MR1417 - ADDITION-Plus study: Ten year follow-up of a randomised controlled trial of an individually-tailored behaviour change intervention among people with recently diagnosed type 2 diabetes under intensive UK general practice care

The Anglo-Danish-Dutch Study of Intensive Treatment In People with Screen Detected Diabetes in Primary Care (ADDITION study) was initiated in the financial year 2001/02 with the initial recruitment phase starting in March 2002 and ending in December 2006. Prior to the start of the main functions of both studies, a feasibility study was implemented in 1999, however all patient data and outputs from this earlier project were held separate from the main studies and were destroyed upon finalisation of the feasibility study.

The ADDITION-Plus study is a trial nested within the Cambridge arm of the ADDITION study, once the end of the recruitment phase had been reached, patients from 34 general practises in Eastern England agreed to take part in the study. Once the cohort had been formed, a total of 478 men and women diagnosed with diabetes and a mean age of 60 years were individually randomised to receive (i) intensive treatment (n=239), or (ii) intensive treatment plus a theory-based behaviour change intervention led by a facilitator external to the general practice team (n=239). The aim is to assess whether a theory-based behaviour change intervention delivered by trained and quality-assured lifestyle facilitators can achieve and maintain improvements in physical activity, dietary change, medication adherence and smoking cessation in people with recently diagnosed diabetes.

Both the ADDITION and ADDITION-plus studies began with a recruitment and screening phase running from March 2002 to March 2006. The University of Cambridge has conducted follow-up analyses at 5 year intervals. The 5-year follow-up was conducted in late 2009, with the 10 year follow-up up conducted in late 2014. The University is in the process of work for a 15-year follow-up which began in December 2019. For this purpose, ADDITION-plus researchers request access to available HES data for the cohort described below from 2001/02 to latest available.

The primary objectives of the ADDITION studies were:

• To evaluate whether population-based screening for undiagnosed type 2 diabetes was feasible in a primary care setting

• To assess whether subsequent optimised intensive treatment of diabetes and associated risk factors among screen-detected patients were feasible in primary care and benefited the patients, and

• To quantify the harms associated with screening. Previous screening studies have focused on people with pre-diabetes and how to prevent progression from pre-diabetes to diabetes.

The study consists of a screening phase followed by a pragmatic trial of intensive treatment compared to routine care in four centres (Denmark, Cambridge UK, the Netherlands and Leicester UK).

Diabetes meets many of the criteria for screening. A high proportion of people have undiagnosed diabetes, many individuals present with complications when they are diagnosed, and there is a long “latent” period between when the disease starts and when a person experiences symptoms and presents at their GP. People with diabetes have an increased risk of developing cardiovascular disease (heart attack or stroke). It is logical to suggest that if diabetics were identified earlier in the disease trajectory and treated before symptoms developed, the risk of them suffering from an early death or experiencing a heart attack or stroke would be reduced. However, there is little evidence to support this, and there are a number of uncertainties concerning the potential benefits of population-based screening for type 2 diabetes. The ADDITION study was set -up to answer some of these uncertainties.

ADDITION Plus is a study involving patients within the intensive treatment programme of the ADDITION study in Cambridge, and of patients with diabetes recently diagnosed by their doctor in general practices in Cambridge and surrounding counties. These two characteristics make up each half of the full ADDITION-plus cohort.

In addition to their treatment programme, half of the patients were given additional support and advice to encourage changes to lifestyle. The aim is to help patients achieve sustained changes to their physical activity and diet, to support cessation of smoking if applicable and to help patients to take their medication as prescribed. The other half of the patients were not given any additional support. The four behaviours (physical activity, diet, smoking and adherence to medication) were measured along with other measures such as weight, blood pressure, blood glucose and cholesterol after one year and five years.

Cohort

For background, the ADDITION study overall recruited approximately 36000 people (the ADDITION-Cambridge screening cohort, MR798/NIC-147750-8GS7S) in or near Cambridge to be invited to screening. Following screening in 343 practices, 3,057 eligible participants with screen-detected diabetes agreed to take part in the ADDITION treatment trial. In Cambridge, confirmation of the diagnosis of diabetes and baseline assessments were carried out in the clinical research facilities in Ely, Wisbech and Addenbrooke’s with additional measurements in Peterborough and Huntingdon. As a result of the screening programme 867 people in the East Anglia region were made aware that they had diabetes. These 867 patients make up the ADDITION Cambridge main cohort (MR1406/NIC-28744-S4F8H).

For the ADDITION-plus cohort, 239 participants from the ADDITION Cambridge main cohort (MR1406) detailed above were recruited for this sub-study alongside 239 additional participants recruited from the Cambridge area, who had not previously been involved in any ADDITION related studies. These 478 participants form the full ADDITION-plus cohort for which this Data Sharing Agreement pertains to (MR1417).

Hospital Episode Statistics (HES) data for the 239 participants recruited from the ADDITION Cambridge main cohort has already been obtained by the University under the valid agreement NIC-28744-S4F8H. Therefore the data requested under this agreement, as with the List Clean data previously disseminated, relates only to the additional 239 participants recruited from the Cambridge area for the purpose of the ADDITION-plus study only.

The University of Cambridge would also like to include the processing of HES data held under the agreement NIC-28744-S4F8H for the purpose of the ADDITION-plus study in this request. Only HES data relating to the member of the full ADDITION-plus cohort will be processed for the purpose of this request. Any data relating to participants of other ADDITION studies will not be used for the purpose of this request.

Data Summary

The University of Cambridge has previously received record-level identifying data, for the 239 participants not previously involved in any ADDITION related studies, via the MRIS List Clean Service for the purpose of ensuring the accuracy of participant demographic details for several follow-ups.

Under this version of the agreement, University of Cambridge require access to record-level identifying data for the same sub-cohort of participants (239) from the following HES products:

HES Accident and Emergency

HES Outpatients

HES Admitted Patient Care

Historical HES data is being requested for these datasets as University of Cambridge requires data from 2001/02 (the time of recruitment) to present, therefore all available annual data is requested under each of the datasets above. The HES data being requested will be pseudonymised using the Study ID, however this data will be linked to demographic details previously disseminated under the MRIS – List Clean Report, therefore this data is considered identifiable.

Section 251 support has been granted to permit the study to obtain up to date address details for all participants in order to inform them of the 10 year follow up (and if they wish, speak to the study team about the study); invite them to fill in questionnaires, and to update the study team of their current GP practice in order to complete the follow up, as well as obtaining linked Hospital Episode Statistics (HES) data.

The University of Cambridge also holds Sentinel Stroke National Audit Programme (SSNAP) (access to this provided via the Royal College of Physicians) and is in the process of obtaining MINAP (Myocardial Ischemia National Audit Project) data (access provided by the National Institute for Cardiovascular Outcomes Research [NICOR]).

University of Cambridge is the lead site for a Europe-wide study called ADDITION-Europe, which pools data from Cambridge and three other centres across Europe. However, none of the data from MR798, MR1406, or MR1417 will be shared with the wider European study, it will be used in Cambridge only. For clarity, only the University of Cambridge will have access to the data shared under this Agreement.

This study will allow an assessment of the long term effects of the differences in intensity of treatment achieved during the first five years after diagnosis. One of the objectives of the current follow up phase is to evaluate whether follow-up through GP notes can be replaced with follow-up through HES for a potential 15-year follow-up of this cohort. Follow-up through GP notes is a very time-consuming process both for the researchers and for the many GP surgeries involved in ADDITION, and has taken about 2 years. This is exacerbated by the gradual movement of participants to new surgeries over the study, meaning that for 10-year follow-up many GP surgeries who were contacted for notes were never originally involved with the study. The amount of work involved in follow-up through GP notes will only increase as the time since the start of the study increases.

The HES data is being requested for the ADDITION-plus ten year follow-up study which is assessing the effectiveness and cost effectiveness of adding a facilitator-led theory-based behaviour change intervention to intensive treatment of screen- and clinically- detected type 2 diabetes patients.

This study aims to collect follow up information on cardiovascular events and risk factors, treatment and mortality for the cohort of participants of the ADDITION-plus study who enrolled in the UK to contribute to 10 & 15 year follow-up work.

In order to assess whether follow-up through routine sources alone would be practical, the University of Cambridge research team need to work out whether the events captured through HES data match up with the events captured through GP note follow-up. The research team need HES data covering the whole duration of the study. The first ADDITION-plus participants were recruited in March 2002, and so data from 2001/02 is required (where available) to the present (therefore 15 years of data is required).

With the HES data, the University of Cambridge plan to extract study outcome data (such as inpatient admission for a heart attack). This will then form part of the master dataset which will also include similar outcome data that has been extracted from other sources (MINAP, SSNAP, Questionnaires & GP records). All identifiers will be removed from the master dataset once this has been formed.

The University of Cambridge also plan to use all data sources to provide data for the health economics assessment of cost utility of the study intervention. Again, this will be at aggregated level, with small numbers suppressed in line with the HES Analysis Guide.

The University of Cambridge contacted all ADDITION-plus participants in the UK with a self-report questionnaire to assess health behaviour and patient-reported outcomes.

Prior to sending any questionnaires, the participant list was cross-checked with available records from MRIS using the List Clean service, to minimise the risk of sending questionnaires to participants who are deceased.

The purposes for data processing are:

(1) cross reference the currently-held address and GP data with those held by NHS Digital (via MRIS) to ensure the participant records are up to date and to allow follow-up of participants who have moved since last follow-up (this has already been completed).

(2) identify CVD (cardiovascular disease) outcomes recorded in HES data by scrutinising inpatient, outpatient and A&E admissions for this cohort, to quantify completeness of follow-up. Events identified through HES records will be compared with events identified through already-collected self-reported data and GP record data, and with data from Myocardial Ischaemia National Audit Project (MINAP) and the Sentinel Stroke National Audit Programme (SSNAP), to establish whether it is feasible to conduct future follow-ups of this cohort through HES data alone.

(3) assess cost-effectiveness of screening using HES data by identifying and costing all hospital admissions, including hospital admissions for reasons other than the primary end point.

Purpose (1) data was needed to complete 10-year follow-up of this cohort with as high a follow-up rate as possible, in order to inform the management of newly diagnosed patients and to establish the size and nature of the benefits of detecting and treating diabetes earlier. Ten year follow-up of the trial participants will add to the existing research base concerning early treatment of type 2 diabetes and inform NHS policy decisions concerning whether population-based diabetes screening programmes should be established in Europe and worldwide. As detailed, this has already been completed.

Purpose (2) will allow the study to establish whether future follow-up of this cohort can be conducted using HES data alone, which could greatly simplify the process of data collection and reduce costs and time needed for future follow-up.

Purpose (3) will allow the study to quantify the total healthcare costs following diabetes diagnosis, and establish whether they are reduced by early intensive treatment.

These studies are part of the University of Cambridge and as such the legal basis to hold and use personal data is covered under the General Data Protection Regulation “Article 6(1) (e) processing is necessary for the performance of a task carried out in the public interest”. To hold special categories of personal data our lawful basis is for pursuing scientific research under “Article 9(2)(j) processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) based on Union or Member State law which shall be proportionate to the aim pursued, respect the essence of the right to data protection and provide for suitable and specific measures to safeguard the fundamental rights and the interests of the data subject.”

The University of Cambridge will submit a file containing the identifiers of the 239 participants recruited to NHS Digital using its secure electronic file transfer system. This will include:

STUDY_ID

NHS number

Date of Birth

Sex

Postcode

HES records of these participants will be extracted. No additional filters will be applied to the data, nor any additional derived fields provided.

The HES data flowing from NHS Digital will be pseudonymised, however, the University of Cambridge will use the Study ID to link the HES data to the data previously disseminated (i.e. the List clean data and the data the University of Cambridge holds). By means of this re-identification, the HES data to be disseminated is therefore considered Identifiable.

The data will be downloaded at the University of Cambridge's MRC Epidemiology Unit and transferred immediately to an independent, physically-separated network that is isolated from public network systems and can only be accessed locally, with a managed access system including both password and procedural controls. This separate database is still on the Unit

premises but is known as the "private network" where all of the Unit's patient data is stored. It is not connected to the internet and can only be accessed by being at the Unit. Access to this network must be approved by both local senior management and the ADDITION study CI. All study team members accessing the data have a contract of employment with the Unit.

Purpose (1):

The address and GP data has been used to re-contact participants who were lost to follow up, and to contact their registered GP surgery to enable the collection of recent endpoint and clinical measures. Where NHS Digital data identifies changed contact details for participants, the data has left the MRC Epidemiology Unit in one of two ways:

(a) Individual letters have been sent to participants informing them of the 10-year follow-up process, reminding them of their right to withdraw from the study, providing them with an up to date study Participant Information Sheet, and inviting them to fill in study questionnaires. Data shared in this way will necessarily involve sharing the participant's name and address, but will not include any other data from NHS Digital, nor will the letters include linkage to any other study data.

(b) Individual letters have been sent to GPs asking them to provide data from relevant consenting participants' medical notes. Data shared in this way must be sufficient to allow the GP surgery to identify the correct person, but will be limited to the participant's name, date of birth, and NHS number. The letters will not include any other study data.

Address and GP details have been linked to the University of Cambridge's existing contact records via a study identifier. The address and GP data supplied by NHS Digital will never be released for analysis; it has only been used for locating participants and their GPs to enable them to be asked if they wish to take part in the 10 & 15 year follow-up work.

Personal identifying data provided for this purpose has been stored on a physically separate server in the Unit offices and can only be accessed and used on site by those who have permission within the research team. Other than this contact with individual participants who were previously lost to follow-up and their GPs, no personal identifying data will be used

by or given to any other third party, and no record-level data will be shared outside the Unit.

The List Clean data has been used to update the University of Cambridge ADDITION participant contact database with current address details. Addresses have been updated to send participants information on the current wave of data collection. Unless a participant specifically requests to opt out, following receipt of the current participant information sheet, their consent to hold and collect future study-related health data continues based on their original consent taken at baseline and the study's current s251 approval.

The University of Cambridge will retain updated addresses in the database to enable continued correspondence with participants to keep them updated on future waves of data collection. The team also plan to continue sending Christmas cards and newsletters to participants as has been done throughout the course of the study and will use the updated

addresses for this purpose also. The team will also hold further public meetings to disseminate the results of the study and would invite participants through letter to these.

The list clean data is not linked to study questionnaires, MINAP or SSNAP data. The ADDITION contacts database with personal identifying data is held separately to any outcome data and linked only by ID numbers.

Purpose (2):

The ADDITION research team at Cambridge plan to use HES data to verify data from other sources and to ensure completeness of the data set where a source has missing data. The study's main outcome is whether someone has had a cardiovascular event. Whilst the team have access to participant's GP records there may be cases where hospital discharge summaries relating to an event (such as a heart attack) have not been filed with the notes. In these cases, the event may be captured on one of the requested datasets to enable the study to have accurate and complete data. This will prevent incorrect conclusions being drawn based on incomplete data.

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

Identifying data provided for this purpose will be kept on a physically separate server in the Unit offices and can only be accessed and used on site by those who have permission within the research team. No personal identifying data will be used by or given to any other third party, and no record-level HES data will be shared outside the Unit.

Aggregated data with small numbers suppressed in line with the HES analysis guidelines will be published in papers describing the proportion of events identified through each data source, to aid other researchers in deciding on the best sources to use for data collection; all data published will be aggregated with small numbers suppressed in line with the HES analysis guidelines. Data used during this analysis will be pseudonymised- HES events will be classified into broad categories and only month and year of event will be released for analysis intended for publication.

Purpose (3):

HES data will be used to identify all hospital admissions, and linked to NHS cost codes to quantify the total healthcare costs following diabetes diagnosis, and establish whether they are reduced by early intensive treatment. This data will be linked to the Unit's existing records (self-report data and data directly collected by the study team, e.g. heights/weights/blood sugar levels) via the study identifier.

Data used during analysis will be pseudonymised- only month and year of event will be released for analysis intended for publication, and details of HES events will be reduced to the minimum necessary to be able to identify the appropriate NHS cost code.

Pseudonymised data may be released from the Unit's physically separate server onto the Units main network, and may be accessed on site or by remote access, data is not downloaded locally.

All individuals with access to this data will either be substantively employed by the University of Cambridge, or will have a Visiting Worker honorary contract and be subject to the same policies, procedures and sanctions as substantive employees.

The data from NHS Digital will not be used for any other purpose other than that outlined in this Agreement.

No personal identifying HES data will be used by or given to any other third party, and no record-level HES data will be shared outside the Unit.

All outputs and publications contain only aggregated data with small numbers suppressed in line with the HES Analysis Guide.

The outputs achieved from using the List Cleaning service are:

(1) Informed living participants of the 10-year follow-up process; reminded them of their right to withdraw from the study; provided them with an up to date study Participant Information Sheet, and invited them to fill in study questionnaires;

(2) Requests have been sent to participants' GPs asking them to provide data from consenting participants' medical notes;

(3) Medical notes passed to the ADDITION Plus Study Research Assistant who is to review the notes to look for CVD endpoints and clinical measures.

(4) Use HES data as another method to ascertain CVD outcomes to ensure completeness of the data.

Outputs 1, 2 & 3 has been completed as of Spring 2020.

The aim of contacting participants is to enable the study to continue collecting information to be used in the ADDITION-Europe study.

It is essential for the study to have up to date events as the statistical power of the analyses depend on the number of known events. Less common outcomes can only be studied with sufficiently long follow-up and event numbers;

The ADDITION-Plus study has so far led to the publication of 10 papers in peer-reviewed scientific journals. Data from ADDITION-Plus has also contributed to 5 PhD theses. The primary analysis of 1-year outcomes was published in Diabetologia (Griffin et al. (2014) Diabetologia, 57, 1308-19). The results of the 5-year analysis have been submitted for publication. Analysis of the 10-year data will be submitted to a similar leading medical journal. Throughout the next three years, secondary analyses and mechanistic analyses will be submitted for publication to leading medical or disease-specific peer-reviewed journals. All publications will be open access, in line with the University of Cambridge open-access policy, and can be accessed by clinicians, academics, policy makers and interested members of the public. Outputs presented and/or reported will contain aggregate level data with small numbers supressed in line with the HES analysis guide. No personal identifying data will be released or published. Findings will also be used to advise policy makers on the cost-effectiveness of adding lifestyle intervention to treatment packages for people who are newly diagnosed with diabetes, and the level of behavioural change that is associated with reduction in risk of cardiovascular outcomes. When the study is completed, meta-data will be added to the online data dictionary http://epi meta.medschl.cam.ac.uk/includes/add/add.html. Bonafide researchers can then apply to access the anonymised data set to conduct their own secondary analyses subject to the terms of the MRC Data Sharing Policy http://epi-meta.medschl.cam.ac.uk/data_sharing_policy.html and appropriate collaborative/data sharing agreements. The University of Cambridge will also send all ADDITION Plus participants a summary of the results in the form of a newsletter.

The primary benefits of using NHS Digital’s List Cleaning service are that it enables ADDITION-Plus to continue to collect information about its participants fairly and transparently giving participants the option to withdraw should they wish and mitigates the risk of attempting to contact deceased participants and potentially causing distress to living relatives.

The benefits of collecting information about the participants are achieved through the wider study. The 10 year follow up main results paper, which includes data collected after re-contact of participants and their GPs which was made possible by the list clean data in this application, looks at the difference in cardiovascular events between the control and intervention arms. This will add to the evidence of treatment and screening options for diabetes and will influence NHS policy makers and clinician decisions nationally on the best way to care for this population.

Through finalising work contributing to the 10 year follow up main results paper, HES data for the preliminary years to the 10 year follow-up will be used to look at the difference in cardiovascular events between the control and intervention arms. This will add to the evidence of treatment and screening options for diabetes and will influence NHS policy makers and clinician decisions nationally on the best way to care for this population.

The HES data will also be used to inform the cost-utility analysis. Completeness of information on the use of health services is crucial to enable a true cost to the NHS to be determined and hence influence implementation decisions on the course of treatment for the population. This could lead to reductions in NHS spending on treatments without proven effectiveness or to invest in treatments that will generate savings in the future through reduced NHS service use.

Finally, the HES data will also be used to look at how researchers can maximise efficiency in research data collection to minimise burden on the NHS. The HES data will be cross-checked with that which has already obtained from individual primary care records. The results of the agreement of these two sources will be analysed by the team. If HES provides a complete dataset of outcomes, in future, there will be no need to burden GP practices with requests for medical notes for research purposes as secondary care data will be proven reliable and

comprehensive. This will free up substantial practice staff time which can be then refocused on their main activities - providing frontline care to their patients.

It is estimated that 1 in 16 UK adults has (diagnosed or undiagnosed) type 2 diabetes, and this creates a substantial burden of suffering and health service use. Treatment of type 2 diabetes and related complications (cardiovascular disease, amputation, blindness, kidney failure) accounts for 10% of the NHS budget. This is expected to rise as the number of people in the UK who have type 2 diabetes is estimated to rise to 6.25 million by 2035.

Type 2 diabetes is frequently asymptomatic, with the true onset occurring several years before diagnosis. While detection of the condition may be improving, around 30-50% of people with diabetes remain undiagnosed, and when patients are diagnosed, around 20-30% have evidence of diabetic complications. Long-term follow-up of the wider ADDITION trial will inform the management of newly diagnosed patients and to establish the size and nature of the benefits of detecting and treating diabetes earlier.

The ADDITION (Anglo-Danish-Dutch Study of Intensive Treatment In People with Screen Detected Diabetes in Primary Care) Plus study is a randomised controlled trial (i.e. an equal chance of being selected in the control group or intervention group) to see if changes in behaviour (physical activity, diet, smoking and adherence to medication) can be achieved among individuals recently diagnosed with type 2 diabetes via an intervention delivered by facilitators trained in behaviour change theory and techniques, and whether behaviour change is associated with reductions in risk factors for cardiovascular disease.

All participants were receiving intensive multifactorial treatment (i.e. treating several conditions all at once) as part of the parent trial (ADDITION) and half were randomised to also receive the behavioural intervention. The initial trial included follow-up at one and five years incorporating self-report and objective assessment of four behavioural outcomes (physical activity, diet, smoking and medication adherence) and cardiovascular risk factors. The lifestyle intervention did not improve modelled (i.e. predicted) cardiovascular risk, but it did improve functional status and health utility. These latter variables as well as acquired skills and knowledge relating to behaviour change could have implications for longer-term cost-effectiveness.

Ten-year follow up of the ADDITION-Plus trial will establish the longer-term effectiveness and cost-effectiveness of delivering a behavioural intervention among intensively-treated individuals with recently diagnosed type 2 diabetes in primary care. It will also allow the University to evaluate the impact of this intervention on cardiovascular endpoints (such as heart attacks and strokes, as well as mortality) which were not previously captured. These data would also provide a unique opportunity to investigate the association between changes in objectively measured behaviours in the first five years after diagnosis and cardiovascular outcomes over ten years, and to estimate the amount of behavioural change that interventions need to achieve to have a significant impact on cardiovascular outcomes. Results will add evidence to decisions about treatment from diagnosis and the balance between treatment and disease burden.

ADDITION Plus investigators contribute to the organisation and delivery of diabetes care both locally and nationally (e.g. guideline development, managed care networks, expert review group for diabetes QOF indicators, National Screening Committee Advisory Group, NICE committees, NHS Health Checks advisory panel, Diabetes Prevention Programme advisory panel) and therefore have established mechanisms for influencing policy and practice in these and related fields. Results from this study will help inform care early in the course of the disease.

The primary benefits of using NHS Digital's List Cleaning service are that it has enabled ADDITION-Cambridge to continue to collect information about its participants fairly and transparently giving participants the option to withdraw should they wish. Furthermore, knowing which participants are deceased has enabled the study team to avoid attempting contact and potentially causing distress to living relatives.

Being able to track and trace participants has meant that the study team has been able to collect more information from the respective GP practices (as stated in Output 2 above) which will greatly improve the power of the study, thus making it more useful for planning NHS resource allocation and best practice in the treatment of type 2 diabetes. The extra information collected as a result of the track and trace data will be included in subsequent publications.