Evaluating the NHS Diabetes Prevention Programme (NHS DPP): the DIPLOMA research programme (Diabetes Prevention – Long term Multimethod Assessment)

This Data Sharing Agreement permits access to two datasets: National Diabetes Audit (NDA) and Non-Diabetic Hyperglycaemia (NDH). These datasets are collected from the same sources at the same time. NHS England treats them as separate datasets because one (NDA) is for collected for the purpose of auditing diabetes and the other (NDH) is by definition a dataset of people without diabetes.

The University of Manchester requires extracts of NDA and NDH data for the purpose of the DIPLOMA (Diabetes Prevention - Long term Multimethod Assessment) research programme, the aim of which is to provide a comprehensive assessment of the implementation, delivery and outcomes of the NHS Diabetes Prevention Programme (NHS DPP) to inform commissioning.

Type 2 diabetes is a common health condition that can cause serious health problems, reduce people’s quality of life and reduce life expectancy, as well as costing a lot of money to treat. Some people are at higher risk of diabetes and can be identified with a blood test. If people at risk make changes to their lifestyle (more exercise, losing weight), they can substantially reduce their chances of getting type 2 diabetes. The NHS and Diabetes UK have introduced a new scheme called Healthier You: the NHS Diabetes Prevention Programme (NHS DPP). People in England who are at risk of type 2 diabetes, referred to as having Non-Diabetic Hyperglycaemia (NDH), will be offered a practical course which will help them change their lifestyle. The aim is that this course will improve peoples’ health and reduce their diabetes risk.

The people managing the NHS DPP need to know if it really works to prevent diabetes and whether it is a good use of NHS resources. The DIPLOMA programme was designed to help answer these questions.

Personal data can be lawfully processed for this purpose because it is a task carried out in the public interest (General Data Protection Regulation Article 6 (1) (e)). The task concerns the enactment of the NHS Diabetes Prevention Programme and the performance concerns the use of this programme in the population and sub-groups of the population of England.

Processing of the data is necessary for scientific research purposes and will be used to evaluate the effectiveness of the NHS Diabetes Prevention Programme (General Data Protection Regulation Article 9 (2) (j)). The DIPLOMA (Diabetes Prevention - Long term Multimethod Assessment) project has been funded by the National Institute for Health Research to evaluate the NHS Diabetes Prevention Programme and is comprised of eight work packages.

The data required are from the National Diabetes Audit (NDA) dataset which contains details of individuals with all types of diabetes and the Non-Diabetic Hyperglycaemia (NDH) which contains data on individuals at risk of Type 2 diabetes with.

The datasets will be used to identify which individuals with NDH go on to develop type 2 diabetes.

The dataset called the Minimum Data Set (MDS) will be linked with the NDA and NDH data. All data will be pseudonymised and linked using common pseudo-IDs applied by NHS England. The linkage of these datasets will enable the University of Manchester to track which NDH patients were offered a place in the NHS DPP; which patients accepted the offer, and which patients were not offered a place.

The MDS contains participant information that providers of the DPP are contractually obliged to collect in order to receive payment. It contains the date that participants were referred to the programme, as well as information about how many sessions they attended, as well as certain outcome measures, as recorded by the providers of the programme. Importantly however, this dataset contains no information on the wider eligible population who may benefit from the programme, only those that have been referred. To accurately evaluate the efficacy of the NHS DPP, it is important to know the characteristics of the eligible population.

The NDA dataset also contains information about the characteristics of the people with NDH or diabetes and, in combination with the MDS, this can be used to identify trends such as inequalities in access to the NHS DPP and variations in outcomes of the NHS DPP linked to certain characteristics. The data contains demographic and socio-economic characteristics of patients which can be used to assess the characteristics of those with the condition and whether any activity delivered is more or less prominent based on these characteristics. This data will also provide national level data to facilitate the calculation of statistics on the prevalence of NHS DPP participants, the characteristics of those who go on to develop Type 2 diabetes, as well as inform parameters to be used in a decision-analytic model, which will be used to assess the long term cost-effectiveness of the programme.

The data subjects are all individuals in the NDA dataset aged over 15 years from 2017/18 to 2020/21 with type 2 diabetes and all individuals in the NDH dataset aged over 15 years from 2017/18 to 2020/21 with non-diabetic hyperglycaemia.

Under this Agreement the University of Manchester request three additional data fields: Type 2 Diabetes diagnosis date, CCG code and birth date. The University of Manchester request the full diagnosis date of type 2 diabetes, rather than the year of diagnosis (currently provided in the datasets) as the analysis is being conducted in terms of financial years. This is to match how the DPP was rolled out across England, in waves, with each new wave starting in April. Secondly, the University of Manchester request CCG code as the DPP is delivered at the CCG level. Currently the University of Manchester have appended the data with GP/CCG lists, but the provision of CCG in the dataset itself would be beneficial. Lastly, the University of Manchester request year of birth. The University of Manchester currently have a data field ‘age’, however, as the University of Manchester do not have the audit year for the data records, this makes it difficult to define age consistently across the three datasets (NDA, NDH and MDS). This could be rectified with the year of birth field.

The data will be used in three of eight work packages in the overall DIPLOMA research project. These are:

• Work package 1: Access and Equity – the aim is to assess the accessibility of the NHS DPP and identify inequalities in access.

• Work package 5: Comparative Effectiveness – the aim is to examine whether the NHS DPP leads to a reduction in the prevalence of Type 2 Diabetes and other outcomes related to Type 2 Diabetes compared to those without access to the NHS DPP.

• Work package 7: Economic Evaluation – the aim is to explore the cost-effectiveness of the NHS DPP, from the perspective of the NHS and Personal Social Services.

Further details on these work packages are provided below.

Quality will be assessed in terms of equitable use of the NHS Diabetes Prevention Programme (for Work Package 1); the effectiveness of reducing the prevalence of Type 2 diabetes (Work Package 5), and whether the programme is considered to be cost-effective (Work Package 7).

The DIPLOMA was originally funded by the National Institute of Health Research (NIHR) to be delivered by researchers at the University of Manchester from 2017-2021. The research project has since been extended (DIPLOMA 2), to include an additional component, evaluating the Digital component of the Diabetes Prevention Programme. Its aim is to produce a comprehensive evaluation of the NHS DPP, comprised of eight distinct but related work packages, using various research techniques and different sources of data. The DIPLOMA research programme is designed to provide:

(a) feedback regularly to NHS DPP stakeholders on the delivery and outcomes of the programme to support ongoing development and quality improvement,

(b) a rigorous longer-term assessment of the success of the NHS DPP in meeting the aim of reducing diabetes prevalence in a way that is cost-effective and sustainable for the NHS.

The University of Manchester is the sole data controller. There are two joint Principal Investigators leading the DIPLOMA work programme. Both are substantive employees of the University of Manchester. The Contract with the Department of Health and Social Care for undertaking the evaluation of the NHS DPP was signed with Salford Royal NHS Foundation Trust (SRFT) as the Contractor. The SRFT are subcontracting all of the work to the University of Manchester. It is agreed between SRFT and University of Manchester that all the research and decisions on how this is conducted, including the data management, are the University of Manchester’s responsibility, whilst SRFT’s role is essentially to administer the contract. The University of Manchester is reporting directly to the NIHR, not through SRFT.

The University of Manchester is the sole organisation processing the data. No other organisations process the data for this purpose.

The Department of Health and Social Care and Public Health England are in liaison with the DIPLOMA evaluation. These advise on approaches taken and provide updates on the NHS DPP. The findings of the DIPLOMA study will be distributed to these wider organisations but no data will be made available to them.

A Professor employed by the University of Oxford is providing consultancy for work package 7, as he is an expert in developing economic decision models for diabetes. He will advise the University of Manchester DIPLOMA researchers involved in work package 7, on the methods used, but will not have any access to the data. His role is strictly advisory.

A Professor from the University of Sheffield is also funded as a consultant on the DIPLOMA project, because of his clinical expertise. He will not have access to the data under this Agreement and has purely an advisory role.

The data requested will not be used or accessible to other work packages within the project beyond those mentioned in this data request (Work packages 1, 5 and 7). The anonymous findings (aggregated with small number suppression) may be used to inform the work of other work packages. The data is restricted to only those with type 2 diabetes and those identified as at risk of type 2 diabetes (Non-diabetic hyperglycaemia). Data has been further minimised by only selecting the fields required for analysis.

The following provides additional detail on the three work packages for which the data will be used:

Work Package 1 – Access and Equity:

This work package is interested in the accessibility of the DPP. The work package will use the NDA and NDH data as well as national survey data (accessed via the UK data archive) to explore the characteristics of patients who enter the NHS DPP, and those who are eligible and do not enter. Inequalities in health have been shown across England. There are many reasons why inequalities may exist, but one of the reasons might be because there is inequality in a patient's access to health care. Access may be related to multiple factors: availability of the NHS DPP in the area; awareness of NHS DPP among patients and professionals, acceptability of NHS DPP to professionals and patients; and the perceived ‘costs’ of NHS DPP (such as taking time off work).

To understand any issues in access to the NHS DPP and the impacts of the programme on diabetes incidence, assessments will be made to understand whether inequalities in protected characteristics exist for:

1. the identification of patients eligible for the programme (‘NHS DPP eligible’ patients)

2. the referral of patients to the programme (‘NHS DPP attenders’ and ‘non-attenders’)

3. programme delivery and completion

4. the effectiveness of the programme

The evaluation concerns the population that are non-diabetic hyperglycaemic (NDH), this patient group will start to be collected from 2017-2018 as part of the NHS Diabetes Prevention Programme fields (DPPOFF_COD; DPPATT_COD and DPPCOMP_COD).

The University of Manchester intends to compare the patients identified as non-diabetic hyperglycaemic (NDH) to populations identified in national surveys to identify whether identification of patients is representative of the general population. The national surveys are the UK Household Longitudinal Survey (UKHLS); the Health Survey for England (HSE), and the English Longitudinal Study for Ageing (ELSA). These population surveys are accessed via the UK Data Archive and are not controlled by NHS England.

Comparisons of uptake and completion of the NHS Diabetes Prevention Programme, by patient characteristics will inform whether uptake and completion of the programme differs by certain types of patients.

The data will be analysed in order to measure differences in the prevalence of prediabetes (NDH) by age-groups, gender, ethnicity and area deprivation, as well as differences in the treatment availability and uptake by age-groups, gender, ethnicity and area deprivation. In order for this to occur, pseudonymised patient-level data is required.

There will be two main strands of analyses:

1. Identifying the population with NDH

The data will be used to describe the characteristics of patients with NDH. This will involve presenting the share of patients in the data by the following patient characteristics:

[AGE]: Year of age

[DERIVED_CLEAN_SEX]: Gender

[IMD_QUINTILE]: IMD quintile

[DERIVED_CLEAN_ETHNICITY]: Ethnicity

[LD]: Learning disability flag*

*The learning disability flag is required as this is an NHS protected characteristic. As such, and in line with the other protected characteristics requested such as ethnicity, age etc., the analysis seeks to understand whether access to the DPP is associated with this characteristic.

The distribution of people with NDH by their age, gender, IMD deprivation quintile, ethnicity and disability status will be examined. Comparisons of these distributions with data from three population survey datasets (UKHLS, HSE and ELSA) and to those patients under the NHS diabetes prevention programme (the ‘Minimum Data Set’/MDS). The datasets will be pooled and statistical tests will be conducted to ascertain whether distributions of people by age, gender, IMD deprivation quintile, ethnicity and disability differ between the datasets. The analytic approach will not involve linking individuals across datasets.

Additional analyses will stratify patients with NDH on the basis of various clinical measures concerning blood pressure, body mass index, and cholesterol to further see whether survey data is a good approximation of the population with NDH.

These additional analyses will inform whether certain types of patients with NDH are over- or under-represented in the NDA/NDH data in comparison to survey data.

2. Identifying whether inequalities in programme participation exist by patient characteristics.

Additional analyses will test for associations between NHS Diabetes Prevention Programme participation and explanatory variables.

The distribution of NHS Diabetes Prevention Programme outcome variables by patient characteristics (age, gender, ethnicity, IMD deprivation and disability status) and additional explanatory variables would be processed in order to see whether there are particular groups of patients ceasing to attend and complete the programme.

The University of Manchester request years 2017/18 to latest available at the time of NDA and NDH data to carry out this work with the linked Minimum Dataset. The linked datasets will be pseudonymised as none of the datasets contain patient identifying information.

To ensure comparability with survey data the full geographic spread of patients in the data are requested. No alternative data would enable the proposed evaluation to be performed. This is because the data requested covers all general practice. In addition, the analysis requires individual-level data to enable comparisons of patients by age, gender, ethnicity, IMD deprivation, and disability status.

Work Package 5 - Comparative Effectiveness:

Work Package 5 seeks to examine whether the NHS DPP achieves its primary objective of reducing the prevalence of Type 2 Diabetes and other outcomes related to Type 2 Diabetes such as hospitalisation, primary care visits, HbA1c and BMI/weight levels, and death, compared to those without access to the NHS DPP. To do this, the University of Manchester will use NDA/NDH data for the years 2017/18 to 2020/21 (from when the NHS DPP programme has been implemented) of NDA/NDH data, with the linked Minimum Dataset. The DPP is implemented nationally and therefore data is required at national level, with regions where the NHS DPP was not implemented or not part of Wave 1 being used as comparators.

The conversion rate from non-diabetic hyperglycaemia (NDH) to Type 2 Diabetes will be compared between people referred to the NHS DPP, those not referred and those declining referral. Matched cohort designs will be used to compare the rates of conversion to diabetes within a year of NDH diagnosis, across these three groups. The University of Manchester intend to match both within practices and also across practices (accounting for their time of participation in the DPP) to better account for unmeasured confounding. Preliminary work will examine overall conversion rates and the prevalence of NDH nationally and regionally.

Specific Aims:

1. To estimate the prevalence of Non-diabetic hyperglycaemia (NDH), i.e. people at risk of diabetes.

2. To identify the characteristics associated with NDH (e.g. deprivation, age, sex).

3. To calculate the conversion rate from NDH to Type 2 Diabetes, nationally and regionally.

4. To assess whether the NHS DPP is more effective than usual care in reducing conversion of NDH to diabetes, thus eventually reducing diabetes prevalence in England.

CPRD data will also be used for this comparison. However, CPRD data does not provide national coverage and the numbers are relatively small. The data will not be linked to anything at the patient level. At the practice level data will be linked to practice characteristics that are freely available through NHS England and the ONS: practice list size, rural/urban, practice location deprivation (IMD 2015). There is no additional identification risk as linkages are not at the patient level.

Work Package 7 – Economic Evaluation:

Work Package 7 aims to explore the cost-effectiveness of the NHS DPP, from the perspective of the NHS and Personal Social Services. It is important to evaluate if the NHS DPP represents good value for money, as the resources to provide healthcare are limited, and so it is important that funds are allocated to the most effective intervention or prevention strategies.

To evaluate the cost-effectiveness of the NHS DPP, the analysis will take on two main strands:

1) An economic evaluation using observed data from the MDS and

2) A long-term decision analytic model, evaluating the theoretical impact of the NHS DPP over the lifetime of participants, compared to those who do not enrol in the programme.

The NDA/NDH data is required to inform parameters in the long-term decision model. Decision-analytic models are frequently used within economic evaluations, as often the benefits of a programme are not entirely captured within the relatively short time horizon that data is captured. Therefore, decision-analytic models are used to extrapolate beyond observed data, to evaluate final outcomes which are of economic interest. This often involves modelling a hypothetical cohort over a lifetime time horizon. Inputs into decision-analytic models include parameters such as the probabilities of different events occurring (referred to as transition probabilities), the costs associated with events or resource use, and utility scores, which can be used to estimate quality adjusted life years.

Whilst several models evaluating prevention strategies for Type 2 diabetes have been developed, there is currently no model that evaluates the NHS DPP, and as such data from the NDA/NDH is required to inform parameters, such as the number and characteristics of patients being offered the NHS DPP and the uptake rate. Information on the time taken to develop Type 2 diabetes is also needed to produce survival/time to event analysis curves, and use these rates to calculate transition probabilities, to be used within the model. As a standard, the benefits of the NHS DPP will be evaluated in terms of the quality adjusted life years generated by the programme, in comparison to not attending. As such, the data within the NDA/NDH will be used to produce quality of life estimates using a validated mapping tool and regression analyses to predict EQ-5D index values from the data.

The NDA is the only dataset with national coverage of those diagnosed with NDH. CPRD is no longer reflective of the general population, and not all GP practices that have participated in the NHS DPP are captured within it. The required information is also not available within the MDS, as this dataset only has information on patients who have taken up the invitation to enrol on the programme, and therefore does not inform the number, or the characteristics of those eligible (patients with non-diabetic hyperglycaemia) nor the number of people who were offered but refused to attend (available in the NDA by looking at the differences between numbers offered and accepted). The publicly available extract of the NDA is insufficient in the data it provides, as it does not provide the characteristics of those offered the NHS DPP, only the summary characteristics of those diagnosed with NDH, across limited variables, such as BMI and age groups.

Therefore, pseudonymised patient level data for the years 2017/18 to 2020/21 are required, with the linked Minimum Dataset. This linkage will facilitate the analysis of the programme, showing the characteristics of the eligible population (from the NDA/NDH), the subset within this who have participated in the NHS DPP, along with information on their adherence to the programme (from the MDS). As the NHS DPP has been rolled out across England, NDA data from across the whole of England is required.

• Work Package 1 aims to produce a series of reports and journal articles investigating social inequalities in NDH and how they are affected by the NHS DPP. Journals that will be targeted for these publications will be: Diabetic Medicine and British Journal of General Practice. Journal article 1 submission date will be around December 2021.

• Work Package 5 will disseminate the projects outputs through a variety of media, including conference presentations and conventional academic publications, seminars and short accessible reports for stakeholders, and plain English summaries and podcasts for patients and the public. The research team are experienced at writing for a wide variety of audiences. The University of Manchester will work closely with the University of Manchester’s Stakeholder Advisory Group and the Research Advisory Group to maximise the utility of the dissemination. The University of Manchester also plan to disseminate the research findings by publications in peer-reviewed journals and presentations in relevant conferences organised by the Society for Academic Primary Care (SAPC), the Royal College of General Practitioners (RCGP), the Royal College of Physicians (RCP) and the Royal Statistical Society (RSS). When reporting, the principles outlined in the STROBE and RECORD statements will be followed. Data will not reported in the outputs only summary statistics (descriptive statistics) and model estimates. The University of Manchester plan to submit a paper around December 2021 with another targeting the outputs from this work package by October 2022.

• Work Package 7 intends to prepare a report of the findings and recommendations of the cost-effectiveness of the NHS DPP. All reports are submitted to the Director for the NHS Diabetes Programme at NHS England, with the final report expected in September 2022. Results of the analyses will also be presented to the Expert Reference Group, which has representation from the three DPP funders: NHS England, Public Health England and Diabetes UK. A report will also be prepared which outlines the findings of the long-term economic model, which subject to acceptance, will be published in a high impact peer review journal such as Value in Health or Diabetes, Obesity and Metabolism. Presentations will be given at national and international conferences to audiences of academics and health professionals. These conferences will include the Mt Hood Diabetes Challenge Network conference, where the long term economic model will be presented, and the Health Economists’ Study Group (HESG) conference to present initial findings of the cost-effectiveness of the NHS DPP (both subject to abstract acceptance).

Reports will be provided for the DIPLOMA research programme on the progress and outcome of the research. Target report publication dates are as follows:

a) The DIPLOMA research programme final report: September 2022.

Reports and journal articles will be published online and made available to all. Submitted versions of all journal articles will be freely available on the University of Manchester library website.

With the addition of NDA 2020/21 data to this agreement, the University of Manchester expects similar outputs as outlined above in respect of this latest available data.

Researchers at the University of Manchester were able to access the data on the 3rd November 2020. It then took considerable time to format and validate the dataset, and to receive a working linkage file between the NDA/NDH and MDS datasets. Since then, the university was subject to a cyber attack (May 2023) and as a precautionary measure access to the data safe haven (where the NDA data is stored) was revoked for a year, meaning a pause in the planned data analysis. Access is in the process of being reinstated, but it is because of this delay that the University of Manchester are seeking an extension of one year.

The University of Manchester has one manuscript currently in submission and another planned to be submitted soon, that requires access to and the analysis of the NDA/NDH dataset. This is needed in case any further analyses are requested during the journal peer review process.

To date, there have already been three papers published reporting analysis of this dataset, as well as presentations to the Health Services Research UK conference (July 2021), Health Economists’ Study Group (June 2023) and at two university seminars (University of Manchester Centre seminar, UCL Priment Seminar series). These analyses have helped to demonstrate the effectiveness of the NHS Diabetes Prevention programme and will be used as evidence towards its continued provision by policymakers.

The University still expects to meet the expected benefits described in this agreement.

Publications:

Ross J, Cotterill S, Bower P, Murray E. Influences on Patient Uptake of and Engagement With the National Health Service Digital Diabetes Prevention Programme: Qualitative Interview Study J Med Internet Res 2023;25:e40961 doi: 10.2196/40961

Antonia M. Marsden, Mark Hann, Emma Barron, Jamie Ross, Jonathan Valabhji, Elizabeth Murray, Sarah Cotterill, Comparison of weight change between face-to-face and digital delivery of the English National Health service diabetes prevention programme: An exploratory non-inferiority study with imputation of plausible weight outcomes, Preventive Medicine Reports, Volume 32, 2023, https://doi.org/10.1016/j.pmedr.2023.102161.

Rhiannon E. Hawkes, Lisa M. Miles, David P. French, What behaviour change technique content is offered to service users of the nationally implemented English NHS Digital Diabetes Prevention Programme: Analysis of multiple sources of intervention content, Preventive Medicine Reports, Volume 32, 2023, https://doi.org/10.1016/j.pmedr.2023.102112.

Miles LM, Hawkes RE, French DP. How the Behavior Change Content of a Nationally Implemented Digital Diabetes Prevention Program Is Understood and Used by Participants: Qualitative Study of Fidelity of Receipt and Enactment. J Med Internet Res. 2023 Jan 11;25:e41214. doi: 10.2196/41214.

Hawkes RE, Miles LM, French DP. Fidelity to Program Specification of the National Health Service Digital Diabetes Prevention Program Behavior Change Technique Content and Underpinning Theory: Document Analysis. J Med Internet Res. 2022 Apr 27;24(4):e34253. doi: 10.2196/34253.

This Data Sharing Agreement permits access to two datasets: National Diabetes Audit (NDA) and Non-Diabetic Hyperglycaemia (NDH). These datasets are collected from the same sources at the same time. NHS Digital treats them as separate datasets because one (NDA) is for collected for the purpose of auditing diabetes and the other (NDH) is by definition a dataset of people without diabetes.

The University of Manchester requires extracts of NDA and NDH data for the purpose of the DIPLOMA (Diabetes Prevention - Long term Multimethod Assessment) research programme, the aim of which is to provide a comprehensive assessment of the implementation, delivery and outcomes of the NHS Diabetes Prevention Programme (NHS DPP) to inform commissioning.

Type 2 diabetes is a common health condition that can cause serious health problems, reduce people’s quality of life and reduce life expectancy, as well as costing a lot of money to treat. Some people are at higher risk of diabetes and can be identified with a blood test. If people at risk make changes to their lifestyle (more exercise, losing weight), they can substantially reduce their chances of getting type 2 diabetes. The NHS and Diabetes UK have introduced a new scheme called Healthier You: the NHS Diabetes Prevention Programme (NHS DPP). People in England who are at risk of type 2 diabetes, referred to as having Non-Diabetic Hyperglycaemia (NDH), will be offered a practical course which will help them change their lifestyle. The aim is that this course will improve peoples’ health and reduce their diabetes risk.

The people managing the NHS DPP need to know if it really works to prevent diabetes and whether it is a good use of NHS resources. The DIPLOMA programme was designed to help answer these questions.

Personal data can be lawfully processed for this purpose because it is a task carried out in the public interest (General Data Protection Regulation Article 6 (1) (e)). The task concerns the enactment of the NHS Diabetes Prevention Programme and the performance concerns the use of this programme in the population and sub-groups of the population of England.

Processing of the data is necessary for scientific research purposes and will be used to evaluate the effectiveness of the NHS Diabetes Prevention Programme (General Data Protection Regulation Article 9 (2) (j)). The DIPLOMA (Diabetes Prevention - Long term Multimethod Assessment) project has been funded by the National Institute for Health Research to evaluate the NHS Diabetes Prevention Programme and is comprised of eight work packages.

The data required are from the National Diabetes Audit (NDA) dataset which contains details of individuals with all types of diabetes and the Non-Diabetic Hyperglycaemia (NDH) which contains data on individuals at risk of Type 2 diabetes with.

The datasets will be used to identify which individuals with NDH go on to develop type 2 diabetes.

As a contracted data processor for NHS England, NHS Digital processes a separate dataset called the Minimum DataSet (MDS). With NHS England’s permission, NHS Digital will link the NDA and NDH data with the MDS and will disseminate linked MDS data alongside the NDA/NDH data to the University of Manchester. All data will be pseudonymised and linked using common pseudo-IDs applied by NHS Digital. The linkage of these datasets will enable the University of Manchester to track which NDH patients were offered a place in the NHS DPP; which patients accepted the offer, and which patients were not offered a place.

The MDS contains participant information that providers of the DPP are contractually obliged to collect in order to receive payment. It contains the date that participants were referred to the programme, as well as information about how many sessions they attended, as well as certain outcome measures, as recorded by the providers of the programme. Importantly however, this dataset contains no information on the wider eligible population who may benefit from the programme, only those that have been referred. To accurately evaluate the efficacy of the NHS DPP, it is important to know the characteristics of the eligible population.

The NDA dataset also contains information about the characteristics of the people with NDH or diabetes and, in combination with the MDS, this can be used to identify trends such as inequalities in access to the NHS DPP and variations in outcomes of the NHS DPP linked to certain characteristics. The data contains demographic and socio-economic characteristics of patients which can be used to assess the characteristics of those with the condition and whether any activity delivered is more or less prominent based on these characteristics. This data will also provide national level data to facilitate the calculation of statistics on the prevalence of NHS DPP participants, the characteristics of those who go on to develop Type 2 diabetes, as well as inform parameters to be used in a decision-analytic model, which will be used to assess the long term cost-effectiveness of the programme.

The data subjects are all individuals in the NDA dataset aged over 15 years from 2017/18 to 2020/21 with type 2 diabetes and all individuals in the NDH dataset aged over 15 years from 2017/18 to 2020/21 with non-diabetic hyperglycaemia.

Under this Agreement the University of Manchester request three additional data fields: Type 2 Diabetes diagnosis date, CCG code and birth date. The University of Manchester request the full diagnosis date of type 2 diabetes, rather than the year of diagnosis (currently provided in the datasets) as the analysis is being conducted in terms of financial years. This is to match how the DPP was rolled out across England, in waves, with each new wave starting in April. Secondly, the University of Manchester request CCG code as the DPP is delivered at the CCG level. Currently the University of Manchester have appended the data with GP/CCG lists, but the provision of CCG in the dataset itself would be beneficial. Lastly, the University of Manchester request year of birth. The University of Manchester currently have a data field ‘age’, however, as the University of Manchester do not have the audit year for the data records, this makes it difficult to define age consistently across the three datasets (NDA, NDH and MDS). This could be rectified with the year of birth field.

The data will be used in three of eight work packages in the overall DIPLOMA research project. These are:

• Work package 1: Access and Equity – the aim is to assess the accessibility of the NHS DPP and identify inequalities in access.

• Work package 5: Comparative Effectiveness – the aim is to examine whether the NHS DPP leads to a reduction in the prevalence of Type 2 Diabetes and other outcomes related to Type 2 Diabetes compared to those without access to the NHS DPP.

• Work package 7: Economic Evaluation – the aim is to explore the cost-effectiveness of the NHS DPP, from the perspective of the NHS and Personal Social Services.

Further details on these work packages are provided below.

Quality will be assessed in terms of equitable use of the NHS Diabetes Prevention Programme (for Work Package 1); the effectiveness of reducing the prevalence of Type 2 diabetes (Work Package 5), and whether the programme is considered to be cost-effective (Work Package 7).

The DIPLOMA was originally funded by the National Institute of Health Research (NIHR) to be delivered by researchers at the University of Manchester from 2017-2021. The research project has since been extended (DIPLOMA 2), to include an additional component, evaluating the Digital component of the prevention programme. Its aim is to produce a comprehensive evaluation of the NHS DPP, comprised of eight distinct but related work packages, using various research techniques and different sources of data. The DIPLOMA research programme is designed to provide:

(a) feedback regularly to NHS DPP stakeholders on the delivery and outcomes of the programme to support ongoing development and quality improvement,

(b) a rigorous longer-term assessment of the success of the NHS DPP in meeting the aim of reducing diabetes prevalence in a way that is cost-effective and sustainable for the NHS.

The University of Manchester is the sole data controller. There are two joint Principal Investigators leading the DIPLOMA work programme. Both are substantive employees of the University of Manchester. The Contract with the Department of Health and Social Care for undertaking the evaluation of the NHS DPP was signed with Salford Royal NHS Foundation Trust (SRFT) as the Contractor. The SRFT are subcontracting all of the work to the University of Manchester. It is agreed between SRFT and University of Manchester that all the research and decisions on how this is conducted, including the data management, are the University of Manchester’s responsibility, whilst SRFT’s role is essentially to administer the contract. The University of Manchester is reporting directly to the NIHR, not through SRFT.

The University of Manchester is the sole organisation processing the data. No other organisations process the data for this purpose.

The Department of Health and Social Care and Public Health England are in liaison with the DIPLOMA evaluation. These advise on approaches taken and provide updates on the NHS DPP. The findings of the DIPLOMA study will be distributed to these wider organisations but no data will be made available to them.

A Professor employed by the University of Oxford is providing consultancy for work package 7, as he is an expert in developing economic decision models for diabetes. He will advise the University of Manchester DIPLOMA researchers involved in work package 7, on the methods used, but will not have any access to the data. His role is strictly advisory.

A Professor from the University of Sheffield is also funded as a consultant on the DIPLOMA project, because of his clinical expertise. He will not have access to the data under this Agreement and has purely an advisory role.

The data requested will not be used or accessible to other work packages within the project beyond those mentioned in this data request (Work packages 1, 5 and 7). The anonymous findings (aggregated with small number suppression) may be used to inform the work of other work packages. The data is restricted to only those with type 2 diabetes and those identified as at risk of type 2 diabetes (Non-diabetic hyperglycaemia). Data has been further minimised by only selecting the fields required for analysis.

The following provides additional detail on the three work packages for which the data will be used:

Work Package 1 – Access and Equity:

This work package is interested in the accessibility of the DPP. The work package will use the NDA and NDH data as well as national survey data (accessed via the UK data archive) to explore the characteristics of patients who enter the NHS DPP, and those who are eligible and do not enter. Inequalities in health have been shown across England. There are many reasons why inequalities may exist, but one of the reasons might be because there is inequality in a patient's access to health care. Access may be related to multiple factors: availability of the NHS DPP in the area; awareness of NHS DPP among patients and professionals, acceptability of NHS DPP to professionals and patients; and the perceived ‘costs’ of NHS DPP (such as taking time off work).

To understand any issues in access to the NHS DPP and the impacts of the programme on diabetes incidence, assessments will be made to understand whether inequalities in protected characteristics exist for:

1. the identification of patients eligible for the programme (‘NHS DPP eligible’ patients)

2. the referral of patients to the programme (‘NHS DPP attenders’ and ‘non-attenders’)

3. programme delivery and completion

4. the effectiveness of the programme

The evaluation concerns the population that are non-diabetic hyperglycaemic (NDH), this patient group will start to be collected from 2017-2018 as part of the NHS Diabetes Prevention Programme fields (DPPOFF_COD; DPPATT_COD and DPPCOMP_COD).

The University of Manchester intends to compare the patients identified as non-diabetic hyperglycaemic (NDH) to populations identified in national surveys to identify whether identification of patients is representative of the general population. The national surveys are the UK Household Longitudinal Survey (UKHLS); the Health Survey for England (HSE), and the English Longitudinal Study for Ageing (ELSA). These population surveys are accessed via the UK Data Archive and are not controlled by NHS Digital.

Comparisons of uptake and completion of the NHS Diabetes Prevention Programme, by patient characteristics will inform whether uptake and completion of the programme differs by certain types of patients.

The data will be analysed in order to measure differences in the prevalence of prediabetes (NDH) by age-groups, gender, ethnicity and area deprivation, as well as differences in the treatment availability and uptake by age-groups, gender, ethnicity and area deprivation. In order for this to occur, pseudonymised patient-level data is required.

There will be two main strands of analyses:

1. Identifying the population with NDH

The data will be used to describe the characteristics of patients with NDH. This will involve presenting the share of patients in the data by the following patient characteristics:

[AGE]: Year of age

[DERIVED_CLEAN_SEX]: Gender

[IMD_QUINTILE]: IMD quintile

[DERIVED_CLEAN_ETHNICITY]: Ethnicity

[LD]: Learning disability flag*

*The learning disability flag is required as this is an NHS protected characteristic. As such, and in line with the other protected characteristics requested such as ethnicity, age etc., the analysis seeks to understand whether access to the DPP is associated with this characteristic.

The distribution of people with NDH by their age, gender, IMD deprivation quintile, ethnicity and disability status will be examined. Comparisons of these distributions with data from three population survey datasets (UKHLS, HSE and ELSA) and to those patients under the NHS diabetes prevention programme (the ‘Minimum DataSet’/MDS). The datasets will be pooled and statistical tests will be conducted to ascertain whether distributions of people by age, gender, IMD deprivation quintile, ethnicity and disability differ between the datasets. The analytic approach will not involve linking individuals across datasets.

Additional analyses will stratify patients with NDH on the basis of various clinical measures concerning blood pressure, body mass index, and cholesterol to further see whether survey data is a good approximation of the population with NDH.

These additional analyses will inform whether certain types of patients with NDH are over- or under-represented in the NDA/NDH data in comparison to survey data.

2. Identifying whether inequalities in programme participation exist by patient characteristics.

Additional analyses will test for associations between NHS Diabetes Prevention Programme participation and explanatory variables.

The distribution of NHS Diabetes Prevention Programme outcome variables by patient characteristics (age, gender, ethnicity, IMD deprivation and disability status) and additional explanatory variables would be processed in order to see whether there are particular groups of patients ceasing to attend and complete the programme.

The University of Manchester request years 2017/18 to latest available at the time of NDA and NDH data to carry out this work with the linked Minimum Dataset. The linked datasets will be pseudonymised as none of the datasets contain patient identifying information.

To ensure comparability with survey data the full geographic spread of patients in the data are requested. No alternative data would enable the proposed evaluation to be performed. This is because the data requested covers all general practice. In addition, the analysis requires individual-level data to enable comparisons of patients by age, gender, ethnicity, IMD deprivation, and disability status.

Work Package 5 - Comparative Effectiveness:

Work Package 5 seeks to examine whether the NHS DPP achieves its primary objective of reducing the prevalence of Type 2 Diabetes and other outcomes related to Type 2 Diabetes such as hospitalisation, primary care visits, HbA1c and BMI/weight levels, and death, compared to those without access to the NHS DPP. To do this, the University of Manchester will use NDA/NDH data for the years 2017/18 to 2020/21 (from when the NHS DPP programme has been implemented) of NDA/NDH data, with the linked Minimum Dataset. The DPP is implemented nationally and therefore data is required at national level, with regions where the NHS DPP was not implemented or not part of Wave 1 being used as comparators.

The conversion rate from non-diabetic hyperglycaemia (NDH) to Type 2 Diabetes will be compared between people referred to the NHS DPP, those not referred and those declining referral. Matched cohort designs will be used to compare the rates of conversion to diabetes within a year of NDH diagnosis, across these three groups. The University of Manchester intend to match both within practices and also across practices (accounting for their time of participation in the DPP) to better account for unmeasured confounding. Preliminary work will examine overall conversion rates and the prevalence of NDH nationally and regionally.

Specific Aims:

1. To estimate the prevalence of Non-diabetic hyperglycaemia (NDH), i.e. people at risk of diabetes.

2. To identify the characteristics associated with NDH (e.g. deprivation, age, sex).

3. To calculate the conversion rate from NDH to Type 2 Diabetes, nationally and regionally.

4. To assess whether the NHS DPP is more effective than usual care in reducing conversion of NDH to diabetes, thus eventually reducing diabetes prevalence in England.

CPRD data will also be used for this comparison. However, CPRD data does not provide national coverage and the numbers are relatively small. The data will not be linked to anything at the patient level. At the practice level data will be linked to practice characteristics that are freely available through NHS Digital and the ONS: practice list size, rural/urban, practice location deprivation (IMD 2015). There is no additional identification risk as linkages are not at the patient level.

Work Package 7 – Economic Evaluation:

Work Package 7 aims to explore the cost-effectiveness of the NHS DPP, from the perspective of the NHS and Personal Social Services. It is important to evaluate if the NHS DPP represents good value for money, as the resources to provide healthcare are limited, and so it is important that funds are allocated to the most effective intervention or prevention strategies.

To evaluate the cost-effectiveness of the NHS DPP, the analysis will take on two main strands:

1) An economic evaluation using observed data from the MDS and

2) A long-term decision analytic model, evaluating the theoretical impact of the NHS DPP over the lifetime of participants, compared to those who do not enrol in the programme.

The NDA/NDH data is required to inform parameters in the long-term decision model. Decision-analytic models are frequently used within economic evaluations, as often the benefits of a programme are not entirely captured within the relatively short time horizon that data is captured. Therefore, decision-analytic models are used to extrapolate beyond observed data, to evaluate final outcomes which are of economic interest. This often involves modelling a hypothetical cohort over a lifetime time horizon. Inputs into decision-analytic models include parameters such as the probabilities of different events occurring (referred to as transition probabilities), the costs associated with events or resource use, and utility scores, which can be used to estimate quality adjusted life years.

Whilst several models evaluating prevention strategies for Type 2 diabetes have been developed, there is currently no model that evaluates the NHS DPP, and as such data from the NDA/NDH is required to inform parameters, such as the number and characteristics of patients being offered the NHS DPP and the uptake rate. Information on the time taken to develop Type 2 diabetes is also needed to produce survival/time to event analysis curves, and use these rates to calculate transition probabilities, to be used within the model. As a standard, the benefits of the NHS DPP will be evaluated in terms of the quality adjusted life years generated by the programme, in comparison to not attending. As such, the data within the NDA/NDH will be used to produce quality of life estimates using a validated mapping tool and regression analyses to predict EQ-5D index values from the data.

The NDA is the only dataset with national coverage of those diagnosed with NDH. CPRD is no longer reflective of the general population, and not all GP practices that have participated in the NHS DPP are captured within it. The required information is also not available within the MDS, as this dataset only has information on patients who have taken up the invitation to enrol on the programme, and therefore does not inform the number, or the characteristics of those eligible (patients with non-diabetic hyperglycaemia) nor the number of people who were offered but refused to attend (available in the NDA by looking at the differences between numbers offered and accepted). The publicly available extract of the NDA is insufficient in the data it provides, as it does not provide the characteristics of those offered the NHS DPP, only the summary characteristics of those diagnosed with NDH, across limited variables, such as BMI and age groups.

Therefore, pseudonymised patient level data for the years 2017/18 to 2020/21 are required, with the linked Minimum Dataset. This linkage will facilitate the analysis of the programme, showing the characteristics of the eligible population (from the NDA/NDH), the subset within this who have participated in the NHS DPP, along with information on their adherence to the programme (from the MDS). As the NHS DPP has been rolled out across England, NDA data from across the whole of England is required.

• Work Package 1 aims to produce a series of reports and journal articles investigating social inequalities in NDH and how they are affected by the NHS DPP. Journals that will be targeted for these publications will be: Diabetic Medicine and British Journal of General Practice. Journal article 1 submission date will be around December 2021.

• Work Package 5 will disseminate the projects outputs through a variety of media, including conference presentations and conventional academic publications, seminars and short accessible reports for stakeholders, and plain English summaries and podcasts for patients and the public. The research team are experienced at writing for a wide variety of audiences. The University of Manchester will work closely with the University of Manchester’s Stakeholder Advisory Group and the Research Advisory Group to maximise the utility of the dissemination. The University of Manchester also plan to disseminate the research findings by publications in peer-reviewed journals and presentations in relevant conferences organised by the Society for Academic Primary Care (SAPC), the Royal College of General Practitioners (RCGP), the Royal College of Physicians (RCP) and the Royal Statistical Society (RSS). When reporting, the principles outlined in the STROBE and RECORD statements will be followed. Data will not reported in the outputs only summary statistics (descriptive statistics) and model estimates. The University of Manchester plan to submit a paper around December 2021 with another targeting the outputs from this work package by October 2022.

• Work Package 7 intends to prepare a report of the findings and recommendations of the cost-effectiveness of the NHS DPP. All reports are submitted to the Director for the NHS Diabetes Programme at NHS England, with the final report expected in September 2022. Results of the analyses will also be presented to the Expert Reference Group, which has representation from the three DPP funders: NHS England, Public Health England and Diabetes UK. A report will also be prepared which outlines the findings of the long-term economic model, which subject to acceptance, will be published in a high impact peer review journal such as Value in Health or Diabetes, Obesity and Metabolism. Presentations will be given at national and international conferences to audiences of academics and health professionals. These conferences will include the Mt Hood Diabetes Challenge Network conference, where the long term economic model will be presented, and the Health Economists’ Study Group (HESG) conference to present initial findings of the cost-effectiveness of the NHS DPP (both subject to abstract acceptance).

Reports will be provided for the DIPLOMA research programme on the progress and outcome of the research. Target report publication dates are as follows:

a) The DIPLOMA research programme final report: September 2022.

Reports and journal articles will be published online and made available to all. Submitted versions of all journal articles will be freely available on the University of Manchester library website.

With the addition of NDA 2020/21 data to this agreement, the University of Manchester expects similar outputs as outlined above in respect of this latest available data.

Researchers at the University of Manchester were able to access the data on the 3rd November 2020. It has taken considerable time to format and validate the dataset, and to receive a working linkage file between the NDA/NDH and MDS datasets . The University of Manchester have started preliminary analyses and writing up these results for publication, but no manuscripts have yet been submitted.

To date, the University of Manchester have presented preliminary analyses to the Health Services Research UK conference (July 2021), and at two university seminars (University of Manchester Centre seminar, UCL Priment Seminar series).

Whilst no manuscripts have been submitted to date, the University still expects to meet the expected benefits described in this agreement.

The University of Manchester will ensure that any future extension of this application will refer to NHS Digital’s DARS Standard for Expected Measurable Benefits for any benefits achieved.

This Data Sharing Agreement permits access to two datasets: National Diabetes Audit (NDA) and Non-Diabetic Hyperglycaemia (NDH). These datasets are collected from the same sources at the same time. NHS Digital treats them as separate datasets because one (NDA) is for collected for the purpose of auditing diabetes and the other (NDH) is by definition a dataset of people without diabetes.

The University of Manchester requires extracts of NDA and NDH data for the purpose of the DIPLOMA (Diabetes Prevention - Long term Multimethod Assessment) research programme, the aim of which is to provide a comprehensive assessment of the implementation, delivery and outcomes of the NHS Diabetes Prevention Programme (NHS DPP) to inform commissioning.

Type 2 diabetes is a common health condition that can cause serious health problems and reduce people’s quality of life, as well as costing a lot of money to treat. Some people are at higher risk of diabetes and can be identified with a blood test. If people at risk make changes to their lifestyle (more exercise, losing weight), they can substantially reduce their chances of getting type 2 diabetes. The NHS and Diabetes UK have introduced a new scheme called Healthier You: the NHS Diabetes Prevention Programme (NHS DPP). People in England who are at risk of type 2 diabetes, referred to as having Non-Diabetic Hyperglycaemia (NDH), will be offered a practical course which will help them change their lifestyle. The aim is that this course will improve peoples’ health and reduce their diabetes risk.

The people managing the NHS DPP need to know if it really works to prevent diabetes and whether it is a good use of NHS resources. The DIPLOMA programme was designed to help answer these questions.

Personal data can be lawfully processed for this purpose because it is a task carried out in the public interest (General Data Protection Regulation Article 6 (1) (e)). The task concerns the enactment of the NHS Diabetes Prevention Programme and the performance concerns the use of this programme in the population and sub-groups of the population of England.

Processing of the data is necessary for scientific research purposes and will be used to evaluate the effectiveness of the NHS Diabetes Prevention Programme (General Data Protection Regulation Article 9 (2) (j)). The DIPLOMA (Diabetes Prevention - Long term Multimethod Assessment) project has been funded by the National Institute for Health Research to evaluate the NHS Diabetes Prevention Programme and is comprised of eight work packages.

The data required are from the National Diabetes Audit (NDA) dataset which contains details of individuals with all types of diabetes and the Non-Diabetic Hyperglycaemia (NDH) which contains data on individuals at risk of Type 2 diabetes with.

The datasets will be used to identify which individuals with NDH go on to develop type 2 diabetes.

As a contracted data processor for NHS England, NHS Digital processes a separate dataset called the Minimum DataSet (MDS). With NHS England’s permission, NHS Digital will link the NDA and NDH data with the MDS and will disseminate linked MDS data alongside the NDA/NDH data to the University of Manchester. All data will be pseudonymised and linked using common pseudo-IDs applied by NHS Digital. The linkage of these datasets will enable the University of Manchester to track which NDH patients were offered a place in the NHS DPP; which patients accepted the offer, and which patients were not offered a place.

The MDS contains participant information that providers of the DPP are contractually obliged to collect in order to receive payment. It contains the date that participants were referred to the programme, as well as information about how many sessions they attended, as well as certain outcome measures, as recorded by the providers of the programme. Importantly however, this dataset contains no information on the wider eligible population who may benefit from the programme, only those that have been referred. To accurately evaluate the efficacy of the NHS DPP, it is important to know the characteristics of the eligible population.

The NDA dataset also contains information about the characteristics of the people with NDH or diabetes and, in combination with the MDS, this can be used to identify trends such as inequalities in access to the NHS DPP and variations in outcomes of the NHS DPP linked to certain characteristics. The data contains demographic and socio-economic characteristics of patients which can be used to assess the characteristics of those with the condition and whether any activity delivered is more or less prominent based on these characteristics. This data will also provide national level data to facilitate the calculation of statistics on the prevalence of NHS DPP participants, the characteristics of those who go on to develop Type 2 diabetes, as well as inform parameters to be used in a decision-analytic model, which will be used to assess the long term cost-effectiveness of the programme.

The data subjects are all individuals in the NDA dataset aged over 15 years from 2017/18 to 2019/20 with type 2 diabetes and all individuals in the NDH dataset aged over 15 years from 2017/18 to 2019/20 with non-diabetic hyperglycaemia.

The data will be used in three of eight work packages in the overall DIPLOMA research project. These are:

• Work package 1: Access and Equity – the aim is to assess the accessibility of the NHS DPP and identify inequalities in access.

• Work package 5: Comparative Effectiveness – the aim is to examine whether the NHS DPP leads to a reduction in the prevalence of Type 2 Diabetes and other outcomes related to Type 2 Diabetes compared to those without access to the NHS DPP.

• Work package 7: Economic Evaluation – the aim is to explore the cost-effectiveness of the NHS DPP, from the perspective of the NHS and Personal Social Services.

Further details on these work packages are provided below.

Quality will be assessed in terms of equitable use of the NHS Diabetes Prevention Programme (for Work Package 1); the effectiveness of reducing the prevalence of Type 2 diabetes (Work Package 5), and whether the programme is considered to be cost-effective (Work Package 7).

DIPLOMA is funded by the National Institute of Health Research (NIHR) and will be delivered by researchers at the University of Manchester over four years (2017-2021). Its aim is to produce a comprehensive evaluation of the NHS DPP, comprised of eight distinct but related work packages, using various research techniques and different sources of data. The DIPLOMA research programme is designed to provide:

(a) feedback regularly to NHS DPP stakeholders on the delivery and outcomes of the programme to support ongoing development and quality improvement,

(b) a rigorous longer-term assessment of the success of the NHS DPP in meeting the aim of reducing diabetes prevalence in a way that is cost-effective and sustainable for the NHS.

The University of Manchester is the sole data controller. There are two joint Principal Investigators leading the DIPLOMA work programme. Both are substantive employees of the University of Manchester. The Contract with the Department of Health and Social Care for undertaking the evaluation of the NHS DPP was signed with Salford Royal NHS Foundation Trust (SRFT) as the Contractor. The SRFT are subcontracting all of the work to the University of Manchester. It is agreed between SRFT and University of Manchester that all the research and decisions on how this is conducted, including the data management, are the University of Manchester’s responsibility, whilst SRFT’s role is essentially to administer the contract. The University of Manchester is reporting directly to the NIHR, not through SRFT.

The University of Manchester is the sole organisation processing the data. No other organisations process the data for this purpose.

The Department of Health and Social Care and Public Health England are in liaison with the DIPLOMA evaluation. These advise on approaches taken and provide updates on the NHS DPP. The findings of the DIPLOMA study will be distributed to these wider organisations but no data will be made available to them.

A Professor employed by the University of Oxford is providing consultancy for work package 7, as he is an expert in developing economic decision models for diabetes. He will advise the University of Manchester DIPLOMA researchers involved in work package 7, on the methods used, but will not have any access to the data. His role is strictly advisory.

A Professor from the University of Sheffield is also funded as a consultant on the DIPLOMA project, because of his clinical expertise. He will not have access to the data under this Agreement and has purely an advisory role.

The data requested will not be used or accessible to other work packages within the project beyond those mentioned in this data request (Work packages 1, 5 and 7). The anonymous findings (aggregated with small number suppression) may be used to inform the work of other work packages. The data is restricted to only those with type 2 diabetes and those identified as at risk of type 2 diabetes (Non-diabetic hyperglycaemia). Data has been further minimised by only selecting the fields required for analysis.

The following provides additional detail on the three work packages for which the data will be used:

Work Package 1 – Access and Equity:

This work package is interested in the accessibility of the DPP. The work package will use the NDA and NDH data as well as national survey data (accessed via the UK data archive) to explore the characteristics of patients who enter the NHS DPP, and those who are eligible and do not enter. Inequalities in health have been shown across England. There are many reasons why inequalities may exist, but one of the reasons might be because there is inequality in a patient's access to health care. Access may be related to multiple factors: availability of the NHS DPP in the area; awareness of NHS DPP among patients and professionals, acceptability of NHS DPP to professionals and patients; and the perceived ‘costs’ of NHS DPP (such as taking time off work).

To understand any issues in access to the NHS DPP and the impacts of the programme on diabetes incidence, assessments will be made to understand whether inequalities in protected characteristics exist for:

1. the identification of patients eligible for the programme (‘NHS DPP eligible’ patients)

2. the referral of patients to the programme (‘NHS DPP attenders’ and ‘non-attenders’)

3. programme delivery and completion

4. the effectiveness of the programme

The evaluation concerns the population that are non-diabetic hyperglycaemic (NDH), this patient group will start to be collected from 2017-2018 as part of the NHS Diabetes Prevention Programme fields (DPPOFF_COD; DPPATT_COD and DPPCOMP_COD).

The University of Manchester intends to compare the patients identified as non-diabetic hyperglycaemic (NDH) to populations identified in national surveys to identify whether identification of patients is representative of the general population. The national surveys are the UK Household Longitudinal Survey (UKHLS); the Health Survey for England (HSE), and the English Longitudinal Study for Ageing (ELSA). These population surveys are accessed via the UK Data Archive and are not controlled by NHS Digital.

Comparisons of uptake and completion of the NHS Diabetes Prevention Programme, by patient characteristics will inform whether uptake and completion of the programme differs by certain types of patients.

The data will be analysed in order to measure differences in the prevalence of prediabetes (NDH) by age-groups, gender, ethnicity and area deprivation, as well as differences in the treatment availability and uptake by age-groups, gender, ethnicity and area deprivation. In order for this to occur, pseudonymised patient-level data is required.

There will be two main strands of analyses:

1. Identifying the population with NDH

The data will be used to describe the characteristics of patients with NDH. This will involve presenting the share of patients in the data by the following patient characteristics:

[AGE]: Year of age

[DERIVED_CLEAN_SEX]: Gender

[IMD_QUINTILE]: IMD quintile

[DERIVED_CLEAN_ETHNICITY]: Ethnicity

[LD]: Learning disability flag*

*The learning disability flag is required as this is an NHS protected characteristic. As such, and in line with the other protected characteristics requested such as ethnicity, age etc., the analysis seeks to understand whether access to the DPP is associated with this characteristic.

The unit of analyses will be individual. The distribution of people with NDH by their age, gender, IMD deprivation quintile, ethnicity and disability status will be examined. Comparisons of these distributions with data from three population survey datasets (UKHLS, HSE and ELSA) and to those patients under the NHS diabetes prevention programme (the ‘Minimum DataSet’/MDS). The datasets will be pooled and statistical tests will be conducted to ascertain whether distributions of people by age, gender, IMD deprivation quintile, ethnicity and disability differ between the datasets. The analytic approach will not involve linking individuals across datasets.

Additional analyses will stratify patients with NDH on the basis of various clinical measures concerning blood pressure, body mass index, and cholesterol to further see whether survey data is a good approximation of the population with NDH.

These additional analyses will inform whether certain types of patients with NDH are over- or under-represented in the NDA/NDH data in comparison to survey data.

2. Identifying whether inequalities in programme participation exist by patient characteristics.

Additional analyses will test for associations between NHS Diabetes Prevention Programme participation and explanatory variables.

The distribution of NHS Diabetes Prevention Programme outcome variables by patient characteristics (age, gender, ethnicity, IMD deprivation and disability status) and additional explanatory variables would be processed in order to see whether there are particular groups of patients ceasing to attend and complete the programme.

The University of Manchester request years 2017/18 to latest available at the ti of NDA and NDH data to carry out this work with the linked Minimum Dataset. The linked datasets will be pseudonymised as none of the datasets contain patient identifying information.

To ensure comparability with survey data the full geographic spread of patients in the data are requested. No alternative data would enable the proposed evaluation to be performed. This is because the data requested covers all general practice. In addition, the analysis requires individual-level data to enable comparisons of patients by age, gender, ethnicity, IMD deprivation, and disability status.

Work Package 5 - Comparative Effectiveness:

Work Package 5 seeks to examine whether the NHS DPP achieves its primary objective of reducing the prevalence of Type 2 Diabetes and other outcomes related to Type 2 Diabetes such as hospitalisation, primary care visits, HbA1c and BMI/weight levels, and death, compared to those without access to the NHS DPP. To do this, the University of Manchester will use NDA/NDH data for the years 2017/18 to 2019/20 (from when the NHS DPP programme has been implemented) of NDA/NDH data, with the linked Minimum Dataset. The DPP is implemented nationally and therefore data is required at national level, with regions where the NHS DPP was not implemented or not part of Wave 1 being used as comparators.

The conversion rate from non-diabetic hyperglycaemia (NDH) to Type 2 Diabetes will be compared between people referred to the NHS DPP, those not referred and those declining referral. Matched cohort designs will be used to compare the rates of conversion to diabetes within a year of NDH diagnosis, across these three groups. We intend to match both within practices and also across practices (accounting for their time of participation in the DPP) to better account for unmeasured confounding. Preliminary work will examine overall conversion rates and the prevalence of NDH nationally and regionally.

Specific Aims:

1. To estimate the prevalence of Non-diabetic hyperglycaemia (NDH), i.e. people at risk of diabetes.

2. To identify the characteristics associated with NDH (e.g. deprivation, age, sex).

3. To calculate the conversion rate from NDH to Type 2 Diabetes, nationally and regionally.

4. To assess whether the NHS DPP is more effective than usual care in reducing conversion of NDH to diabetes, thus eventually reducing diabetes prevalence in England.

CPRD data will also be used for this comparison. However, CPDR data does not provide national coverage and the numbers are relatively small. The data will not be linked to anything at the patient level. At the practice level data will be linked to practice characteristics that are freely available through NHS Digital and the ONS: practice list size, rural/urban, practice location deprivation (IMD 2015). There is no additional identification risk as linkages are not at the patient level.

Work Package 7 – Economic Evaluation:

Work Package 7 aims to explore the cost-effectiveness of the NHS DPP, from the perspective of the NHS and Personal Social Services. It is important to evaluate if the NHS DPP represents good value for money, as the resources to provide healthcare are limited, and so it is important that funds are allocated to the most effective intervention or prevention strategies.

To evaluate the cost-effectiveness of the NHS DPP, the analysis will take on two main strands:

1) An economic evaluation using observed data from the MDS and

2) A long-term decision analytic model, evaluating the theoretical impact of the NHS DPP over the lifetime of participants, compared to those who do not enrol in the programme.

The NDA/NDH data is required to inform parameters in the long-term decision model. Decision-analytic models are frequently used within economic evaluations, as often the benefits of a programme are not entirely captured within the relatively short time horizon that data is captured. Therefore, decision-analytic models are used to extrapolate beyond observed data, to evaluate final outcomes which are of economic interest. This often involves modelling a hypothetical cohort over a lifetime time horizon. Inputs into decision-analytic models include parameters such as the probabilities of different events occurring (referred to as transition probabilities), the costs associated with events or resource use, and utility scores, which can be used to estimate quality adjusted life years.

Whilst several models evaluating prevention strategies for Type 2 diabetes have been developed, there is currently no model that evaluates the NHS DPP, and as such data from the NDA/NDH is required to inform parameters, such as the number and characteristics of patients being offered the NHS DPP and the uptake rate. Information on the time taken to develop Type 2 diabetes is also needed to produce survival/time to event analysis curves, and use these rates to calculate transition probabilities, to be used within the model. As is standard, (according to the NICE Reference Case), the benefits of the NHS DPP will be evaluated in terms of the quality adjusted life years generated by the programme, in comparison to not attending. As such, the data within the NDA/NDH will be used to produce quality of life estimates using a validated mapping tool and regression analyses to predict EQ-5D index values from the data.

The NDA is the only dataset with national coverage of those diagnosed with NDH. CPRD is no longer reflective of the general population, and not all GP practices that have participated in the NHS DPP are captured within it. The required information is also not available within the MDS, as this dataset only has information on patients who have taken up the invitation to enrol on the programme, and therefore does not inform the number, or the characteristics of those eligible (patients with non-diabetic hyperglycaemia) nor the number of people who were offered but refused to attend (available in the NDA by looking at the differences between numbers offered and accepted). The publicly available extract of the NDA is insufficient in the data it provides, as it does not provide the characteristics of those offered the NHS DPP, only the summary characteristics of those diagnosed with NDH, across limited variables, such as BMI and age groups.

Therefore, pseudonymised patient level data for the years 2017/18 to 2019/20 are required, with the linked Minimum Dataset. This linkage will facilitate the analysis of the programme, showing the characteristics of the eligible population (from the NDA/NDH), the subset within this who have participated in the NHS DPP, along with information on their adherence to the programme (from the MDS). As the NHS DPP has been rolled out across England, NDA data from across the whole of England is required.

The purpose of the data processing activities performed on these data is to produce statistical outputs to be used in peer-reviewed journal articles, presentations and reports for the DIPLOMA research programme. These statistical outputs will be of three types: 1) descriptive tables summarising the data, 2) graphs and figures summarising the data, and 3) regression results tables. Outputs will contain only data which is aggregated, with any small numbers suppressed in line with the following disclosure control policy (applicable if one of something is necessarily one person e.g. length of stay – one day as an inpatient can only relate to one person in many data sets):

• Zeros should be shown.

• 1-7 to be rounded to 5.

• Any other numbers rounded to nearest 5.

• Rounding unnecessary for averages etc.

• Percentages calculated from rounded values

• If zeros need to be suppressed, round to 5.

The overall research project includes a number of work packages. The number of outputs, in terms of articles/presentations/reports, will be determined by the findings of the analysis and by the academic peer-review process. There is a PPI (Patient and Public Involvement) forum which will be engaged in an advisory capacity in producing these reports and their dissemination, they have also provided guidance on the fair processing statement, which will need to be published following approval of the NDA/NDH data request. The work packages each have aims to output the results of their analyses:

• Work Package 1 aims to produce a series of reports and journal articles investigating social inequalities in NDH and how they are affected by the NHS DPP. Journals that will be targeted for these publications will be: Diabetic Medicine and British Journal of General Practice. Journal article 1 submission date will be around February 2021.

• Work Package 5 will disseminate the projects outputs through a variety of media, including conference presentations and conventional academic publications, seminars and short accessible reports for stakeholders, and plain English summaries and podcasts for patients and the public. The research team are experienced at writing for a wide variety of audiences. We will work closely with our Stakeholder Advisory Group and the Research Advisory Group to maximise the utility of our dissemination. We also plan to disseminate the research findings by publications in peer-reviewed journals and presentations in relevant conferences organised by the Society for Academic Primary Care (SAPC), the Royal College of General Practitioners (RCGP), the Royal College of Physicians (RCP) and the Royal Statistical Society (RSS). When reporting, the principles outlined in the STROBE and RECORD statements will be followed. Data will not reported in the outputs only summary statistics (descriptive statistics) and model estimates. We plan to target our outputs from this work package by the end of 12 months from data availability.

• Work Package 7 intends to prepare a report of the findings and recommendations of the cost-effectiveness of the NHS DPP. All reports are submitted to the Director for the NHS Diabetes Programme at NHS England, with the final report expected in March 2021. Results of the analyses will also be presented to the Expert Reference Group, which has representation from the three DPP funders: NHS England, Public Health England and Diabetes UK. A report will also be prepared which outlines the findings of the long-term economic model, which subject to acceptance, will be published in a high impact peer review journal such as Value in Health or Diabetes, Obesity and Metabolism. Presentations will be given at national and international conferences to audiences of academics and health professionals. These conferences will include the Mt Hood Diabetes Challenge Network conference 2021, where the long term economic model will be presented, and the Health Economists’ Study Group (HESG) in Summer 2021 conference to present initial findings of the cost-effectiveness of the NHS DPP (both subject to abstract acceptance).

Reports will be provided for the DIPLOMA research programme on the progress and outcome of the research. Target report publication dates are as follows:

a) The DIPLOMA research programme final report: March 2021.

Reports and journal articles will be published online and made available to all. Submitted versions of all journal articles will be freely available on the University of Manchester library website.

With the addition of NDA 2019/20 data to this agreement, the University of Manchester expects similar outputs as outlined above in respect of this latest available data.

Researchers at the University of Manchester were able to access the data on the 3rd November. As such, the University has only been able to work on formatting and validating the dataset, preparing for preliminary analyses.

At this stage the University has not been able to produce any of the outputs listed within this agreement and as such has been unable to realise any of the expected benefits from this processing. However, the University still expects to meet the expected benefits described in this agreement.

This Data Sharing Agreement permits access to two datasets: National Diabetes Audit (NDA) and Non-Diabetic Hyperglycaemia (NDH). These datasets are collected from the same sources at the same time. NHS Digital treats them as separate datasets because one (NDA) is for collected for the purpose of auditing diabetes and the other (NDH) is by definition a dataset of people without diabetes.

The University of Manchester requires extracts of NDA and NDH data for the purpose of the DIPLOMA (Diabetes Prevention - Long term Multimethod Assessment) research programme, the aim of which is to provide a comprehensive assessment of the implementation, delivery and outcomes of the NHS Diabetes Prevention Programme (NHS DPP) to inform commissioning.

Type 2 diabetes is a common health condition that can cause serious health problems and reduce people’s quality of life, as well as costing a lot of money to treat. Some people are at higher risk of diabetes and can be identified with a blood test. If people at risk make changes to their lifestyle (more exercise, losing weight), they can substantially reduce their chances of getting type 2 diabetes. The NHS and Diabetes UK have introduced a new scheme called Healthier You: the NHS Diabetes Prevention Programme (NHS DPP). People in England who are at risk of type 2 diabetes, referred to as having Non-Diabetic Hyperglycaemia (NDH), will be offered a practical course which will help them change their lifestyle. The aim is that this course will improve peoples’ health and reduce their diabetes risk.

The people managing the NHS DPP need to know if it really works to prevent diabetes and whether it is a good use of NHS resources. The DIPLOMA programme was designed to help answer these questions.

Personal data can be lawfully processed for this purpose because it is a task carried out in the public interest (General Data Protection Regulation Article 6 (1) (e)). The task concerns the enactment of the NHS Diabetes Prevention Programme and the performance concerns the use of this programme in the population and sub-groups of the population of England.

Processing of the data is necessary for scientific research purposes and will be used to evaluate the effectiveness of the NHS Diabetes Prevention Programme (General Data Protection Regulation Article 9 (2) (j)). The DIPLOMA (Diabetes Prevention - Long term Multimethod Assessment) project has been funded by the National Institute for Health Research to evaluate the NHS Diabetes Prevention Programme and is comprised of eight work packages.

The data required are from the National Diabetes Audit (NDA) dataset which contains details of individuals with all types of diabetes and the Non-Diabetic Hyperglycaemia (NDH) which contains data on individuals at risk of Type 2 diabetes with.

The datasets will be used to identify which individuals with NDH go on to develop type 2 diabetes.

As a contracted data processor for NHS England, NHS Digital processes a separate dataset called the Minimum DataSet (MDS). With NHS England’s permission, NHS Digital will link the NDA and NDH data with the MDS and will disseminate linked MDS data alongside the NDA/NDH data to the University of Manchester. All data will be pseudonymised and linked using common pseudo-IDs applied by NHS Digital. The linkage of these datasets will enable the University of Manchester to track which NDH patients were offered a place in the NHS DPP; which patients accepted the offer, and which patients were not offered a place.

The MDS contains participant information that providers of the DPP are contractually obliged to collect in order to receive payment. It contains the date that participants were referred to the programme, as well as information about how many sessions they attended, as well as certain outcome measures, as recorded by the providers of the programme. Importantly however, this dataset contains no information on the wider eligible population who may benefit from the programme, only those that have been referred. To accurately evaluate the efficacy of the NHS DPP, it is important to know the characteristics of the eligible population.

The NDA dataset also contains information about the characteristics of the people with NDH or diabetes and, in combination with the MDS, this can be used to identify trends such as inequalities in access to the NHS DPP and variations in outcomes of the NHS DPP linked to certain characteristics. The data contains demographic and socio-economic characteristics of patients which can be used to assess the characteristics of those with the condition and whether any activity delivered is more or less prominent based on these characteristics. This data will also provide national level data to facilitate the calculation of statistics on the prevalence of NHS DPP participants, the characteristics of those who go on to develop Type 2 diabetes, as well as inform parameters to be used in a decision-analytic model, which will be used to assess the long term cost-effectiveness of the programme.

The data subjects are all individuals in the NDA dataset aged over 15 years from 2017/18 to 2018/19 with type 2 diabetes and all individuals in the NDH dataset aged over 15 years from 2017/18 to 2018/19 with non-diabetic hyperglycaemia.

The data will be used in three of eight work packages in the overall DIPLOMA research project. These are:

• Work package 1: Access and Equity – the aim is to assess the accessibility of the NHS DPP and identify inequalities in access.

• Work package 5: Comparative Effectiveness – the aim is to examine whether the NHS DPP leads to a reduction in the prevalence of Type 2 Diabetes and other outcomes related to Type 2 Diabetes compared to those without access to the NHS DPP.

• Work package 7: Economic Evaluation – the aim is to explore the cost-effectiveness of the NHS DPP, from the perspective of the NHS and Personal Social Services.

Further details on these work packages are provided below.

Quality will be assessed in terms of equitable use of the NHS Diabetes Prevention Programme (for Work Package 1); the effectiveness of reducing the prevalence of Type 2 diabetes (Work Package 5), and whether the programme is considered to be cost-effective (Work Package 7).

DIPLOMA is funded by the National Institute of Health Research (NIHR) and will be delivered by researchers at the University of Manchester over four years (2017-2021). Its aim is to produce a comprehensive evaluation of the NHS DPP, comprised of eight distinct but related work packages, using various research techniques and different sources of data. The DIPLOMA research programme is designed to provide:

(a) feedback regularly to NHS DPP stakeholders on the delivery and outcomes of the programme to support ongoing development and quality improvement,

(b) a rigorous longer-term assessment of the success of the NHS DPP in meeting the aim of reducing diabetes prevalence in a way that is cost-effective and sustainable for the NHS.

The University of Manchester is the sole data controller. There are two joint Principal Investigators leading the DIPLOMA work programme. Both are substantive employees of the University of Manchester. The Contract with the Department of Health and Social Care for undertaking the evaluation of the NHS DPP was signed with Salford Royal NHS Foundation Trust (SRFT) as the Contractor. The SRFT are subcontracting all of the work to the University of Manchester. It is agreed between SRFT and University of Manchester that all the research and decisions on how this is conducted, including the data management, are the University of Manchester’s responsibility, whilst SRFT’s role is essentially to administer the contract. The University of Manchester is reporting directly to the NIHR, not through SRFT.

The University of Manchester is the sole organisation processing the data. No other organisations process the data for this purpose.

The Department of Health and Social Care and Public Health England are in liaison with the DIPLOMA evaluation. These advise on approaches taken and provide updates on the NHS DPP. The findings of the DIPLOMA study will be distributed to these wider organisations but no data will be made available to them.

A Professor employed by the University of Oxford is providing consultancy for work package 7, as he is an expert in developing economic decision models for diabetes. He will advise the University of Manchester DIPLOMA researchers involved in work package 7, on the methods used, but will not have any access to the data. His role is strictly advisory.

A Professor from the University of Sheffield is also funded as a consultant on the DIPLOMA project, because of his clinical expertise. He will not have access to the data under this Agreement and has purely an advisory role.

The data requested will not be used or accessible to other work packages within the project beyond those mentioned in this data request (Work packages 1, 5 and 7). The anonymous findings (aggregated with small number suppression) may be used to inform the work of other work packages. The data is restricted to only those with type 2 diabetes and those identified as at risk of type 2 diabetes (Non-diabetic hyperglycaemia). Data has been further minimised by only selecting the fields required for analysis.

The following provides additional detail on the three work packages for which the data will be used:

Work Package 1 – Access and Equity:

This work package is interested in the accessibility of the DPP. The work package will use the NDA and NDH data as well as national survey data (accessed via the UK data archive) to explore the characteristics of patients who enter the NHS DPP, and those who are eligible and do not enter. Inequalities in health have been shown across England. There are many reasons why inequalities may exist, but one of the reasons might be because there is inequality in a patient's access to health care. Access may be related to multiple factors: availability of the NHS DPP in the area; awareness of NHS DPP among patients and professionals, acceptability of NHS DPP to professionals and patients; and the perceived ‘costs’ of NHS DPP (such as taking time off work).

To understand any issues in access to the NHS DPP and the impacts of the programme on diabetes incidence, assessments will be made to understand whether inequalities in protected characteristics exist for:

1. the identification of patients eligible for the programme (‘NHS DPP eligible’ patients)

2. the referral of patients to the programme (‘NHS DPP attenders’ and ‘non-attenders’)

3. programme delivery and completion

4. the effectiveness of the programme

The evaluation concerns the population that are non-diabetic hyperglycaemic (NDH), this patient group will start to be collected from 2017-2018 as part of the NHS Diabetes Prevention Programme fields (DPPOFF_COD; DPPATT_COD and DPPCOMP_COD).

The University of Manchester intends to compare the patients identified as non-diabetic hyperglycaemic (NDH) to populations identified in national surveys to identify whether identification of patients is representative of the general population. The national surveys are the UK Household Longitudinal Survey (UKHLS); the Health Survey for England (HSE), and the English Longitudinal Study for Ageing (ELSA). These population surveys are accessed via the UK Data Archive and are not controlled by NHS Digital.

Comparisons of uptake and completion of the NHS Diabetes Prevention Programme, by patient characteristics will inform whether uptake and completion of the programme differs by certain types of patients.

The data will be analysed in order to measure differences in the prevalence of prediabetes (NDH) by age-groups, gender, ethnicity and area deprivation, as well as differences in the treatment availability and uptake by age-groups, gender, ethnicity and area deprivation. In order for this to occur, pseudonymised patient-level data is required.

There will be two main strands of analyses:

1. Identifying the population with NDH

The data will be used to describe the characteristics of patients with NDH. This will involve presenting the share of patients in the data by the following patient characteristics:

[AGE]: Year of age

[DERIVED_CLEAN_SEX]: Gender

[IMD_QUINTILE]: IMD quintile

[DERIVED_CLEAN_ETHNICITY]: Ethnicity

[LD]: Learning disability flag*

*The learning disability flag is required as this is an NHS protected characteristic. As such, and in line with the other protected characteristics requested such as ethnicity, age etc., the analysis seeks to understand whether access to the DPP is associated with this characteristic.

The unit of analyses will be individual. The distribution of people with NDH by their age, gender, IMD deprivation quintile, ethnicity and disability status will be examined. Comparisons of these distributions with data from three population survey datasets (UKHLS, HSE and ELSA) and to those patients under the NHS diabetes prevention programme (the ‘Minimum DataSet’/MDS). The datasets will be pooled and statistical tests will be conducted to ascertain whether distributions of people by age, gender, IMD deprivation quintile, ethnicity and disability differ between the datasets. The analytic approach will not involve linking individuals across datasets.

Additional analyses will stratify patients with NDH on the basis of various clinical measures concerning blood pressure, body mass index, and cholesterol to further see whether survey data is a good approximation of the population with NDH.

These additional analyses will inform whether certain types of patients with NDH are over- or under-represented in the NDA/NDH data in comparison to survey data.

2. Identifying whether inequalities in programme participation exist by patient characteristics.

Additional analyses will test for associations between NHS Diabetes Prevention Programme participation and explanatory variables.

The distribution of NHS Diabetes Prevention Programme outcome variables by patient characteristics (age, gender, ethnicity, IMD deprivation and disability status) and additional explanatory variables would be processed in order to see whether there are particular groups of patients ceasing to attend and complete the programme.

The University of Manchester request years 2017/18 and 2018/19 of NDA and NDH data to carry out this work with the linked Minimum Dataset. The linked datasets will be pseudonymised as none of the datasets contain patient identifying information.

To ensure comparability with survey data the full geographic spread of patients in the data are requested. No alternative data would enable the proposed evaluation to be performed. This is because the data requested covers all general practice. In addition, the analysis requires individual-level data to enable comparisons of patients by age, gender, ethnicity, IMD deprivation, and disability status.

Work Package 5 - Comparative Effectiveness:

Work Package 5 seeks to examine whether the NHS DPP achieves its primary objective of reducing the prevalence of Type 2 Diabetes and other outcomes related to Type 2 Diabetes such as hospitalisation, primary care visits, HbA1c and BMI/weight levels, and death, compared to those without access to the NHS DPP. To do this, the University of Manchester will use NDA/NDH data for the years 2017/18 and 2018/19 (from when the NHS DPP programme has been implemented) of NDA/NDH data, with the linked Minimum Dataset. The DPP is implemented nationally and therefore data is required at national level, with regions where the NHS DPP was not implemented or not part of Wave 1 being used as comparators.

The conversion rate from non-diabetic hyperglycaemia (NDH) to Type 2 Diabetes will be compared between people referred to the NHS DPP, those not referred and those declining referral. Matched cohort designs will be used to compare the rates of conversion to diabetes within a year of NDH diagnosis, across these three groups. We intend to match both within practices and also across practices (accounting for their time of participation in the DPP) to better account for unmeasured confounding. Preliminary work will examine overall conversion rates and the prevalence of NDH nationally and regionally.

Specific Aims:

1. To estimate the prevalence of Non-diabetic hyperglycaemia (NDH), i.e. people at risk of diabetes.

2. To identify the characteristics associated with NDH (e.g. deprivation, age, sex).

3. To calculate the conversion rate from NDH to Type 2 Diabetes, nationally and regionally.

4. To assess whether the NHS DPP is more effective than usual care in reducing conversion of NDH to diabetes, thus eventually reducing diabetes prevalence in England.

CPRD data will also be used for this comparison. However, CPDR data does not provide national coverage and the numbers are relatively small. The data will not be linked to anything at the patient level. At the practice level data will be linked to practice characteristics that are freely available through NHS Digital and the ONS: practice list size, rural/urban, practice location deprivation (IMD 2015). There is no additional identification risk as linkages are not at the patient level.

Work Package 7 – Economic Evaluation:

Work Package 7 aims to explore the cost-effectiveness of the NHS DPP, from the perspective of the NHS and Personal Social Services. It is important to evaluate if the NHS DPP represents good value for money, as the resources to provide healthcare are limited, and so it is important that funds are allocated to the most effective intervention or prevention strategies.

To evaluate the cost-effectiveness of the NHS DPP, the analysis will take on two main strands:

1) An economic evaluation using observed data from the MDS and

2) A long-term decision analytic model, evaluating the theoretical impact of the NHS DPP over the lifetime of participants, compared to those who do not enrol in the programme.

The NDA/NDH data is required to inform parameters in the long-term decision model. Decision-analytic models are frequently used within economic evaluations, as often the benefits of a programme are not entirely captured within the relatively short time horizon that data is captured. Therefore, decision-analytic models are used to extrapolate beyond observed data, to evaluate final outcomes which are of economic interest. This often involves modelling a hypothetical cohort over a lifetime time horizon. Inputs into decision-analytic models include parameters such as the probabilities of different events occurring (referred to as transition probabilities), the costs associated with events or resource use, and utility scores, which can be used to estimate quality adjusted life years.

Whilst several models evaluating prevention strategies for Type 2 diabetes have been developed, there is currently no model that evaluates the NHS DPP, and as such data from the NDA/NDH is required to inform parameters, such as the number and characteristics of patients being offered the NHS DPP and the uptake rate. Information on the time taken to develop Type 2 diabetes is also needed to produce survival/time to event analysis curves, and use these rates to calculate transition probabilities, to be used within the model. As is standard, (according to the NICE Reference Case), the benefits of the NHS DPP will be evaluated in terms of the quality adjusted life years generated by the programme, in comparison to not attending. As such, the data within the NDA/NDH will be used to produce quality of life estimates using a validated mapping tool and regression analyses to predict EQ-5D index values from the data.

The NDA is the only dataset with national coverage of those diagnosed with NDH. CPRD is no longer reflective of the general population, and not all GP practices that have participated in the NHS DPP are captured within it. The required information is also not available within the MDS, as this dataset only has information on patients who have taken up the invitation to enrol on the programme, and therefore does not inform the number, or the characteristics of those eligible (patients with non-diabetic hyperglycaemia) nor the number of people who were offered but refused to attend (available in the NDA by looking at the differences between numbers offered and accepted). The publicly available extract of the NDA is insufficient in the data it provides, as it does not provide the characteristics of those offered the NHS DPP, only the summary characteristics of those diagnosed with NDH, across limited variables, such as BMI and age groups.

Therefore, pseudonymised patient level data for the years 2017 to 2019 are required, with the linked Minimum Dataset. This linkage will facilitate the analysis of the programme, showing the characteristics of the eligible population (from the NDA/NDH), the subset within this who have participated in the NHS DPP, along with information on their adherence to the programme (from the MDS). As the NHS DPP has been rolled out across England, NDA data from across the whole of England is required.

The purpose of the data processing activities performed on these data is to produce statistical outputs to be used in peer-reviewed journal articles, presentations and reports for the DIPLOMA research programme. These statistical outputs will be of three types: 1) descriptive tables summarising the data, 2) graphs and figures summarising the data, and 3) regression results tables. Outputs will contain only data which is aggregated, with any small numbers suppressed in line with the following disclosure control policy (applicable if one of something is necessarily one person e.g. length of stay – one day as an inpatient can only relate to one person in many data sets):

• Zeros should be shown.

• 1-7 to be rounded to 5.

• Any other numbers rounded to nearest 5.

• Rounding unnecessary for averages etc.

• Percentages calculated from rounded values

• If zeros need to be suppressed, round to 5.

The overall research project includes a number of work packages. The number of outputs, in terms of articles/presentations/reports, will be determined by the findings of the analysis and by the academic peer-review process. There is a PPI (Patient and Public Involvement) forum which will be engaged in an advisory capacity in producing these reports and their dissemination, they have also provided guidance on the fair processing statement, which will need to be published following approval of the NDA/NDH data request. The work packages each have aims to output the results of their analyses:

• Work Package 1 aims to produce a series of reports and journal articles investigating social inequalities in NDH and how they are affected by the NHS DPP. Journals that will be targeted for these publications will be: Diabetic Medicine and British Journal of General Practice. Journal article 1 submission date will be around October 2020.

• Work Package 5 will disseminate the projects outputs through a variety of media, including conference presentations and conventional academic publications, seminars and short accessible reports for stakeholders, and plain English summaries and podcasts for patients and the public. The research team are experienced at writing for a wide variety of audiences. We will work closely with our Stakeholder Advisory Group and the Research Advisory Group to maximise the utility of our dissemination. We also plan to disseminate the research findings by publications in peer-reviewed journals and presentations in relevant conferences organised by the Society for Academic Primary Care (SAPC), the Royal College of General Practitioners (RCGP), the Royal College of Physicians (RCP) and the Royal Statistical Society (RSS). When reporting, the principles outlined in the STROBE and RECORD statements will be followed. Data will not reported in the outputs only summary statistics (descriptive statistics) and model estimates. We plan to target our outputs from this work package by the end of 12 months from data availability.

• Work Package 7 intends to prepare a report of the findings and recommendations of the cost-effectiveness of the NHS DPP. All reports are submitted to the Director for the NHS Diabetes Programme at NHS England, with the final report expected in March 2021. Results of the analyses will also be presented to the Expert Reference Group, which has representation from the three DPP funders: NHS England, Public Health England and Diabetes UK. A report will also be prepared which outlines the findings of the long-term economic model, which subject to acceptance, will be published in a high impact peer review journal such as Value in Health or Diabetes, Obesity and Metabolism. Presentations will be given at national and international conferences to audiences of academics and health professionals. These conferences will include the Mt Hood Diabetes Challenge Network, Chicago conference 2020, where the long term economic model will be presented, and the Health Economists’ Study Group (HESG) in January 2021 conference to present initial findings of the cost-effectiveness of the NHS DPP (both subject to abstract acceptance).

Reports will be provided for the DIPLOMA research programme on the progress and outcome of the research. Target report publication dates are as follows:

a) The DIPLOMA research programme interim report:

Twice annually (in April and October), subject to data transfer timing, UoM anticipate the October 2020 interim report will contain findings from the evaluation of the requested data.

b) The DIPLOMA research programme final report: March 2021.

Reports and journal articles will be published online and made available to all. Submitted versions of all journal articles will be freely available on the University of Manchester library website.

Yielded Benefits is not a requirement for new applications.