-going) ICL DFU intends to examine demand and capacity measures for A&E and admissions, and the impact that pressure on resources might have on safety and patient outcomes. By profiling hospital trusts in terms of demand, patient mix and outcomes, researchers will better understand key NHS metrics and patterns of service use and thereby help managers manage demand. (Jun 2017) Regarding the travel time analysis, using Lower Super Output Areas would enable us to study the effect of distance from home to hospital on patient outcomes. This also allows geographical access to services to be estimated, as researchers can calculate how far patients must travel for their treatment both now and after any future service reorganisation. (Dec 2017) ICL DFU analysis of their mortality alerting system will allow us to improve the alerting process and provide a better indication of how hospitals should investigate them to reduce mortality (including what are the key contributing factors to the alerts and to subsequent improvement in mortality by the hospitals). (Dec 2016) The modelling of health trajectories in stroke patients will improve risk stratification and understanding of the medium-term prognosis and needs. This will also allow better econometric modelling of NHS service use. (Jul 2018) References [1] CQC Quarterly publication of individual outlier alerts for high mortality: Explanatory text (URL available at http://www.cqc.org.uk/public/about-us/monitoring-mortality-trends) [2] Investigation into Mid Staffordshire NHS Foundation trust. Healthcare Commission 2009. Outcomes for patients and mortality rates. Pages 20 - 25 http://www.midstaffspublicinquiry.com/sites/default/files/Healthcare_Commission_report_on_Mid_Staffs.pdf [3] Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry 2013. Volume 1. Pages 458 - 466 http://www.midstaffspublicinquiry.com/report. [4] Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry 2013. Executive Summary. Recommendation 262: http://www.midstaffspublicinquiry.com/report). [5] Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry 2013. Executive Summary. Recommendation 271: http://www.midstaffspublicinquiry.com/report. [6] Development of the new Summary Hospital-level Mortality Indicator. Department of Health Website. http://www.dh.gov.uk/health/2011/10/shmi-update/ [7] Indicator Specification: Summary Hospital-level Mortality Indicator. http://www.ic.nhs.uk/SHMI [8] Review into the quality of care and treatment provided by 14 hospital trusts in England: overview report Professor Sir Bruce Keogh KBE. http://www.nhs.uk/NHSEngland/bruce-keogh-review/Documents/outcomes/keogh-review-final-report.pdf 2) Support the provision of a management information systems (Dr Foster Analysis Toolkit) for the NHS Expected benefits include: • Enabling NHS acute trusts to measure, compare and benchmark key quality indicator trends – focusing on risk-adjusted measures of mortality, readmissions and length of stay in hospital. • Providing evidence to instigate clinical audit and investigations related to quality of care, such as highlighting potential poor clinical coding or quality/efficiency concerns. • Validating other mortality indicators – such as HSMR, Custom alerts and crude mortality. • Enabling NHS acute trusts and commissioners to use performance information to identify, quantify and act on opportunities to improve efficiency of health services. • Understanding areas of best practice amongst our customers and facilitate interactions with other customers who are not performing as well to support quality and efficiency improvement. • Helping clinicians and managers by providing independent and authoritative analysis of the variations that exist in acute hospital care in a way that is meaningful for them and that is understandable to patients and the public. • Highlighting topics of interest to the health industry and wider public to enable discussion and improvement in healthcare provision. • Publication of articles around variations of healthcare within the NHS is in the public interest and supports the government agenda for transparency by promoting choice and accountability within the NHS. • Maintaining the focus of the organisations on improvement. • Raising public and professional awareness through the Dr Foster's Hospital Guide regarding issues that affect the quality and efficiency of care provided by the NHS by publishing new information about variation in outcomes at the level of individual hospitals. In recent years, the guide has focussed on issues of clinical and managerial concern such as weekend care, overcrowding, management of chronic conditions and variations in access to elective care. In each case, the approach has been to identify effects that are known from the academic literature and to show their impact here and now in English NHS hospitals. By publishing this information Dr Foster Limited support the improvement of healthcare in England. How will these benefits be measured: Benefits are ongoing as the outputs described above are used within NHS Trusts’ internal monthly reporting and quality processes. Dr Foster Intelligence Ltd (DFI) services allow performance of NHS Provider Trusts to be monitored and trended over time and therefore provide customers with the ability to measure changes in quality and performance particularly in instances where customers have been alerted and they have worked with them to understand the causes of worse than expected performance. DFI intends to provide an online customer survey within the Dr Foster Analytics Tool to capture customer feedback and associated benefits, this data will form the foundation for improving their services and enable them to provide HSCIC, and other relevant bodies, with tangible evidence to support their ongoing use of HES data. DFI welcomes the opportunity to work with HSCIC to ensure information captured can support their ongoing supply and use of HES data. When will these be achieved: As a majority of benefits are achieved on an ongoing basis, it is not possible to outline a specific target date for achievement of the benefits outlined as they are reliant on a range of factors outside of ICL DFU and DFI’s control. However, whenever there are areas of particular concern about performance against key indicators, the 2 parties act immediately to alert relevant stakeholders and offer their assistance in better understanding and addressing them.

Outputs:

1) Research into variations in quality of healthcare by provider: background to proposed work Imperial College London Dr Foster Unit (ICL DFU) work programme is designed to develop and validate indicators of quality and safety of healthcare, show variations in performance by unit and socio-demographic stratum and develop methods for risk prediction, risk adjustment and outlier detection. The unit’s work focuses on quality of care and patient safety, including healthcare-acquired infections (surgical wound infections and urinary tract infections) and safety indicators. Collaborative projects with clinical colleagues have helped develop and validate healthcare quality indicators other than mortality, including bariatric surgery, primary angioplasty rates, indicators for stroke care, obstetric care, orthopaedic redo rates and returns to theatre. ICL DFU is currently working on the following analyses: ‘Biggest bang per buck’ elements of treatment pathways for chronic diseases. By mapping out NHS hospital contacts and modelling the variation across units, the unit will determine the elements (e.g. readmissions, missed OPD appointments, surgery that could have been done as a day case) with the most potential for improvement. This forms part of the unit’s work with Imperial’s NIHR funded Patient Safety Translational Research Centre on the use of information for service improvement. (Dec 2017) Drivers of unscheduled return to theatre (or reoperation) in elective hip and knee replacements: correlation between Return To Theatre (RTT) and revision rates by surgeon; volume-outcome relation for RTT; risk of RTT following revision rates. The objective is to better understand these key metrics for the specialty: revision rates are of major interest to surgeons and are on the NHS Choices website. The unit has recently established that there is greater non-random variation in RTT rates between surgeons than between hospitals. (on-going) Predictors of readmissions and A&E attendance in patients with chronic diseases (heart failure, COPD, cancer). Readmissions are the focus of much attention worldwide in efforts to reduce costs and improve outcomes, but little is known about the role of A&E attendance (not ending in admission) in observed variations in readmission rates. The study has revealed that earlier OPD nonattendance is a strong risk factor for readmission. The objective is again to better understand readmissions as an indicator and to suggest reformulation if desirable. (Jun 2017) Travel time. Due to the well-documented relation between patient volume and outcomes, there is a growing drive to centralise certain services such as for stroke and elective surgery. Treatment rates for many conditions such as thoracic aortic disease (TAD) vary around the country. Using Lower Super Output Areas of the patient’s residence and the hospital postcode, researchers will first calculate how far patients currently travel for their TAD treatment and then the travel distance that would be incurred were surgical services retained only at large centres. The effect on outcomes will also be assessed. (Dec 2017) Modelling Health trajectories for Stroke patients ICL DFU is currently undertaking a study which involves the evaluation of patients who had a stroke and following them up for 5 years. The study involves people who had a stroke for the first time. Previous studies have been criticised for including patients with recurrent stroke. Based on previous research, ICL DFU has tracked back their chosen stroke patients for 10 years to ascertain whether the stroke event under observation was the first or recurrent. Moreover, ICL DFU has to evaluate important cardiovascular co-morbidities by looking at the patients hospital diagnosis made in the previous years. The study aims to identify stroke patients who are initially stable but later become high users of health care resources. ICL DFU also plans to look at pattern of causes of subsequent hospitalisation in the same cohort of patients. The study requires tracking back patients 10 years and following up for 5 years from the time of their index stroke event. (Jul 2018) Recent pressures on A&E and breaches of the 4-hour wait have led to concerns over pressure on A&E and inpatient capacity. ICL DFU intends to examine capacity measures for A&E and inpatient admissions, and the impact that pressure on resources might have on safety and patient outcomes with a view to better understanding key NHS metrics and patterns of service use to better match supply to need. (Dec 2016) ICL DFU is working in collaboration with the University of Manchester and supported by the Care Quality Commission, to improve understanding of the unit’s mortality alerts and to evaluate their impact as an intervention to reduce avoidable mortality within English NHS hospital trusts, focusing on two conditions commonly attributed to mortality alerts acute myocardial infarction and septicaemia. The aim of this study is to provide a descriptive analysis of all alerts, their relationships with other measures of quality and their impact on reducing avoidable mortality. (Dec 2016) International comparisons of service use and outcomes. England and the USA. The unit holds data from Centre for Medicare and Medicaid Services enrollees and from the Nationwide Inpatient Sample from the USA. Researchers have previously set out the methodological issues with using administrative data from multiple countries. This study will compare patient casemix, rates of outcomes such as infections and readmissions, and rates of surgery, for example in patients near the end of their life (overtreatment is a growing concern) between the two countries. The objective is to highlight areas of better or poorer performance by the NHS compared with the USA. ICL DFU has an extract of the Italian data and will be using HES data to compare hospital use for patients with heart failure in England compared with Italy. (on-going) Examples of key published research that have used HES data include: Palmer WL, Bottle A and Aylin P. Association between day of delivery and obstetric outcomes: observational study. BMJ 2015; 351: h5774. Bottle A, Goudie R, Cowie MR, Bell D, Aylin P, 2015, Relation between process measures and diagnosis-specific readmission rates in patients with heart failure, HEART, Vol: 101, Pages: 1704-1710, ISSN: 1355-6037 Aylin P; Alexandrescu R; Jen MH; Mayer EK; Bottle A. Day of week of procedure and 30-day mortality for elective surgery: retrospective analysis of hospital episode statistics. BMJ 2013;346:f2424. Palmer WL; Bottle A; Davie C; Vincent CA; Aylin P. Dying for the Weekend: A Retrospective Cohort Study on the Association Between Day of Hospital Presentation and the Quality and Safety of Stroke Care. Arch Neurol. 2012;69:1296-1303. Aylin P; Bottle A; Majeed A. Use of administrative data or clinical databases as predictors of risk of death in hospital: comparison of models. BMJ 2007;334:1044. Aylin P, Yunus A, Bottle A, Majeed A, Bell D. Weekend mortality for emergency admissions. A large, multicentre study. Qual Saf Health Care. 2010;19:213-217 Jen MH, Bottle A, Majeed A, Bell D, Aylin P. Early in-hospital mortality following trainee doctors' first day at work. PLoS One. 2009;4:e7103. For full publication list see unit website: http://www1.imperial.ac.uk/publichealth/departments/pcph/research/drfosters/unit_publications/ 2) Support the provision of a management information systems (Dr Foster Analysis Toolkit) for the NHS Dr Foster Intelligence Limited (DFI) is an independent healthcare information company. It provides a research grant to ICL DFU to develop indicators and methodologies to assist in the analysis of healthcare performance. ICL DFU works in collaboration with DFI to provide the NHS with a number of management information systems via the Dr Foster Analysis Toolkit. The main output created are benchmarked or standardised healthcare indicators & analysis such as mortality (SHMI/HSMR), LOS(Length of Stay), admission trends, readmission rates, patient safety indicators, referral patterns, market share analysis etc. As stated previously, outputs are to be used solely for the purposes of providing a management information function to the NHS. Outputs are provided via: • Dr Foster Analysis Toolkit – Use of Role Based Access to determine the level of data end users can see within the tool. • Value added services - Tabulations, Reports, Spreadsheets, Presentations, Articles & Projects. Outputs will be used by customers to investigate Clinical Quality, Performance and Business Development, specifically: • Assess and manage clinical quality and patient safety within NHS Organisations • Identify pathways where there is potential for improvement • Identify areas of best practice either within the Provider Trust or local/national health economies • Better understand how they compare to other Provider Trusts with similar case mixes • Identify improvements in operational efficiency • Understand patient outcomes • Identify and understand market activity • Monitor the impact of implemented changes • Identify variations in outcomes 3) Provision of a patient re-identification service for the NHS ICL DFU provides a patient re-identification service for the NHS which allows NHS provider trusts to investigate issues around quality and safety of care within their organisation, which have arisen out of performance alerts arising out of ICL DFU analyses (e.g. mortality alerts), or arising from DFI performance tools using ICL DFU methods. Authorised individuals within Provider Trusts are able to identify their own patients indicated in the DFI healthcare performance tools. From April 2015 to April 2016, there were over 3,600 successful logins from 75 NHS provider organisations. 64 provider trusts have used it more than 12 times per year (once a month) and one trust has used the re-identification service 425 times within this period. The re-identification service allows ICL DFU to supply NHS provider trusts with NHS Number and LOPATID using DFI healthcare performance tools without passing these identifiers on to DFI. No patient identifiers will ever be passed to DFI or any other organisation except the NHS provider trust from where the data originated. The patient identifiable data are kept separate to the anonymised and sensitive data. They are held on a different system to clinical data. All patient identifiable data are securely deleted on a rolling 3 year programme. The re-identification service is maintained by ICL DFU and is in full compliance of CAG approval reference:15/CAG/0005

Processing:

Imperial College London Dr Foster Unit (ICL DFU) uses hospital administrative data in the form of HES bespoke/monthly extracts to identify measures of quality and safety of healthcare. The unit’s work focuses on quality of care and patient safety, including healthcare-acquired infections, mortality and safety indicators. ICL DFU holds 2 databases to store data – A Research database and a Patient Identifiable database to provide a Re-Identification service for NHS provider trusts. Patient identifiers are stored separately to the unit’s research database which holds the HES extracts (including sensitive fields). ICL DFU researchers have no access to identifiable fields. Only two named data managers have access to the patient identifiable fields within the unit. The purpose of holding the patient identifiers for the last 3 years is to allow hospitals to further investigate any alerts around poor or good performance and to help improve the quality and safety of healthcare delivery. The HES extracts (including sensitive fields) are stored in the Research database where researchers are able to access the data to do their analyses. The HES extracts (including sensitive fields) are loaded on to the Research database with a unique identifier (fosid) being generated and added to the datasets. A new Extract_hesid for Dr Foster Intelligence Limited (DFI) is also generated using the SHA-256 hashing algorithm, compliant with the e-GIF Technical Standards Catalogue Version 6.2 based on the original Extract_hesid. An extract is taken from ICL DFU patient identifier server and copied to the server which is used to provide the Re-Identification service for the NHS Acute Trusts. Further data processing are carried out on the onward supply of data by DFI who have dedicated staff and processes as per below: • Linkage into spells and superspells, which can often span across financial years • HRG, Tariff and other PBR related fields, using the HRG Grouper software • Various clinical groupings, including CCS Diagnoses, Ambulatory Care Sensitive (ACS) conditions and Procedure Groups • Quality outcomes, including mortality, emergency readmission within 28 days, Long Length of stay and patient safety indicators • Patient-level predicted risks for these outcomes, based on national Logistic Regression models which are executed using R statistical software and updated monthly • Various other national benchmarks, including Length of stay percentiles and Standardised Admission Ratio benchmarks • Numerous efficiency-based metrics, including average length of stay, day case rate and potential bed days saved • Prescribed Specialised Services (PSS) groups, using the PSS Grouper software This process guarantees both DFI and ICL DFU are working from exactly the same data (both in terms of underlying patient linkage and derived fields), which is necessary for their joint projects. No record level data will be transferred outside of the EEA, either under this agreement or any related sub-licence.

Objectives:

Imperial College London Doctor Foster Unit (ICL DFU) uses HES data to identify measures of quality and safety in healthcare. Their research themes are around developing and validating indicators of quality and safety of healthcare, particularly by GP practice, consultant, and NHS Trust, showing variations in performance by unit, patient risk subgroups and risk prediction, risk adjustment and outlier detection for such indicators and variations and any other methodological aspects as they arise. ICL DFU works in collaboration with Dr Foster Limited (DFI) to provide a management information function in the form of analysis for healthcare organisations. ICL DFU calculate a wide range of healthcare indicators (over 100) and as such require HES data to provide a wide array of relevant indicators to give end users as complete a picture of hospital performance as possible to allow UK healthcare and Social care organisations to effectively: • Monitor quality of services provided • Identify efficiency opportunities • Identify pathways where services can be improved for the benefit of patients A data period of 15 years of historical data is essential to enable both ICL DFU and Dr Foster Limited to: 1. Obtain longitudinal data on prior admissions for patients. Risk modelling will also require access to variables on prior admissions including previously recorded co-morbidities. 2. Create, update and maintain statistical risk models to enable the regular production of risk adjusted measures of mortality, quality and efficiency (including HSMR and Cusum alerts as used by NHS organisations and regulators) An example of a longitudinal study which ICL DFU is currently undertaking a study which involves the evaluation of patients who had a stroke and following them up for 5 years. The study involves people who had a stroke for the first time. Previous studies have been criticised for including patients with recurrent stroke. Based on previous research, ICL DFU has tracked back their chosen stroke patients for 10 years to ascertain whether the stroke event under observation was the first or recurrent. Moreover, ICL DFU has to evaluate important cardiovascular co-morbidities by looking at the patients hospital diagnosis made in the previous years. The study aims to identify stroke patients who are initially stable but later become high users of health care resources. ICL DFU also plans to look at pattern of causes of subsequent hospitalisation in the same cohort of patients. The study requires tracking back patients 10 years and following up for 5 years from the time of their index stroke event. Both ICL DFU and DFI require the full HES datasets to increase the power of predictive models for rare diseases, procedures and events (e.g. ICL DFU and DFI build standard casemix adjustment models for 259 diagnosis groups and 200 procedure groups which include some rarer conditions). At a high level the analyses break down into the following: • Quality measures of healthcare services by providers/area/clinical interest/trend analysis • Variations in health outcomes • Health inequalities and needs analysis • Predictions • Performance data and changes in clinical practice • Management information • Efficiency Monitoring • Benchmarking • Contract Management and Variance Analysis • Activity Monitoring • National Target Performance • Pathway design, redesign and improvement. • Practice Performance Monitoring • Capacity and utilisation management • Cross checking of commissioning data • Systems to support and monitor the pattern of healthcare usage • Overall data quality A bespoke extract with lesser fields and lesser frequency will not suffice given that ICL DFU/DFI require the most up-to-date information to inform trusts of potential issues around quality. A soon-to-be-published NIHR-funded review of a subset of mortality alerts sent between 2011 and 2013 (and subsequently followed up by CQC), found that Trusts reported areas of care that could be improved in 70% (108/154) of the alerts and that all were implementing action plans to address these issues. ICL DFU/DFI research has found on average, an associated reduction in mortality of 55% in the 12 months following a notified alert, suggesting timeliness of data may be key to saving lives. Patient identifiers The Regulation 5 of the Health Service (Control of Patient Information) Regulations 2002 (s251) support letter confirms the final approval to receive confidential patient information for ICL DFU research database and identifiers to provide re-identification service to Dr Foster Intelligence Limited (DFI) customers and ALL NHS trusts. Identifiable data processed under CAG [15/CAG/0005] will be retained for a maximum of three years after which it should be destroyed or irreversibly pseudonymised on a rolling basis. The purpose of holding the patient identifiers is to allow hospitals to further investigate any alerts around poor or good performance and to help improve the quality and safety of healthcare delivery. ICL DFU does this by providing a re-identification service to acute NHS providers who are Dr Foster Limited’s customers and ALL NHS Trusts. DFI has no access to the patient re-identification service. No patient identifiers will ever be passed to DFI or any other organisation except the NHS provider trust from where the data originated. For this purpose, ICL DFU have developed a re-identification service whereby authorised individuals within NHS Provider Trusts are able to identify their own patients indicated in the Dr Foster Limited Analysis Toolkit. This service allows supply of Provider trusts’ NHS Number and LOPATID using Dr Foster Limited Analysis Toolkit without passing these fields on to DFI. The re-identification service is maintained by ICL DFU. Sensitive fields Sensitive fields will only be available at a record level to NHS Provider Trusts (or approved regulatory bodies with express authority to demand such data, e.g. the CQC) and are specifically required for the purpose of conducting root cause analysis where there is a legitimate relationship with the patient. Where a legitimate relationship does not exist data will be available at an aggregate level in line with HSCIC HES Analysis Guide, HSCIC Small Numbers Procedure and ONS Guidelines, with any sensitive fields suppressed. Consultant Code ICL DFU and DFI provide consultancy from their analyses to authorised users within trusts to enable reconciliation with local information systems and the instigation of clinical audits and case note reviews. Analyses by consultant activity are fed back to the NHS through a range of Management Information Systems provided by DFI in the forms of aggregation of teams into 'departments' or other hierarchies. Requirements for analyses by consultant activity are consistent with NHS needs and policy direction (to publish at consultant level). Consultant code is also used in research e.g. analysing volume and outcome relations for elective surgery. Some exclusions are applied e.g. Invalid codes, dental consultant etc. Patient’s general medical practitioner Patient’s general medical practitioner is used to examine variations by GP practice and to enable mapping to practice level such as The Quality and Outcomes Framework (QOF) and practice staffing data etc. NHS Provider Trusts are able to identify the registered GP who referred the patient. This is essential to understanding rates of admission and rates of readmission by GP practice which may reflect issues of community and primary care. Person referring patient Analyses by the person referring patient activities are fed back to the NHS Provider Trusts through a range of Management Information Systems provided by Dr Foster Limited. These analyses allow NHS Provider Trusts to identify the person who referred the patient for calculation of referral rates. Understanding referral rates by GP practice and consultant can help to identify issues of quality of primary care. ICL DFU is part-funded by a grant from DFI. On approval of this application, a sub-licence model between the HSCIC and Imperial College will exist to permit ICL DFU to supply derived pseudonymised data together with specific clear text sensitive fields (as stated within this application) to DFI. The unit works in collaboration with DFI to provide a management information function in the form Dr Foster Analysis Toolkit. This purpose is fulfilled by analysis of HES data made available to customers via the following services provided by DFI: 1. Licensed subscriber of Dr Foster Analysis Toolkit a. Directly – i. NHS Provider Trust holding a subscription to the Dr Foster Analysis Toolkit are able to view data at a record level, with an option to use the patient re-identification service for approved individuals; or ii. other NHS organisations holding a subscription to Dr Foster Analysis Toolkit are able to view aggregated analysis to prevent any patients being identified in accordance with guidance provided by HSCIC. b. Indirectly – non-NHS organisation that hold a subscription to the tool supply NHS organisations with aggregate small number suppressed analyses. 2. Value Added Services As an information intermediary, DFI responds to customer requests for analyses of HSCIC data, whose scopes are by their nature bespoke and customised to local needs. An established specialist team of Analysts provides statistical analysis for interpreting complex data and producing analysis on behalf of customers. It should be stated that this team, which is project based, conduct annual training on handling sensitive records and are highly conversant in national guidelines to protect patient confidentiality, where there is any doubt the DFI Head of Information Governance or SIRO will provide guidance and if required contact HSCIC. DFI also provides analysis for publication for the benefit of the public and NHS e.g. Hospital Guide, and to support benefit to health and social care. Such analytical content may be published directly by DFI or within academic journals or articles to journalistic/media entities in the form of text, tables, and other data visualisation such as diagrams/graphs using aggregate information based on HES analysis. DFI is aware that publications, whether inside or outside the NHS, must adhere to strict guidelines in terms of disclosure, and will ensure any such publications are aggregated and comply with small number suppression in line with the HES Analysis Guide/ONS Guidelines and other relevant legislation and standards as defined in the Terms and Conditions of the Data Sharing Agreement.


Project 3 — DARS-NIC-147827-NC2TC

Opt outs honoured: N, Y

Sensitive: Sensitive

When: 2016/04 (or before) — 2017/02.

Repeats: Ongoing

Legal basis: Section 42(4) of the Statistics and Registration Service Act (2007) as amended by section 287 of the Health and Social Care Act (2012)

Categories: Identifiable

Datasets:

  • MRIS - Scottish NHS / Registration
  • MRIS - Cause of Death Report

Objectives:

This study is analysing the risk of cancer or advanced adenomas with varying frequency of colonoscopic surveillance for patients identified with intermediate grade adenomas. The overall research objectives are to: - Examine the optimal frequency of surveillance in people found to have intermediate grade colorectal adenomas.* - Examine he risks and benefits to the patient with respect to prevention of cancer and the development of advanced adenomas; anxiety, morbidity and mortality; costs and cost-effectiveness and implications for the NHS.* * aims for which we require data from ONS using the MRIS.


Project 4 — DARS-NIC-148056-T6T5Z

Opt outs honoured: N

Sensitive: Sensitive

When: 2017/09 — 2018/05.

Repeats: Ongoing

Legal basis: Informed Patient consent to permit the receipt, processing and release of data by the HSCIC

Categories: Identifiable

Datasets:

  • MRIS - Members and Postings Report
  • MRIS - Cause of Death Report
  • MRIS - Cohort Event Notification Report
  • MRIS - Flagging Current Status Report

Benefits:

• Safety of Airwave. The primary objective of the Study is to ascertain whether or not there is any link between use of Airwave and the long-term health of its users. Were such a link identified, it would be relevant and important to both Airwave users and management to understand in what circumstances risks might be increased. Alternatively, a finding of no demonstrable effect would be reassuring to the Airwave user community and would place any current and future concerns about possible health effects into proper context based on objective evidence. • Helping future generations. The study is generating new knowledge of benefit not only to police officers and staff as individuals, but to the wider community and to society as a whole. Analyses of data and samples will help to better understand the risks and causes of future diseases and ill health, and thus inform improved preventive and treatment strategies. There are many special aspects of the Airwave cohort including the occupational setting, given the particular nature of police duties and working patterns, the relatively young age of the cohort, and the inclusion of large numbers of women as well as men, which make this study uniquely valuable. The results generated from the use of this resource will inform future policy and practice both for the betterment of police force health and for the health of the public more generally. • Clarity in respect of health effects of long-term use of radio frequency technology. Continuing to gather the data necessary to undertake and report on the Study’s analyses of Airwave use and health will allow the potential long-term health effects to be better understood, and to place any future claims of possible harm into proper context based on the evidence. The findings would be relevant to, and inform, strategic decisions about future investment in radio communications systems within the Police Service. • Responsibility to the health and welfare of the workforce. Policing is a highly complex occupation with specific patterns of working and occupational risks with potential health effects that are not well understood. The Study has established a ‘broad and deep’ biomedical resource with which to continue to monitor the health and well-being of the workforce, and to help understand the causes and risks of ill-health and disease. Results will inform possible preventive approaches and best practice for maintenance of a healthy and engaged workforce.

Outputs:

The aim of the study is to estimate the risk of all cancers and certain mortality outcomes in relation to Airwave use. Cancer and death notifications will be used to determine prevalent cases at baseline and subsequent incident cases for each outcome (e.g. head and neck cancers) under study. Survival analyses will be performed to investigate the association between each outcome and level of Tetra exposure and the risk of incident cases for each disease using multivariable Cox models. The results of the analyses will be published in peer-reviewed scientific journals and in summary form on the study website. The peer-reviewed journals targeted are likely to be similar to those that have already published in (Environmental Research). The data will be used to compile progress reports for the Study funder (the Home Office). However, all such outputs will report aggregated results only, and no individual will ever be identified. Results will be published on aggregate level with small numbers suppressed. It will not be possible to identify the individuals. The recruitment phase was completed on the 31st of March 2015. However, the follow up and data analysis phase continue. The target date for submission of a scientific outcomes paper (in a peer reviewed journal) on any possible long-term health implications for Police personnel related to use of Airwave (the main purpose of the Study) will be December 2017. Recent publications: 13th July 2016 Acute Exposure to Terrestrial Trunked Radio(TETRA) has effects on the electroencephalogram and electrocardiogram, consistent with vagal nerve stimulation http://dx.doi.org/10.1016/j.envres.2016.06.031 28th April 2016 Validation of objective records and misreporting of personal radio use in a cohort of British police forces (the Airwave Health Monitoring Study) http://dx.doi.org/10.1016/j.envres.2016.04.018 06th September 2014 The Airwave Health Monitoring Study of police Officers and staff in Great Britain: Rationale, design and methods http://dx.doi.org/10.1016/j.envres.2014.07.025

Processing:

NHS Digital data is only accessed by substantive employees of Imperial College London and only for the purposes described in this document. Directly identifiable data is kept separate from the study data, NHS Digital data is only linked to the AIRWAVE study data and no other datasets held by the applicant. The standard ONS terms and conditions will be adhered to. Airwave data from each police participant will be collected from monthly downloads of relevant data from the Home Office, giving information on Airwave exposure at individual level. These will be combined with questionnaire data about participants' use of Airwave to derive an exposure metric. Health outcome data will be assessed by linking information on individual participants to national records on mortality and cancer incidence, and from absence records supplied by the police force employers. Data, once received, is stored and analysed at Servers in South Kensington campus on servers located in a secured area. Users at St Marys connect to the servers via remote desktop. All IT infrastructure is owned and managed by Imperial college, there are no shared resources, and all network traffic is contained within Imperial College. The servers will be designated as holding identifiable data or anonymised data. All servers are secured to only accepting connections from specified users and workstations. Identifiable data can only be accessed from dedicated workstations that sit alongside the users college PC. These workstations can only be used to connect to the “identifiable servers”, there is no internet access available. Data uploading/downloading can be further restricted to specified users and PCs. All data transfers are recorded and kept for audit purposes by a staff member who has the role of “internal auditor”. The internal auditor monitors compliance of the Information Governance Policy including all agreements the groups have with external groups. The NHS data sharing agreement covers the Imperial college campus, and relies on each group having an environment that conforms to the NHS toolkit standard. Data received will be linked to the existing participant database. The scope of IGT policy whose code is EE133887-SPHTR (Imperial College London - School of Public Health Medical Trials and Research) is the Imperial College network. It uses the Imperial College infrastructure to create isolated enclaves that are used to form the security zones of the network. The IGT policy requires that, “Where possible all servers should be held within Imperial College’s Data centre, and subject to its security policy (currently aiming towards ISO27001). Any group not able to place a server in the datacentre will need to seek approval from the Security Manager.” The Airwave Study activities that are bound by EE133887-SPHTR will use servers located in the Data centre. Users working according to EE133887-SPHTR will be based at the College site in Norfolk Place and will access the Imperial College Data centre at South Kensington according to the security requirements defined in EE133887-SPHTR; the IGT policy therefore covers both the South Kensington and Norfolk Place sites. From time-to-time, consolidated pseudonymised extracts of the database are created and these are used by researchers to investigate the questions addressed by the Study. Those extracts follow the same security rules of the main database and will be kept in the same location. Other than in exceptional cases, namely resolving linkage questions or to contact research participants, data used by researchers is delinked from personal identifiers such as name and address. All researchers complete a detailed written confidentiality agreement with the College, and ONS Linkage Short Declaration of Use. When the Study is completed and closed to further analysis, the data will be archived securely during the life time of the data sharing agreement and for such time as is necessary to provide proper audit for published research. The data will be subsequently securely destroyed. No third parties will be allowed to access any data provided under this agreement. The applicant will supply NHS Digital with name, address (including postcode) and date of birth for linkage. This will ensure that any new members not already flagged by NHS Digital are linked and remove any members who have subsequently opted out of the study.

Objectives:

The Airwave Health Monitoring Study was established in 2003 to evaluate possible health risks associated with the use of Terrestrial trunked radio (TETRA), a digital communication system used by the police forces and other emergency services in Great Britain since 2001. It is a long-term observational study following up the health of the police force with respect to TETRA exposure, and ability to monitor both cancer and non-cancer health outcomes. It addresses needs raised in a report by the Advisory Group on Non-Ionising Radiation (AGNIR) on the possible health effects from TETRA. There are currently c. 53,000 participants in the Study. The aim of the study is to estimate the risk of all cancers, certain mortality outcomes and various non-fatal, non-malignant health disorders in relation to Airwave use. As well as the focus on cancer incidence, the study will investigate non-cancer health outcomes (including cognitive, neuropsychiatric and neuro-degenerative effects which may be linked to sickness absence and early retirements), as the mechanisms of any putative health effect related to TETRA use are unknown. The cohort consists of police force employees from Great Britain and c. 53,000 participants are enrolled at the present. The study population will be flagged for mortality and cancer incidence using the cancer and mortality data at NHS Digital and Information Services Division (ISD) of the Scottish Health Service, based on personal identifiers such as full name, date of birth and NHS number.


Project 5 — DARS-NIC-148071-QHNM8

Opt outs honoured: Y

Sensitive: Sensitive

When: 2016/04 (or before) — 2017/02.

Repeats: Ongoing

Legal basis: Approved researcher accreditation under section 39(4)(i) and 39(5) of the Statistical Registration Service Act 2007

Categories: Identifiable

Datasets:

  • MRIS - Cause of Death Report
  • MRIS - Cohort Event Notification Report
  • MRIS - Scottish NHS / Registration

Objectives:

The data supplied bt the NHS IC to Imperial College London will be used only for the approved Medical Research Project - MR700 - SINGLE SIGMOIDOSCOPY SCREENING IN PREVENTION OF BOWEL CANCER


Project 6 — DARS-NIC-148230-KHMHH

Opt outs honoured: N

Sensitive: Sensitive, and Non Sensitive

When: 2016/04 (or before) — 2016/08.

Repeats: Ongoing

Legal basis: Informed Patient consent to permit the receipt, processing and release of data by the HSCIC

Categories: Identifiable

Datasets:

  • MRIS - Cause of Death Report
  • MRIS - Cohort Event Notification Report

Objectives:

The data supplied will be used only for the approved medical research project - MR1009 - High Risk Period for Patients with Heart Failure: A Population-Based Study


Project 7 — DARS-NIC-204903-P1J7Q

Opt outs honoured: Y

Sensitive: Sensitive, and Non Sensitive

When: 2016/04 (or before) — 2018/05.

Repeats: Ongoing

Legal basis: Section 251 approval is in place for the flow of identifiable data

Categories: Identifiable

Datasets:

  • Hospital Episode Statistics Accident and Emergency
  • Hospital Episode Statistics Admitted Patient Care
  • Hospital Episode Statistics Critical Care

Benefits:

Expected measurable benefits to health and/or social care including target date: The focus of the work under this application is to enable key public health issues associated with environmental factors at a small area. This would by definition therefore not look at care of individuals, but would related to informing public health policy and considering population health risks. The key points in demonstrating the benefit to health and social care therefore are :- - The core SAHSU programme of work is funded by Public Health England, and this includes the service to Public Health England to assist PHE in fulfilling their duties - Individual projects must meet be aligned with the Terms of Reference of the SAHSU programme, which includes addressing environmental and health issues. As part of this application, SAHSU will be amending these conditions to include an explicit requirement to reference that projects must be for the promotion of health - All projects are approved by the PHE Programme Board, which includes membership from DH. In order to assist with standardising approaches across application processes, SAHSU have offered membership of the approving group to the Data Access Request Service. Whilst individual detailed project related benefits cannot be stated at this time, three examples are given below of how outputs have and will be used to inform health policy. Traffic pollution and health in London study The results of this study will allow for a better understanding of the health problems caused by air pollution and noise from traffic in London. Findings are likely to have a high media-profile and have the potential to influence air pollution policy and regulatory practices in London and the UK. Possible reproductive and other health effects associated with Municipal Waste Incinerators (MWIs) in England, Wales and Scotland (incinerators) The study was commissioned to extend the evidence base and to provide further information to the public about any potential reproductive and infant health risks from MWIs and to extend the evidence base with respect to exposures and any potential reproductive and infant health risks from MWIs. Health effects of large airports – the London Heathrow example (Heathrow) The results of this study allow for better understanding of the health problems caused by noise from aircraft in London. The findings had a high media-profile and are directly relevant to the Davies commission and decisions on whether to build a third run-way at Heathrow. Results of this study have been reported widely in local, national and international media and been raised as questions at Prime-ministers question. The results of SAHSU studies are placed in the public domain via peer-reviewed publications and are used to inform the behaviour of health care providers and to inform national public health policy. All studies equally feedback their results into the relevant policy leads within PHE.

Outputs:

Specific outputs expected, including target date: There are three main types of output from this application :- 1. The maintenance of the research database, and associated record level pseudonymised extracts for use by individual researchers within SAHSU. Such extracts are delivered either through the RIF or the database team. This database will have the ability to provide data for the specific projects which have been approved as set out above. Note that all extracts are subject to the constraints within this agreement, eg: that data will only be held and processed at the processing / storage address. Such work is on-going through the lifetime of this agreement. 2. Individual project research outputs. All such outputs would be aggregate data, and be used within journal papers, research reports and results, presentations at conferences. Such research outputs would be placed into the public domain. Examples of previous outputs and projects are given below. 3. Individual analyses for Public Health England (PHE). Again, all such outputs would be in aggregated form, but would be provided directly to PHE. Examples of outputs produced to date Traffic pollution and health in London study There have been three papers published to date: • Halonen, JI et al. Road traffic noise is associated with increased cardiovascular morbidity and mortality and all-cause mortality in London. European Heart Journal 2015. • Gulliver, J et al. Development of an open-source road traffic noise model for exposure assessment. Environmental Modelling & Software. 2015. • Halonen JI et al. Is long -term exposure to traffic pollution associated with mortality? A spatial analysis in London. Environmental Pollution. 2015. Future outputs from the study will be disseminated via peer reviewed publication and academic conferences presentations. The outputs of the project will be published in peer-reviewed journals during the course of the study and after completion, which is expected by 2016. Possible reproductive and other health effects associated with Municipal Waste Incinerators (MWIs) in England, Wales and Scotland (incinerators) There have been two papers published to date: • Font A et al. Using Metal Ratios to Detect Emissions from Municipal Waste Incinerators in Ambient Air Pollution Data. Atmospheric Environment. 2015 • Ashworth DC, et al. Comparative assessment of particulate air pollution exposure from municipal solid waste incinerator emissions. Journal of Environmental and Public Health. 2013 Study information will be provided on the SAHSU website for the public. Health effects of large airports – the London Heathrow example (Heathrow) The results of this study have been published in the peer-reviewed BMJ (http://www.bmj.com/content/347/bmj.f5432 ). All outputs from individual projects will be anonymised (data will only be shared where aggregated with small numbers suppressed in line with the HES Analysis Guide). To confirm, no record level data is provided to any third party organisation and no commercial use is permitted.

Processing:

Processing activities: Processing is consistent across all three purposes, given that they all require the use of the same research database, In summary :- - identifiable data are encrypted and held in a secure area of the database on the SAHSU private network. Access to the identifiable data is limited to a small database team within SAHSU. - The identifiable data is held on a separate encrypted file system with access limited to the database team only. The pseudonymised output (CSV) file is then loaded into Oracle for use by researchers. - Separation is maintained between the database team, who handle data encryption and see identifiable data, and researchers, who only have access to pseudonymised data. NHS numbers, addresses and postcodes are encrypted and replaced by pseudonyms and held in confidential tables to which only the database team has access. Record separation occurs during data loading; the original files are stored on a separate encrypted file system on a separate server. Identifiers are not held on the same server as the clinical data. Further processing standardises names, data types, correct dates, links in geography via the postcode, performs encryption and pseudonymisation and carries out dataset specific bespoke processing (e.g. the detection of potential duplicates). At this point the data is only accessible by the database team, and is fully protected by encryption and pseudonymisation. The next phase creates production tables and sets them up for use by the researchers, granting appropriate permissions and setting up auditing. There then follows an extensive set of quality control checks; these are documented in the database. Finally, documentation is automatically generated. When the process is complete the load tables are dumped to the encrypted file system for reference and then removed. It is therefore practicable to reload SAHSU data at intervals to enhance security and to add processing improvements to pre-existing data (e.g. improved data processing, enhanced security, improved pseudonymisation). It is normal SAHSU practice to reload datasets each time a fresh year is received to ensure the latest processing is uniformly applied to all data. SAHSU operates a hierarchy of data access permission based on user role: 1. General level access to aggregated health data, such as that publicly available from data providers websites e.g. district level mortality counts; 2. SAHSU researcher with access to small area data that is not pseudonymised and non-sensitive; 3. SAHSU researcher level access to pseudonymised sensitive data where required for specific projects; 4. Database team access to identifiable information supplied by data providers e.g. to pseudonymise the data. All access to confidential data is password protected. Data may only be extracted by the database team, or by using a “self-select” tool called Rapid Inquiry Facility (RIF). Note that this tool allows the user to select the nature of data to be extracted, but only at in an aggregated form. The RIF is an automated tool that uses both database and Geographic Information System (GIS) technologies. The purpose of the RIF is to rapidly address epidemiological and public health questions using routinely collected health and population data. This allows SAHSU to respond rapidly, with expert advice to ad hoc queries from the funding departments about unusual clusters of disease, particularly in the neighbourhood of industrial installations. The RIF can perform risk analysis around putative hazardous sources and can be used for disease mapping. It generates standardised rates and relative risks for any given health outcome, for specified age and year ranges, for any given geographical area. This facility was initially designed as a tool for SAHSU staff to analyse routinely collected health data in relation to environmental exposures in the European Health and Environment Information System (EUROHEIS) project and has also been used by SAHSU to provide aggregated information as part of the US Centers for Disease Control (CDC) environmental public health tracking. In all cases data is extracted using the Rapid Enquiry Facility (RIF) or by the database team, and all extracts are supervised by the database manager. All data extracts are logged and cross checked by the database manager prior to extraction. The RIF will also permit data extraction and will also enforce identical checks. The checks carried out prior to data extraction are: • Projects is approved by the SAHSU liaison committee; • User is under contract to Imperial; • If required, access to event data (date of birth and/or death) has been justified; • SAHSU confidentiality form has been signed ; • The user has been information governance inducted and trained. In addition to the data provided by the HSCIC, SAHSU hold the following record level identifiable datasets: • ONS Births and Still births • ONS Cancer Incidence • Welsh Cancer Intelligence and Surveillance Unit • ONS Mortality • National Congenital Anomaly Register (NCAR from ONS) • Local Congenital Anomaly registries affiliated with BINOCAR • Terminations grounds “E” • NN4B • NCCHD (National Community Child Health Database) Linkage of data between datasets is only permitted with: • Approval via a substantial amendment to SAHSU ethics approval • Approval via a substantial amendment from HRA CAG • Explicit written permission from the data providers concerned. To date, amendments to the s.251 support have been sought and granted for the following three projects requiring specific data linkage: 1. Traffic pollution and health in London; 2. Incinerators; 3. Small area variation in coronary heart disease incidence, mortality and survival and their risk factors and determinants in England. Any projects requiring linkage beyond that covered by this application would also be subject to an amendment for DAAG’s consideration. It would also require support from ethics, and be covered by an amendment to the existing s251. Data minimisation As part of this application, the data required has been rationalised and HES data currently held by SAHSU no longer covered by this agreement will be securely destroyed. The remaining years are required by a number of studies, but the totality of years of data is also required by a single project - a study relating to health at major airports. Whilst the initial study is complete, the data is required to be retained in order to respond to any queries relating to the research (a common requirement for published research). National data is also required in order to provide the ad hoc rapid response service to Public Health England. Given that PHE coverage is across England, and the requirements cannot be predicted in advance, national data is required given that response times do not permit time to request, receive and process individual extracts of data.

Objectives:

Objective for processing: The Small Area Health Statistics Unit (SAHSU) is a long-standing and internationally-recognised centre of excellence assessing the risk of exposure to environmental pollutants to the health of the population, with an emphasis on the use and interpretation of routine health statistics at small-area level. SAHSU was established in 1987 as a recommendation of the Black enquiry into the incidence of leukaemia and lymphoma in children and young adults near the Windscale/Sellafield nuclear power plant. SAHSU has a particular role nationally in carrying out environmental health surveillance of the population in relation to environmental contaminants and point sources of industrial pollution, based on routinely collected health data. This is a highly specialised area of work requiring excellence in computing, database management, geographical information systems (GIS), statistics, environmental exposure assessment and epidemiology. The set of skills and expertise that has been established and built up in SAHSU is a unique resource both nationally and worldwide. SAHSU has a programme of work established which is defined by the following terms of reference :- 1) To develop and maintain databases of health data, environmental exposures as required to meet specific need, and social confounding factors at the small area level; 2) To carry out substantive research studies on environment and health issues including studies of the relationship between socio-economic factors and health, in collaboration with other scientific groups as necessary; 3) In collaboration with other scientific groups, to build up reliable background information on the distribution of environmental exposure, socio-economic data and disease amongst small areas; 4) To develop methodology for analysing and interpreting health outcomes related to small areas; 5) To act as a centre of expertise, disseminating information on developments in spatial epidemiological methods to national and regional groups; 6) To respond rapidly, with expert advice, to ad hoc queries from the core funding bodies (DH and PHE) about unusual clusters of disease, particularly in the neighbourhood of industrial installations To deliver against that Terms of Reference, SAHSU will utilise the data to meet the following purposes: Purpose 1 – maintenance of the SAHSU health research database At the core of the overall programme of work is the maintenance of the SAHSU research database, which combines multiple datasets and has both ethics approval and s251 approval in place Purpose 2 – to carry out a programme of research projects and studies. The research programme includes both methods development and investigation of priority questions in environment and health. A key aim is to improve the science base underlying translation of knowledge on the effects of the environment on health into policy. SAHSU conducts national research studies on environmental factors that may affect health ranging from exposures to electromagnetic fields (such as from electricity power lines) to traffic-related air pollution and noise, using nationally collected patient data including mortality, hospital admissions, cancer registrations and births data. Additional small area analyses are conducted that help support the general remit of the unit e.g. investigating differential hospital admission rates in ethnically diverse small areas. Additionally, SAHSU provides national expertise in cluster and small area statistical methods and has close links with Public Health England including input into their environmental public health tracking programme. Approval of individual projects All SAHSU studies are controlled via the PHE-SAHSU Liaison Committee. New study concepts must initially be approved by either the Director or Assistant director of SAHSU prior to the outline study proposal being created. Consideration is given to whether the study is adequately covered by SAHSU’s existing ethical approval and if not, separate ethical approval must be sought. Once ethical approval is confirmed the outline study proposal is reviewed by the SAHSU-PHE liaison committee who, once approved then take the study for formal minuted approval from the appropriate PHE programme board (attended by a member of the Department of Health). Projects are only approved where they are within the constraints of the SAHSU programme terms of reference. Purpose 3 – to provide rapid ad hoc support to PHE and DH about unusual clusters of disease, particularly in the neighbourhood of industrial installations Such work is carried out on the instruction of DH or PHE, and is approved by the Director of SAHSU. By its nature, such requirements cannot be detailed in advance, but the only outputs would be aggregate with small numbers suppressed, and provided to DH and PHE. SAHSU have a single s251 support in place to cover the above. Any project which may have additional requirements (for example to require the linkage of an additional dataset beyond that previously agreed) must seek an amendment to the existing s251. It would also be outside the scope of this application, and therefore an amendment would be put before DAAG for consideration.


Project 8 — DARS-NIC-278518-F3H0X

Opt outs honoured: N

Sensitive: Non Sensitive

When: 2016/12 — 2017/02.

Repeats: One-Off

Legal basis: Informed Patient consent to permit the receipt, processing and release of data by the HSCIC

Categories: Anonymised - ICO code compliant

Datasets:

  • Hospital Episode Statistics Critical Care
  • Hospital Episode Statistics Admitted Patient Care

Benefits:

1. To add to the accuracy of the cost-effectiveness and the incremental net benefit of an endovascular strategy for repair of ruptured abdominal aortic aneurysm: 2016-2017. 2. To benefit all future patients admitted to hospital with a diagnosis of ruptured abdominal aortic aneurysm. 3 To drive organisational changes, so that endovascular repair becomes available to all patients with ruptured abdominal aortic aneurysm

Outputs:

1. Clinical and cost-effectiveness of an endovascular strategy versus open repair at 3-years after repair of ruptured abdominal aortic aneurysm [for major journal publication, Health Technology Assessment (HTA) report and reporting to the National Institute for Health and Care Excellence (NICE) for April 2017. 2. The main findings to 3 years will be presented to The Vascular Society of Great Britain and Ireland on 1st December 2016. The major outcomes also will be reported at other national and international conferences (International VEITHsymposium in New York, The British Society of Endovascular Therapy and the European Society for Vascular Surgery), on the IMPROVE trial websites and via social media outlets such as Twitter. All outputs will consist of aggregate data only with small numbers suppressed in line with HES analysis guide.

Processing:

The data will be processed in the offices of the Vascular Surgery Research Group in Room 4N12 at Charing Cross Hospital (Imperial College Healthcare NHS Trust) on a non-networked dedicated computer with built-in BitLocker encryption (i.e. full disk encryption) in a secure office only accessible to the IMPROVE trial manager, who is employed by Imperial College London, but has a Research Passport and Letter of Access with Imperial College Healthcare NHS Trust. The sole trial manager has had IGT training and will comply with the principles of the Data Protection Act 1998 at all times when processing/storing personal information. Identifiable personal information will *only* be viewed by the trial manager on this specific computer (i.e. no mobile or remote working). HES datasets will not be saved on any other trial database in any other location. Once the data has been checked and cleaned, it will be fully anonymised on this same stand-alone computer and all patient identifying details removed (only the trial ID will be carried through during data analysis). Procedure codes received from HES will be summarised into clinically helpful categories (e.g. aorta and distal artery related, damaged bowel or abdominal wall related, etc.) and an anonymised file containing four digit Trial ID and these summary categories of complications will be analysed by the statistician, (based at the University of Cambridge) and two health economists (based at the London School of Hygiene and Tropical Medicine (LSHTM). Both the University of Cambridge and LSHTM have formal subcontracts with Imperial College London, to undertake the statistical and health economic analysis for the trial, respectively. Both the statistician and health economists will only view and analyse the anonymised data file at Charing Cross Hospital at the Imperial College site and under the controls/policies of Imperial College London.

Objectives:

To provide a robust alternative assessment of the use of health service resources by patients enrolled in IMPROVE (ISRCTN48334791), a randomised trial of an endovascular strategy versus open repair for patients with a clinical diagnosis of ruptured abdominal aortic aneurysm. Since the primary outcome of the trial, 30-day mortality, was not different between the two randomised groups (BMJ 2014;348:f7661), accurate longer-term clinical and cost-effectiveness evaluations of the 2 treatments have particular relevance for the NHS. Imperial College London's current source of information on the use of health services after hospital discharge comes primarily from the specialist vascular centres participating in the trial (where patients had their ruptured aneurysms repaired). Re-admissions to non-trial hospitals are not captured. To enhance the generalisability and robustness of clinical and cost-effectiveness evaluations, IMPROVE trial Management Committee wish to cross-check both aneurysm-related re-interventions and hospital re-admissions (any hospital for any reason) data, to validate and supplement the information collected from relevant trial hospitals with HES admissions/procedures data. The aneurysm-related re-interventions include procedures directly related to the aorta and distal arteries (L procedure codes) and procedures which arise as a result of damaged bowel or abdominal wall during aneurysm repair (including some G, H and T procedure codes). HES data for admissions, procedures, length of hospital stay are requested for specified patients, to provide details of hospital resources used between discharge and 3 years post-operative period. Ethical approvals are in place to support the collection of data. Admitted patient care and critical care datasets from 2009/10 (01/09/2009 onwards) to 2012/13 were received in 2015 as one-off files as part of this DSA and the data crosschecked against aneurysm-related re-interventions captured by participating hospitals: this covered 1-year follow-up for the majority of patients (randomised Sep-2009 to July-2013). Imperial College London identified two major re-interventions in the HES dataset (1 had occurred at a non-participating hospital and the other overlooked by one of the participating centres). After 3 years of follow-up, Imperial College London again need to cross check the data received from trial centres, this time separating re-interventions on the aorto-iliac and distal arteries from those due to damage to the abdominal viscera or aorta wall incurred by either the event or repair of the ruptured aneurysm. Imperial College London will submit details for patients who were recruited to the IMPROVE trial at England hospitals and were alive at primary discharge, with post-operative consent (around 350 patients). Imperial will provide the NHS number, Trial ID, sex, DOB and the date of discharge after index repair date for each patient. For these patients Imperial would like to request all hospital admission dates, discharge dates, diagnosis codes, provider codes, until 3 years following discharge from index repair (or death if this occurs before 3 years). Imperial would also like for these admissions to be linked to certain procedure codes to ascertain aneurysm-related re-interventions, if there were any. To be able to cover 3-year period for all patients Imperial request to receive admitted patient care and critical care datasets from 2009/10 to 2015/16 (current year needed to 22nd July 2016). Imperial plan on using Trial ID and HES patient identifier to link APC and CC datasets and do not require any sensitive/identifiable fields. The pseudonymised data received from NHS Digital will not be linked to identifiable data which is held on a separate unlinked database.


Project 9 — DARS-NIC-28095-S9N3P

Opt outs honoured: N, Y

Sensitive: Sensitive, and Non Sensitive

When: 2016/04 (or before) — 2016/11.

Repeats: Ongoing

Legal basis: Section 251 approval is in place for the flow of identifiable data, Health and Social Care Act 2012

Categories: Anonymised - ICO code compliant, Identifiable

Datasets:

  • Hospital Episode Statistics Accident and Emergency
  • Hospital Episode Statistics Admitted Patient Care
  • Hospital Episode Statistics Critical Care
  • Hospital Episode Statistics Outpatients

Benefits:

ICL DFU works with the Care Quality Commission (CQC), contributing to its surveillance remit using the same methods and data. The unit generates monthly mortality alerts since 2007, based on high thresholds [1]. This was pivotal in alerting the then Healthcare Commission (HCC) to problems at the Mid Staffordshire NHS Foundation Trust between July and November 2007[2]. The resulting Public Inquiry recognised the role that the unit’s surveillance system of mortality alerts had to play in identifying Mid Staffs as an outlier [3]. Key recommendations, [4] reflecting the unit’s work, are that all healthcare provider organisations should develop and maintain systems which give effective real-time information on the performance of each of their services, specialist teams and consultants in relation to mortality, patient safety and minimum quality standards. A further recommendation is that summary hospital-level mortality indicators should be recognised as official statistics [5]. If ICL DFU is given continued access to the data, this monitoring tool that detected Mid Staffs will continue to monitor patient outcomes at acute hospitals and be ready to detect any future outliers. The unit will be able to assist the investigation of variations in outcomes at a local level by providing Local Patient ID, NHS Number and Consultant Code from the unit’s analyses to authorised users within trusts to enable reconciliation with local information systems and the instigation of clinical audits and case note reviews. ICL DFU mortality outlier outputs are used by CQC within their Hospital Inspection framework.(on-going) As a result of the unit’s leading role in the development of hospital mortality measures, in 2010 ICL DFU was invited to contribute to a DoH Commissioned expert panel (Steering Group for the National Review of the Hospital Standardised Mortality Ratio) [6] to develop a national indicator of hospital mortality. The resultant Summary-level Hospital Mortality Indicator (based in part on their HSMR methods) is now a public indicator used by all acute trusts. [7] Professor Sir Bruce Keogh suggests that a relatively “poor” SHMI should trigger further analysis or investigation by the hospital Board. The recent review (published in July 2013) into the quality of care and treatment provided by 14 hospital trusts with consistently high mortality in either measure led to 11 out of the 14 trusts identified being immediately placed on special measures. The review also informs the way in which hospital reviews and inspections are to be carried out with the recommendation that mortality is used as part of a broad set of triggers for conducting future inspections [8]. ICL DFU continues to advise the Health and Social Care Information Centre on methodological issues around the Summary level Hospital Mortality Index (SHMI), and carry out analyses relating to this measure to assist in its development.(ongoing) The unit’s research on specific aspects of care has received a high media profile and has been highly cited. Their research on weekend mortality in emergency care, analysis of mortality associated with the junior doctor changeover and work on elective procedures and mortality by day of the week resulted in front page broad sheet coverage, and radio and TV interviews. (ongoing) https://www1.imperial.ac.uk/publichealth/departments/pcph/research/drfosters/inthemedia/ The unit’s “Out of hours” work has been a key driver in moving NHS towards 7/7 care. Headlines include, “NHS Services – open seven days a week: every day counts” and, “Sunday Times Safe Weekend Care”. As a result of the unit’s published research into the junior doctor changeover, Bruce Keogh introduced a week's shadowing where newly qualified doctors worked alongside more senior ones for a week before they start work in August. The Academy of Medical Royal Colleges published proposals (16th April 2014) suggesting all Foundation Year 1 posts should begin on the first Wednesday in August as has always been the case, but other training posts should begin in September.(on-going) As part of the ‘biggest bang per buck’ analysis, econometric modelling will suggest which elements of the patient pathway are the most costly. Combining this with modelling of variation by unit will suggest priorities for improvement. Outputs will benefit managers, commissioners and patients. (Dec 2017) Analyses of return to theatre and joint revision for elective hip and knee surgery will help orthopaedic surgeons, commissioners and patients understand these key quality markers for this specialty and devise appropriate improvement projects, for instance by determining which patients are at the highest risk and therefore need more rigorous follow-up. (on-going) ICL DFU intends to examine demand and capacity measures for A&E and admissions, and the impact that pressure on resources might have on safety and patient outcomes. By profiling hospital trusts in terms of demand, patient mix and outcomes, researchers will better understand key NHS metrics and patterns of service use and thereby help managers manage demand. (Jun 2017) Regarding the travel time analysis, using Lower Super Output Areas would enable us to study the effect of distance from home to hospital on patient outcomes. This also allows geographical access to services to be estimated, as researchers can calculate how far patients must travel for their treatment both now and after any future service reorganisation. (Dec 2017) ICL DFU analysis of their mortality alerting system will allow us to improve the alerting process and provide a better indication of how hospitals should investigate them to reduce mortality (including what are the key contributing factors to the alerts and to subsequent improvement in mortality by the hospitals). (Dec 2016) The modelling of health trajectories in stroke patients will improve risk stratification and understanding of the medium-term prognosis and needs. This will also allow better econometric modelling of NHS service use. (Jul 2018) References [1] CQC Quarterly publication of individual outlier alerts for high mortality: Explanatory text (URL available at http://www.cqc.org.uk/public/about-us/monitoring-mortality-trends) [2] Investigation into Mid Staffordshire NHS Foundation trust. Healthcare Commission 2009. Outcomes for patients and mortality rates. Pages 20 - 25 http://www.midstaffspublicinquiry.com/sites/default/files/Healthcare_Commission_report_on_Mid_Staffs.pdf [3] Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry 2013. Volume 1. Pages 458 - 466 http://www.midstaffspublicinquiry.com/report. [4] Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry 2013. Executive Summary. Recommendation 262: http://www.midstaffspublicinquiry.com/report). [5] Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry 2013. Executive Summary. Recommendation 271: http://www.midstaffspublicinquiry.com/report. [6] Development of the new Summary Hospital-level Mortality Indicator. Department of Health Website. http://www.dh.gov.uk/health/2011/10/shmi-update/ [7] Indicator Specification: Summary Hospital-level Mortality Indicator. http://www.ic.nhs.uk/SHMI [8] Review into the quality of care and treatment provided by 14 hospital trusts in England: overview report Professor Sir Bruce Keogh KBE. http://www.nhs.uk/NHSEngland/bruce-keogh-review/Documents/outcomes/keogh-review-final-report.pdf 2) Support the provision of a management information systems (Dr Foster Analysis Toolkit) for the NHS Expected benefits include: • Enabling NHS acute trusts to measure, compare and benchmark key quality indicator trends – focusing on risk-adjusted measures of mortality, readmissions and length of stay in hospital. • Providing evidence to instigate clinical audit and investigations related to quality of care, such as highlighting potential poor clinical coding or quality/efficiency concerns. • Validating other mortality indicators – such as HSMR, Cusum alerts and crude mortality. • Enabling NHS acute trusts and commissioners to use performance information to identify, quantify and act on opportunities to improve efficiency of health services. • Understanding areas of best practice amongst our customers and facilitate interactions with other customers who are not performing as well to support quality and efficiency improvement. • Helping clinicians and managers by providing independent and authoritative analysis of the variations that exist in acute hospital care in a way that is meaningful for them and that is understandable to patients and the public. • Highlighting topics of interest to the health industry and wider public to enable discussion and improvement in healthcare provision. • Publication of articles around variations of healthcare within the NHS is in the public interest and supports the government agenda for transparency by promoting choice and accountability within the NHS. • Maintaining the focus of the organisations on improvement. • Raising public and professional awareness through the Dr Foster Hospital Guide regarding issues that affect the quality and efficiency of care provided by the NHS by publishing new information about variation in outcomes at the level of individual hospitals. In recent years the guide has focussed on issues of clinical and managerial concern such as weekend care, overcrowding, management of chronic conditions and variations in access to elective care. In each case, the approach has been to identify effects that are known from the academic literature and to show their impact here and now in English NHS hospitals. By publishing this information Dr Foster Limited support the improvement of healthcare in England. How will these benefits be measured: Benefits are ongoing as the outputs described above are used within NHS Trusts’ internal monthly reporting and quality processes. Dr Foster Ltd services allow performance of NHS Provider Trusts to be monitored and trended over time and therefore provide customers with the ability to measure changes in quality and performance particularly in instances where customers have been alerted and they have worked with them to understand the causes of worse than expected performance. Dr Foster Ltd intends to provide an online customer survey within the Dr Foster Analytics Tool to capture customer feedback and associated benefits, this data will form the foundation for improving their services and enable them to provide HSCIC, and other relevant bodies, with tangible evidence to support their ongoing use of HES data. Dr Foster Limited welcomes the opportunity to work with HSCIC to ensure information captured can support our ongoing supply and use of HES data. When will these be achieved: As a majority of benefits are achieved on an ongoing basis, it is not possible to outline a specific target date for achievement of the benefits outlined as they are reliant on a range of factors outside of ICL DFU and Dr Foster Limited’s control. However, whenever there are areas of particular concern about performance against key indicators, the 2 parties act immediately to alert relevant stakeholders and offer their assistance in better understanding and addressing them.

Outputs:

1) Research into variations in quality of healthcare by provider: background to proposed work ICL DFU work programme is designed to develop and validate indicators of quality and safety of healthcare, show variations in performance by unit and socio-demographic stratum and develop methods for risk prediction, risk adjustment and outlier detection. The unit’s work focuses on quality of care and patient safety, including healthcare-acquired infections (surgical wound infections and urinary tract infections) and safety indicators. Collaborative projects with clinical colleagues have helped develop and validate healthcare quality indicators other than mortality, including bariatric surgery, primary angioplasty rates, indicators for stroke care, obstetric care, orthopaedic redo rates and returns to theatre. ICL DFU is currently working on the following analyses: ‘Biggest bang per buck’ elements of treatment pathways for chronic diseases. By mapping out NHS hospital contacts and modelling the variation across units, the unit will determine the elements (e.g. readmissions, missed OPD appointments, surgery that could have been done as a day case) with the most potential for improvement. This forms part of the unit’s work with Imperial’s NIHR funded Patient Safety Translational Research Centre on the use of information for service improvement. (Dec 2017) Drivers of unscheduled return to theatre (or reoperation) in elective hip and knee replacements: correlation between Return To Theatre (RTT) and revision rates by surgeon; volume-outcome relation for RTT; risk of RTT following revision rates. The objective is to better understand these key metrics for the specialty: revision rates are of major interest to surgeons and are on the NHS Choices website. The unit has recently established that there is greater non-random variation in RTT rates between surgeons than between hospitals. (on-going) Predictors of readmissions and A&E attendance in patients with chronic diseases (heart failure, COPD, cancer). Readmissions are the focus of much attention worldwide in efforts to reduce costs and improve outcomes, but little is known about the role of A&E attendance (not ending in admission) in observed variations in readmission rates. The study has revealed that earlier OPD nonattendance is a strong risk factor for readmission. The objective is again to better understand readmissions as an indicator and to suggest reformulation if desirable. (Jun 2017) Travel time. Due to the well-documented relation between patient volume and outcomes, there is a growing drive to centralise certain services such as for stroke and elective surgery. Treatment rates for many conditions such as thoracic aortic disease (TAD) vary around the country. Using Lower Super Output Areas of the patient’s residence and the hospital postcode, researchers will first calculate how far patients currently travel for their TAD treatment and then the travel distance that would be incurred were surgical services retained only at large centres. The effect on outcomes will also be assessed. (Dec 2017) Modelling Health trajectories for Stroke patients ICL DFU is currently undertaking a study which involves the evaluation of patients who had a stroke and following them up for 5 years. The study involves people who had a stroke for the first time. Previous studies have been criticised for including patients with recurrent stroke. Based on previous research, ICL DFU has tracked back their chosen stroke patients for 10 years to ascertain whether the stroke event under observation was the first or recurrent. Moreover, ICL DFU has to evaluate important cardiovascular co-morbidities by looking at the patients hospital diagnosis made in the previous years. The study aims to identify stroke patients who are initially stable but later become high users of health care resources. ICL DFU also plans to look at pattern of causes of subsequent hospitalisation in the same cohort of patients. The study requires tracking back patients 10 years and following up for 5 years from the time of their index stroke event. (Jul 2018) Recent pressures on A&E and breaches of the 4-hour wait have led to concerns over pressure on A&E and inpatient capacity. ICL DFU intends to examine capacity measures for A&E and inpatient admissions, and the impact that pressure on resources might have on safety and patient outcomes with a view to better understanding key NHS metrics and patterns of service use to better match supply to need. (Dec 2016) ICL DFU is working in collaboration with the University of Manchester and supported by the Care Quality Commission, to improve understanding of the unit’s mortality alerts and to evaluate their impact as an intervention to reduce avoidable mortality within English NHS hospital trusts, focusing on two conditions commonly attributed to mortality alerts acute myocardial infarction and septicaemia. The aim of this study is to provide a descriptive analysis of all alerts, their relationships with other measures of quality and their impact on reducing avoidable mortality. (Dec 2016) International comparisons of service use and outcomes. England and the USA. The unit holds data from Centre for Medicare and Medicaid Services enrollees and from the Nationwide Inpatient Sample from the USA. Researchers have previously set out the methodological issues with using administrative data from multiple countries. This study will compare patient casemix, rates of outcomes such as infections and readmissions, and rates of surgery, for example in patients near the end of their life (overtreatment is a growing concern) between the two countries. The objective is to highlight areas of better or poorer performance by the NHS compared with the USA. ICL DFU has an extract of the Italian data and will be using HES data to compare hospital use for patients with heart failure in England compared with Italy. (on-going) Examples of key published research that have used HES data include: Palmer WL, Bottle A and Aylin P. Association between day of delivery and obstetric outcomes: observational study. BMJ 2015; 351: h5774. Bottle A, Goudie R, Cowie MR, Bell D, Aylin P, 2015, Relation between process measures and diagnosis-specific readmission rates in patients with heart failure, HEART, Vol: 101, Pages: 1704-1710, ISSN: 1355-6037 Aylin P; Alexandrescu R; Jen MH; Mayer EK; Bottle A. Day of week of procedure and 30-day mortality for elective surgery: retrospective analysis of hospital episode statistics. BMJ 2013;346:f2424. Palmer WL; Bottle A; Davie C; Vincent CA; Aylin P. Dying for the Weekend: A Retrospective Cohort Study on the Association Between Day of Hospital Presentation and the Quality and Safety of Stroke Care. Arch Neurol. 2012;69:1296-1303. Aylin P; Bottle A; Majeed A. Use of administrative data or clinical databases as predictors of risk of death in hospital: comparison of models. BMJ 2007;334:1044. Aylin P, Yunus A, Bottle A, Majeed A, Bell D. Weekend mortality for emergency admissions. A large, multicentre study. Qual Saf Health Care. 2010;19:213-217 Jen MH, Bottle A, Majeed A, Bell D, Aylin P. Early in-hospital mortality following trainee doctors' first day at work. PLoS One. 2009;4:e7103. For full publication list see unit website: http://www1.imperial.ac.uk/publichealth/departments/pcph/research/drfosters/unit_publications/ 2) Support the provision of a management information systems (Dr Foster Analysis Toolkit) for the NHS Dr Foster Limited is an independent healthcare information company. It provides a research grant to our unit to develop indicators and methodologies to assist in the analysis of healthcare performance. ICL DFU works in collaboration with Dr Foster Limited to provide the NHS with a number of management information systems via the Dr Foster Analysis Toolkit. The main output created are benchmarked or standardised healthcare indicators & analysis such as mortality (SHMI/HSMR), LOS(Length of Stay), admission trends, readmission rates, patient safety indicators, referral patterns, market share analysis etc. As stated previously, outputs are to be used solely for the purposes of providing a management information function to the NHS. Outputs are provided via: • Dr Foster Analysis Toolkit – Use of Role Based Access to determine the level of data end users can see within the tool. • Value added services - Tabulations, Reports, Spreadsheets, Presentations, Articles & Projects. Outputs will be used by customers to investigate Clinical Quality, Performance and Business Development, specifically: • Assess and manage clinical quality and patient safety within NHS Organisations • Identify pathways where there is potential for improvement • Identify areas of best practice either within the Provider Trust or local/national health economies • Better understand how they compare to other Provider Trusts with similar case mixes • Identify improvements in operational efficiency • Understand patient outcomes • Identify and understand market activity • Monitor the impact of implemented changes • Identify variations in outcomes

Processing:

ICL DFU uses hospital administrative data in the form of HES/MMES to identify measures of quality and safety of healthcare. The unit’s work focuses on quality of care and patient safety, including healthcare-acquired infections, mortality and safety indicators. ICLDFU holds 2 databases to store data – A Research database and a Patient Identifiable database to provide a Re-Identification service for NHS provider trusts. Patient identifiers are stored separately to the unit’s research database which holds the standard HES extracts and sensitive fields. Imperial’s researchers have no access to identifiable fields. Only two named data managers have access to the patient identifiable fields within the unit. The purpose of holding the patient identifiers for the last 3 years is to allow hospitals to further investigate any alerts around poor or good performance and to help improve the quality and safety of healthcare delivery. The standard HES extracts and sensitive fields are stored in the Research database where researchers are able to access the data to do their analyses. The standard extracts are loaded on to the Research database with a unique identifier (fosid) being generated and added to the datasets. A new Extract_hesid (for Dr Foster Limited) is also generated using the SHA-256 hashing algorithm, compliant with the e-GIF Technical Standards Catalogue Version 6.2 based on the original Extract_hesid. An extract is taken from ICL DFU patient identifier server and copied to the server which is used to provide the Re-Identification service for the NHS Acute Trusts. Further data processing are carried out on the onward supply of data by Dr Foster Ltd who have dedicated staff and processes as per below: • Linkage into spells and superspells, which can often span across financial years • HRG, Tariff and other PBR related fields, using the HRG Grouper software • Various clinical groupings, including CCS Diagnoses, Ambulatory Care Sensitive (ACS) conditions and Procedure Groups • Quality outcomes, including mortality, emergency readmission within 28 days, Long Length of stay and patient safety indicators • Patient-level predicted risks for these outcomes, based on national Logistic Regression models which are executed using R statistical software and updated monthly • Various other national benchmarks, including Length of stay percentiles and Standardised Admission Ratio benchmarks • Numerous efficiency-based metrics, including average length of stay, day case rate and potential bed days saved • Prescribed Specialised Services (PSS) groups, using the PSS Grouper software This process guarantees both Dr Foster Limited and ICL DFU are working from exactly the same data (both in terms of underlying patient linkage and derived fields), which is necessary for their joint projects. No record level data will be transferred outside of the EEA, either under this agreement or any related sub-licence.

Objectives:

ICL DFU uses HES data to identify measures of quality and safety in healthcare. Their research themes are around developing and validating indicators of quality and safety of healthcare, particularly by GP practice, consultant, and NHS Trust, showing variations in performance by unit, patient risk subgroups and risk prediction, risk adjustment and outlier detection for such indicators and variations and any other methodological aspects as they arise. Refer to section (Expected measurable benefits to health and/or social care including target date) to demonstrate the benefits ICL DFU work have brought to the Health and Social Care. Patient identifiers The Regulation 5 of the Health Service (Control of Patient Information) Regulations 2002 (s251) support letter confirms the final approval to receive confidential patient information for ICL DFU research database and identifiers to provide re-identification service to DFI customers and ALL NHS trusts. Identifiable data processed under CAG [15/CAG/0005] will be retained for a maximum of three years after which it should be destroyed or irreversibly pseudonymised on a rolling basis. The purpose of holding the patient identifiers is to allow hospitals to further investigate any alerts around poor or good performance and to help improve the quality and safety of healthcare delivery. ICL DFU does this by providing a re-identification service to acute NHS providers who are Dr Foster Limited’s customers. Dr Foster Limited has no access to the patient re-identification service. No patient identifiers will ever be passed to Dr Foster Limited or any other organisation except the NHS provider trust from where the data originated. For this purpose, we have developed a re-identification service whereby authorised individuals within NHS Provider Trusts are able to identify their own patients indicated in the Dr Foster Analysis Toolkit. This service allows us to supply Provider trusts’ NHS Number and LOPATID using Dr Foster Analysis Toolkit without passing these fields on to Dr Foster Limited. The re-identification service is maintained by ICL DFU. Sensitive fields Sensitive fields will only be available at a record level to NHS Provider Trusts (or approved regulatory bodies with express authority to demand such data, e.g. the CQC) and are specifically required for the purpose of conducting root cause analysis where there is a legitimate relationship with the patient. Where a legitimate relationship does not exist data will be available at an aggregate level in line with HSCIC HES Analysis Guide, HSCIC Small Numbers Procedure and ONS Guidelines, with any sensitive fields suppressed. Consultant Code ICL DFU and Dr Foster Limited provide consultant from our analyses to authorised users within trusts to enable reconciliation with local information systems and the instigation of clinical audits and case note reviews. Analyses by consultant activity are fed back to the NHS through a range of Management Information Systems provided by Dr Foster Limited in the forms of aggregation of teams into 'departments' or other hierarchies. Requirements for analyses by consultant activity are consistent with NHS needs and policy direction (to publish at consultant level). Consultant code is also used in research e.g. analysing volume and outcome relations for elective surgery. Some exclusions are applied e.g. Invalid codes, dental consultant etc. Patient’s general medical practitioner Patient’s general medical practitioner is used to examine variations by GP practice and to enable mapping to practice level such as The Quality and Outcomes Framework (QOF) and practice staffing data etc. NHS Provider Trusts are able to identify the registered GP who referred the patient. This is essential to understanding rates of admission and rates of readmission by GP practice which may reflect issues of community and primary care. Person referring patient Analyses by the person referring patient activities are fed back to the NHS Provider Trusts through a range of Management Information Systems provided by Dr Foster Limited. These analyses allow NHS Provider Trusts to identify the person who referred the patient for calculation of referral rates. Understanding referral rates by GP practice and consultant can help to identify issues of quality of primary care. ICL DFU is part-funded by a grant from Dr Foster Limited. On approval of this application, a sub-licence model between the HSCIC and Imperial College similar to that previously in place, which permits Imperial to supply derived pseudonymised data together with specific clear text sensitive fields (as stated within this application) to Dr Foster Limited. The unit works in collaboration with Dr Foster Limited to provide a management information function in the form Dr Foster Analysis Toolkit. This purpose is fulfilled by analysis of HES data made available to customers via the following services provided by Dr Foster Ltd: 1. Licensed subscriber of Dr Foster Analysis Toolkit a. Directly – i. NHS Provider Trust holding a subscription to the Dr Foster Analysis Toolkit are able to view data at a record level, with an option to use the patient re-identification service for approved individuals; or ii. other NHS organisations holding a subscription to Dr Foster Analysis Toolkit are able to view aggregated analysis to prevent any patients being identified in accordance with guidance provided by HSCIC. b. Indirectly – analyses are provided to NHS organisations via a non-NHS organisation that holds a subscription to the tool, only being able to view aggregate small number suppressed data. 2. Value Added Services As an information intermediary, Dr Foster Ltd responds to customer requests for analyses of HSCIC data, whose scopes are by their nature bespoke and customised to local needs. An established specialist team of Analysts provides statistical analysis for interpreting complex data and producing analysis on behalf of customers. It should be stated that this team, which is project based, conduct annual training on handling sensitive records and are highly conversant in national guidelines to protect patient confidentiality, where there is any doubt the Dr Foster Ltd Head of Information Governance or SIRO will provide guidance and if required contact HSCIC. Dr Foster limited also provides analysis for publication for the benefit of the public and NHS e.g. Hospital Guide, and to support benefit to health and social care. Such analytical content may be published directly by Dr Foster Ltd or within academic journals or articles to journalistic/media entities in the form of text, tables, and other data visualisation such as diagrams/graphs using aggregate information based on HES analysis. Dr Foster Ltd is aware that publications, whether inside or outside the NHS, must adhere to strict guidelines in terms of disclosure, and will ensure any such publications are aggregated and comply with small number suppression in line with the HES Analysis Guide/ONS Guidelines and other relevant legislation and standards as defined by Schedule 3 of the Data Sharing agreement.


Project 10 — DARS-NIC-287804-H1T1R

Opt outs honoured: N

Sensitive: Sensitive, and Non Sensitive

When: 2016/04 (or before) — 2017/02.

Repeats: Ongoing

Legal basis: Informed Patient consent to permit the receipt, processing and release of data by the HSCIC

Categories: Identifiable

Datasets:

  • MRIS - Cause of Death Report
  • MRIS - Cohort Event Notification Report
  • MRIS - Members and Postings Report

Benefits:

1. To enable discrimination between aneurysm-related and other causes of death in the mid-term, as well as full patient follow up for mortality, following endovascular or open repair of ruptured abdominal aortic aneurysm: 2015-2017. 2. To benefit all future patients admitted to hospital with a diagnosis of ruptured abdominal aortic aneurysm (from 2015) 3. To drive organizational changes, so that endovascular repair becomes available to all patients with ruptured abdominal aortic aneurysm (from 2015)

Outputs:

Mortality following endovascular strategy versus open repair 3-years after repair of a rupture of an abdominal aortic aneurysm [for major journal publication, HTA report and reporting to NICE] for December 2016 This major outcome also will be reported at conferences and via relevant charities, patient groups and social media outlets.

Processing:

The data will be processed in the offices of the Vascular Surgery Research Group in Room 4N12 at Charing Cross Hospital (Imperial College Healthcare NHS Trust) on a non-networked dedicated computer with built-in BitLocker encryption (i.e. full disk , encryption) password-protected computer in a secure office only accessible to the trial manager, who is employed by Imperial College London, but has a Research Passport and Letter of Access with Imperial College Healthcare NHS Trust. The trial manager has had IGT training and will comply with the principles of the Data Protection Act 1998 at all times when processing/storing personal information. Identifiable personal information will *only* be viewed by the trial manager on this specific computer (i.e. no mobile or remote working) Data will then be fully pseudonymised by removing all identifiable information, and a file containing four digit Trial ID and category (10 used) of underlying cause of death will be analysed by the statistician, who is based at the University of Cambridge, but he will carry out the analysis at Imperial College Charing Cross Campus under the controls and policies of Imperial College).

Objectives:

To facilitate complete reporting of mortality for a randomised trial of an endovascular strategy versus open repair for patients with a clinical diagnosis of ruptured abdominal aortic aneurysm (IMPROVE ISRCTN48334791), which is supported by the National Institute for Health Research. The primary outcome of this trial is 30-day mortality and secondary outcomes are 1 year and 3 year mortality and aneurysm-related mortality. The most recent data file was received for these patients in December 2015. To enable full patient follow up for a minimum of 3 years, data are needed for deaths through July 2016 (last patient in the study will complete 3-year follow-up on 21st July 2016). The trial included critically ill patients often in great pain. Either patients or a relative/carer/welfare guardian on behalf of the patient have signed a consent form before aneurysm repair. After recovery patients provided fully informed consent for continued participation in the trial, which included consent to access “their data from NHS Information Centre and NHS Central Register NHS information”.


Project 11 — DARS-NIC-302604-S7H2N

Opt outs honoured: N

Sensitive: Sensitive, and Non Sensitive

When: 2016/04 (or before) — 2017/05.

Repeats: Ongoing

Legal basis: Informed Patient consent to permit the receipt, processing and release of data by the HSCIC

Categories: Identifiable, Anonymised - ICO code compliant

Datasets:

  • MRIS - Cause of Death Report
  • MRIS - Scottish NHS / Registration
  • MRIS - Cohort Event Notification Report
  • MRIS - Members and Postings Report

Objectives:

The data supplied by the NHS IC to University of Cambridge will be used only for the approved Medical Research Project identified above.


Project 12 — DARS-NIC-345991-H2F5N

Opt outs honoured: N, Y

Sensitive: Sensitive

When: 2016/04 (or before) — 2016/08.

Repeats: Ongoing

Legal basis: Section 251 approval is in place for the flow of identifiable data

Categories: Anonymised - ICO code compliant, Identifiable

Datasets:

  • Hospital Episode Statistics Accident and Emergency
  • Hospital Episode Statistics Admitted Patient Care
  • Hospital Episode Statistics Critical Care
  • Hospital Episode Statistics Outpatients

Benefits:

1) Ongoing benefits from our previous work The Imperial unit’s methodological research forms the basis of a near Real-Time Monitoring System called Quality Investigator (QI) produced by Dr Foster Intelligence, currently used by 70% of English NHS acute trusts to assist them in monitoring a variety of casemix adjusted outcomes at the level of diagnosis group and procedure group [1]. Dr Foster Intelligence is an independent healthcare information company to develop indicators and methodologies to assist in the analysis of healthcare. It provides a research grant to the unit to develop indicators and methodologies to assist in the analysis of healthcare performance. The unit works with the Care Quality Commission (CQC), contributing to its surveillance remit using the same methods and data. We generate monthly mortality alerts, based on high thresholds [2], which we have been running since 2007. This was pivotal in alerting the then Healthcare Commission (HCC) to problems at the Mid Staffordshire NHS Foundation Trust between July and November 2007[3]. The resulting Public Inquiry recognised the role that our surveillance system of mortality alerts had to play in identifying Mid Staffs as an outlier [4]. Key recommendations, [5] reflecting our unit’s work, are that all healthcare provider organisations should develop and maintain systems which give effective real-time information on the performance of each of their services, specialist teams and consultants in relation to mortality, patient safety and minimum quality standards. A further recommendation is that summary hospital-level mortality indicators should be recognised as official statistics [6]. If Imperial are given continued access to the data, this monitoring tool that detected Mid Staffs will continue to monitor patient outcomes at acute hospitals and be ready to detect any future outliers. Imperial will be able to assist the investigation of variations in outcomes at a local level by providing Local Patient ID, NHS Number and Consultant Code from our analyses to authorised users within trusts to enable reconciliation with local information systems and the instigation of clinical audits and case note reviews. Our mortality outlier outputs are used by CQC within their Hospital Inspection framework. As a result of our leading role in the development of hospital mortality measures, in 2010 we were invited to contribute to a DoH Commissioned expert panel (Steering Group for the National Review of the Hospital Standardised Mortality Ratio) [7] to develop a national indicator of hospital mortality. The resultant Summary-level Hospital Mortality Indicator (based in part on our HSMR methods) is now a public indicator used by all acute trusts. [8] Professor Sir Bruce Keogh suggests that a relatively “poor” SHMI should trigger further analysis or investigation by the hospital Board. The recent review (published in July 2013) into the quality of care and treatment provided by 14 hospital trusts with consistently high mortality in either measure led to 11 out of the 14 trusts identified being immediately placed on special measures. The review also informs the way in which hospital reviews and inspections are to be carried out with the recommendation that mortality is used as part of a broad set of triggers for conducting future inspections [9]. We continue to advise the Health and Social Care Information Centre on methodological issues around the Summary­level Hospital Mortality Index (SHMI), and carry out analyses relating to this measure to assist in its development. The unit’s research on specific aspects of care has received a high media profile and has been highly cited. Our research on weekend mortality in emergency care, analysis of mortality associated with the junior doctor changeover and work on elective procedures and mortality by day of the week resulted in front page broad sheet coverage, and radio and TV interviews (http://www1.imperial.ac.uk/publichealth/departments/pcph/research/drfosters/inthemedia/). The unit’s “Out of hours” work has been a key driver in moving NHS towards 7/7 care. Headlines include, “NHS Services – open seven days a week: every day counts” and, “Sunday Times Safe Weekend Care”. As a result of our published research into the junior doctor changeover, Bruce Keogh introduced a week's shadowing where newly qualified doctors worked alongside more senior ones for a week before they start work in August. The Academy of Medical Royal Colleges published proposals (16th April 2014) suggesting all Foundation Year 1 posts should begin on the first Wednesday in August as has always been the case, but other training posts should begin in September. 2) Expected benefits from the proposed work For our future research, analyses of return to theatre for elective hip and knee surgery will help orthopaedic surgeons, commissioners and patients understand these key quality markers for this specialty and devise appropriate improvement projects. We intend to examine capacity measures for A&E and admissions, and the impact that pressure on resources might have on safety and patient outcomes with a view to better understanding key NHS metrics and patterns of service use to better match supply to need. As part of the ‘biggest bang per buck’ analysis, econometric modelling will suggest which elements of the patient pathway are the most costly. Combining this with modelling of variation by unit will suggest priorities for improvement. Outputs will benefit managers, commissioners and patients. Previous work using HES showed higher mortality risk for asthma in those living in areas further from a hospital than those near it. Using Lower Super Output Areas would enable studies into the effect of distance from home to hospital on patient outcomes and the estimation of hospital catchment areas. This allows geographical access to services to be estimated, as we can calculate how far patients must travel for their treatment. Using larger geographical areas than LSOAs would incur too much measurement error. Our analysis on our mortality alerting system will allow us to improve the alerting process, and provide a better indication of how to investigate them (including what are the key contributing factors in the alerts). Our development of maternity indicators are expected to help monitor quality, and if similar findings around weekend differences in outcomes are discovered, may help to drive improvements in this area. References [1] Real Time Monitoring (RTM). Enabling providers and commissioners to benchmark and monitor clinical outcomes. http://drfosterintelligence.co.uk/solutions/nhs-hospitals/real-time-monitoring-rtm/ [2] CQC Quarterly publication of individual outlier alerts for high mortality: Explanatory text (URL available at: http://www.cqc.org.uk/public/about-us/monitoring-mortality-trends) [3] Investigation into Mid Staffordshire NHS Foundation trust. Healthcare Commission 2009. Outcomes for patients and mortality rates. Pages 20 - 25 http://www.midstaffspublicinquiry.com/sites/default/files/Healthcare_Commission_report_on_Mid_Staffs.pdf [4] Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry 2013. Volume 1. Pages 458 - 466 http://www.midstaffspublicinquiry.com/report. [5] Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry 2013. Executive Summary. Recommendation 262: http://www.midstaffspublicinquiry.com/report). [6] Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry 2013. Executive Summary. Recommendation 271: http://www.midstaffspublicinquiry.com/report. [7] Development of the new Summary Hospital-level Mortality Indicator. Department of Health Website. http://www.dh.gov.uk/health/2011/10/shmi-update/ [8] Indicator Specification: Summary Hospital-level Mortality Indicator. http://www.ic.nhs.uk/SHMI [9] Review into the quality of care and treatment provided by 14 hospital trusts in England: overview report Professor Sir Bruce Keogh KBE. http://www.nhs.uk/NHSEngland/bruce-keogh-review/Documents/outcomes/keogh-review-final-report.pdf

Outputs:

1) Research into variations in quality of healthcare by provider: background to proposed work The Dr Foster Unit at Imperial College use hospital administrative data in the form of HES/MMES to provide measures of quality and safety of delivery of healthcare by provider, or in some instances, by area or time. The unit’s work focuses on quality of care and patient safety, including healthcare acquired infections (surgical wound infections and urinary tract infections) and safety indicators. Collaborative projects with clinical colleagues have helped develop and validate healthcare quality indicators other than mortality, including bariatric surgery, primary angioplasty rates, indicators for stroke care, obstetric care, orthopaedic redo rates and returns to theatre. The proposed work will continue Imperial’s principal themes: i) developing and validating indicators of quality and safety of healthcare, particularly by consultant and hospital; ii) show variations in performance by unit and socio demographic stratum; iii) risk prediction and risk adjustment of such indicators and variations and any other methodological aspects as they arise. Imperial plan the following analyses: ‘Biggest bang per buck’ elements of treatment pathways for chronic diseases. By mapping out NHS hospital contacts and modelling the variation across units, we will determine the elements (e.g. readmissions, missed OPD appointments, surgery that could have been done as a day case) with the most potential for improvement. This forms part of our work with Imperial’s NIHR-funded Patient Safety Translational Research Centre on the use of information for service improvement. Oct 2017. Drivers of return to theatre (reoperation: RTT) in elective hip and knee replacements: correlation between RTT and revision rates by surgeon; volume-outcome relation for RTT; risk of RTT following revision rates. The objective is to better understand these key metrics for the specialty: revision rates are of major interest to surgeons and are on the NHS Choices website. Dec 2015 Predictors of readmissions and A&E attendance in patients with chronic diseases (heart failure, COPD, cancer). Readmissions are the focus of much attention worldwide in efforts to reduce costs and improve outcomes, but little is known about the role of A&E attendance (not ending in admission) in observed variations in re-admission and re-operation rates. We have recently found that earlier OPD non-attendance is a strong risk factor for readmission. Previous frequent emergency admissions are also highly predictive. The objective is again to better understand readmissions as an indicator and to suggest reformulation if desirable. Nov 2015. Imperial intend to examine capacity measures for A&E and admissions, and the impact that pressure on resources might have on safety and patient outcomes with a view to better understanding key NHS metrics and patterns of service use to better match supply to need. This will require a historical perspective, to look at changes to health service policy, and its impact on capacity. Jun 2016 International comparisons of service use and outcomes: England and the USA. We hold data from Centre for Medicare and Medicaid Services enrolees and from the Nationwide Inpatient Sample from the USA. We have previously set out the methodological issues with using administrative data from multiple countries. We will compare patient casemix, rates of outcomes such as infections and readmissions, and rates of surgery, for example in patients near the end of their life (overtreatment is a growing concern) between the two countries. The objective is to highlight areas of better or poorer performance by the NHS compared with the USA. Oct 2017. Imperial are working in collaboration with Professor Aneez Esmail from the University of Manchester and supported by the CQC, to improve understanding of our mortality alerts and to evaluate their impact as an intervention to reduce avoidable mortality within English NHS hospital trusts, focusing on two conditions commonly attributed to mortality alerts - acute myocardial infarction and septicaemia. We aim to provide a descriptive analysis of all alerts, their relationships with other measures of quality and their impact on reducing avoidable mortality. Oct 2016. Imperial are testing a hypothesis that pregnant women who undergo non-obstetric surgery have an increased risk of adverse pregnancy outcomes compared with those not undergoing surgery. We propose to analyse data collected between 2002 and 2013 and identify patients who underwent non- obstetric surgery whilst pregnant. Previous years of data are used to validate and complement maternal parity counts. A preliminary analysis suggests that we will be able to identify around 85,000 such patients out of a total of 4 million pregnancies. We aim to investigate adverse pregnancy outcomes occurring in this group; outcomes we will analyse include miscarriage, stillbirth, preterm labour, low birth weight, prolonged length of neonatal stay and neonatal death prior to discharge from hospital. With the data obtained from our study and subsequent statistical analysis, we aim to examine variation in practice and outcomes, and provide an evidence base with which we can counsel women who face the prospect of undergoing surgery during pregnancy. April 2015. Imperial are developing indicators of maternity care, based on HES which include perinatal mortality, complications following birth, caesarean rates and perineal tears. Imperial propose to examine variation by trust, and by day of the week. September 2015. Examples of key published research that have used HES/SUS data include: Aylin P; Alexandrescu R; Jen MH; Mayer EK; Bottle A. Day of week of procedure and 30 day mortality for elective surgery: retrospective analysis of hospital episode statistics. BMJ 2013;346:f2424. Palmer WL; Bottle A; Davie C; Vincent CA; Aylin P. Dying for the Weekend: A Retrospective Cohort Study on the Association Between Day of Hospital Presentation and the Quality and Safety of Stroke Care. Arch Neurol. 2012;69:1296-1303. Aylin P; Bottle A; Majeed A. Use of administrative data or clinical databases as predictors of risk of death in hospital: comparison of models. BMJ 2007;334:1044. Aylin P, Yunus A, Bottle A, Majeed A, Bell D. Weekend mortality for emergency admissions. A large, multicentre study. Qual Saf Health Care. 2010;19:213-217 Jen MH, Bottle A, Majeed A, Bell D, Aylin P. Early in-hospital mortality following trainee doctors' first day at work. PLoS One. 2009;4:e7103. For full publication list see unit website: http://www1.imperial.ac.uk/publichealth/departments/pcph/research/drfosters/unit_publications/ 2. Supporting a management information systems for the NHS Dr Foster Intelligence is an independent healthcare information company. It provides a research grant to our unit to develop indicators and methodologies to assist in the analysis of healthcare performance. We work in collaboration with Dr Foster Intelligence to provide the NHS with a number of management information systems including: • Quality Investigator (QI) ­ A web-based solution that monitors quality outcomes and patient safety in NHS trusts by assessing clinical, process and coding factors. It is currently used by 70% of acute provider trusts and 50 CCGs (CCGs are not provided with the patient records module, and small numbers are suppressed, hence have no access to the re­identification service) in assisting them in monitoring a variety of casemix adjusted outcomes at the level of diagnosis group and procedure group. • Practice and Provider Monitor (PPM) ­ A strategic planning tool and a joint information resource for providers and commissioners which enables users to quickly identify opportunities for improving operational and clinical outcomes. • TrustView – A dashboard developed in collaboration with NHS organisations, to provide a top level view of benchmarked trust performance around key clinical quality and clinical efficiency metrics. It gives high level, pertinent and timely data on overall trust performance to senior decision makers within trusts. • Care Quality Tracker ­ An online quality monitoring solution designed with and for NHS acute trusts that acts as an early warning system. • Mortality Comparator ­ Compare the two leading mortality indicators in England – SHMI and HSMR. Uncover, investigate and understand variations against peers. Imperial provide standard pseudonymised data extracts about the health care and treatment patients have received in any English NHS hospital in the form of Hospital Episodes Statistics – inpatient and day case admissions, outpatient appointments and Accident and Emergency attendances to Dr Foster Intelligence. These data are supplied by the Health and Social Care Information Centre (HSCIC) to Imperial under license and approved through HSCICs own Data Access Advisory Group (DAAG). The HSCIC has agreed to a sublicense between Imperial and Dr Foster Intelligence. The license permits Dr Foster Intelligence to provide products or services based on the standard extract only to public bodies (including NHS, CQC, Monitor, TDA, DH, PHE and local authorities) with appropriate small number suppression. It prohibits Dr Foster Intelligence from selling services or products derived from the data to commercial companies. Imperial can confirm that no identifiers will ever be disclosed to Dr Foster Intelligence (DFI); in particular LOPATID and NHS number will not be disclosed to DFI. Imperial provide a re­identification service whereby authorised individuals within NHS Provider Trusts are able to identify their own patients indicated in the DFI tools. This service allows Imperial to supply Provider trusts’ NHS Number and LOPATID using DFI tools without passing these fields on to DFI. The re­identification service is maintained by Imperial College. DFI have no access to the data held on it.

Processing:

The Dr Foster Unit at Imperial College use hospital administrative data in the form of HES/MMES to provide measures of quality and safety of delivery of healthcare by provider, or in some instances, by area or time. The unit’s work focuses on quality of care and patient safety, including healthcare acquired infections (surgical wound infections and urinary tract infections) and safety indicators. We plan to store the identifiers LOPATID and NHS number separately to our research database which will include the standard HES extracts and sensitive fields. Imperial’s own researchers have no access to identifiable fields. Only two named data managers will have access to the patient identifiable fields within the unit. An extract will be generated from the patient identifiable database and will be loaded to the Re-identification server to provide the service described below. Current CAG approval allows us to hold identifiers (LOPATID and NHS number) for inpatients admitted to NHS provider trusts who are customers of DFI (Dr Foster Limited, trades as Dr Foster Intelligence, registered Company No. 3812015). We require sensitive fields for all records. Sensitive fields Consultant Code We provide consultant from our analyses to authorised users within trusts to enable reconciliation with local information systems and the instigation of clinical audits and case note reviews. Analyses by consultant activity are fed back to the NHS through a range of Management Information Systems provided by Dr Foster Intelligence (DFI) in the forms of aggregation of teams into 'departments' or other hierarchies. Requirements for analyses by consultant activity are consistent with NHS needs and policy direction (to publish at consultant level). Consultant code is also used in research e.g. analysing volume and outcome relations for elective surgery. Some exclusions are applied e.g. Invalid codes, dental consultant etc. Patient’s general medical practitioner Patient’s general medical practitioner is used to examine variations by GP practice and to enable mapping to practicelevel such as The Quality and Outcomes Framework (QOF) and practice staffing data etc. Person referring patient We provide analyses by person referring patient activity which are fed back to the NHS through a range of Management Information Systems provided by Dr Foster Intelligence (DFI) Patient identifiers In our approved Section 251 application (CAG Reference: 15/CAG/0005), we have reduced the number of patient identifiable fields to 2 fields - NHS Number and Local Patient ID number (LOPATID). We have been granted permission to hold NHS Number and Local Patient ID number (LOPATID) to assist local NHS trusts investigating issues around quality and safety of care within their organisation, which have arisen out of Dr Foster Intelligence healthcare performance tools using our methods. We will no longer be holding Homeadd and Date of birth. Historical data processed under PIAG 2-07(d)/2007 will be irreversibly pseudonymised in line with this application. This will be carried out as soon as possible and confirmation will be sent to CAG and HSCIC if required. New identifiable data processed under CAG [15/CAG/0005] will be retained for a maximum of three years after which it should be destroyed or irreversibly pseudonymised on a rolling basis. The purpose of holding the patient identifiers is to allow hospitals to further investigate any alerts around poor or good performance, and to help improve the quality and safety of healthcare delivery. Imperial do this by providing a re-identification service to which Dr Foster Intelligence has no access. Imperial only provide this service to acute NHS providers who are customers of DFI. The provision of a re-identification service to non-customers of DFI allowing them to respond to mortality alerts issued by the academic unit has been deferred, pending further information from the applicant, in relation to the disclosure of confidential patient information to provide re-identification service for all trusts. No identifiers will ever be passed to Dr Foster Intelligence or any other organisation except the NHS provider trust from where the data originated. For this purpose we have developed a re-identification service whereby authorised individuals within NHS Provider Trusts are able to identify their own patients indicated in the DFI tools. This service allows us to supply Provider trusts’ NHS Number and LOPATID using DFI tools without passing these fields on to DFI. The re-identification service is maintained by Imperial College. DFI have no access to the data held on it. In the last 12 months, there were over 3,600 successful logins from 119 provider organisations. 74 provider trusts use it more than 12 times per year (once a month). One trust has used the service 285 times in the year. (See Supplementary evidence from NHS trusts attesting usefulness of Re-Identification service). The standard extracts will be loaded on to Imperial’s systems and a unique identifier (fosid) will be generated and added to the datasets. A new Extract_hesid (for Dr Foster) will be generated using SHA-256 hashing algorithm, compliant with the e-GIF Technical Standards Catalogue Version 6.2 based on the original Extract_hesid. No record level data will be transferred outside of the EEA, either under this agreement or any related sub-licence Pseudonymised data dating back to 1996/7 has been requested for three reasons. • To examine historical trends of treatment practice (e.g. Faiz et al. Traditional and Laparoscopic Appendectomy in Adults Outcomes in English NHS Hospitals Between 1996 and 2006, ANNALS OF SURGERY 2008;248:800-806) and the historical impact of changes in policy (e.g. proposal to examine capacity issues). • To obtain longitudinal data on prior admissions for patients (e.g. to determine prior diagnosis of cancer in Bottle A et al. Association between patient and general practice characteristics and unplanned first-time admissions for cancer: observational study, BRITISH JOURNAL OF CANCER 2012;107:1213-1219 or to validate and complement parity status in maternity fields for non-obstetric surgery proposals). Risk modelling will also require access to variables on prior admissions including previously recorded co-morbidities. • To increase the power of predictive models for rare diseases, procedures and events (e.g. standard casemix adjustment models for 259 diagnosis groups and 200 procedure groups which includes some rarer conditions).

Objectives:

To use hospital administrative data to provide measures of quality of delivery of healthcare by providers, or in some instances, by area and to show variations in quality by provider and to support management information function for the NHS


Project 13 — DARS-NIC-370843-R6V8T

Opt outs honoured: N

Sensitive: Sensitive

When: 2016/04 (or before) — 2016/08.

Repeats: Ongoing

Legal basis: Informed Patient consent to permit the receipt, processing and release of data by the HSCIC

Categories: Identifiable

Datasets:

  • MRIS - Flagging Current Status Report
  • MRIS - Scottish NHS / Registration
  • MRIS - Cause of Death Report

Benefits:

The results of the applicants studies, which will be placed in the public domain via peer reviewed publications, are expected to provide reliable and robust scientific evidence to: 1. address current gaps in scientific evidence regarding the possible health effects of long-term mobile phone use. 2. inform UK health policy on use of mobile phones and newly emerging RF technologies. Specifically, the detailed research outputs on mobile phone use, and potential associated health risks will allow the UK Chief Medical Officer to review the current precautionary advice regarding mobile phone use, and if required, update this advice. 3. identify specific ways to reduce RF exposure levels, if required, and thus provide more specific health advice to the UK public.

Outputs:

Research outputs from the applicants access to the data will be placed in the public domain by presentations at scientific conferences (presentations, abstracts, posters) and publication in peer-reviewed journals in aggregated and anonymised form. Timing of analyses for specific disease outcomes is dependent upon sufficient statistical power, and thus accrual of sufficient cases of disease over time. The applicant expects to start publishing results of epidemiological analyses of mobile phone use and chronic disease outcomes in 2016. However, this will be an ongoing process especially with the investigation of rare disease outcomes, e.g. salivary gland tumours and amyotrophic lateral sclerosis, which will require much longer follow-up time.

Processing:

Data will be received in from the applicant, and run through the patient status and list cleaning services to a) retrieve any missing identifiers in the cohort where possible (NHS numbers, DOBs, Postcodes, Addresses) and b) update the cohort to the latest data for these variables. The cleaned cohort will then be linked to HES Inpatient, Critical Care Outpatient, and Accident & Emergency data. It will also be linked to ONS mortality data including cancer information, and the information for Scottish patients retrieved. The sensitive and identifiable output will be returned to the applicant. The HES data will be run on a bi-annual basis there afterwards, and the ONS/Cancers run on an annual basis. Health event/mortality data supplied by HSCIC will be linked, using a randomly assigned unique ID number for each participant, to other UK COSMOS data on mobile phone usage, other RF exposures, other environmental exposures e.g. noise and air pollution, green space etc, health and lifestyle to allow epidemiological analyses of exposure and health outcomes. Any personal identifiers such as Name, Address, Postcode that may be supplied by HSCIC will be stored separately from health data. Access to personal identifying information is limited to the COSMOS/SCAMP database team for processing and one COSMOS researcher from the COSMOS research team (to enable participant enquiries or withdrawal requests to be actioned). All COSMOS researchers are employees of Imperial College, who have signed strict non-disclosure agreements for the use of the SAHSU private network and the COSMOS study database. The applicant will also obtain health outcome data from NHS Scotland, NHS Wales (both approved) and the Office for National Statistics (ONS births, approved), as appropriate, under data sharing agreements. Data from these datasets will be matched through the use of anonymised unique record-level identifiers, assigned to each verified study participant on receipt of the datasets by the COSMOS database team, before being provided as pseudonymised data to the COSMOS research team. Name, Address, Postcode are required to verify quality of matching, to improve quality of demographic data, and to ascertain any change in details over the course of this longitudinal study. Over time, these personally identifiable data fields from ‘Latest Patient Information’ which the applicant is requesting as part of ‘Patient Tracking’ may become the most up to date records as the study progresses. Should the applicant find from the data that UK COSMOS study participants’ details have changed, the applicant may use updated names and addresses provided by HSCIC to re-contact study subjects for follow-up questionnaires and/or update information sent to mobile phone network operators to maximise the matching rate to their databases in order to receive mobile phone usage data, which are crucial for ongoing accurate exposure assessment. Similarly, updated name, address, postcode information is provided by network operators (where available and for those consenting) for participants as the study progresses. Written agreements to ensure confidentiality and non-disclosure of data are in place between Imperial College London and mobile phone network operators. The applicant has ethical approval and individual consent from participants to use the personally identifiable data fields (Name, Address, Postcode) from ‘Latest Patient Information’ and from mobile phone network operators for all the purposes stated above. The study requires HES data linked to the applicants cohort participants for all requested years as on the following basis: (i) For prospective epidemiological analyses: HES data provides a more comprehensive medical history, providing information on underlying conditions and treatment of these conditions. This will allow us to perform statistical analysis to enable us to investigate whether a link exists between mobile phone use and future adverse health outcomes. (ii) For retrospective epidemiological analyses: The applicant has collected mobile phone use exposure data for the applicants cohort participants going back to 1985. They will conduct historical analyses of mobile phone use and health outcomes in HES data from 1997 onwards. Imperial College London, and the Department of Epidemiology and Biostatistics, where this study is being undertaken, have considerable experience over a number of years in receiving, holding and analysing sensitive and identifiable data from a wide range of sources. Through this experience, Imperial has developed a robust and effective Information Governance infrastructure, policy and procedures and appropriate culture for the secure handling of these types of data. In general, information governance and especially anonymisation is taken very seriously at ICL and does conform to the Information Commissioner’s Code of Practice. It is important to note that the UK COSMOS researchers and database where the raw identifiable data is received and processed are kept separated. On receipt of the datasets from HSCIC, the COSMOS/SCAMP Database Manager will separate identifiable data for individuals from their health data, by the use of pseudonyms and records identifiers. The COSMOS study researchers then receive this pseudonymised dataset for the purpose of performing statistical analysis for research purposes only. Researchers undertaking epidemiological analysis therefore only have access to pseudonymised data. Access to personal or identifiable personal data requires the use of specialised cryptography and SQL, is restricted to a limited number of specified staff within the COSMOS team, who (for example) may need to check back with raw data provided, for the purpose of ensuring that a participant is accurately identified and linked to their personal information or for the purposes of identifying records to be deleted after a withdrawal request. The SAHSU/COSMOS/SCAMP Information Governance Policy is written to be compliant with ISO/IEC 17799:2005 & ISO/IEC 27001:2005 and has been internally reviewed and risk assessed in accordance with Imperial College policy.

Objectives:

There is extensive public and scientific interest that exposure to RF electromagnetic fields from mobile telephony might increase disease risk. Results from epidemiological studies on RF and disease risk published to date have been inconsistent. A large prospective cohort study of mobile phone users with long-term follow up was required as the best way to resolve current uncertainties The UK cohort study of mobile phone use and health (COSMOS) is a valuable national resource to improve the applicants understanding of environmental exposures and health in the UK. The applicant aims to investigate whether there is any link between long-term use of mobile phones and other radio frequency (RF) technologies and human health, to provide maximum benefit to public health. The UK COSMOS study has recruited a cohort of approximately 105,000 adult participants. It is the UK’s fourth largest cohort study. Data collection also including other environmental exposures e.g. noise and air pollution, information on health, lifestyle, and demographics have already been collected at baseline, and collection continues via prospective follow-up of participants for the next 20-30 years. The UK COSMOS study forms part of the International COSMOS prospective cohort study and together the applicant has recruited 290,000 adult mobile phone users across Europe. This research has been endorsed as a priority by agencies worldwide, including the Department of Health and the World Health Organization. It is important to note that the data requested from HSCIC will be used for research purposes only and that COSMOS is an observational study, with no clinical intervention. The objectives for processing of the Data are therefore: 1. To conduct long-term health monitoring of the UK COSMOS cohort via linkage to national health and mortality datasets, to capture chronic health outcomes as they occur. 2. To pool record-level data of the UK COSMOS cohort (as limited and bespoke datasets containing non-sensitive pseudonymised data) with data for the international COSMOS cohort, to enable pooled epidemiological and statistical analyses on mobile phone use, other RF exposures, other environmental exposures and a wide range of health outcomes. 3. To conduct epidemiological analyses on other environmental exposures (e.g. noise, air pollution) and health outcomes. The COSMOS International Research Group (UK, Sweden, Finland, the Netherlands, Denmark, France) was established and produced a joint international study protocol in 2005 and has collaborated since that time. These collaborators are : • Department of Biostatistics and Epidemiology, Institute of Cancer Epidemiology, Danish Cancer Society, Copenhagen , Denmark • Karolinska Institutet, Stockholm, Sweden • Tampere School of Public Health, University of Tampere, Finland • Institute for Risk Assessment Sciences (IRAS), University of Utrecht, Utrecht, The Netherlands • Section of Environment and Radiation, International Agency for Research on Cancer (IARC), Lyon, France Details of these collaboration establishments have been provided on the study website since its inception. The pooling of data from all of these countries is essential to achieve the scientific aims of the COSMOS study and to provide the required statistical power for meaningful analysis of the associated health outcome events. NB – the study will not share whole datasets as provided by HSCIC. Instead, only a bespoke dataset will be shared with the collaborating country leading the group on a certain topic or health outcome (e.g., mobile phone use and cancer). This dataset will include only those variables required to perform the relevant statistical analysis, i.e., main dependent and independent variables as well as covariates/potential confounders (such as age, body mass index, sex etc.) which need to be adjusted for in the statistical models. The bespoke anonymised dataset will not contain variables listed as identifiable and sensitive by the HSCIC (as listed in this application). Data will not be shared beyond this group of collaborators without prior approval, which would be sought through a supplementary application to HSCIC. These pooled analyses would therefore require selected and limited record-level data (as non-sensitive and anonymised in line with the ICO Anonymisation Code of Practice) to be sent outside of the UK to the applicants international COSMOS research partners. These data would also comply with the HES Analysis Guide, including the small numbers requirements. At this stage in this long-term cohort study on the health effects of RF-EMF exposure, the UK group will investigate cardiovascular disease, the Karolinska Institutet (Sweden) cancer risk and the Institute for Risk Assessment Sciences (The Netherlands) reproductive health. However, this may be subject to change and various partners may aid each other in these analyses or other analyses e.g. on neurodegenerative diseases as the number of years of follow-up of the cohort increases and rare health events accrue. The applicant confirms that to protect the confidentiality of UK COSMOS participants, Imperial College will only provide non-sensitive anonymised health data to researchers outside the UK COSMOS research team. This means that: • No sensitive personal information will ever be passed to other researchers outside the UK COSMOS team. • Any variables required for analyses by other researchers that may derive from a personal identifier (e.g. date of birth, mothers date of birth, date of death or health event, cause of death, or type of health event) will be calculated into categories as appropriate by the UK COSMOS research team, so that it is not identifiable by other researchers from the data extract provided by UK COSMOS, i.e. it will be anonymised with regard to the research teams receiving these data. For example, date of birth will be used to calculate age or age-bands. Date of event will be converted to days of follow-up (i.e., days from the start of the COSMOS study to date of event) because this is important information for the study that this is available for inclusion in survival analysis models used to calculate hazard ratios and to estimate health risks associated with RF-EMF exposure. Cause of death or a health event will be coded and included under broad terms/variables such as all-cause mortality, fatal and non-fatal cardiovascular events, cardiac events, cancer etc. For example, if a person had coronary bypass surgery, it will be included in the dataset under cardiac events with 1 or 0 listed for each participant, representing yes or no for a cardiac event. Therefore, broad binomial or dichotomous variables are created which is also the only format in which it is usable in the statistical models. All dates and details of a health event such as ICD codes are removed from any data extract prepared for other researchers. Any release or sharing of study data to international COSMOS collaborators will be subject to approval of the UK COSMOS publication committee. The publication committee consists of the UK Principle Investigators and senior COSMOS International Group members, and decides whether the research question is appropriate. The model for obtaining participant consent for the study was extensive. It took shape over a period of 7 years, alongside the development of the study itself. This included three rounds of testing the consent and associated explanatory materials by 3 rounds of national focus groups, to ensure that information provided and the form of consent were clear and easily comprehensible. These focus groups included adults from across the UK, from all socio-economic groups and ethnic minority groups. These trials indicated that information was best presented in plain English, avoiding unnecessary jargon or technical terms, where these might confuse or cause concern among participants. The resulting materials were subsequently considered by the relevant Medical Research Ethics Committee against the expectations and requirements that were prevalent at that time (2009) and approved for use. This information and form of consent were used throughout the participant recruitment period which completed in 2012.


Project 14 — DARS-NIC-383203-Q8B9L

Opt outs honoured: Y

Sensitive: Sensitive

When: 2016/09 — 2018/02.

Repeats: One-Off

Legal basis: Approved researcher accreditation under section 39(4)(i) and 39(5) of the Statistical Registration Service Act 2007

Categories: Identifiable

Datasets:

  • Office for National Statistics Mortality Data

Benefits:

1) Ongoing benefits from our previous work The Imperial Unit’s methodological research forms the basis of a near real-time monitoring system, currently used by 70% of English NHS Acute Trusts to assist them in monitoring a variety of casemix-adjusted outcomes at the level of diagnosis and procedure groups. The unit works with the CQC, contributing to its surveillance remit using the same methods and data. From our monitoring system, the Unit at Imperial College generates monthly mortality alerts, based on high thresholds, which we have been running since 2007. This was pivotal in alerting the then Healthcare Commission (HCC) to problems at the Mid Staffordshire NHS Foundation Trust between July and November 2007. The resulting Public Inquiry recognised the role that our surveillance system of mortality alerts had to play in identifying Mid Staffs as an outlier. Key recommendations, reflecting our unit’s work, are that all healthcare provider organisations should develop and maintain systems which give effective real-time information on the performance of each of their services, specialist teams and consultants in relation to mortality, patient safety and minimum quality standards. A further recommendation is that summary hospital-level mortality indicators should be recognised as official statistics. If the Dr Foster Unit at Imperial College are given continued access to the data, this monitoring tool that detected Mid Staffs will continue to monitor patient outcomes at acute hospitals and be ready to detect any future outliers. The Dr Foster Unit at Imperial College will be able to assist the investigation of variations in outcomes at a local level by providing a set of fields from our analyses to authorised users within trusts to enable reconciliation with local information systems and the instigation of clinical audits and case note reviews. Our mortality outlier outputs are used by CQC within their Hospital Inspection framework. As a result of our leading role in the development of hospital mortality measures, in 2010 we were invited to contribute to a Department of Health (DoH) Commissioned expert panel (Steering Group for the National Review of the Hospital Standardised Mortality Ratio) to develop a national indicator of hospital mortality. The resultant Summary-level Hospital Mortality Indicator (based in part on our HSMR methods) is now a public indicator used by all acute trusts. Professor Sir Bruce Keogh suggests that a relatively “poor” SHMI should trigger further analysis or investigation by the hospital Board. The recent review (published in July 2013) into the quality of care and treatment provided by 14 hospital trusts with consistently high mortality in either measure led to 11 out of the 14 trusts identified being immediately placed on special measures. The review also informs the way in which hospital reviews and inspections are to be carried out with the recommendation that mortality is used as part of a broad set of triggers for conducting future inspections. We continue to advise the Health and Social Care Information Centre on methodological issues around the Summary­level Hospital Mortality Index (SHMI), and carry out analyses relating to this measure to assist in its development. The unit’s research on specific aspects of care has received a high media profile and has been highly cited. Our research on weekend mortality in emergency care, analysis of mortality associated with the junior doctor changeover and work on elective procedures and mortality by day of the week resulted in front page broadsheet coverage, and radio and TV interviews The unit’s “Out of hours” work has been a key driver in moving NHS towards 7/7 care. Headlines include, “NHS Services – open seven days a week: every day counts” and, “Sunday Times Safe Weekend Care”. As a result of our published research into the junior doctor changeover, Bruce Keogh introduced a week's shadowing where newly qualified doctors worked alongside more senior ones for a week before they start work in August. The Academy of Medical Royal Colleges published proposals (16th April 2014) suggesting all Foundation Year 1 posts should begin on the first Wednesday in August as has always been the case, but other training posts should begin in September. Another example of our research that have used HES-ONS Mortality data include one-year survival and readmission in heart failure patients and risk of post-operative death by cause of death over time in patients undergoing general surgery. 2) Expected benefits from the proposed work For future research, it is important to be able to capture deaths occurring following discharge from hospital to assess the full mortality burden relating to that hospitalisation. Out of hospital deaths are particularly useful for surgical outcomes, e.g. for the calculation of total 30-day post-operative death rates, as the effect of premature discharge (in terms of mortality) would otherwise go unnoticed. Longer-term follow-up of hospitalised patients, e.g. using one year survival, necessitates being able to capture all deaths, not just those occurring in hospital. For this reason, the Summary Hospital-level Mortality Indicator (SHMI) specification requires out of hospital post discharge deaths. As described above in relation to the project on heart failure and COPD, mortality is a “competing risk” for important non-fatal outcomes such as readmission. Accurate prediction of the risk of these other outcomes will help with risk stratification and health service planning, and is not possible without total mortality. Knowledge of the cause of death is particularly important for quality improvement. The relation between the cause(s) of death and the reason(s) for admission is of particular interest too. The place of death, including whether it was an NHS institution, is necessary to monitor end-of-life services. Of particular interest is the proportion of patients who die at home. Previous work using HES showed higher mortality risk for asthma in those living in areas further from a hospital than those near it. Using Lower Super Output Areas would enable studies into the effect of distance from home to hospital on patient outcomes and the estimation of hospital catchment areas. This allows geographical access to services to be estimated, as we can calculate how far patients must travel for their treatment. This is of growing importance given the current drive to centralise services, particularly for surgery. Using larger geographical areas than LSOAs would incur too much measurement error when calculating the distance between the patient’s home and the hospital. Ongoing analysis of the mortality alerting system will allow the Dr Foster Unit to improve the alerting process, reducing the number of false positives and unnecessary effort spend by hospitals investigating them. It will also provide advice to hospitals who receive the mortality alerts on how to follow them up and learn, for example, which are the key contributing factors in the alerts. The proposed analysis of variations in treatment and outcomes in TAD patients will shed light on which patients are underserved by current surgical practice, which patients are most likely to benefit from treatment, and what might be the effect of centralisation of surgery.

Outputs:

1) Research into variations in quality of healthcare by provider: background to proposed work The Dr Foster Unit at Imperial College use hospital administrative data in the form of HES/MMES/ONS Mortality data to provide measures of quality and safety of delivery of healthcare by provider, or in some instances, by area or time. The unit’s work focuses on quality of care and patient safety, including healthcare acquired infections and safety indicators. Collaborative projects with clinical colleagues have helped develop and validate healthcare quality indicators other than mortality, including bariatric surgery, primary angioplasty rates, indicators for stroke care, obstetric care, orthopaedic redo rates and returns to theatre. Our proposed work will continue our principal themes: i) developing and validating indicators of quality and safety of healthcare, particularly by consultant and hospital; ii) show variations in performance by unit and socio demographic stratum; iii) risk prediction and risk adjustment of such indicators and variations and any other methodological aspects as they arise. The Dr Foster Unit at Imperial College require data dating back to 2000 for two reasons: • To examine historical trends of treatment practice (e.g. Faiz et al. Traditional and Laparoscopic Appendectomy in Adults Outcomes in English NHS Hospitals Between 1996 and 2006, ANNALS OF SURGERY 2008;248:800-806) and the historical impact of changes in policy (e.g. proposal to examine capacity issues). • To increase the power of predictive models for rare diseases, procedures and events (e.g. we build standard casemix adjustment models for 259 diagnosis groups and 200 procedure groups which includes some rarer conditions). The Dr Foster Unit at Imperial College plan the following analyses: They are working in collaboration with the University of Manchester and supported by the Care Quality Commission (CQC), to improve understanding of our mortality alerts and to evaluate their impact as an intervention to reduce avoidable mortality within English NHS hospital trusts, focusing on two conditions commonly attributed to mortality alerts - acute myocardial infarction and septicaemia. The Dr Foster Unit at Imperial College aim to provide a descriptive analysis of all alerts, their relationships with other measures of quality and their impact on reducing avoidable mortality by Oct 2016. The Dr Foster Unit at Imperial College have a two-year NIHR-funded project looking at predictors of readmissions and one-year mortality (in and out of hospital) in patients with chronic diseases (heart failure and COPD), following on from our two recently published studies on readmissions in heart failure patients (first two listed below). Mortality acts as a “competing risk” for readmission, and it is therefore essential to know whether a patient has been discharged alive but subsequently dies and is therefore no longer at risk of readmission by Jun 2017. The Dr Foster Unit at Imperial College have begun work using HES with the University of Leicester on thoracic aortic disease (TAD) looking at variations in rates of surgery and mortality between centres. There seem to be wide variations in the rates of treatment for this condition, but it is unclear how this impacts on outcomes. In-hospital mortality only captures part of the effect. With the recent growth in the number of endovascular procedures (TEVARs), post-discharge deaths are vital to assess the impact of these procedures and of TAD services in general by Aug 2016. International comparisons of service use and outcomes: England and the USA. The Dr Foster Unit at Imperial College hold data from Centre for Medicare and Medicaid Services enrolees and from the Nationwide Inpatient Sample from the USA. They have previously set out the methodological issues with using administrative data from multiple countries. They will compare patient casemix, rates of outcomes such as post-operative mortality, infections and readmissions, for example in patients near the end of their life (overtreatment is a growing concern) between the two countries. The objective is to highlight areas of better or poorer performance by the NHS compared with the USA by Oct 2017. Examples of key published research that have used HES/ONS data include: Bottle A; Goudie R; Cowie MR; Bell D; Aylin P. Relation between process measures and diagnosis-specific readmission rates in patients with heart failure. Heart 2015; Jun 11 (epub). Bottle A, Aylin P, Bell D. Effect of the readmission primary diagnosis and time interval in heart failure patients: analysis of English administrative data. Eur J Heart Fail 2014; 16(8): 846–853. Aylin P; Alexandrescu R; Jen MH; Mayer EK; Bottle A. Day of week of procedure and 30 day mortality for elective surgery: retrospective analysis of hospital episode statistics. BMJ 2013;346:f2424. Palmer WL; Bottle A; Davie C; Vincent CA; Aylin P. Dying for the Weekend: A Retrospective Cohort Study on the Association Between Day of Hospital Presentation and the Quality and Safety of Stroke Care. Arch Neurol. 2012;69:1296-1303. Aylin P, Yunus A, Bottle A, Majeed A, Bell D. Weekend mortality for emergency admissions. A large, multicentre study. Qual Saf Health Care. 2010;19:213-217 For full publication list see unit website: http://www1.imperial.ac.uk/publichealth/departments/pcph/research/drfosters/unit_publications/

Processing:

This process comprises the following steps: The Dr Foster Unit at Imperial College use hospital administrative data in the form of HES/Monthly Managed Extract Service (MMES)/ONS to identify measures of quality and safety of healthcare. The unit’s work focuses on quality of care and patient safety, including healthcare acquired infections, mortality and safety indicators. The Dr Foster Unit at Imperial College hold 2 databases to store data – A Research database and a Patient Identifiable Database to provide a Re-Identification service for NHS provider trusts. The ONS Mortality data will be stored in the Research database where named researchers with Approved Researcher status will be able to access the data to do their analyses. Patient identifiers are stored separately to our research database which holds the standard HES extracts and sensitive fields. Imperial’s researchers have no access to identifiable fields. Only two named data managers have access to the patient identifiable fields within the unit. The purpose of holding the patient identifiers for the last 3 years is to allow hospitals to further investigate any alerts around poor or good performance, and to help improve the quality and safety of healthcare delivery. No record level will be transferred outside of the EEA under this agreement, and is only processed and stored at the addresses given within this application. ONS Data supplied under this Agreement may be linked with HES Data Supplied under NIC-12628 for the purposes of cross HES-ONS mortality analysis. Data may be linked using Encrypted_HESID. All individuals with access to the data are employees of Imperial College London. All outputs (including those shared with collaborators) are aggregated with small numbers suppressed in line with the HES analysis guide. The ONS data provided under this agreement will not be shared with any third party, and (for the avoidance of doubt) specifically not shared with commercial companies including (but not limited to) Dr Foster Limited.

Objectives:

The Dr Foster Unit at Imperial College have the following objectives: • To use hospital administrative data to provide measures of quality of delivery of healthcare by providers, or in some instances, by area and to show variations in quality by provider and to support management information function for the NHS. • To compare hospitals' mortality rates for in-hospital deaths with rates for all deaths (to evaluate the effect of differential discharge policies). • To calculate total post-operative mortality rates, e.g. when comparing operative techniques such as laparoscopy and open approaches. • To access potential quality of care issues by comparing the cause of death with the reason(s) for admission, e.g. for surgical patients who are discharged within 30 days of the procedure but who die at home and whether the death is related to their disease process or to complications of treatment? • The Dr Foster Unit at Imperial College are requesting Office for National Statistics (ONS) deaths for Hospital Episode Statistics (HES) 2012/13 and 2014/15 plus 30 days in order to capture all deaths (in and out of hospital) within 30 days of admission or procedure for patients in hospital in March 2015, i.e., so that all patients have the full 30 days of follow-up in the data. Please note that whilst Dr Foster Unit and Dr Foster Limited share a similar name, and the Dr Foster Unit is named in recognition of the funding provided by Dr Foster Limited, the Dr Foster Unit is legally part of Imperial College. It thus is physically and logically separate to Dr Foster Limited, and the agreement and uses within solely relate to activities of Imperial College.


Project 15 — DARS-NIC-39944-D4M0D

Opt outs honoured: N

Sensitive: Non Sensitive

When: 2016/09 — 2016/11.

Repeats: One-Off

Legal basis: Health and Social Care Act 2012

Categories: Anonymised - ICO code compliant

Datasets:

  • Hospital Episode Statistics Admitted Patient Care

Benefits:

The preliminary research for this project has identified, through a survey of 150 GPs across England, that currently only 37% of GPs with access to risk stratification think it is actually useful to their practice. Nevertheless, they are financially incentivised to use it though the NHS Enhanced Service specification on Admission Avoidance, which requires GP to identify the top 2% high risk patients on their list, and provide them with special services. There is a need to try and improve the usefulness of risk stratification. This research aims: - to improve the accuracy of risk prediction; and - to provide healthcare professionals with the tools to do the analysis and provide patient-centred, effective care. 1) The research aims to identify new and improved methods to perform risk prediction. Improved risk stratification methods will allow GPs to target their interventions more accurately, to those patients who need it most. This will improve care for high-risk patients and reduce their need for emergency admissions. Preventing emergency admissions will have a positive impact on all three elements of the Triple Aim of healthcare: Patients receive high quality care, the negative experience of an emergency hospitalisation is avoided, and costly emergency care is prevented. 2) If a software tool or manual is developed based on new and improved methods, GPs and CCGs will be able to conduct their own risk prediction analysis. Currently, the majority of GPs (82%, according to our survey) receive only a list of names for the high risk patients, with no additional analysis or insights. If they run their own risk prediction analysis they can understand what is driving the high risk of each patient and see the full picture. This information can then be used to personalise interventions and improve care.

Outputs:

· A PhD on data analysis methods to segment patients - submission Jun 2017 · A paper published in a peer reviewed journal focused on a health professionals audience (e.g. BMJ or Health Affairs) - expected initial submission Dec 2016. This paper will describe the research, and if applicable, provide a reference to where the software tool or manual (as per below) can be found. This will ensure a widespread dissemination of both the knowledge and the actual method. · A software tool or manual for risk stratification, based on the method that was found to be most accurate - March 2017 (Note - this is contingent on the results of the analysis). If a method is found that significantly improves on current methods, Imperial College London will develop either a software solution or a manual that describes how to do the analysis in an existing software package (e.g. SPSS). This will be made available on a not-for-profit basis to CCGs and GPs for use in their practice. Imperial College will share results with the Data team at NHS England. This has already been discussed with the Chief Data Officer, and Imperial expect to share with him the results of the study, to ensure it reaches the right people. This will include any negative results. HES data will only be used to identify the optimal methods for risk prediction, but that the actual tool will contain no HES data. Instead, it will be run from GPs’ or CCGs’ own data.

Processing:

The data will be used to test the different predictive models. Data from the years 2011-2014 will be used to predict emergency readmissions in 2015. The following steps will be taken to create a dataset to which the different predictive models can be applied: • A range of predictor variables (such as hospital admissions, demographics including age and gender, and diagnoses of specific long-term conditions) will be extracted from the 2011-2014 data • These predictor variables will be used to create a patient-level file, with on each line a patient, his/her predictive variables, and whether or not the patient had an emergency admission in 2015 • The patient-level file will be split into a training and a test dataset, consisting of either 10% and 10%, or 50% and 50% of the population (depending on computing power available). The training data will be used to train the predictive methods, and the test dataset will be used to evaluate the predictive power of the developed models in a new population. • The models will be tested for predictive power using the area under the receiver operator characteristic (ROC) curve, and the positive predictive value. • In addition, the different methods will be compared by looking at the high-risk population (e.g. top 5%) they identify.

Objectives:

While predicting the risk of emergency admissions is becoming ingrained in healthcare delivery and financing in England, the models that are being used have limited predictive power. Most models are linear regressions, based on simple, static predictor variables. As such, there is significant scope to improve on these methods. The aim of this project is to develop a new method to predict risk in healthcare, going beyond linear regression, by using data science methods. The aim is to explore the different options such as artificial neural networks and to explore which method offers the most potential to revolutionise risk prediction. Simple risk algorithms that require no advanced data analysis will also be explored. Using HES data, the different techniques will be analysed and compared. If successful, the new risk prediction models will be developed into a tool or manual for the NHS to use.


Project 16 — DARS-NIC-67398-K2Y3T

Opt outs honoured: N

Sensitive: Non Sensitive

When: 2016/12 — 2017/05.

Repeats: One-Off

Legal basis: Health and Social Care Act 2012

Categories: Anonymised - ICO code compliant

Datasets:

  • Hospital Episode Statistics Accident and Emergency
  • Hospital Episode Statistics Admitted Patient Care
  • Hospital Episode Statistics Outpatients

Benefits:

The measurable benefits to health and social care are expected to be as follows: 1. The structure of interhospital transfers in the NHS. The transfer of a patient from one hospital to another often occurs at critical periods in a patient’s journey where they can no longer be optimally cared for by their current hospital. Recent centralization of specialist services has increased the need for transfer to another hospital to receive specialist care. This transfer process is a period of increased patient risk, where an often critically ill patient is transferred by ambulance over significant distances and whose care is handed over to an entirely new team of clinicians. Which patients need to be transferred, when and to which hospital remains poorly understood, as do their health outcomes relative to those who do not need to be transferred to receive the same specialist care. Through the publication of this work in relevant academic journals, insights into the movement of patients from one hospital to another may be achieved by clinicians and commissioners. This knowledge may be used to both understand the factors which influence patient and physician choice, and also incorporate these factors into future service design. By understanding the circumstances that lead to patient transfer the aim is to identify patient groups that are particularly likely to undergo interhospital transfer and to focus on the development of local and national strategies to ensure optimal transfer of care for these specific groups. It is expected that this work will be completed within three months of receipt of the data. 2. The impact of patient choice in maternity care on local service supply and demand. Patient choice is an increasingly important factor of care delivery in the NHS. The factors underlying patient choice remain poorly understood, in part because of the many patient and provider factors that influence decision making. Expectant mothers can freely choose which hospital they would like to deliver their maternity care. Maternity care is delivered frequently across the country and as it is generally focused solely on the process of giving birth, the variability in patient and provider factors is far less than for other clinical scenarios. This therefore serves as an excellent setting to model the factors which underlie patient choice. In the context of maternity care, where patients can freely choose where their care is delivered, certain providers may be repeatedly favoured or avoided by expectant mothers in response to a range of factors including individual previous experience, geography or waiting times. This may lead to demand for certain providers becoming too great to be met, while others have unused capacity. Identifying and predicting these factors allows providers locally and nationally to correct imbalance in the supply and demand relationship for maternity care, thereby optimising the effectiveness of maternity provision nationally. It is expected that this work will be completed within three months of receipt of the data. 3. The structure of care networks for patients following trauma, stroke and cardiovascular events, comparing regions with established care networks to those without. The introduction of defined care networks for the treatment of trauma, stroke and cardiovascular disease in parts of the NHS have demonstrated significant improvements in patient outcomes where they have been implemented. In the case of stroke care, networks have been extremely successful in London and Manchester where they have been introduced. The rest of the country currently does not have the same effective network structure. Using the principles of community detection analysis and Markov models is would be possible to identify for the London and Manchester stroke networks whether their structure optimally reflects the distribution of disease and pattern of clinical practice in the geographic areas they cover. Outside of these two networks it would be possible to examine whether similar network structures already informally exist elsewhere in the country, and develop a nationwide stroke network, in a manner like that which was created for the highly successful national trauma network. This knowledge would inform the development of a national stroke network so that the benefits already obtained from its implementation in London and Manchester may be available nationally. Publication of these findings in high impact health policy journals will bring this work to the attention of key stakeholders nationally and locally. It is expected that this work will be completed within nine months of receipt of the data. 4. A network analysis demonstrating the interdependence of secondary care providers in the NHS followed by predictive modelling of patient flows to secondary care providers in response to changing organizational capacity. Demand for health care within the National Health Service continues to rise, and does so in a stochastic fashion. Each hospital has a finite capacity to provide safe care, and therefore a threshold over which harm is more likely to result. The likelihood of the demand being placed on a hospital exceeding the care it can safely provide is dependent upon the local incidence of disease and its intrinsic capacity to provide care, but is also critically dependent on the performance of its neighbouring hospitals. If a hospital is unable to meet the demands placed on it, the burden of care provision falls to its neighbouring hospitals, which therefore see an increase in the demands placed on their services. Hospitals with many nearby hospitals may be less vulnerable to this pattern of behaviour and would therefore be said to have a low degree of interdependence, while a pair of hospitals with no nearby neighbours would be highly interdependent on the behaviour of one another. This principle when applied across hospitals the National Health Service will identify areas of high interdependence within the health care network. Areas of high interdependence of care providers are expected to be less resilient to increases in demand for care or reduction in the capacity to provide care. Identifying these vulnerabilities will assist NHS England in identifying hospitals who require additional investment to ensure the ongoing delivery of high quality patient care. It is intended that the predictive models developed from this work will be published in high impact health policy or general medical journals to reach the widest possible interested audience. Additionally, the methodological insights from this work will be disseminated either in the form of a further journal article or white paper for NHS Improvement and to detail the application of these techniques. It is expected that this work will be completed within 12 months of receipt of data. The proposed work in focusing on the interconnectedness of healthcare providers, represents an exciting, novel and important means by which the efficiency and equity of health care provision may be examined in a new light, with a high likelihood of lasting improvement to the NHS as a whole.

Outputs:

-Specific outputs expected, including target dates: All outputs will contain only aggregate data with small numbers suppressed in line with the HES Analysis Guide. The study will yield a PhD Thesis between October 2019 and October 2020 in addition to published academic papers as follows: 1. The structure of interhospital transfers in the NHS. 2. The impact of patient choice in maternity care on local service supply and demand. 3. The structure of care networks for patients following trauma, stroke and cardiovascular events, comparing regions with established care networks to those without. 4. A network analysis demonstrating the interdependence of secondary care providers in the NHS. 5. Predictive modelling of patient flows to secondary care providers in response to changing organizational capacity. Each of these papers will be targeted at health policy or health informatics journals including; The British Medical Journal Annals of Surgery The Lancet British Journal of Obstetrics and Gynaecology Health Affairs International Journal of Systems Science. It is intended that all work will be presented at academic conferences prior to publication, including; Health Systems Global Symposium – UK International Health Policy Conference – UK. Results will also be directly reported to NHS England and NHS Improvement where appropriate.

Processing:

The Department of Surgery and Cancer confirms that the data under this application would only be used for the project described in this document. Individuals working on this project would only be permitted to access data relating to that project, as identified within the application. Access is granted to the data only to named individuals working on the project under authorized user names. Such access is password controlled (with a password reset required on a regular refresh). Only substantive employees of Imperial College London will use the disseminated data and only for the purposes described in this document. The raw data will be handled only within the Department of Surgery and Cancer to support academic research. The data will be received from NHS Digital and stored on a secure server hosted at the South Kensington campus of Imperial College. Access to data on this server is restricted to authorized individuals only. The data is accessed and processed by researchers who are based in rooms with keyless combination locks that are always locked when not in use. This access is password-based and permitted solely to registered users logging on via permitted IP addresses. Record level data will not be distributed to different parts of the organization. The data will not be made available to third party individuals, institutions or companies. No other data will be linked to this data though data will be compared at aggregate levels if required. The data will be processed as part of the above mentioned research project within the Department of Surgery and Cancer. It will be queried using data analytical tools such as SPSS, STATA, SAS, Microsoft Excel, Matlab, Python etc. to aid in answering specific research questions. Data visualizations will be done to present insights gained using suitable tools including Tableau, Inkscape and others. The specific processing activities will be as follows: 1. A patient level database for interhospital transfers of patients will be constructed, in addition to a range of utilization and outcome variables (e.g. length of stay, additional procedures, readmissions). A patient level database of maternity care will be constructed to examine patient choice in relation to delivery location. In both cases, directed unipartite networks of care transitions from one provider to another will be constructed and the characteristics of the network, providers and patients will be analysed using linear and logistic regression. 2. A patient level database of presentations to acute hospitals will be constructed. This database will be used to identify the probability of presentation of a patient in a particular geographic location to a particular centre with a particular diagnosis. These values will be used to undertake computational community detection algorithms to identify geographically nested networks of care providers to compare to existing predetermined care networks and guide the implementation of novel, more efficient networks of care. 3. A patient level database of outpatient, inpatient and A&E presentations will be created. The aggregate interaction across datasets between a particular geographic region (e.g. postcode, LSOA or primary care provider) and a secondary care institution will be used to generate a unipartite network of acute hospitals linked to one another by the strength of their shared patient activities. The network will then be interrogated to identify how patterns of patient flow will change in response to increased patient demand and altered provider capacity. Clusters of vulnerability in the network will be identified and optimal avenues for intervention will be suggested.

Objectives:

The Department of Surgery and Cancer, based at Imperial College London, is requesting data for use in the following research project: The Power of Connections: Mapping the Behaviour of Health Care Networks The purpose of this study is to examine how care providers in England are connected by virtue of the patients that flow between them. This request for data will, through the application of network analysis, provide insights into the factors determining how patients flow through the network and where the network may be particularly vulnerable will be identified. Strategies to improve the efficiency, equity and safety of the network may be developed and tested using predictive modelling in order to identify to optimal routes for investment and restructuring of the health care providers. This project will use the following data: HES OP 2011/12-2014/15, HES A&E 2011/12-2014/15 and HES APC 2011/12-2014/15. These four years of data are necessary to provide an adequate picture of health care utilization and capture less common events.


Project 17 — DARS-NIC-72318-M4W8J

Opt outs honoured: N

Sensitive: Non Sensitive

When: 2017/09 — 2017/11.

Repeats: One-Off

Legal basis: Health and Social Care Act 2012

Categories: Anonymised - ICO code compliant

Datasets:

  • Hospital Episode Statistics Admitted Patient Care

Benefits:

By highlighting the differences in cost associated with the different treatments to key decision makers through the dissemination strategy outlined above, it is the hope and expectation that decisions will be taken to adopt the intervention type that offers the highest value resulting in potentially significant cost savings. Savings of £820 per patient with index cholecystectomy have been estimated (Gutt CN et al. 2013). With 72,572 (http://www.rcseng.ac.uk/healthcare-bodies/nscc/data-tools) non-operative admissions with gallstone disease in 2014, the potential for savings of £59,509,040 exists. Therefore, the opportunity cost of reallocating resources towards higher value services is great and, although this work will not guarantee such efficiency savings, it will contribute to beginning conversations with policy makers and clinicians to optimise treatment and begin change management. This conversation would utilise evidence produced by this work. This project would not only provide knowledge of the cost of persistent interval cholecystectomy but also an understanding of how best to promote a change to index cholecystectomy. By providing a novel model of efficient de-adoption (which would be a specific output of this research) potential benefits may be extended to other low value procedures both surgical (e.g. arthroscopy in osteoarthritis) and non-surgical (e.g. use of antibiotics when not indicated.) The aim with this work is to explore the practice, purpose and experience of deadoption and to develop new tools and insights to help guide those trying to navigate this space. The expectation is that the papers would be published by October 2018, thereby impacting clinical activity by October 2019.

Outputs:

The following outputs will be produced: Models: Q3/2018 - A model for efficient de-adoption will be developed as part of this study Publications: It is intended that this study will lead to the following peer-reviewed publications which will be targeted for Health Affairs, the Lancet and the BMJ: Q3/2018 - Modelling of deadoption of low value procedures Q3/2018 - Adoption of high value procedures and geographical network analysis of diffusion of innovation Q3/2018 - Cost implications of non-deadoption of low value procedures Presentations: It is intended that this study will lead to presentations at the following conferences: Q2/2018 - The Association of Surgeons of Great Britain and Ireland - surgical conference Q2/2018 - Health + Care, Commissioning in Healthcare - conferences directed at healthcare commissioners Q3/2018 - The Association of Upper Gastrointestinal Surgeons - surgical conference Q3/2018 - Road to Rightcare, Overuse Conferences, World Congress on Health Economics - academic meetings Academic outputs: This study will contribute to a PhD thesis which will be published online. Target audience: The outputs of this study are aimed at those who will make use of the findings to decide the best course of care for patients. This includes surgeons who would be performing these operations, clinical commissioners who decide on priorities for funding and healthcare leads who can influence guidelines. This study is part of the Centre for Health Policy at Imperial College London which helps advise on global health policy, the Patient Safety Translational Research Centre which is one of 3 centres in the UK which translates research into clinical practice and the Global Health and Development Group which were formally part of NICE International which helped advise for local and global standards for clinical practice. All data which is used for outputs will be anonymous summary aggregate data. All outputs will contain only aggregate level data with small numbers suppressed in line with the HES analysis guide. No raw data will be transferred outside the BDAU SE and neither the data nor outputs will be used for commercial purposes.

Processing:

NHS Digital will securely transfer a pseudonymised extract of HES data and linked month and year of death to Imperial College London. Imperial College London will store the data on a server in the BDAU Secure Environment (SE). Data access is strictly controlled by the BDAU through a robust dataset registration process. No one other than BDAU staff can authorise access to the data. Access to the data will be restricted to one researcher, a PhD student, and that researcher’s supervisors if necessary (usually not required), only for the purpose outlined in this Data Sharing Agreement. The student and the supervisors are bound to the policies, procedures and equivalent controls of the BDAU SE and Imperial College London as substantive employees of the College. The raw data provided by NHS Digital will be analysed solely in the BDAU SE. Any further analysis done outside the BDAU SE (usually for visualisation purposes for output) will be done using data that has been aggregated with small numbers suppressed in line with the HES Analysis Guide. The data will be analysed to investigate the two interventions and associated cost, utilisation and outcome patterns. This will involve statistical analysis using standard and innovative statistical programmes inside the BDAU SE. Cost data sourced from the freely available ‘National Schedule of Reference Costs’ will be integrated into the raw dataset on an intervention level. Each intervention will be costed according to its relevant healthcare resource group (HRG); which is a reimbursement tariff of the average unit cost to the NHS of providing a defined service in a given financial year. This will not increase the risk of re-identifying individuals. Results will be graphed and compared at an aggregate level. The importance of longitudinal data (10 years) in this scenario is to capture the change in clinician & institution behaviour following the publication of a Cochrane review in 2006 (Gurusamy et al. 2006) which recognised interval cholecystectomy as being low value. Today interval cholecystectomy is still being used despite growing evidence to the contrary. The goal of creating a model of deadoption requires longitudinal data in order to illicit the changes in rates of use and to identify whether changes have been sustained. Without longitudinal observation an incomplete picture would be shown and the study would be unable to formulate recommendations to supply side policy change which is an ambition of this project. there will be no linkage with other record level data and no attempt to re-identify the data.

Objectives:

Imperial College London’s Big Data and Analytical Unit (BDAU) requires an extract of HES data with linked month and year of death (where applicable) for use in a research study: ’ Evaluating the Rate of Deadoption of Interval Cholecystectomy, a Low Value Intervention, and Diffusion of Index Cholecystectomy, a High Value Intervention’. This study aims to investigate the respective values of two ways of treating patients with the same specific health condition. This study is being undertaken by a PhD student within Imperial College London and will contribute to a PhD thesis in addition to other outputs intended to maximise the benefit of the work. The Big Data and Analytical Unit (BDAU) is a multidiscipline team within Imperial College London which collaborates with a large network of researchers across the college with the aim of ensuring the maximum use, impact and dissemination of research using healthcare data. Finding efficiency savings in health care provision is paramount given the pressures on national health care budgets worldwide. This provides motivation to identify and reduce the use of health care interventions that deliver only marginal benefits, be it through overuse, misuse or waste, that could be substituted by less costly alternatives without affecting safety and quality of care. A greater emphasis on value is key and achieving high value for patients must become the goal of health care delivery. A clinical definition of low value interventions has been established as care in the absence of a clear medical basis for use or when the benefit of therapy does not outweigh risks; this encompasses terms such as medical overuse and over-diagnosis. The importance of identifying and studying low value healthcare services is motivated by the concept of ‘opportunity cost’, i.e. that disinvestments in low value procedures and services from the healthcare budget leads to the opportunity for further investments in higher value services. That is, a reduction in low value service results in improved value of care overall. This study aims to investigate the relationship between two interventions for a cholecystectomy – a surgical procedure to remove the gallbladder. Interval cholecystectomy is a low value intervention, while the other, index cholecystectomy is considered a high value intervention. Interval cholecystectomy is the choice to discharge patients following index admission and readmit them for an elective operation whereas index cholecystectomy is performed during index admission. These two interventions would be analysed to inspect the patterns of deadoption and adoption respectively. Cost analysis will be used to compare the two interventions and this will take into account the impact of adverse events, readmissions and excess mortality to ensure that costs and impacts are both analysed. These analyses will then inform outputs which will be used to help change current practices and improve patient care in respect of this particular condition. The study requires the hospital admissions data of any individual who had a procedure (defined by specific procedure codes) indicating a cholecystectomy treated by trusts which had 10 or more operations per year for the relevant procedure codes. Details of all hospital admissions for these individuals over a period of up to 10 years will be required because the study will take into account the possible relationships between the cholecystectomy and other admissions to ascertain the true costs of each type of intervention.