UK Prospective Diabetes Study (UKPDS) Legacy Study: long-term follow-up of participants into electronic health records

The purpose of this application is to link participants in the UK Prospective Diabetes Study (UKPDS) trial to all death and hospitalisation health records in order to measure the effect of the treatments in this study on death, major medical illnesses, and costs of treatment.

The use of the data requested in this agreement is in accordance with Article 6(1)e as processing is necessary for a task carried out in the public interest. The aim of the study is to provide reliable evidence about the very long-term effects of blood pressure and glucose-lowering treatments on important health outcomes in people with diabetes. This promotes the public interest aiming to improve the treatment of estimated 5 million people with diabetes by the year 2025. Further details can be found on the UKPDS website (https://www.dtu.ox.ac.uk/OurTrials/UKPDSLegacy)

Processing of the data is in accordance with Article 9(2)(j) exemption, i.e. that the processing of the data is necessary for scientific research. The scientific aim of the study is to provide reliable evidence about the very long-term effects of blood pressure and glucose-lowering treatments on important health outcomes. This will help to inform people with diabetes about the long term consequence of blood pressure lowering and glucose lowering treatment, so they can make better decisions about their care. Further details can be found on the UKPDS website (https://www.dtu.ox.ac.uk/OurTrials/UKPDSLegacy)

The research team have taken the opinion of the South East Scotland Research Ethics Service (18/SS/0127), who have granted approval to the study in its current form, and of the Confidentiality Advisory Group (CAG) (18/CAG/0182).

The University of Oxford propose to link participants in the trial to their health data held within hospitalisation and death records curated by NHS England. By doing so, The University of Oxford will be able to determine the effects of higher versus lower blood pressure and glucose on future health and health care resource use. The University of Oxford will compare the incidence of major disease in different arms of the trial, in participants with different baseline factors (both clinical and genetic), and with difference disease occurrence during follow-up.

The UKPDS was a randomised, multi-centre trial of glucose-lowering and antihypertensive therapies in 5,102 patients with newly diagnosed type 2 diabetes that ran in 23 clinical centres for twenty years from 1977 to 1997.

UKPDS has been a major project, that has influenced diabetes guidelines in the UK and around the world (www.dtu.ox.ac.uk/ukpds). Both the Health Economic Research Unit and the Diabetes Trials Unit have a strong research interest in determining the effects of treatments for diabetes, and have published many influential studies.

The UKPDS trial investigators would therefore now like to evaluate the effects of the randomised treatments on dementia and other measures of long-term health. Therefore, the University of Oxford propose the extended follow up of all UKPDS participants into Electronic Health Records and other routinely collected health data.

The purpose of the project is to:

1. To determine whether participants randomly allocated to tight, rather than less tight, blood pressure control have a lower risk of dementia.

2. To determine whether participants randomly allocated to intensive, rather than conventional, glucose control have a lower risk of dementia.

3. To determine whether tight blood pressure control or intensive glucose control reduces the long-term risk of major vascular diseases in diabetes.

4. To determine whether tight blood pressure control or intensive glucose control reduces long-term health resource use and total burden of disease in diabetes.

5. To investigate use of health care resources in secondary care by patients with diabetes.

University of Oxford decided against requesting Mental Health data because it was not available for such a long time, and varied in coverage over the potential period of follow up. It was therefore determined that HES and mortality data will hold enough dementia data for the purposes highlighted above.

The study’s purpose is to examine the effect of baseline clinical variables on major health events, including major vascular events (including, but not limited to strokes of different types and myocardial infarction), cancers, renal disease and dementia. In order to ascertain these events, chiefly though ICD-10 coding. ICD-10 consists of a tabular list of diseases, the study team propose that it accesses hospital admission, and death records. The study team have found in previous work that the majority of cases of chronic disease (including dementia) can be found in admitted patient care, and death records. Data is needed on individuals in order to adjust for differing factors at baseline. The chief comparison will be between participants allocated to blood pressure and glucose lowering regimens and participants allocated to control regimens.

The University of Oxford propose that the linkage of individual participants by their identifying information (name, NHS number, date of birth, sex) should be performed by NHS England. The University of Oxford propose that NHS England return data to the university at the Nuffield Department of Population Health at the University of Oxford with each participant’s trial identity number with data from the linkage, i.e. a pseudonymised level of data.

The study team propose that to obtain data from each participant from the time they began to take part in the trial (the first participant was randomised in 1977) to the date of linkage. This will allow them to: (i) compare the occurrence of events recorded by the study during follow up with events recorded by Electronic Health Records; (ii) compare the very long-term incidence of major disease by randomised treatments and baseline factors. Further data releases will allow the study team to continue follow up into the extremely long term.

Participants were resident and recruited from across the UK, who may have moved between nations since recruitment, and therefore the data held will be linked across the country.

Linking participants to their data held by NHS England is the least intrusive way of achieving the stated purposes; in fact there is no other way that this could be performed without leading to considerable bias in the assessment of important outcomes. The University of Oxford believe that re-approaching participants for further consent would lead to potentially very large biases in ascertainment of major health conditions (because of non-responder biases particularly amongst those in poorest health, or have died), potentially lead to distress, and therefore have sought permission from the Health Research Authority (HRA) CAG for this linkage. The University of Oxford propose to link only to those datasets that contain information on medical diagnoses.

The Data Controller and Processor will be the University of Oxford. The Primary Investigator is employed by The University of Oxford. All processing of NHS England data will be within the University of Oxford. Analyses will be performed within the University of Oxford department the Nuffield Department of Population Health, by researchers within that department and researchers with contracts with that department. Advice on analysis and manuscript will be provided by researchers within the Diabetes Trials Unit. The project is funded by the Nuffield Department of Population Health.

The funders of the study and individual outside the study team will only have access to tabular data with small numbers suppressed. This so that people who wish to meta-analyse studies, who are currently not predictable (it may be no-one), are able to do so. The type of data that would be shared would be very similar to what would be in a published paper. Sharing this level of data is very important for openness of the scientific endeavour and would suppress small numbers.

The Nuffield Department of Population Health (NDPH), University of Oxford will transfer participant identifiers with linked trial ID numbers to NHS England through a secure route approved by the receiving bodies. The entire UKPDS cohort (5102) will be transferred, in case of relocation of Scottish participants to England. The University of Oxford will receive data back to the Nuffield Department of Population Health in an encrypted format via NHS England encrypted transfer solution. The data will be encrypted during receiving, storage and processing. Each participant will be identified by trial identifier only at this stage, not with name, date of birth etc. The University of Oxford will construct a dataset from the original UKPDS trial dataset with covariates for this analysis, where each participant is identified only by the anonymised trial identifier, including baseline clinical, genetic, and event-based data. Data on individual participants held in the Nuffield Department of Population Health will be linked to data on the same individual provided by NHS England. Analyses will be performed with these data to achieve the stated scientific aims of estimating the effects of treatment in the UKPDS on major health outcomes, and health economic outcomes. All analyses will be performed using pseudo-anonymised trial identifiers.

All data will be transferred, handled and processed in agreement with the NHS England Data Sharing Framework Contract. Data will be received by the NDPH, University of Oxford. Using existing ICD-10 code lists for different clinical phenotypes, the study team will define the date and the nature of disease events for each participant. The study team will link the list of processed disease outcomes with existing datasets of baseline clinical, genetic and biomarkers data, and the occurrence of disease events within trial. Data are pseudonymised prior to analysis.

The Nuffield Department of Population Health, University of Oxford will link this dataset with the information received from NHS England with pseudonymised trial identifiers.

NDPH has successfully acquired analysed and appropriately stored data from HES for previous large long-term studies such as HPS2-THRIVE and HPS3-REVEAL. NDPH researchers are experienced in handling confidential and participant sensitive data and have appropriate training in Information Governance.

The NDPH servers are protected against unauthorised external access by an appropriate strength firewall. Access to patient identifiable information is protected by the appropriate authentication procedures (user IDs and passwords). Authentication is only given to personnel with a need to access the required data. Only personnel involved in the long-term follow-up study for UKPDS (processing and analysing data) will have access to this data, which will be members of the statistical and health economics teams NDPH has a Corporate Level Security Policy that has been fully adopted by management and will apply fully to the long-term follow-up study. The data protection Registration Number is Z575783X. NDPH investigators are fully aligned with all data management and security policies.

Identifiers will need to be retained whilst linkages are made between UKPDS datasets (the University of Oxford anticipate this will take up to a year) before all data are identified primarily with the study ID, in order to pseudo-anonymise the long-term follow-up dataset. This may not be necessary, but if there are linkage problems, or inconsistencies in the data provided, the identifiers may need to be provided again to NHS England.

The study team will link data from individuals to existing trial databases, which hold information on the baseline clinical, genetic, biochemical, and Health Economic characteristics of participants, and in-trial health events.

Data processing will only be carried out by substantive employees of the data processor(s) and or data controller who have been appropriately trained in data protection and confidentiality.

No data will be stored at premises which are owned by an organisation which is not named in the agreement. All information is stored securely by University of Oxford and is kept confidential. Access to the computer database is by unique combinations of usernames and passwords and only authorised study personnel can access information about participants. The building is secure with authorised swipe card access only. No individuals will be identified in any study reports.

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

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

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

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

The University of Oxford expect at least 5 publications to address each of the following questions:

1. To determine whether participants randomly allocated to tight, rather than less tight, blood pressure control have a lower risk of dementia.

2. To determine whether participants randomly allocated to intensive, rather than conventional, glucose control have a lower risk of dementia.

3. To determine whether tight blood pressure control or intensive glucose control reduces the long-term risk of major vascular diseases in diabetes

4. To determine whether tight blood pressure control or intensive glucose control reduces long-term health resource use and total burden of disease in diabetes

5. To investigate use of health care resources in secondary care by patients with diabetes

6. To investigate the concordance of outcomes recorded within a trial and administrative data

The University of Oxford are close to completing the manuscripts describing the analyses in points 3 & 4, these have been presented at a major diabetes conference (EASD) and are being written up with the intention of achieving publication in Q4 2023/Q1 2024.

The UKPDS trial investigators will publish these outputs in major clinical journals (e.g.,NEJM, Lancet, BMJ, Stroke etc.) and present them in international meetings, with publication targets annually over the next five years.

It is anticipated these would contribute to national guidelines on reducing risk in people with diabetes, and will be communicated to charities (e.g., Diabetes UK) that are active in giving advice to people with diabetes.

The data will be aggregated with small number suppression in accordance with the HES analysis guide. The University of Oxford will not present data on individual participants. however, will present actual and modelled data in graphical and tabular format in line with the HES analysis guidelines on suppression.

The Diabetes Trials Unit and NDPH contribute widely to health policy, particularly in the area of vascular risk prevention. They contribute to debate with academic papers, conference participation, lectures to the public and advice to government (including NHS England). The University of Oxford will share all outputs through all of the listed channels: website and newsletters, open lectures and talks, exhibition at public events, posters, press/media engagement and other public promotion of the research, stakeholder mailing list.

Most people in the UK with type 2 diabetes require treatment incorporating a healthy lifestyle combined with pharmacological therapies to attain good control of blood glucose and blood pressure. The short to medium term benefits of good (relative to less good) control have been proven in trials, but longer term benefits require more evidence. Therefore, the very long-term effects of these agents are of great interest to patients and potential patients, payers, and clinicians in primary and secondary care. University of Oxford would like to continue work on this clinical trial, the UKPDS (UK Prospective Diabetes Study) with routinely collected NHS outcomes reflecting the primary endpoint of the trial.

The results of these analysis, which will concentrate on the effects of blood glucose and blood lowering on major health events, will inform and determine policies in clinical guidelines and whether to fund new therapies to prevent or delay the complications of diabetes. University of Oxford are particularly interested in dementia, cancer, and other major diseases which are more common in people with diabetes compared with people without diabetes. The planned health economic analyses may lead to a better understanding of the long-term economic benefits of starting treatments at different stages of life.

The research team believe that the results of this research will be of benefit to 3.5 million people with type 2 diabetes in the UK considering the long-term benefits and harms of vascular secondary preventative medication, for example those in mid-life who are considering intensive control of their diabetes or blood pressure lowering medication.

Determining precisely the magnitude of any benefit in controlling blood glucose and blood pressure in delaying or preventing dementia is a priority for public health.

If the control of vascular risk in mid-life could be shown to prevent or delay dementia in type 2 diabetes, this would have considerable implications for the global burden of vascular-mediated dementia and for public policy. It would expand the number of people eligible for intervention at a younger age; and identify new methods to target interventions (currently targeted based on absolute risk of heart attack and stroke, not dementia).

The cost-savings estimated from risk factor control (£60 million saved for every year’s delay in dementia in the 1% of the general population with vascular risk factors) have been estimated solely from observational data (nice.org.uk/Guidance/NG16). Data from the analyses might substantially modify these estimates, giving NICE and NHS commissioners better evidence to weigh the benefits of glycaemic and blood pressure control related to type 2 diabetes.

The UK Prospective Diabetes Study (UKPDS) was a landmark randomised, multicentre trial of glycaemic therapies in 5,102 patients with newly diagnosed type 2 diabetes. It ran for twenty years (1977 to 1997) in 23 UK clinical sites and showed conclusively that the complications of type 2 diabetes, previously often regarded as inevitable, could be reduced by improving blood glucose and/or blood pressure control. The study has led to 122 manuscripts as of October 2023, covering many aspects of diabetes care (https://www.dtu.ox.ac.uk/ukpds/) The key benefits for patients with type 2 diabetes was to demonstrate that lower blood pressure and more intensive glucose control were of benefit. These treatments are the cornerstone of modern diabetes practice.