A community-based primary prevention programme for type 2 diabetes mellitus integrating identification and lifestyle intervention for prevention: a cluster randomised controlled trial

Article history

The research reported in this issue of the journal was funded by PGfAR as project number RP-PG-0606-1272. The contractual start date was in August 2007. The final report began editorial review in January 2015 and was accepted for publication in November 2015. As the funder, the PGfAR programme agreed the research questions and study designs in advance with the investigators. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PGfAR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Melanie Davies has acted as consultant, advisory board member and speaker for Novo Nordisk, Sanofi-Aventis, Eli Lilly and Company, Merck Sharp & Dohme, Boehringer Ingelheim, AstraZeneca and Janssen, and as a speaker for Mitsubishi Tanabe Pharma Corporation. She has received grants in support of investigator and investigator-initiated trials from Novo Nordisk, Sanofi-Aventis and Eli Lilly and Company. She has received grants and support from the National Institute for Health Research (NIHR) during the conduct of this study. Alastair Gray reports grants from NIHR during the conduct of the study. Kamlesh Khunti reports that he has acted as a consultant and speaker for Novartis, Novo Nordisk, Sanofi-Aventis, Eli Lilly and Company and Merck Sharp & Dohme. He has received grants in support of investigator and investigator-initiated trials from Novartis, Novo Nordisk, Sanofi-Aventis, Eli Lilly and Company, Pfizer, Boehringer Ingelheim and Merck Sharp & Dohme. Kamlesh Khunti has received funds for research, honoraria for speaking at meetings and has served on advisory boards for Eli Lilly and Company, Sanofi-Aventis, Merck Sharp & Dohme and Novo Nordisk.

Copyright statement

© Queen’s Printer and Controller of HMSO 2017. This work was produced by Davies et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Identification

Type 2 diabetes mellitus is at one end of a continuous glucose control spectrum, with normal glucose control at the other. In between these two extremes there is a clinically important and much-researched state in which glucose levels are elevated but not over the threshold for diagnosis of T2DM. This state of glucose control has historically been termed prediabetes mellitus (PDM), impaired glucose regulation (IGR) or intermediate hyperglycaemia. Individuals with these elevated glucose levels are significantly more likely to develop T2DM than those with normal blood glucose.

The tests that can be used to identify those at high risk of T2DM include the oral glucose tolerance test (OGTT) and, more recently, glycated haemoglobin (HbA_1c). There are usually two blood measurements taken for an OGTT: fasting plasma glucose (which is the glucose measured after 12 hours of fasting before glucose is taken) and the 2-hour post-challenge plasma glucose (which is measured 2 hours after the glucose is taken). The World Health Organization (WHO) defines impaired glucose tolerance (IGT) as a 2-hour post-glucose reading of > 7.8 mmol/l but < 11.1 mmol/l, whereas impaired fasting glucose (IFG) is defined as fasting plasma glucose concentration of > 6.1 mmol/l but < 7 mmol/l. 8 IFG and IGT can occur as isolated, mutually exclusive conditions or together. Estimates of progression to T2DM within 1 year suggest that those with isolated IGT have over five times the risk, those with isolated IFG have seven times the risk and those with both IGT and IFG have over 12 times the risk than normoglycemic individuals. 9 Both the terms PDM and IGR are commonly used to describe the presence of IFG and/or IGT as defined by the WHO.

In 2011, after the start of this work, WHO revised the criteria for the diagnosis of T2DM to include the use of HbA_1c. 10 This precipitated a shift in clinical practice, with the use of OGTT in the diagnosis of T2DM being gradually phased out and HbA_1c becoming the dominant method of classification. As it is potentially burdensome and confusing to define categories on a continuous glucose spectrum with different measures, this change necessitated a discussion around the definition of an HbA_1c-defined PDM category analogous to that of IGT or IFG. Although the WHO found insufficient evidence for the use of HbA_1c in the definition of PDM, statements from the American Diabetes Association (ADA) and an international expert committee recommended that HbA_1c be used to signify a high-risk state at levels of between 5.7% and 6.4%, and 6.0% and 6.4%, respectively. 11,12 The National Institute for Health and Care Excellence (NICE) have since adopted the recommendation of 6.0% to 6.4% as an alternative to fasting or 2-hour glucose in the identification of PDM. 13 Follow-up studies have shown similar rates of progression to diabetes mellitus from the HbA_1c-defined prediabetic state as seen for IFG. 14

In most countries, around 15% of adults have PDM based on WHO criteria;15 this figure rises in some minority populations and with age. For example, in elderly populations up to 50% of individuals are estimated to have PDM. 16 Of those with PDM, an estimated 4–12% develop T2DM per year, with the highest rates seen among those with both IGT and IFG. 15,17 Evidence from prospective studies suggests that approximately 25–40% of individuals with PDM go on to develop diabetes mellitus over a 3- to 8-year period, and as many as 70% will eventually develop T2DM over the course of their lifetime. 17

Along with an increased risk of T2DM, the risk of CVD and premature mortality is also elevated in individuals with PDM. 15,18 For example, those with PDM have been shown to be 50% more likely to die of CVD than people with normal blood glucose control. 19 Interestingly, as the prevalence of PDM is four times greater than T2DM, there are likely to be more premature deaths attributable to PDM than to diabetes mellitus. 19,20

Research published by Diabetes UK proposed that, although the terms IGR, IFG and IGT may be useful when talking to health-care professionals, the term ‘prediabetes’ was found by focus groups to be preferable when talking to the public. It was noted that people identified with this term and felt that it adequately portrayed the seriousness of the condition and its future risk. 21,22 Recent recommendations suggest that individuals with either IGT, IFG or elevated HbA_1c be referred to as ‘persons at high risk of T2DM’; however, at the onset of this study the term PDM was commonly used and thus has been adopted throughout this report. The term PDM is used to refer interchangeably to IGT, IFG and/or elevated HbA_1c, according to any recommended definition. 12,13,23 Throughout this report we shall use the term PDM to include IGT-, IFG- or HbA_1c-identified high risk of T2DM.

Regardless of the invasive biochemical test used to define risk status, universal screening for T2DM risk status is problematic for several important reasons. First, screening tests are relatively expensive and there is limited appetite in an era of restricted health-care budgets for screening apparently healthy individuals for disease risk. This is consistent with a review of the evidence commissioned by the Health Technology Assessment programme, which concluded that screening for T2DM meets most of the National Screening Committees’ key criteria, although it fails on several, including a lack of adequate staffing and facilities. 24 In addition to cost and resource, there is high variation in the risk of developing both T2DM and CVD across categories of PDM, regardless of the assessment method used. For instance, data from the Finnish Diabetes Prevention Study (DPS) showed that the risk of T2DM in those with IGT more than doubled with the presence of other readily identifiable risk factors. 25 It is also known that the risk of CVD increases linearly with increasing levels of dysglycaemia, and there is no distinct threshold that justifies the use of distinct risk categories. 26 Given these factors, there has been much international focus on developing pragmatic systematic approaches for identifying and stratifying individuals with an elevated risk of T2DM for referral into diabetes-prevention initiatives. These have primarily focused around the use of risk-score technology.

Prevention programmes

Clinical trials have unequivocally demonstrated that lifestyle interventions reduce the risk of progressing to T2DM by 40–60% in those with PDM, specifically IGT. 31 For example, the Finnish DPS found that the risk of T2DM was reduced by 58% in those with an intensive lifestyle intervention compared with usual care over a 3-year period. 32 Identical findings were reported for Diabetes Prevention Program (DPP) conducted in the USA. 33 Similar and consistent results have been observed in many different and diverse countries including India,34 Japan35 and China. 36 Lifestyle interventions aimed at the prevention of T2DM have been based on promoting moderate- to vigorous-intensity physical activity, generally 150 minutes per week, and a healthy diet aimed at weight maintenance for normal-weight individuals or weight loss for overweight or obese individuals. For example, DPS had five intervention goals: (1) a reduction in body weight of ≥ 5%; (2) < 30% of energy intake derived from fat; (3) < 10% of energy intake derived from saturated fat; (4) at least 15 g of fibre per 1000 kcal; and (5) at least 30 minutes of moderate-intensity physical activity per day. 32 Interestingly, there was not a single case of T2DM over the 3-year study period in those who achieved four of these goals. 32

Successful lifestyle-change programmes have also been shown to have so-called legacy effects whereby the effect persists well after the active intervention has ceased. DPS, DPP and the Chinese Da Qing diabetes mellitus prevention study all found sustained reductions in the incidence of T2DM relative to the control group after 7–20 years of follow-up. 37–39 These findings suggest that once individuals are enabled to successfully change and self-regulate their lifestyle behaviours, benefits can be sustained long after active lifestyle interventions have ceased.

Economic modelling studies have consistently demonstrated that lifestyle-based interventions are likely to be cost-effective and may even be cost saving in some populations. 40 When considering the whole process from screening to treatment, Gillies et al. 41 estimated that screening for T2DM and PDM followed by tailored treatment to each group was more cost-effective than screening for T2DM alone in the UK, with lifestyle interventions being more cost-effective than pharmaceutical therapy for prevention [£6242 vs. £7023 per quality-adjusted life-year (QALY) gained].

The consistent clinical effectiveness and cost-effectiveness of lifestyle interventions aimed at decreasing the risk of T2DM is unsurprising given that unhealthy lifestyle practices associated with modern ‘obesogenic’ environments are the primary causal factor for T2DM. The prevalence of T2DM has been estimated to have increased by a factor of six over the past couple of centuries, ruling out genetic change as a direct causal factor. 42 Although there are some genetic factors that increase the risk of T2DM, they can be expressed only in combination with unhealthy modern environments. Given the centrality of lifestyle factors in the pathophysiology of T2DM, and considering the strong evidence of efficacy for lifestyle intervention, the promotion of lifestyle change is central to the prevention of T2DM. This has been recognised by NICE, which recommends that those identified as being at high risk of T2DM should be referred to a lifestyle intervention before pharmaceutical agents, such as metformin (e.g. Glucophage, Merck), are considered. 13

Summary

In conclusion, the prevention of T2DM is a recognised national and international health-care priority. However, in order to make prevention a reality rather than an aspiration within primary care, there is an urgent need to develop methods and pathways that can facilitate the identification of those with PDM and provide methods of delivering lifestyle intervention that are both evidence-based and suitable for mass implementation within the existing health-care infrastructure. Our programme grant was aimed at addressing this unmet need within the context of primary care in the UK.

The main objectives of the programme grant were:

• to develop and validate a risk score to identify those who require diagnostic testing to identify undiagnosed T2DM and those at high risk of future T2DM and CVD in a multiethnic population

• to use this risk score to identify and engage those at highest risk of T2DM and offer them a lifestyle self-management programme with the aim of reducing the risk of progression to T2DM and reducing cardiovascular risk

• to pilot and test a lifestyle self-management programme based on group care, targeting the five key areas, using information currently collated from the European Union-funded Diabetes in Europe Prevention using Lifestyle, physical Activity and Nutritional intervention (DEPLAN) project

• to develop a training and quality-assurance programme for community-based health trainers, who may include health-care professionals, to deliver the initial programme and provide ongoing support to those at highest risk of T2DM

• to evaluate the lifestyle self-management programme and its cost-effectiveness

• to explore how a two-stage screening programme and prevention intervention can be implemented in primary care.

Introduction

A major opportunity exists to drastically reduce the incidence of T2DM, a disease that has a huge impact on patients and health-care systems worldwide. Large, high-quality clinical trials31,33,39 show that relatively modest changes in diet and physical activity reduce the incidence of T2DM by > 50% for people with PDM. Indeed, within-trial data show that the rate of progression to T2DM after 7 years of follow-up was reduced to almost zero for people who had succeeded in making five modest lifestyle changes. 39 The main drivers of diabetes mellitus prevention appear to be weight loss and physical activity. 53,54 However, a substantial challenge remains in translating these findings into routine clinical practice. The intensive and prohibitively expensive interventions used in clinical trials to ensure lifestyle change need to be translated into practical, affordable interventions that are deliverable in real-world health-care systems and that, nevertheless, retain a reasonable degree of effectiveness. 29

Since the publication of the original diabetes mellitus prevention clinical trials between 1996 and 2001, a number of translational or ‘real-world’ diabetes mellitus prevention programmes55,56 have aimed to translate the evidence. 32–34,36 A meta-analysis of the evidence on translational interventions was published in 2010,55 although this review excluded 15 studies that were conducted in non-health-care settings. A more recent meta-analysis was published in 2012. 46 However, the authors focused only on translation of evidence from the US DPP and also included studies where up to half of the population already had diabetes mellitus. Other systematic reviews of diabetes mellitus prevention interventions have either not included a meta-analysis54,56–60 or not focused on translational studies. 31,54,57,58,61–65 Overall, the systematic reviews conducted to date indicate that real-world diabetes mellitus prevention programmes vary widely in their effectiveness, although most produce lower levels of weight loss than the more intensive interventions used in the clinical efficacy trials. 55

To consolidate the evidence we undertook a systematic review of studies, considering the effectiveness of translational interventions for prevention of T2DM in high-risk populations. The primary aim was to conduct a meta-analysis of the effectiveness of pragmatic interventions on weight loss. If sufficient data were available, a secondary aim was to consider other diabetes mellitus risk factors using similar methods.

Methods

Results

Identification of studies

Results relating to the identification and selection of eligible trials are summarised in Figure 1. Searches yielded 8492 citations, and 5196 unique titles and/or abstracts were screened for eligibility. Following full-text retrieval of 167 potentially relevant papers, 20 additional papers were identified from reference lists, making a total of 187. Authors for 14 studies were then contacted in order to clarify eligibility criteria and/or for additional outcome data. Replies were received for 13 studies, 10 of which were subsequently included in the 29 studies50,62,69–95 (42 papers50,51,62,69–107) that met the review criteria.

FIGURE 1.

Flow chart of selection of studies from search to final inclusion. DM, diabetes mellitus; SR, systematic review.

Meta-analysis

Twenty-five studies involving 5785 participants (estimated 36% male) were included in the meta-analysis for mean weight change at 12 months. One study was excluded from the primary meta-analysis, as weight change was not recorded as a study outcome. 78 Three were excluded from all analyses as one study reported only 15-month data,92 and two were excluded as they reported only 18-month data. 83,91 Two studies included in the meta-analysis had two intervention arms,50,81 meaning that 27 study groups were analysed.

The pooled result of the meta-analysis (Figure 2) shows that lifestyle interventions resulted in a mean weight loss of 2.31 kg (95% CI –2.87 to –1.76 kg; I² = 92.9%).

FIGURE 2.

Forest plot showing mean weight change in each study and the overall pooled estimate. Adapted from Dunkley et al. 52 under Creative Commons public licence 3.0, https://creativecommons.org/licenses/by/3.0/. Additional studies (Janus et al. ,93 Kanaya et al. 94 and Penn et al. 95) have been added; therefore, the subtotal line and overall line in the plots have changed, as they now include additional data. Boxes and horizontal lines represent mean weight change and 95% CI for each study. Size of box is proportional to weight of that study result. Diamonds represent the 95% CI for pooled estimates of effect and are centred on pooled mean weight change. CPC, carbohydrate reduction and hunger focus post core; TPC, traditional post core.

Sensitivity analysis, restricted to RCTs only, indicated a mean weight change (–2.5 kg, 95% CI –3.8 to –1.2 kg) that is similar to the overall result. Additional analysis comparing the difference in weight lost between the treatment and control arms, for RCTs only, suggests that, on average, the intervention arm lost an extra –1.79 kg (95% CI –2.78 kg to –0.80 kg; p < 0.001). Furthermore, sensitivity analyses that included studies scoring ++ for external validity demonstrated a slightly greater weight loss in higher-quality studies (–2.8 kg, 95% CI –4.1 to –1.5 kg). However, there was some evidence of publication bias (p = 0.033, Egger’s test; see Figure 3 for funnel plot).

FIGURE 3.

Funnel plot with pseudo 95% confidence limits assessing publication bias for the primary outcome weight change.

All other outcomes showed an improvement at 12 months (Table 2), with all of these reaching statistical significance with the exception of HDL cholesterol. The pooled result for the 19 studies (20 study groups) that reported BMI demonstrates that lifestyle interventions resulted in a mean decrease in BMI of 0.98 kg/m² (95% CI –1.28 to –0.68 kg/m²; I² = 95.2%). A collective decrease in waist circumference of 3.36 cm (95% CI –4.33 to –2.39 cm; I² = 97.5%) was found across the studies that reported the measure (n = 18). The pooled result for the studies that conveyed HbA_1c percentages (n = 8) indicated that lifestyle intervention corresponded to a 0.11% (95% CI –0.19% to –0.03%; I² = 97.1%) decrease in HbA_1c. Significant reductions in fasting glucose of 0.10 mmol/l (95% CI –0.18 to –0.02 mmol/l; I² = 86.0%) and 0.36 mmol/l (95% CI –0.66 to –0.06 mmol/l; I² = 92.8%) in 2-hour glucose were suggested for lifestyle intervention. Total cholesterol was reported for 16 study groups, for which the pooled result of the direct pairwise meta-analysis indicated a 0.18 mmol/l (95% CI –0.23 to –0.13 mmol/l; I² = 39.8%) mean decrease in total cholesterol at 12 months for those in receipt of lifestyle intervention. A slightly smaller reduction in LDL cholesterol of 0.15 mmol/l (95% CI –0.22 to –0.07 mmol/l; I² = 66.0%) was demonstrated for the intervention groups by the pooled result for the meta-analysis including 10 study groups. The pooled result for HDL cholesterol indicated a 0.02 mmol/l (95% CI –0.002 to 0.04 mmol/l; I² = 92.8%) increase for the intervention groups across 16 study groups; however, this was not a statistically significant finding. The overall effect of lifestyle intervention on triglycerides showed a 0.1 mmol/l (95% CI –0.18 to –0.01 mmol/l; I² = 99.2%) decrease in triglycerides measurement spanning 14 study groups. Significant combined effects of intervention on BP were demonstrated, with a decrease in systolic BP of 4.02 mmHg (95% CI –5.66 to –2.37 mmHg; I² = 77.1%) over 16 study groups and 3.88 mmHg (95% CI –5.24 to 2.52 mmHg; I² = 83.7%) in diastolic BP across 12 study groups. Across the nine studies that reported incident diabetes mellitus, the pooled incidence rate was 35 cases per 1000 person-years (95% CI 24 to 53 cases per 1000 person-years), which gives the number needed to treat as 29.

TABLE 2 - Pooled effects for intervention groups showing change from baseline to 12 months

Not change from baseline.

Outcome	Number of study groups	Pooled effect	95% CI	p-value	I 2	Publication bias p-value
Weight (kg)	27	–2.31	–2.87 to –1.76	< 0.001	92.9%	0.033
BMI (kg/m2)	20	–0.98	–1.28 to –0.68	< 0.001	95.2%	0.067
Waist circumference (cm)	20	–3.36	–4.33 to –2.39	< 0.001	97.5%	0.136
HbA1c (%)	8	–0.11	–0.19 to –0.03	0.009	86.7%	0.961
Fasting glucose (mmol/l)	19	–0.10	–0.18 to –0.02	0.014	86.0%	0.344
2-hour glucose (mmol/l)	12	–0.36	–0.66 to –0.06	0.018	92.8%	0.156
Total cholesterol (mmol/l)	16	–0.18	–0.23 to –0.13	< 0.001	39.8%	0.776
LDL cholesterol (mmol/l)	10	–0.15	–0.22 to –0.07	< 0.001	66.0%	0.278
HDL cholesterol (mmol/l)	16	0.02	–0.002 to 0.04	0.082	92.8%	0.931
Triglycerides (mmol/l)	14	–0.10	–0.18 to –0.01	0.022	99.2%	0.585
Systolic BP (mmHg)	16	–4.02	–5.66 to –2.37	< 0.001	77.1%	0.018
Diastolic BP (mmHg)	12	–3.88	–5.24 to –2.52	< 0.001	83.7%	0.005
Incident diabetes mellitus (per 1000 person-years)a	9	35.3	23.6 to 52.7	< 0.001	79.5%	0.117

High levels of heterogeneity were demonstrated for all secondary outcomes. Further significant evidence of publication bias was apparent for the reporting of systolic (p = 0.018) and diastolic (p = 0.005) BP outcomes via Egger’s test. No other significant evidence of publication bias was detected.

Discussion

The 25 translational diabetes mellitus prevention programmes included in our meta-analysis significantly reduced weight in their intervention arms by a mean 2.3 kg at 12 months’ follow up. Where data were available, we found significant reductions in other diabetes mellitus and cardiovascular risk factors, including blood glucose, BP and some cholesterol measures. The pooled diabetes mellitus incidence rate in the intervention arms was 35 per 1000 person-years (number needed to treat 29). Outcome data on changes in the key lifestyle behaviour targets (physical activity and diet) were poorly reported.

Summary

Overall, the interventions were effective, but there was wide variation in effectiveness. More research is needed to establish optimal strategies for maximising both cost-effectiveness and longer-term maintenance of the lifestyle changes that these programmes can achieve.

Introduction

Risk scores are a way of stratifying a population for targeted screening. They use data from risk factors to calculate an individual’s score; a higher score reflects higher risk. Risk scores can be applied either to an individual as a questionnaire (these scores generally require only data from non-invasive risk factors, which would be known by members of the public) or to a population. Population risk scores are usually developed for use in primary care where a piece of software is used to calculate the score for everyone listed on the electronic medical records using routinely stored data. Screening invitations can then be sent to those at the highest risk.

Over the past decade, a plethora of risk scores have been developed and validated for detecting those at risk of T2DM. One of the first risk scores developed in this field was the FINDRISC score. This risk score was developed for use in Finland; it is questionnaire based and designed to be completed by members of the public to detect those at risk of developing T2DM in the future. 27 It includes eight questions relating to age, BMI, waist circumference, BP, history of high blood glucose, family history of diabetes mellitus, physical activity and consumption of vegetables, fruits or berries. This score has been shown to have acceptable levels of discrimination and, since its development in 2003, it has been validated for use in Greece,114 Bulgaria,115 Italy,116 Spain117 and Sweden. 118 It was decided not to validate this score for use in this project for a number of reasons. First, the FINDRISC was not developed for detecting those with existing undiagnosed PDM (IFG and/or IGT) and T2DM. Second, it is well reported that risk scores that have been developed for a particular population tend to have low validity when used on another. In addition, the FINDRISC does not include ethnicity, which is an important risk factor when assessing risk in a multiethnic population such as in the UK. 28,119,120 Third, the questionnaire nature of this risk score and the inclusion of patient-specific risk factors that would not be available routinely in primary care meant that this risk score could not be implemented in primary care for population-based stratification.

The Cambridge Diabetes Risk Score (CDRS) addresses some, but not all, of these issues. 121 This score was developed to detect undiagnosed T2DM and it collects data on age, sex, BMI, steroid and antihypertensive medication, and family and smoking history. This score would be suitable for use in primary care but it does not detect current undiagnosed PDM and it does not reflect the higher incidence of T2DM in those from black and minority ethnic (BME) groups. The FINDRISC identifies people who are at risk of developing T2DM in the next 10 years, with the CDRS detecting current undiagnosed T2DM only. To date, there is no evidence base for intervening in such a group for the prevention of T2DM. The evidence from the large pivotal trials for preventing T2DM is in people with IGT. 31 The ultimate aim of this programme of work is to develop and test a pragmatic intervention, taking the learning of the previous trials, delivered in a UK primary care setting. Therefore, we wished to identify people who have PDM rather than those at risk of developing diabetes mellitus in the future. Hence, it was decided to derive and statistically validate a new risk score that detects PDM/T2DM for use in a multiethnic population using data from two existing population-based screening studies from Leicester and Leicestershire. 122,123

The development and validation of the Leicester Practice Risk Score (LPRS) had three phases. Initially, a pilot score was developed and validated, and tested in two general practices (phase one). The aim of the pilot phase was not so much to assess the performance of a risk score per se, but to test the feasibility of a risk-score approach for identifying people with PDM in primary care. Owing to the milestones required for the programme of work, this feasibility testing needed to be completed before the final data set from the large population-based screening study was ready for analysis. A very simple pragmatic score was therefore derived to enable this approach to screening to be tested. Reporting of details of how this score was derived is outside the scope of this report. Following this pilot, complete data from a large-scale population-based screening study [Anglo–Danish–Dutch study of Intensive Treatment In people with screen detected diabetes in primary care (ADDITION)] became available; therefore, the score was redeveloped based on the learning from the pilot study. This score was subsequently used to identify those at high risk for screening within Let’s Prevent (phase two). Following this, the score was updated based on subsequent improvements in data completeness in primary care and the addition of HbA_1c to the diagnostic criteria for T2DM (phase three). Given that this score is published and used in clinical practice, full details of the development and validation are given for the final updated score.

Data sets

Data sets from two existing closely related screening studies were used throughout all three phases, that is, ‘Screening Those At Risk’ (STAR) and ADDITION. These are described briefly below and their shared methodology is outlined in the final section.

Statistical methods

The purpose of the risk scores was to identify those at greatest risk of glucose intolerance, defined as those with either T2DM or PDM (which includes IFG and/or IGT) who, up until screening with an OGTT, had been undiagnosed. All of the scores developed and validated as part of this project used similar methodology. To avoid repetition this is detailed below. Where differences occurred, these are also summarised.

Phase one: pilot risk-score results

The risk score developed to conduct the screening pilot in primary care was derived using the STAR data set. The final model is given in Table 5.

From the coefficients of the final model a crude pilot risk score formula was defined as the sum of:

1. person’s age (in years)

2. twice their BMI (in kg/m²)

3. 10 if they are male (no change if female).

A simplified score was preferred for the pilot study, as this would be easily calculated by researchers; the score for the main study was calculated using a piece of software and, therefore, the simplification of the relative weightings of the score components was not needed.

Using the score to identify the 10% most at risk for invitation to further screening identified 132 people with PDM in the STAR data set and 41 people with T2DM, representing a sensitivity of 20.1% and a specificity of 92.4%. Increasing the threshold so that the top 28% of people at risk were invited for screening increases the sensitivity to 43.6% and reduces the specificity to 75.2%.

Results from the pilot screening study

Phase two: initial Leicester Practice Risk Score results

Phase one showed that it was feasible to use a risk-score approach to identify people with undiagnosed PDM and T2DM in primary care. The next phase was to develop the risk score for use in the main Let’s Prevent study and to assess its validity using the STAR study data set that had been used for the development of the pilot risk score. The starting point for the development of this risk score was different from that for the pilot. The initial plan was to recruit 20 patients from each practice; data on ethnicity would be available in terms of proportions of the main ethnic groups at the practice level but not available from individuals. It became apparent that in order to fulfil the study aim of recruiting those eligible patients judged to be at highest risk of conversion to T2DM, a more efficient sampling plan would be to invite patients for an OGTT, starting from those with the highest risk score and moving down in order of score to a common level across all the practices that would result in the required study sample size. The number of patients invited for OGTT would, therefore, reflect the size and general risk levels of practices, and it would be valid to include a proportion of patients from ethnic minority groups when deriving the risk score.

From the modelling results (Table 7), the initial risk score was defined for the study as:

Risk score = 0.0407 × age (years)

+ 0.296 (if male, no change if female)

+ 0.934 (ethnicity, as practice proportion of South Asians)

+ 0.0859 × BMI (kg/m²)

+ 0.440 (if family history of diabetes mellitus, no change otherwise)

+ 0.374 (if on antihypertensive medication, no change otherwise).

Statistical validation of the initial risk score was carried out by examining its performance temporally on the STAR data and, additionally, by comparing its discrimination with other standard risk scores and the pilot risk score with respect to the area under the ROC curve. This was carried out for the total sample, and then separately for the South Asian and white European cohorts. Table 8 shows that the ROC area under the curve (AUC) of the initial risk score had better discrimination than either the FINDRISC or the CDRS in both of the data sets on which it was tested; however, in common with the other risk scores, it performed worse on the South Asian subset in the STAR study sample.

Using the initial risk score to identify the 10% most at risk for invitation to further screening gave a sensitivity of 19.2% with a specificity of 89.3%; the sensitivity is increased to 46% if the top 25% of at-risk participants are screened.

Phase three: updated Leicester Practice Risk Score results

Given the poor reporting of ethnicity in primary care, the initial score used practice-level ethnicity as a proxy for individual-level ethnicity. This may overinflate the score of white Europeans living in areas with a large South Asian population and vice versa. Recording of ethnicity has since been included in the Quality and Outcomes Framework,130 which has significantly improved the level of completeness for individual-level ethnicity, with > 90% of UK practices now recording ethnicity for all newly registered patients. 131 In addition, HbA_1c has been used to diagnose T2DM since 2011. Therefore, it was decided to develop another score that would include individual-level ethnicity and define T2DM using OGTT or HbA_1c to reflect these important changes to clinical practice. Although this score was developed de novo (given the change in outcome and the definition of an important predictor), the same methodology as used to develop the initial score was employed. As this score is now used in practice, a more thorough description of the development and validation is given. The subsequent section outlines the development of the LPRS software for use in primary care for running this updated risk score.

Table 9 shows the model-building process. Of the variables considered for inclusion, prescription of steroids and statins, smoking status, history of CVD and deprivation were excluded from the final model based on their association with PDM/T2DM.

Table 10 shows the final model produced. Age, sex (male vs. female), BMI, ethnicity (‘other’ vs. white European), antihypertensive therapy (yes vs. no) and family history of diabetes (any type, yes vs. no) were all found to be significant predictors of PDM or T2DM both when modelled separately and together. Adding other variables did not improve the area under the ROC curve. There were no statistically significant two-way interactions, assessing significance at the 1% level, because of the high number of comparisons. Polynomial terms were considered for age and BMI but this did not improve the fit of the model. The area under the ROC curve for the final model was 70.1 (95% CI 68.4 to 71.7). Figure 4 shows the observed vs. the estimated prevalence of PDM and T2DM grouped by the predicted probability. This shows overall good agreement between the observed and predicted estimates. This is reflected in the result of the Hosmer–Lemeshow test based on 10 groups (χ² = 2.4, p = 0.97) and a Brier score of 0.15.

FIGURE 4.

Comparison of the observed vs. the estimated prevalence of PDM or T2DM grouped by decile of predicted probability of PDM or T2DM. Adapted with kind permission from Springer Science+Business Media. Diabetologia, Detection of impaired glucose regulation and/or type 2 diabetes mellitus, using primary care electronic data, in a multiethnic UK community setting, vol. 55, 2012, pp. 959–66, Gray LJ, Davies MJ, Hiles S, Taub NA, Webb DR, Srinivasan BT, Khunti K, figure 1. 113

The performance of the score in differentiating between those who had PDM or T2DM diagnosed using either an OGTT or HbA_1c and those who had normal glucose tolerance in the temporal data set is shown in Table 11 and Figure 5. The score can be used in two ways: either by setting the sensitivity to a certain level or by deciding what percentage of the general practice to invite for further testing. If using an OGTT for diagnosis, then 50% of a general practice would need to be invited for testing to detect T2DM with 80% sensitivity, this is raised slightly to 54% being invited if using HbA_1c. To retain 80% sensitivity for the PDM outcomes, the percentage invited would need to be increased to 60% if using an OGTT and 66% for an HbA_1c between 6.0% (42 mmol/mol) and 6.4% (46 mmol/mol). Inviting the top 10% for testing, 9% of these would have T2DM using an OGTT [PPV 8.9% (95% CI 5.8%, 12.8%)] and 26% would have PDM [PPV 25.9% (95% CI 20.9%, 31.4%)]. Using HbA_1c increases the PPV to 19% for T2DM [PPV 18.6% (95% CI 14.2%, 23.7%)] and 28% for an HbA_1c between 6.0% and 6.4% [PPV 28.3% (95% CI 23.1%, 34.0%)]. If screening for both T2DM and PDM using an OGTT, inviting the top 10% for further testing gives a sensitivity of 17%. The high NPV (81.3%) suggests that this cut point is good for ruling out disease.

FIGURE 5.

Receiver operating characteristic curve for T2DM, PDM and T2DM or PDM using the OGTT and HbA_1c. OGTT: (a) area under ROC curve = 70.6%; (b) area under ROC curve = 66.3%; and (c) area under ROC curve = 68.5%. HbA_1c: (d) area under ROC curve = 69.4%; (e) area under ROC curve = 62.2%; and (f) area under ROC curve = 66.7%. Reproduced with kind permission from Springer Science + Business Media: Diabetologia, Detection of impaired glucose regulation and/or type 2 diabetes mellitus, using primary care electronic data, in a multiethnic UK community setting, vol. 55, 2012, pp. 959–66, Gray LJ, Davies MJ, Hiles S, Taub NA, Webb DR, Srinivasan BT, Khunti K, figure 2. 113

Development of the Leicester Practice Risk Score software

It is well reported that many risk scores are not used in practice. One reason for this may be because little thought is given to implementing them at the development stage. 132 To enable widespread use of the LPRS we developed a piece of software which uses existing medical records within primary care to calculate the LPRS for each patient within a practice population aged 40–75 years, having excluded people with known diabetes mellitus, the terminally ill and those with coded gestational diabetes (as they are already identified as being at higher risk and it is not necessary to screen them). When developing the software it came to light that many patients will have been screened; as it is unnecessary to rescreen these people, the software analyses existing OGTT/glucose/HbA_1c data. This process also identifies any people with ‘missed’ diabetes mellitus, that is, people with glucose results in the diabetes mellitus range who have not been coded as diagnosed with diabetes mellitus. The output is presented in a single Microsoft Excel^® version 10 (Microsoft Corporation, Redmond, WA, USA) spreadsheet that can be used to check records and recall patients for screening for diabetes mellitus. This software can be downloaded from http://leicesterdiabetescentre.org.uk/The-Leicester-Diabetes-Risk-Score.

Discussion

We have developed a simple and sensitive automated screening tool for use in multiethnic populations that will enable primary care practitioners to rank individuals by their risk of having undiagnosed PDM or T2DM and therefore allow targeting of screening resources. Ranking people by risk allows flexibility in the screening strategy chosen; practices can choose to hone in to the top of the list and invite fewer people for screening for a bigger ‘hit’ rate or, if resources allow, to widen their inclusion criteria, giving greater sensitivity at the offset of the specificity.

Although some existing scores have been validated against HbA_1c,133,134 the updated score is the first to be developed that incorporates the new WHO diagnostic criteria into the outcome. Previous work has shown that different cohorts are detected using either an OGTT or a HbA_1c to diagnose T2DM. 135 Previously developed scores may now miss people who meet the new diagnostic criteria. This is also the first computer based score developed in a multiethnic population within the UK to identify prevalent disease. The Cambridge risk score was designed to identify undiagnosed diabetes only and does not adjust for ethnicity. 121 Although not taken into account in the original score, a post hoc study using data from both Caribbean and South Asian populations showed that using alternative ethnic specific cut points could give acceptable levels of prediction for undiagnosed hyperglycaemia in these groups, but that further work needed to be carried out to refine these. 136 The multivariable risk score to predict the 10-year risk of acquiring T2DM (QDScore) predicts the 10-year risk of developing diabetes mellitus and includes similar variables to both the CDRS and scores developed here, but with the addition of deprivation and CVD (both of these were found not to improve the fit of models produced). 137 Compared with the Cambridge Risk score, the QDScore showed greater levels of discrimination, but only detects incident disease. In addition, the algorithm to compute the risk score has not been published and cannot be used to detect PDM. Other scores, including the Leicester Self Assessment score and the FINDRISC score, have been developed, which rely on the person at risk completing a questionnaire themselves and attending the GP practice. 27,138 The score developed here may increase the uptake to screening invitation by removing the need for people to calculate their own risk.

Although the score was developed using high fidelity data from a randomly selected population who all received an OGTT, there are a number of limitations to be taken into account when applying the score. First, the cross-sectional nature of the data limits the score to detecting prevalent undiagnosed disease. The score cannot, therefore, be used to estimate the risk of future disease, although detecting PDM will identify a high-risk group that is likely to develop T2DM in the future. Although this could be viewed as a limitation, screening strategies may want to focus on those who have current undetected disease as a priority. In addition, those scores predicting incident disease may give biased estimates, as those variables that are included in the score are also those that prompt testing. Future work will look at validating the score on a prospective data set. Second, only 22% of those invited for screening in the ADDITION-Leicester study attended. Although this is similar to other studies in similar populations139 and reflects the difficulty in recruiting a multiethnic urban population with wide variations in socioeconomic status into research studies, this may have affected the representativeness of the data that the score has been derived from. For example, those screened were slightly older than those invited. 140 It is difficult to predict the possible implications of the response rate to the initial study on the score produced. Reassuringly, the score contains a similar set of variables to other comparable risk scores. 27,121 Future work will further validate the score on other population-based data sets. There are some limitations with the analysis performed to derive the risk scores. We used a variable selection procedure based on the statistical significance of their association with the outcome to initially select variables for potential inclusion into the models produced. This is not the recommended approach to variable selection and can increase the risk of excluding important variables that become important only after adjustment for other variables. 141 We did, however, reassess each variable excluded in this manner to see if their inclusion in the final model would improve the discrimination of the model. Although the approach taken may have been suboptimal, the variables included in the model are similar to those included in previous scores and the score was shown to work adequately when tested in a temporal data set. There is a hierarchy of data sets to be used for evaluating a risk scores validity: (1) internal; (2) temporal (using an external data set from the same centre); and (3) external (using a truly external data set from a different centre). 142 Here we have used a temporal data set; although these data incorporate an independent population, the data set was run to the same standard operating procedures as the ADDITION study and within the same centre. Future work should focus on assessing the updated risk score in an external data set. The method of dealing with missing covariate data could also have been improved. Here risk scores were developed on complete-case data; in hindsight, a better way to deal with missing data would be to use multiple imputation. 143 Studies have shown that risk scores produced using complete-case data may be biased and could produce scores that perform poorly when used in clinical practice. 125 Future work should consider the robustness of the scores produced to missing data. Finally, the score was developed using data from Leicester (UK). The ethnic makeup of this area means that the ethnicity component of the score is based on data from South Asian participants (mostly of Indian descent). Although there were participants included from other ethnicities in both data sets (such as Chinese, Caribbean and African), there were insufficient data to model separate scores for each ethnicity. South Asians are known to have a high level of risk,144 and, therefore, assuming the same level of risk for all BME groups, may overestimate risk for some, but this was thought to be preferential to underestimating risk or estimating risk based on insufficient data.

Summary

In summary, we have developed a valid and sensitive score for identifying those at the highest risk of prevalent PDM or T2DM within a multiethnic UK population. Using an automated tool is simple to implement and can be used to target screening approaches in a cost-effective manner. For example, in the UK, this tool could be used to complement the NHS Health Check programme as the score has been developed using data that are reflective of the inclusion criteria of the health checks. The results from the screening study using the initial version of the risk score, which incorporated practice-level ethnicity data, are reported in Chapter 6.

Drafting a curriculum

Research shows that lifestyle modification in people with PDM can significantly reduce the risk of developing T2DM. Both the DPP and the DPS demonstrated that, in participants with PDM, a programme addressing weight loss, diet and physical activity was able to reduce progression to T2DM by 58%. 32,33 Although these results were certainly very encouraging, their lack of transferability from research into practice was a valid concern. Many of the programmes called for resource-intensive interventions, which would simply be unachievable in our current health-care system. For example, the DPP saw participants have, on average, 20 individual counselling sessions over the course of 4 years. 33 Many countries would, therefore, be unable to replicate the intervention owing to such a huge demand on resources. It was noted, therefore, that there was an urgent need to develop a prevention programme that was effective, economically viable and could easily be implemented into a real-world UK health-care setting.

Previous pilot work prior to National Institute for Health Research grant

Early development work to establish a prevention programme that focused on key lifestyle targets and that was suitable for those with PDM commenced in 2006 prior to the successful National Institute for Health Research (NIHR) grant award.

Key findings from a qualitative study in subjects with PDM22 showed that many had no prior understanding of PDM and their risk of CVD, felt confused by the diagnosis and wanted clarification of what was meant by being at ‘high risk’. All wanted to know why they had developed PDM. Many recognised that lifestyle changes could delay progression to T2DM but were not confident that these were achievable. Many felt that they did not know what to change in their lifestyle. Although many received an education booklet at the time of diagnosis, they felt that the advice was too general.

Prior to the successful NIHR grant application, the full results of the DESMOND RCT were being finalised. Initial pilot work had already demonstrated significant improvements in outcomes (e.g. HbA_1c, lipids, BP), changes in illness beliefs and increase in physical activity,147 thereby suggesting that any prevention programme developed for PDM could benefit from using a similar approach.

Early development work tested the format and content of a 3-hour prevention session developed by our group. Thirty-four people with PDM (including 10 from a South Asian population) attended four groups. Feedback showed that all felt that the programme helped them to improve their understanding of PDM, including the role of diet and exercise. All participants agreed that they intended to become more physically active and eat a healthier diet. Practical sessions were valued, and personalising risk was deemed helpful.

Exploratory quantitative evaluation showed significantly increased self-efficacy in making lifestyle changes, perceived control and perceived knowledge, as well as a significant decrease in perceived symptom load and consequences. Ten subjects provided pedometer information and reported an increase in both daily step count (by 1690 steps per day) and energy expenditure [2635 metabolic equivalent of task (MET) minutes per week]. These changes showed that the initial programme was setting the necessary preconditions for successful behaviour regulation and change. Qualitative interviews carried out by telephone revealed that many would attend extra sessions on food choices and activity and, therefore, a 6-hour session model was considered for our current format.

Following on from this exploratory work came the Pre-diabetes Risk Education and Physical Activity Recommendation and Encouragement (PREPARE) programme,148 which was a RCT funded by Diabetes UK. A total of 103 individuals were recruited, of whom 30% were from a South Asian ethnic background. It was found that when participants were given structured education that enabled them to set realistic personalised step-per-day goals and to self-monitor their physical activity behaviour with a pedometer, their physical activity levels increased significantly (p < 0.05), by 2600 steps per day (equivalent to around 26 minutes per day of walking activity) and their 2-hour glucose levels decreased by 1.4 mmol/l. A decrease in 2-hour glucose of this magnitude has been associated with around a 60% reduction in the relative risk of developing T2DM. Furthermore, significant improvements in both fasting and 2-hour glucose levels, and increases in physical activity, remained even at 12 months’ follow up. 50 This study therefore suggests that structured education that targets participants’ perceptions of PDM, highlights the importance of physical activity and promotes self-management skills can substantially reduce the risk of developing T2DM. This study provided a foundation on which to build a structured education group intervention to initiate behaviour change and improve glucose tolerance in those with PDM in a UK setting.

Pilot work supported by National Institute for Health Research Grant 1272

The Let’s Prevent programme was developed by a core multidisciplinary team, in collaboration with the DESMOND collaborative and the National Physical Activity Centre in Loughborough (Collaboration in the PREPARE study). 50

In July 2007, a working party gathered to develop the Let’s Prevent intervention. Using the findings from previous pilot work, the intervention was compiled in line with best practice. The programme was based on the fundamentals of the DESMOND programme. DESMOND is currently the only education programme that fully meets the NICE guidelines for structured education in diabetes149 and its RCT demonstrated its clinical effectiveness in improving weight, smoking behaviours and illness beliefs. 48 The DESMOND model is based on an empowerment philosophy that sees the participant as capable and responsible for his or her own health decisions and behaviours. 150 The DESMOND model is also underpinned by adult learning theory and psychological models of learning. These include Leventhal’s common sense model,151 the dual process theory152 and social learning theory. 153

The Let’s Prevent intervention was designed as a group educational programme with a written curriculum suitable for the broadest range of participants, to be deliverable in a community setting for ease of access for patients and to have the potential to be integrated into routine care in the future. As per the DESMOND model, the programme was approximately 6 hours in duration, deliverable in either one full day or as two half-day equivalents. It was designed to be facilitated by two trained health-care professionals (educators) to a group of 5–12 participants with PDM, who had the option of bringing an accompanying person.

The session content was developed by a core multidisciplinary team, founded on a sound evidence base and guided by the following:

• a review of the literature surrounding nutrition, exercise and educational principles

• pragmatic decision-making on best practice where the evidence base was ambiguous, lacking or conflicting

• the philosophy of DESMOND, which calls for non-directive methods of education to be used. It recognises that individuals with PDM have insight and expertise in relation to their own food choices.

The DPP and DPS informed the key messages regarding food. The key food goals within the curriculum were:

• sustained weight reduction of > 5% of body weight

• moderate reduction in total fat intake to < 30% of energy intake

• lower saturated fat intake to < 10% of energy intake

• higher fibre intake of > 15 g per 1000 kcal.

The key physical activity messages were taken from the PREPARE programme. 148 The PREPARE study demonstrated that pedometer use was an effective means of improving glucose tolerance in those with PDM when incorporated as part of an educational programme. 50,154 As in the PREPARE study, Let’s Prevent participants were given a pedometer to encourage regular goal-setting and to facilitate monitoring of their physical activity levels. The physical activity target was for participants to increase their daily step count by 4500 steps, which is equivalent to approximately 45 minutes of walking. It was suggested that this goal be broken down into smaller, more achievable, goals, such as increasing steps by 500 a day every fortnight, in order to provide sufficient time for participants to adjust to their new level of activity.

In addition, participants were given a handbook containing a summary of the key messages and a plethora of other useful resources to help further integrate their knowledge. There was a health profile sheet, which displays a continuum of risk for the following risk factors: fasting blood glucose, 2-hour blood glucose, BP, total cholesterol, HDL cholesterol, BMI, waist circumference and pedometer counts (steps). Glucose targets were adapted from the WHO diagnostic criteria for diabetes;155 BP, cholesterol, BMI and waist circumference were taken from the Department of Health document on cardiovascular risk assessment;156 and step count targets were adapted from those proposed by Tudor-Locke and Bassett. 157 Participants plot their biomedical and anthropometric data along continuums to build up a personalised picture of their risk profiles. They subsequently complete an action plan targeting their selected area of behaviour change to help them to achieve their goals. Tape measures were also provided to encourage regular monitoring.

The broad curriculum content for Let’s Prevent is detailed in Tables 12 and 13.

The Medical Research Council (MRC) Framework for Complex Interventions to Improve Health has internationally recognised criteria to inform the development and evaluation of any lifestyle intervention programmes. 158 The MRC guidance recommends that all interventions undertake a process of development, feasibility and piloting, evaluation and implementation. In order to develop the Let’s Prevent programme, the methodology previously used in the development of the DESMOND programme was used. 159,160 This methodology encompasses a pragmatic iterative approach, similar to an ‘audit cycle’, whereby data are collected and analysed to obtain feedback and suggestions from key stakeholders, including patient and public involvement, health-care providers and facilitators (trainers, educators and interpreters). Each cycle involves identifying potential modifications that could be made; piloting these changes and collecting both qualitative and quantitative data, analysing the data and reflecting on; their implications; and making refinements and adjustments. The cycle is repeated until the pilot stage suggests that the intervention is fit for purpose (Figure 6).

FIGURE 6.

The cycle of programme development.

First phase of the pilot of the Let’s Prevent programme (October–November 2007)

Second phase of the pilot of the Let’s Prevent programme (March–April 2008)

The main purpose of the repilot was to check the changes to the curriculum, resources and retraining of educators. The quantitative and qualitative methodologies used were those adopted in the first pilot. Changes were made to the topic guides used for collecting qualitative data to reflect our primary interest in evaluating the impact and effectiveness of the modifications made in response to the first pilot.

Drafting a curriculum for black and minority ethnic participants

Although T2DM is a significant health problem for the general population, there is a higher risk of T2DM and premature heart disease for individuals from South Asian backgrounds. 170 Therefore, it was imperative to develop a culturally informed and appropriate intervention for this at-risk population, whose dietary practices and frequently reported low levels of physical activity contribute to the onset of T2DM. 144 At the time of developing the Let’s Prevent programme, studies investigating the prevention of diabetes mellitus in migrant populations in the UK were lacking. Given that NICE guidelines state that those identified as having a moderate or high risk of developing T2DM should be offered culturally appropriate information or support in a range of formats and languages, including structured education, in order to assist them to make lifestyle changes,13 it was considered necessary to design and tailor the Let’s Prevent programme to meet the needs of ethnically diverse communities.

Previous qualitative research159,160 helped to modify the DESMOND module for people newly diagnosed with diabetes mellitus from BME groups living in the UK. Although this work did include some work with African and African Caribbean populations, the main focus of the research centred on providing education that met the language, literacy and cultural needs of the South Asian population with T2DM. It appeared that the previous methodology used was helpful in the process of modifying the programme to be relevant to the South Asian community, and so owing to its previous success, the same methodology was employed to inform the development of a Let’s Prevent (BME) curriculum. The Let’s Prevent (BME) programme was developed to address the specific needs of the South Asian population living in the UK, many of whom do not speak English. The key educational messages remained the same as in the standard Let’s Prevent programme; however, minor changes were made to ensure that the food and activity messages were relevant to the South Asian culture. The main change was the identification and development of additional teaching resources (images and models) that reduced reliance on written formats, and that could potentially facilitate delivery to those for whom English is not a first language. Educational resources were developed specifically to take into account the language, literacy and cultural needs of people from a South Asian background with PDM. Many of the images and other visual resources were derived from those already developed for the DESMOND newly diagnosed module, with some additions, such as an image for insulin resistance and supplementary materials to address the increased emphasis on physical activity. As with the generic programme, participants were enabled to assess their personal risk by filling in a personal health profile. The targets were adapted to be appropriate for the South Asian population.

Based on the DESMOND model for BME groups, the programme was designed to be delivered as four sessions of 3 hours, facilitated by two trained health-care professionals and two trained interpreters. Previous work has demonstrated that using interpreters adds considerably to the delivery time and effectively doubles the length of time required per session. 171 The outline curriculum for the BME programme is shown in Table 16.

Piloting and refining of the Let’s Prevent (black and minority ethnic) programme

The methodology used to develop and refine the BME programme mirrored that used for the generic programme, as described earlier in this report. This involved an iterative process including the use of qualitative and quantitative data collection. Owing to the small sample size, we recognised that there would be limitations in undertaking statistical analysis, which would not be sufficiently powered to accurately detect differences. Nevertheless, it was felt that undertaking such an analysis would give an indication of the direction of results, and statistical analyses were performed for this reason. Topic guides were adapted to reflect any additional areas of interest relevant to the target population (see Appendices 8–10) and interpreters were included in the stakeholder groups from whom qualitative data were collected.

Second phase of pilot of the Let’s Prevent (black and minority ethnic) programme (March–April 2008)

As with the repiloting of the standard (general population) programme, the main purpose of the BME repilot was to check the changes to the curriculum, resources and retraining of educators. The methodology used was the same as for the first pilot, with topic guide modifications to reflect our primary interest in evaluating the changes made after the first cycle of piloting (Appendices 8–10).

Development of a quality-development programme

Quality development is a requirement of the NICE criteria for educational interventions. 149 The DESMOND Training and Quality Development programme was an initiative in health-care professional training, which was awarded the Health Service Journal Award 2007. The Training and Quality Development programme was originally developed by a task group of the DESMOND collaborative (a multidisciplinary, multiorganisational group of health-care professionals, academics and people with diabetes mellitus) to support the clinical trial of DESMOND. Through involving educators, trainers and patients in the development process, the programme created training, mentorship and continued professional development that was uniquely integrated with the patient education programme being delivered. This method of quality development has been piloted and now used successfully for the DESMOND newly diagnosed module. Therefore, this training and quality-assurance model was deemed to be transferable to the Let’s Prevent programme.

A quality-development programme was developed for the Let’s Prevent programme to ensure that the educational intervention was delivered in a standardised way, to ensure that core content and learning outcomes were achieved and to ensure that educator behaviour was closely allied to the programme’s philosophies and learning theories (Figure 7).

FIGURE 7.

Quality-development pathway. DOS, DESMOND observation sheet; DOT, DESMOND observer tool; QD, quality development.

The quality-development programme consists of both internal and external processes. The internal processes encourage educators to reflect on their practice using self-reflection sheets and peer-reflection sheets (see Appendix 11), providing educators with a feedback tool for personal development. The external component of the quality-development programme is an adapted DESMOND observation sheet, used to check content and process indicators of observable educator behaviours, and an adapted DESMOND observer tool, used to assess the interaction of the educator and the group (see Appendix 12). These tools can be used by educators to identify areas for personal development and also by external trained assessors to provide quantitative data. These data provide a mechanism by which to accredit educators and inform the ongoing development of the training programmes.

Development of an educator training programme

Despite the existence of many self-management education programmes in the UK for those living with long-term conditions, current evidence suggests that these programmes have variable outcomes. One possible reason, highlighted by NICE in 2003, is the variable nature of the delivery of the programmes by health-care professionals, especially given that none has clearly described the education in terms of skills, attitudes and behaviours demonstrated by the educators, without which monitoring for internal consistency cannot take place. 149 The training programme for educators was developed by a core multidisciplinary team, in partnership with the DESMOND collaborative. The DESMOND collaborative were approached as they have proven to have national success at developing training programmes for educators.

The ‘Train the Educator’ programme for the Let’s Prevent initiative aimed to familiarise educators with the content and resources of the Let’s Prevent curriculum. Adequate training enables educators to become confident and competent to deliver the programme in accordance with its learning theories and philosophy. The training programme was delivered by trainers in a style that reflects the learning theories and philosophy underpinning the Let’s Prevent programme.

The ‘Train the Educator’ programme was designed to allow some flexibility in the way theory and practical sessions are delivered by the trainers. The dual-processing approach ensured that knowledge and skills were elicited rather than taught didactically, and this was further developed through experiential exercises, reflection and debate. A core element of training was to model sessions to trainees who take on the role of participants. By enacting a participant’s perspective, they were simultaneously experiencing the programme and learning from the observation of specific behaviours portrayed by the trainers. This process was interspersed with opportunities for trainees to practice delivering the curriculum in a safe environment, and it also allowed time for reflection, which enabled participants to review their own skills. The process was supported with a ‘Train the Educator’ manual, a curriculum and other resources provided for trainers to retain, read and reinforce their learning.

Summary

Overall, in terms of the Let’s Prevent intervention, both the qualitative and quantitative data revealed a positive change in health behaviours (mainly increased walking and physical activity levels) and beliefs and knowledge related to the prevention of T2DM. The BME adaptations were well received, with individuals seeming to gain value from having culturally appropriate resources. The recruitment strategy also proved feasible. Regular feedback was used throughout with regard to the ‘Train the Trainer’ programme and the Let’s Prevent intervention itself. This allowed for maximum input from a variety of stakeholders. In accordance with the MRC recommendations, the development team took into consideration the suggested improvements and made the necessary amendments. Finally, the intervention was considered fit for purpose and ready to be tested in a RCT.

Aims and objectives

The aim of the Let’s Prevent study was to ascertain if a pragmatic, structured education programme focused on lifestyle modification and behaviour change, in combination with regular telephone support, was able to prevent the development of T2DM in people identified as having PDM through a two-stage screening process within a primary care setting, and to assess if such an intervention was cost-effective.

The objectives of the study were to:

• provide a robust identification strategy for those at high risk of diabetes mellitus

• provide an appropriately structured education programme to target lifestyle modification and behaviour change

• provide a culturally sensitive structured education programme for the BME community

• provide a support and motivational maintenance programme for the duration of the study

• evaluate the programme to provide a low-cost solution for practical health improvement.

The study had two phases. The first phase was a screening phase which identified people at risk of PDM/T2DM through the use of a screening tool that had been developed and validated for use within primary care. For further details of the development of the screening tool, see Chapter 3. The second phase took participants who had been screened and found to have PDM and recruited them into a cluster RCT, designed to test the effectiveness of a structured education programme in preventing progression to T2DM. This chapter will outline the methodology of the two phases.

Phase one: screening for prediabetes mellitus

Inclusion/exclusion criteria

Participants were invited for screening if they fulfilled the following criteria:

• high risk according to the LPRS

• aged 40–75 years if white European, or 25–75 years if South Asian.

Participants were excluded from the screening study if they were/had:

• unable to give informed consent

• pregnant or lactating

• established diabetes

• a terminal illness

• required an interpreter for a language other than Gujarati.

The differential included age ranges for white European and South Asian participants were chosen to reflect the fact that South Asians, as well as being at higher risk than their white European counterparts, also develop the disease at an earlier age. The Southall and Brent Revisited study showed that people from BME communities were diagnosed, on average, 5 years earlier than white Europeans. 172 Therefore, any prevention initiative should begin at an earlier age for these groups; this also reflects NICE guidance for screening those at risk of T2DM. 126

Oral glucose tolerance test

An OGTT was performed at baseline. Participants were sent instructions for the OGTT along with their appointment date and time, which requested that they fast from midnight the night before their OGTT (see Appendix 19). A baseline set of bloods was taken before the participant was given a glucose load of 75 g in the form of Lucozade (Lucozade Ribena Suntory Ltd, Uxbridge, UK) (410 ml for 70 kcal/100 ml Lucozade). Timing of the 2-hour interval was taken from the start of the Lucozade drink. A post-load glucose sample was taken at 120 minutes. All participants who were identified as having PDM took part in the RCT (see Appendices 21–24). The OGTT was repeated at 12, 24 and 36 months to check the participant’s diabetes mellitus status. Participants who did not consent at baseline to an OGTT were not entered into the study on the grounds that their diabetes mellitus status could not be assessed. These individuals were referred back to their GP for standard care.

If a participant had an OGTT result in the range for diabetes mellitus at screening they were recalled as soon as possible for a second, confirmatory test (see Appendix 25). In accordance with WHO criteria, diabetes mellitus was defined as a fasting blood glucose ≥ 7.0 mmol/l and/or 2-hour plasma glucose of ≥ 11.1 mmol/l. If a participant was found to have diabetes mellitus at screening, they were excluded from the study and were referred back to their GP for commencement of standard diabetes mellitus care (see Appendices 26 and 27).

Venous blood tests

Blood tests were collected at baseline and repeated at 12, 24 and 36 months (Table 19).

This amounted to a total of 49 ml of blood for the baseline, 12-, 24- and 36-month blood tests, and 8 ml for the 6-month test.

In-study assessments

In-study assessments were carried out alongside the blood tests at baseline and at 6, 12, 24 and 36 months post randomisation. Details of all clinical assessment and timelines are summarised in Table 20. (For examples of the case report form and patient questionnaire, see Appendices 28 and 29. )

Body mass index was calculated after body weight and height were measured. Weight was measured in light clothing without shoes. Height was measured at baseline and 36 months only. Waist circumference was measured with a soft tape on standing participants, mid-way between the lowest rib and iliac crest. Hip circumference was measured over the widest part of the gluteal region and the waist-to-hip ratio calculated. Three BP recordings were obtained from the arm of the participant in a sitting position after 5 minutes of rest, with at least 2 minutes between readings, and the mean value of the second and third readings was calculated, discounting the first. All clinical measurements were carried out as detailed by Diabetes Research Network standard operating procedures. Seven-day step count was assessed by giving all study participants a sealed piezoelectric pedometer (NL-800). Participants were asked to wear the pedometer fitted to their trunks (placed on right anterior axillary line) for 7 consecutive days during waking hours (see Appendix 30). Participants were provided with a stamped addressed envelope to return the pedometers to the study co-ordinators.

Study questionnaires

A health and dietary questionnaire was completed by the participant with assistance from a trained nurse or research assistant at baseline and at 6, 12, 24 and 36 months (see Appendix 29 for the 36-month questionnaire). The questionnaires covered several areas, as summarised in Table 21.

Participants also self-reported on two questions concerning sleeping pattern (‘How many hours’ sleep did you get last night?’ and ‘On average, how many hours do you sleep in 24 hours?’). This was assessed at baseline, 12, 24 and 36 months. These data were included, as there is accumulating evidence for an association between short sleep time (< 6 hours per 24 hours) and long sleep time (≥ 10 hours per 24 hours) and metabolic dysfunction (i.e. T2DM). 176 Furthermore, lifestyle intervention trials have reported improved sleep hygiene and reduced risk of developing T2DM. 177

Outcome measures

The primary outcome of the screening phase was the proportion of people detected with PDM or T2DM using the LPRS (PPV). Secondary outcomes included the response rate to the invitation to screening. Those with PDM took part in phase two, the diabetes mellitus prevention cluster RCT.

Phase two: diabetes mellitus prevention cluster randomised controlled trial study design

Phase two was a cluster RCT providing a structured intervention for people with PDM. Randomisation was at practice level to negate contamination between individual participants. The practices were randomly assigned 1 : 1 to either the standard core or intervention arm by a researcher who was independent of the study team, using stratification by list size (< 6000, ≥ 6000) and ethnicity (percentage South Asian < 21%, ≥ 21%; median level of per cent of South Asian in the ADDITION-Leicester study123). Although randomisation occurred prior to recruitment, practices and participants were informed of their allocation in the results letters after the screening/baseline measurements were complete to avoid selection bias. Phase two was designed to adhere to internationally recognised criteria for developing complex interventions and for undertaking and reporting cluster RCTs. 158

Participants within the standard core practices were managed by nationally regarded ‘standard care’ guidelines for the condition, which consisted of participants being given an information booklet and general lifestyle advice by their GP or practice nurse. The booklet gave information on risk factors for T2DM and indicated how dietary and lifestyle changes and increased physical activity can be used to prevent progression of the disease. The booklet gave information in accordance with Leventhal’s common sense model,151 to address the causes, consequences, identity, control/treatment and timeline for participants with PDM.

Participants in the intervention practices were given the same written information as the control group and were also invited to attend the ‘Let’s Prevent’ programme, which was a 6-hour structured group education session. In addition, they received a telephone call every 3 months from nursing staff who were trained to offer ongoing support in behaviour change and to encourage participants to achieve their individual goals. Finally, each participant within the intervention arm was invited to attend a 3-hour refresher session once per year. This session was designed to reinforce the key messages from the programme, and allowed participants to re-examine their personalised risk profiles and set new goals/update their action plans if necessary.

The intervention

The structured group education programme was named Let’s Prevent and was a modified version of the DESMOND programme. It was based on published work performed locally for people with PDM. Let’s Prevent educational sessions consisted of one full-day session (6 hours) or two half-day sessions (3 hours each). In the case of the BME groups in which English was not readily spoken, the sessions were delivered as four 3-hour sessions by educators and interpreters (who had also undergone training). The programme followed a detailed written curriculum (see Chapter 4).

Outcome measures

The primary outcome was progression to T2DM at 3 years in people with screen-detected PDM.

The main secondary outcomes included:

• changes in participant’s glucose levels: HbA_1c, blood glucose levels fasting and post-glucose load

• change in cardiovascular risk as calculated by the Framingham risk calculator

• 7-day step count

• presence of metabolic syndrome as defined by the National Cholesterol Education Program Adult Treatment Panel III

• cost-effectiveness of the intervention (for a full description see Chapter 8)

Inclusion/exclusion criteria

Participants were included in the RCT if they were:

• diagnosed with PDM

• aged 40–75 years if of white European ethnicity, or aged 25–75 years if of South Asian ethnicity

• able to attend group education sessions.

Participants were excluded from the study if they:

• were unable to give consent

• were unable to attend group education sessions

• were diagnosed with diabetes mellitus at screening

• required an interpreter for a language other than Gujarati.

Sample size

It was calculated that 748 participants would need to be recruited to participate in the study: 374 in the intensive arm and 374 in the control arm, giving 280 per group after allowing for dropout. A substantial proportion, estimated at ≈30%, was predicted to be from BME groups. Recruitment took place from July 2009 to June 2011.

Assuming a 3-year cumulative conversion rate to T2DM of 35% in the control group, an intraclass correlation of 0.05, an average of 17 participants per practice and a dropout rate of 20%, we calculated that we would need 374 participants per group to detect a 40% risk reduction in the intervention group, that is, data from 44 practices, with 80% power at the 5% significance level. One secondary outcome was the percentage of participants in each group with a 10-year CVD risk > 20% at the end of 3 years. It was estimated that 55% of participants would have a CV risk > 20%. To detect a difference between the two groups of 20% of points in the proportion of participants with a 10-year risk of > 20% with 80% power and two alpha of 5%, and an intrapractice correlation coefficient of 0.0551, the required sample size was calculated to be 180 in the two groups.

Clinical assessments and measures

The clinical assessments and measurements that were taken at baseline during phase one of the trial were repeated at various time points during the RCT (Table 20 and Figure 8). Follow-up assessments occurred at 6, 12, 24 and 36 months post-baseline visit.

FIGURE 8.

Let’s Prevent study flow chart. Adapted from Gray et al. ,146 figure 1, and reproduced courtesy of Biomed Central under Creative Commons public licence 4.0, http://creativecommons.org/licenses/by/4.0/.

Participation in the study consisted of:

• baseline screening visit with questionnaires, urine sample, OGTT, blood samples and clinical assessments (approximately 3 hours)

• 6-month review with questionnaires, blood sampling, body weight and BP measurements (approximately 1 hour)

• first year review with questionnaires, urine sample, OGTT and blood sampling (approximately 3 hours)

• second year review with questionnaires, urine sample, OGTT and blood sampling (approximately 3 hours)

• third year review with questionnaires, urine sample, OGTT and blood sampling (approximately 3 hours).

Participants in the intervention arm additionally received the following:

• a structured education session within 1 month of the initial visit (1 × 6-hour session; 2 × 3-hour generic course; 4 × 3-hour session for a BME group with interpreters)

• an additional group refresher programme (3 hours) each year for those in the intervention arm, to consolidate previous learning and address any issues that had arisen

• a telephone call every 3 months from a nurse trained in motivational techniques, to offer support and encouragement to participants in the intervention arm

• continuous support network, available via telephone and e-mail.

Management of those found to have type 2 diabetes mellitus during the randomised controlled trial

Diagnosis of diabetes mellitus was made in accordance with WHO criteria/guidelines. 8 In all subjects without active symptoms of diabetes mellitus in whom the initial OGTT showed diabetes mellitus, participants were recalled for a second OGTT to confirm diagnosis. Furthermore, after a protocol amendment in January 2013, those who had HbA_1c of ≥ 6.5%, regardless of their fasting and 2-hour blood glucose values, were also called back for a confirmatory HbA_1c test. Therefore, T2DM was diagnosed using only an OGTT prior to January 2013, and with either an OGTT or a HbA_1c test post January 2013 to follow the updated WHO diagnostic criteria for T2DM. 10 Progression to T2DM was the primary outcome, and participants who were identified at follow-up as having T2DM were retained in the trial. Both participants and their GPs were informed of the results of their tests. Participants remained in the study to complete the questionnaires and other biomedical data; however, they did not undertake any further OGTT. All study results were sent to the subjects’ GPs in the form of standard letters, and participants were referred back to their GPs for commencement of diabetes mellitus care. If a participant was diagnosed with T2DM by their GP between follow-up appointments, and they informed the study team, this was confirmed via blood tests taken when participants next attended for follow up. In these circumstances, participants were again excluded from any further OGTT, but their questionnaire and other biomedical data were still collected.

Study personnel and sites

The study was coordinated by the research team based at University Hospitals of Leicester (Leicester Royal Infirmary and Leicester General Hospital). Education sessions and clinical assessments took place at suitable locations within the community to ensure minimum disruption and travel for participants. General practices were approached individually and invited to take part by the primary care research network.

Data storage

Data were recorded directly on to individual participant data collection sheets. These source documents were located in an on-site locked filing cabinet for secure storage along with the questionnaires filled in by the participants. Laboratory reports that have checked and signed off by a nurse and a physician have been retained within the participant’s pack, and were entered onto an access-controlled database on daily basis. Self-evaluation data taken from questionnaires and structured contact report data were entered in batches during the course of the study. Monitoring of the data collection occurred at regular intervals with all data collection activities made in accordance with research governance and International Conference on Harmonisation Good Clinical Practice guidelines.

Data analysis

The study was reported in accordance with the internationally recognised Consolidated Standards of Reporting Trials199 statement for the reporting of RCTs. A statistical analysis plan was written and agreed by the principal investigator before the analysis began. Practice- and participant-level characteristics were compared by intervention group, using either means (SD) or medians (interquartile range) for continuous variables and counts and percentages for nominal variables. Cluster randomisation gives balance with respect to cluster-level covariates but can lead to imbalance in participant-level covariates, and, therefore, differences between the intervention groups were assessed using t-tests and chi-squared tests.

The primary outcome, progression to T2DM, was analysed on an ITT basis, as data were available for all participants. The event rate per 1000 person years was calculated by intervention group. Cox proportional hazards models with intervention group as a covariate were fitted, practices were assumed to have the same frailty (i.e. a shared frailty model was fitted). Hazard ratios (HRs) along with their 95% CIs were presented. The analysis was repeated excluding those from the intervention group who did not attend the education sessions (per-protocol). We had, in addition, assessed the dose–response relationship between level of attendance and the primary outcome; this is an exploratory unplanned analyses which should be viewed as hypothesis-generating.

For all secondary outcomes, those who developed T2DM during the study had their last value from before their diagnosis carried forward for the remainder of the study. This method was used in the previously published PREPARE study. All secondary outcomes were analysed using generalised estimating equation models with an exchangeable correlation structure, which adjusted for clustering. For binary outcomes we used a logit link with a binomial distribution for the outcome, and for continuous outcomes we used an identity link with a normal distribution. The analysis was conducted at each time point. The missing outcomes were not replaced and we derived an average of continuous outcomes over time. This procedure measures the cumulative effect of the intervention and has the maximum number of participants.

A sensitivity analysis was carried out for the main secondary outcomes. The analysis was repeated: (1) excluding those from the intervention group who did not attend the education sessions (per protocol); and (2) imputing any missing values using multiple imputation (ITT).

Subgroup analyses were performed for the primary outcome by PDM status (IGT, IFG, IGT and IFG, HbA_1c 6.0–6.4%); age group (≤ 65 years, > 65 years); BMI (normal, overweight/obese); and risk score (≤ 6.04, > 6.04).

Adjustments were not made for multiple testing. All the results from planned analyses are reported and small p-values are interpreted taking into account the overall pattern of the results. Statistical significance was set at 5%. All analysis was conducted using Stata version 13.

Cost-effectiveness

A within-trial cost-effectiveness analysis was also conducted on the data to evaluate the cost-effectiveness of the Let’s Prevent intervention compared with the standard care arm (see Chapter 8). Cost-effectiveness was measured by the difference in costs divided by the difference in effects. Data were collected from the participants via a case record form, and an economic questionnaire was introduced through the study and given, at least, at patients’ 36-month clinic appointment. Cost-effectiveness was determined by: (1) incidence of T2DM at 36 months, in line with the primary outcome; and (2) QALYs gained. Quality of life was assessed using the health-state descriptive system (15D) questionnaire at baseline and all follow-up points, and the European Quality of Life-5 Dimensions (EQ-5D) instrument was also employed as part of the economic questionnaire. Base-case analysis for QALYs used the EQ-5D results and mapped 15D to EQ-5D scores for the periods without EQ-5D data. Costs included all costs associated with delivering the Let’s Prevent intervention, and all medications and health-care contacts. Missing data in costs and effects were addressed using multiple imputation techniques.

Ethics issues

Main research ethics committee approval and University Hospitals of Leicester Trust R&D approval were sought before commencement of the study. This ensured that all ethics and indemnity issues were adequately dealt with. Primary care approvals were obtained for Leicester City Primary Care Trust (PCT) and Leicestershire County and Rutland PCT. The study was adopted by the South East Midlands Diabetes Research Network, and also the East Midlands and South Yorkshire Primary Care Research Network.

An internal steering committee was established to oversee all activities required to determine safe and effective conduct of the study, and to recommend conclusion of the trial when significant benefits or risks developed, or in the case of the trial needing to be concluded early. The committee met on a regular basis to review data and discuss; however, no issues were identified.

Participants were sent a participant information sheet (see Appendix 18) to read before they registered their interest in taking part in the study by returning a response slip from the introduction letter or telephoning the dedicated study telephone number. This allowed adequate time for the participant to consider participation and actively seek involvement. An appointment was made for the participant to discuss the study and ask any questions that they may have had. A glucose tolerance test and other study assessments were then completed after informed consent had been obtained. All participants signed a consent form (see Appendix 20) to take part in the study. The consenting procedure was undertaken by research nurses and/or a suitably trained researcher before performing any study procedures. Training and assessment of non-medical health-care professionals was provided by the trust’s R&D office.

Data were collected on to case report forms (see Appendices 28 and 29). All data were entered into a restricted-access structured query language server database, held on the PCTs’ servers. This database was protected and access was given to a minimal number of users to deal with participant confidentiality issues. All study records were kept in a locked filing cabinet or room with limited access.

Additional consent was obtained for the taking and storage of samples for future genetic investigation, as well as for follow-up qualitative interviews. These specific requirements were not a condition of entry into the study, and participants could still be involved in the study if they did not wish to partake in the genetic or qualitative components.

Participants who had provided consent to allow the research team to follow up their health status after the study had finished will have their health status checked using sources that include general practice data, secondary care data, and Office for National Statistics mortality data. Data extraction from GP practice databases will be conducted using a Morbidity Query Information Export Syntax search based on participants’ NHS numbers. Short- and long-term outcomes such as weight, BMI, BP, glucose, lipids, CVD and the development of diabetes mellitus will be extracted. The help of the CCGs will be sought to obtain mortality and hospitals admission data using participant’s NHS numbers. The data will then be stored at the University Hospitals of Leicester and will be analysed independently by a statistician at the Leicester Diabetes Centre who will not have any participant identifiable data.

Confidentiality of data was maintained through the coding of participant data and the safe storage of paper and electronic data and audio tapes (which will be labelled with coded ID only).

Summary

The RCT was one of the first of its kind undertaken in the UK to look at the long-term effectiveness of a structured education programme focused on a lifestyle intervention in those patients identified through a risk-screening tool as being at high risk of T2DM. The Let’s Prevent study offered a unique and complete package, consisting of both screening and a subsequent intervention. Such an approach is in accordance with other recommendations from research, such as the IMAGE project. 47

National Institute for Health and Care Excellence guidelines around the prevention of diabetes mellitus also call for the use of lifestyle interventions as an integral part of any diabetes mellitus prevention pathway. 200 Although there has been much evidence to support the use of lifestyle interventions in improving health behaviours, and empirical evidence for both the clinical effectiveness and cost-effectiveness of such interventions have already been established, there is a lack of translational research in this domain. Let’s Prevent specifically targeted implementation within a routine health-care setting and with a multiethnic population for maximum transferability of results.

Introduction

As previously described (see Chapter 3), the LPRS was developed as a pragmatic way of screening primary care databases for those with PDM or undiagnosed T2DM for invitation to screening in a multiethnic UK population. Many risk scores have been developed and validated for identifying those at risk of T2DM,125,132,202 but the majority of these have not been used in practice and, therefore, there is a lack of evidence regarding their use in real-life settings. In addition, the majority of risk scores used in practice to date have identified those at risk of developing diabetes mellitus in the next 10 years rather than those with current PDM or undiagnosed T2DM. A recent systematic review203 found that 18 risk scores had been developed specifically for detecting PDM and undiagnosed T2DM and, of these, only three had published evidence of use in clinical practice. Noble et al. 132 concluded that, although much work had been done to develop risk scores for diabetes mellitus, most of these were rarely used; the authors suggested that the reasons for this may include the fact that they require data that are not routinely recorded or that scores were developed without a specific user or clear use in mind. The aim of this chapter is to report the prevalence of PDM and undiagnosed T2DM found when using a two-step screening strategy involving a risk score compared with a population-based approach.

Methods

The first phase of the Let’s Prevent study involved using the LPRS to identify people at risk of PDM and T2DM for invitation to a screening session where they received an OGTT. Those found to have undiagnosed T2DM were referred back to their GP; those with PDM were included in the cluster randomised prevention trial. For the full methodology of this first phase see Chapter 5.

Statistical methods

The primary outcome of the screening phase was the proportion of people detected with PDM or T2DM. Participants were categorised in accordance with WHO 1999 criteria. 8 Anyone who had an OGTT result in the diabetes mellitus range was recalled for a confirmatory test. In this study PDM was defined as IFG and/or IGT. IFG was defined as a fasting blood glucose concentration of between 6.1 and 6.9 mmol/l inclusive and IGT as a 2-hour blood glucose concentration of between 7.8 and 11 mmol/l inclusive.

Although HbA_1c was not used for diagnosis in this study, we also report data on the following categories using HbA_1c: 6.0–6.4% (42–47 mmol/mol), ≥ 6.5% (48 mmol/mol) and ≥ 6.0% (42 mmol/mol), and those suggested by the ADA, namely 5.7–6.4% (38–47 mmol/mol) and ≥ 5.7% (38 mmol/mol). 204 In addition, we assessed the yield when using a fasting blood glucose concentration of 5.5–6.9 mmol/l to define IFG, as recommended by NICE. 126 We also assessed the response rate to the invitation to screening.

The response rates and prevalence were compared with those found in the ADDITION-Leicester study. 123 As previously described (see Chapter 3), ADDITION-Leicester used a population-based screening approach in the same locality as Let’s Prevent. Random samples of 40- to 75-year-olds (25- to 75-year-olds for those of South Asian ethnicity) from 20 general practices were invited for an OGTT.

Means (SDs) and counts (%) were used to summarise the data. Differences between the screening approaches were assessed using the t-test for continuous variables and the chi-squared test for categorical variables. The prevalence (PPV) of each glucose disorder was calculated. All statistical tests are two-sided and p < 0.05 reflects statistical significance throughout.

Results

A total of 17,972 people from 44 practices were identified as high-risk in the Let’s Prevent Study; 19.2% of those invited attended screening. No difference in response rate was found between Let’s Prevent, where a risk score approach was used, and the population-based approach used in ADDITION (22.0%, p = 0.88). In total 3449 participants were screened (Table 22). The mean age of participants was 63.2 years, 60.8% were male, and the majority were of white European ethnicity (86.7%). Data were not available for those who did not attend screening.

Compared with ADDITION-Leicester, those screened using the risk score approach were significantly older, more likely to be male, heavier, had higher BP, glucose levels and HbA_1c, and were more likely to be taking lipid-lowering and antihypertensive medications. Higher levels of total cholesterol were observed in ADDITION-Leicester, which is probably a reflection of the lower treatment levels with satins in this group (see Table 22).

Overall, 30.1% of those screened had abnormal glucose tolerance when assessed using an OGTT (Table 23). A total of 25.5% of participants were found to have PDM, of whom, the majority were people with IGT (22.4%); 4.5% were found to have undiagnosed T2DM. Using a fasting glucose level of 5.5–6.9 mmol/l to define PDM, as suggested by NICE, led to significantly more cases being identified (36.4% vs. 25.5%). Across all categories of glucose disorder, more cases were picked up when using a risk score approach than had been found in a study using a population-based screening approach.

Using HbA_1c gave higher prevalence across all categories (Table 24). A total of 45.0% of participants had an HbA_1c of ≥ 6.0% (42 mmol/mol). Again, higher rates were seen in Let’s Prevent, which utilised a two-step approach, than in the population-based approach. Using the cut-off points suggested by the ADA204 significantly increased the yield across both screening approaches; for example, 75% of those screened using the two-step approach had a HbA_1c of ≥ 5.7% (38 mmol/mol).

Discussion

Finding effective methods to prevent T2DM is a public health priority, and given the unequivocal success of prevention programmes, there is an increasing need to develop tools in order to identify high-risk individuals who may benefit from these programmes. This study has shown a higher ‘hit rate’ when using a two-step screening approach than when using population screening, and this is in line with previous studies. 205 This approach has also been shown to be cost-effective in modelling studies. 30

Interestingly, although the yield was increased, the uptake to the screening invitation was low and comparable to that found in ADDITION-Leicester, a population-based screening programme carried out in the same area. 123 This finding suggests that informing people that they are at high risk of diabetes mellitus does not increase attendance to screening compared with a generic invitation. This is in contrast to other studies that have shown that risk stratification increases attendance. 205 There are many possible explanations for this. First, the low response rate may be attributable to people not wanting to take part in the clinical trial, rather than the screening. The NHS Health Check programme, which is not part of a research project, has seen uptake rates of around twice the level that we have reported here. 7,206 Second, everyone received an OGTT. The OGTT is costly, time-consuming and inconvenient. 24 Both patients and health-care professionals have reported that the OGTT is a barrier to attending screening. 207 The ADDITION-Europe study used a variety of screening methods to establish a cohort with screen-detected T2DM. 124 Centres that used a three-stepped approach combining a risk score, followed by a random blood glucose test, followed by an OGTT had a significantly greater uptake than centres that used a one-step approach using the OGTT only. 205 Third, both studies have been carried out in Leicester, UK, and the ethnic makeup of this locality may have had an impact on the response to screening. The area has a high prevalence of ethnic minority groups [ranging from 49% in the city of Leicester to 8% in the county of Leicestershire (2011 census; www.ethnicity.ac.uk/medialibrary/briefings/localdynamicsofdiversity/geographies-of-diversity-in-leicestershire.pdf) (accessed 21 July 2016)]. 208 Lower uptake to screening in those of South Asian ethnicity has consistently been observed. 122,123,140 This problem is amplified by the known increased risk of T2DM in this community. 140 The method of risk communication used may also affect uptake, and further investigation into this is warranted. Finally, the method of risk communication used was minimal; potential participants were told that there are risk factors for diabetes mellitus, some of which are modifiable, and that their medical records suggest that they might be at high risk. Informing people of their absolute risk of diabetes mellitus and potential benefits of early diagnosis and intervention may have increased uptake to screening.

In 2011, WHO recommended that the diagnostic criteria for T2DM be revised to include those with a HbA_1c of ≥ 6.5% (48 mmol/mol). 8,10 WHO found insufficient evidence to classify PDM using HbA_1c;10 however, an international expert panel and ADA have suggested that ranges of 6.0% (42 mmol/mol) to 6.4% (46 mmol/mol) and 5.7% (38 mmol/mol) to 6.0% (42 mmol/mol) can be considered. 204 Based on the ADA criteria, up to two-thirds of the population could fall into this category, which would overwhelm prevention initiatives. However, moving away from the OGTT has the potential to increase uptake to screening.

Let’s Prevent invited the top 10% most at-risk patients within a practice for screening. NICE suggest that the top 50% be invited for further testing. 200 Lowering the cut-off point will give potentially lower PPVs than seen here, as the prevalence of disease will be reduced in a lower-risk group. The benefit of using a practice-based risk score for screening is that GP practices can decide where to set the cut-off point based on the resources available. The LPRS is the first tool developed for use in GP practices that includes PDM in the conditions detected. Therefore, there are limited data with which to compare these results. Alongside the development of the risk score, we have also developed a piece of software that enables primary care to integrate this score into routine clinical practice. Using this tool optimises the high-quality data stored in general practice. 209 The LPRS software is freely available for use across the UK and can be downloaded from http://leicesterdiabetescentre.org.uk/The-Leicester-Diabetes-Risk-Score.

There are a number of limitations that must be considered. First, only those at high risk were invited for screening. Therefore, we are unable to assess sensitivity, specificity and the NPV of the score in this setting. Second, the low response rate, although similar to other studies in similar populations140 and which reflects the difficulty in recruiting a multiethnic urban population with wide variations in social economic status into research studies, may have affected the representativeness and generalisability of the data. Owing to ethical constraints, we were not able to collect data on those who were invited but did not attend screening; therefore, we are unable to compare the characteristics of those who attended with those who did not. These data were collected in the ADDITION-Leicester study and, overall, those who attended were older and more likely to be female. 140 Finally, the score used here utilised the percentage of South Asians within the practice as a proxy for individual ethnicity. Since the conception of these studies, the recording of ethnicity within general practice has improved and the updated version of the LPRS includes individuals’ ethnicity (see Chapter 3). Therefore, some South Asian participants may have received a falsely reduced score if they belonged to a practice with a high white European population and vice versa.

Summary

In summary, using a risk score to identify those at high risk of PDM/T2DM identifies a high previously undiagnosed yield using either an OGTT or HbA_1c for diagnosis. Previous population-based screening programmes using universal OGTT have had lower yields. The LPRS is an inexpensive and simple way of targeting screening programmes at those at the highest risk of PDM/T2DM.

Introduction

As previously described in detail (see Chapter 5), the second phase of the Let’s Prevent Study was a cluster RCT in all those found to have PDM, comparing a T2DM prevention programme with standard care. The trial recruited from practices and participants from July 2009 to June 2011. Participants were followed up at 6, 12, 24 and 36 months post baseline. The primary end point was the development of T2DM during the 36-month follow-up. The aim of this chapter is to report the results of the RCT.

Results

Overall, 44 practices were recruited and randomised: 21 to the standard care group and 23 to the intervention group. Screening took place in all 44 practices, and all those found to have PDM were then included in the RCT. At one small practice (list size = 1650), no eligible participants were identified in phase one; therefore, only 43 practices are included in the RCT. The included practices ranged in size from 1470 to 24,000 patients. Six practices were recruited from areas with large South Asian populations (intervention = 4 practices, standard care = 2 practices). The median number of participants recruited per practice was 23 in the standard care arm and 17 in the intervention arm; the number of participants recruited per practice ranged from 2 to 49 (Table 25). Figure 9 shows the flow of participants through the trial. In total, 880 participants were found to have screen-detected PDM and were therefore included in the RCT (433 in the standard care arm, 447 in the intervention arm). At 36 months, 75% (n = 333) of the intervention group were still being followed up compared with 79% (n = 340) in the standard care arm (p = 0.43). Of those who were included in practices randomised to the intervention arm, 101 (22.6%) did not attend the initial education session and, therefore, are excluded from any per-protocol analysis (see Table 27). In total, 130 (29.1%) participants in the intervention arm attended all three education sessions (see Development of type 2 diabetes mellitus by level of attendance). Those who did not attend the initial session were not invited to the refresher sessions but they continued to be followed up.

FIGURE 9.

Flow of practices and participants through the trial. Reprinted from Preventive Medicine, vol. 84, Melanie J Davies, Laura J Gray, Jacqui Troughton, Alastair Gray, Jaakko Tuomilehto, Azhar Farooqi, Kamlesh Khunti, Thomas Yates, A community based primary prevention programme for type 2 diabetes integrating identification and lifestyle intervention for prevention: the Let's Prevent Diabetes cluster randomised controlled trial, pp. 48–56, 2016, figure 1, with permission from Elsevier. 210

Table 25 shows the baseline characteristics of those included in the trial by randomisation. A higher level of deprivation was seen in the intervention group; this was measured using the IMD. The proportion of current smokers was significantly higher in the standard care group than in the intervention group (5.1% vs. 8.5%). The mean (SD) weight, BMI and waist circumference were also significantly higher in the standard care group than in the intervention group.

Development of type 2 diabetes mellitus

Overall, 131 participants developed T2DM over the course of the trial; this equates to 60.32 events per 1000 person-years (95% CI 50.82 to 71.58 events per 1000 person-years) (Table 26 and Figure 10). Lower event rates were seen in the intervention group than in the control group (57.60 events per 1000 person-years and 63.16 events per 1000 person-years, respectively). The HR showed a 26% reduced risk of developing T2DM in the intervention arm compared with standard care, but this did not reach statistical significance (p = 0.18). In a secondary, per-protocol analysis, the treatment effect was greater (35% reduction) when excluding those who did not attend the education programme in the intervention arm; this was also non-significant (p = 0.07). The difference in the event rates is accentuated in the per-protocol analysis (53.04 vs. 63.16, respectively).

TABLE 26 - Development of T2DM. HR (95% CI) takes into account clustering

This is the same as complete case, as there are no missing data for the primary outcome.

Note

Adapted from Preventive Medicine, vol. 84, Melanie J Davies, Laura J Gray, Jacqui Troughton, Alastair Gray, Jaakko Tuomilehto, Azhar Farooqi, Kamlesh Khunti, Thomas Yates, A community based primary prevention programme for type 2 diabetes integrating identification and lifestyle intervention for prevention: the Let’s Prevent Diabetes cluster randomised controlled trial, pp. 48–56, 2016, table 2, with permission from Elsevier. 210

Analysis type	Standard care	Intervention	HR	95% CI	p-value
ITTa			0.74	0.48 to 1.14	0.18
Events, n (%)	67 (15.5)	64 (14.3)
Rate per 1000 person-years (95% CI)	63.16 (49.71 to 80.24)	57.60 (45.09 to 73.59)
Per protocol			0.65	0.41 to 1.03	0.07
Events, n (%)	67 (15.5)	51 (14.7)
Rate per 1000 person-years (95% CI)	63.16 (49.71 to 80.24)	53.04 (40.31 to 69.80)

FIGURE 10.

Kaplan–Meier survival curves by intervention. Green data relate to the standard care group and blue data relate to the intervention group. Trt, treatment group.

Discussion

We have shown that a pragmatic diabetes mellitus prevention programme, aimed at those with PDM identified through a two-stage screening process, can lead to statistically significant improvements in HbA_1c, LDL cholesterol, psychosocial well-being, sedentary time and step count up to 3 years post the initial education programme. The primary outcome of the study was reduction in the progression to T2DM; although non-significant, a large treatment effect of around a 25% reduction was seen in those practices randomised to the education. The treatment effect seems to be related to the level of attendance, with higher attendance being associated with greater reduction in progression to T2DM. This was reflected in the glucose levels recorded, with significant drops seen in fasting glucose, 2-hour glucose and HbA_1c when excluding those who did not attend the education sessions.

Although not as high a reduction in T2DM as seen in more resource-intensive prevention programmes, the reduction in the progression of T2DM seen here is comparable to other pragmatic diabetes mellitus prevention programmes, such as the Indian programme. In the Indian programme, a reduction in T2DM progression of 28.5% was seen in those receiving a lifestyle modification programme. 34 A 58% reduction in T2DM (HR 0.4, 95% CI 0.3 to 0.7) was seen in the Finnish prevention programme, but this involved a significantly higher amount of contact time in those receiving the intervention. 32 Those in the intervention arm of the Finnish study received seven one-to-one sessions with a nutritionist during the first year and then one session every 3 months for the remainder of the study and were offered supervised, progressive, individually tailored circuit-type moderate-intensity resistance training sessions free of charge. 32 In a similar manner, the American DPP study intervention involved 16 lengthy one-to-one counselling sessions, followed by in-person one-to-one contact at least once every 2 months and additional group-based sessions four times annually; this also led to a 58% reduction in T2DM (95% CI 48% to 66%). 33 Such intensive prevention programmes are not achievable in the current UK health-care setting. In contrast, the Let’s Prevent programme consists of a 6-hour initial group course followed by a 1-hour refresher session each year (i.e. a total of 8 hours in group-based sessions over 3 years).

An interesting finding was the intervention attendance was strongly and linearly related to effectiveness; those who attended all three group-based contacts (initial Let’s Prevent programme and both annual follow-on sessions) had an 88% reduction in the risk of T2DM. Although total contact length and frequency have consistently been shown to be related to intervention effectiveness,52 this is the first study to show the strength of this relationship over the longer term and with face-to-face contacts separated by 12 months. This finding suggests that the weak effectiveness of the intervention overall was primarily driven by non-attendance rather than by ineffective interventional components. In total, 77% of participants attended the initial Let’s Prevent structured education programme, with 56% attending at least one follow-on session and 29% attending all three face-to-face group contacts. This demonstrates the difficulty in securing multiple contacts for behavioural interventions within a primary health-care setting and has implications for future prevention programmes. More research is needed to establish optimal methods of increasing uptake to lifestyle interventions within primary care, particularly around maximising adherence to follow-on support after the initial intervention.

Event rates reported in this study, that is, 57.60 events per 1000 person-years in the intervention arm compared with 63.16 events per 1000 person-years in the standard care arm, are consistent with a recent meta-analysis of prospective cohort studies in those with PDM which reported rates ranging from 35.54 per 1000 person-years in those with IFG up to 70.36 per 1000 person-years in those with both IGT and IFG. 14 The event rates reported here are also similar to DPP, which reported 48 per 1000 person-years in the lifestyle intervention group. 33 However, the progression rates in our study are substantially lower than other RCTs employing annual OGTTs in Europe and India, where rates of between 230 and 550 per 1000 person-years under control conditions and rates of between 110 and 393 per 1000 person-years after lifestyle interventions have been reported. 32,34 The sample size for the Let’s Prevent study was based on these European and Indian data and used a conversion rate of 35%. This lower event rate will have affected our power to detect a difference in the primary outcome.

As seen in other trials of a similar design to Let’s Prevent,48,121 improvements in many outcomes were seen in both the intervention and the control groups, although to a greater extent in those receiving the intervention. It could be that informing people that they have PDM is motivational in terms of improving one’s lifestyle or that those in the control group sought medical advice on their condition. Anecdotally, we were also aware of GPs initiating metformin in those included in the control group, even though this is not standard care for those with PDM. In addition, the NICE guidance around identifying and intervening in those at risk of T2DM was published during the RCT, which could have affected the results by increasing the interest in PDM and the prevention of T2DM in primary care. 126

Let’s Prevent was a cluster RCT; trials of this design have inherent issues compared with individually randomised studies. The study was powered based on a cluster size of 17 participants per practice,146 and the sample size was not inflated for variation in cluster size. Differences in cluster size can reduce statistical power. 211 Here the cluster sizes ranged from 2 to 49 (representing a coefficient of variation of 0.69). This was a pragmatic trial and we recruited practices of a range of sizes which led to this large variation in practice size. If this variation had been incorporated into the sample size the total sample size could have been increased by up to 42%, with the design effect increasing from 1.8 to 2.2. 212 This issue with study power could have led to the finding of no significance for the primary outcome. Interestingly, when the study is analysed ignoring the clustering, a significant effect is seen (p = 0.05). Another common issue with cluster randomised studies is that, although balance in practice-level characteristics is seen, balance in participant-level characteristics cannot be ensured. Here differences were seen in deprivation, smoking status and adiposity, with the intervention group more likely to be current smokers and have higher weight, BMI and waist circumference, suggesting a worse risk profile in the intervention group. Again this has been seen in other similar studies, such as DESMOND, where a difference in HbA_1c was seen at baseline,48 and these differences could have affected the power of the study and the results seen; however, when adjusting for smoking status and BMI the interpretation of the data was not changed.

The Let’s Prevent curriculum addressed both diet and physical activity. Significant improvements in both reducing sitting time and step counts were seen. The deleterious effects of sedentary time have recently been reported, with studies showing an increased risk of diabetes mellitus, metabolic syndrome, CVD and mortality in those who sit for long durations. 212,213 Here we see a reduction of around 30 minutes per day; although small and self-reported, this is complemented by objective pedometer data showing an increase in daily step count of 500 steps across the duration of the study. This equates to an increase of around 5 minutes of purposeful walking per day, or 35 minutes per week. This is a similar increase to that found in the Early Activity in Diabetes (ACTID) diet and physical activity intervention for those with newly diagnosed T2DM, which was also run in a primary care setting. Early ACTID achieved a 5.6-minute increase in moderate/vigorous activity in the intervention group compared with the control group. 214 This is a smaller increase than was seen in the PREPARE study on which the Let’s Prevent programme was based. 50 PREPARE saw a 2000-step increase, but was a much smaller study (n = 98) which focused solely on increasing physical activity. To see a dilution effect when an intervention is implemented on a large scale is not uncommon,52 with other, similar, large-scale studies showing no effect on physical activity outcomes. 215

A lower rate of completion was seen for the food frequency questionnaire than for the other self-reported data collected. Increased intake of unsaturated fats was seen with a trend towards decreased fat intake; this might suggest that participants in the intervention group were making healthier food choices or this may be a chance finding. The results also show a trend towards decreased fibre intake in the intervention group, which goes against what is recommend in the education programme. Future research should attempt to incorporate objective measures of diet quality, which could include biomarkers such as circulating vitamin C levels.

Benefits of the education also extended to the psychosocial well-being of the participants, with decreases in anxiety and increases in quality of life and illness perception reported. This is in line with other education programmes; for example, a RCT of the DESMOND programme for those with newly diagnosed T2DM based on the same philosophy as Let’s Prevent saw long-lasting improvements in well-being. 216

Even with the inherent issues of cluster randomised trials and the impact that the variable cluster size and lower progression rate had on the power of the study, we have still managed to show positive findings from a relatively low-resource group intervention. The pragmatic nature of this relatively low-resource intervention, the fact that it was based on a programme that is now delivered as part of routine care for those with established T2DM (DESMOND) and the fact that the study was carried out in primary care are some of the key strengths of this RCT. The study also followed up participants over 3 years, thus providing data for the longer-term effects of such a programme. The NHS Health Check programme and the guidance from NICE to use a two-stage screening programme to identify those at risk of diabetes mellitus means that the number of people identified with PDM is on the rise. 13 Of the 53,799 people screened as part of the Health Check programme in Leicester City CCG, 2675 (5.0%) had PDM. 217 In addition, it is well reported that the use of HbA_1c identifies a much larger disease burden than the OGTT. 135 We know that progression to diabetes mellitus can be prevented in those with PDM,31 but there are few evidence-based interventions that meet the NICE recommended standards which to refer those with PDM. Let’s Prevent could meet this pressing need, and Chapter 9 discusses pathways to implementation of such programmes.

The interpretation of this study should reflect the generalisability of the findings. The sample for the RCT was defined by the two-stage screening programme carried out in phase one of the study. Of those at high risk of having PDM who were invited for screening, only 19% attended. Although this is similar to other studies in similar populations139 and reflects the difficulty in recruiting a multiethnic urban population with wide variations in socioeconomic status into research studies, this may have affected the representativeness of the data. There were lower attendance rates among those from South Asian populations. However, as the screening was part of a research project in which people are asked to consent, provide data, etc., we would expect to see higher rates of uptake in a non-research setting. For example, the NHS Health Check programme has a 40% uptake rate. 206 A full discussion of the uptake to screening is given in Chapter 6. Future studies should look at methods for increasing the uptake to screening, particularly in hard-to-reach groups. All those screened and found to have PDM were included in the RCT. Of those attending practices in the intervention arm, 23% did not attend the initial education session. These participants continued to be followed up and, therefore, the treatment effect seen is representative of what would be expected in clinical practice. The Let’s Prevent programme was offered in two modalities, either as a 1-day (6-hour) programme or as two 3-hour sessions. A BME version of the programme delivered in Gujarati using translators was also offered. With T2DM becoming less a disease of older people and moving into younger populations, the need to deliver prevention interventions flexibly with optimal methods of ensuring continued engagement and adherence, including alternative approaches such as mobile health (mHealth) and electronic health (eHealth) solutions, should form the basis of taking this research further to increase participation. There is also a growing interest in the use of incentives for increasing uptake to interventions that promote health.

Overall, the Let’s Prevent RCT has shown that a relatively low-resource, pragmatic programme fit for implementation in the UK NHS can lead to a reduction in T2DM and improved biomedical and psychosocial outcomes. Future research should focus on increasing attendance to such programmes, which may require offering courses in formats not currently available, such as web-based formats. Chapter 8 assesses the cost-effectiveness of the intervention.

Introduction

In the UK, diabetes mellitus accounts for approximately 10% of the total health-care expenditure, which is projected to increase to around 17% in 2035/366 as the prevalence of T2DM continues to rise. The Let’s Prevent programme is a pragmatic structured education and lifestyle modification intervention (centred on diet and exercise) for T2DM prevention within primary care pathways. The Let’s Prevent programme is based on the DESMOND programme and targets lifestyle behaviour change among individuals with PDM, using simple, non-technical language and visual aids. The DESMOND programme is the first national education programme for people with T2DM to meet NICE criteria149 and was shown to be clinically effective and cost-effective. 48,49

We used data from the Let’s Prevent trial to carry out a within-trial cost-effectiveness analysis of the structured education intervention within primary care. We estimated the incremental cost–utility analysis for the Let’s Prevent programme, with the difference in costs and in effects calculated in relation to the standard care group. The study perspective was that of the NHS, and we considered costs to the health-care service. We captured changes in life expectancy and quality of life using the QALY as the effectiveness measure.

Methods

The trial methodology is described in detail in Chapter 5. Here, only the elements specific to the cost-effectiveness analysis will be given.

Patients randomised to the standard care group received a booklet detailing information on risk factors for T2DM and how physical activity and lifestyle change can be used to prevent or delay the disease. Patients randomised to the intervention group received the same booklet as the standard care arm but were invited to attend an initial 6-hour structured education programme, three monthly nursing support telephone calls, and a 3-hour structured education update programme at 12 and 24 months. Both arms of the trial received follow-up sessions at the same time points and the same data were collected.

Results

Utilities

Mean utility scores are reported in Tables 33 and 34. 15D utilities were available at all time points, with the percentage of missing data ranging between 20% at baseline and 45% at 36 months. EQ-5D utilities are reported at 12, 24 and 36 months with only minimal 12-month data available (n = 20). Mapped EQ-5D uses a published 15D to EQ-5D mapping equation (see Methods) to predict EQ-5D utilities using 15D data. We report the mean utilities of the merged EQ-5D variable pre and post multiple imputation (see Tables 33 and 34). This is shown graphically in Figures 11–13. A significant difference in 15D and in EQ-5D utility levels at baseline is evident, with slightly higher quality of life in the control group.

TABLE 33 - Mean utility scores using complete cases

No data at baseline or 6 months.

Instrument and time point	Intervention		Control
	Mean (SD)	Cases, n	Mean (SD)	Cases, n
15D (reported)
Baseline	0.872 (0.110)	358	0.892 (0.921)	337
6 months	0.886 (0.097)	283	0.893 (0.093)	317
12 months	0.890 (0.092)	300	0.890 (0.097)	314
24 months	0.877 (0.101)	264	0.883 (0.097)	265
36 months	0.884 (0.101)	238	0.876 (0.097)	249
EQ-5D (reported)a
12 months	0.835 (0.197)	16	0.817 (0.174)	4
24 months	0.771 (0.237)	178	0.800 (0.188)	195
36 months	0.797 (0.218)	287	0.783 (0.223)	292
EQ-5D (mapped from 15D)
Baseline	0.783 (0.205)	358	0.820 (0.171)	337
6 months	0.809 (0.180)	283	0.822 (0.172)	317
12 months	0.816 (0.169)	300	0.816 (0.179)	314
24 months	0.791 (0.186)	264	0.802 (0.180)	265
36 months	0.804 (0.187)	238	0.789 (0.179)	249
EQ-5D (reported + mapped)
Baseline	0.783 (0.205)	358	0.820 (0.171)	337
6 months	0.809 (0.180)	283	0.822 (0.172)	317
12 months	0.817 (0.170)	304	0.817 (0.177)	314
24 months	0.783 (0.220)	296	0.804 (0.188)	303
36 months	0.791 (0.215)	261	0.784 (0.208)	276

TABLE 34 - Mean utility scores after multiple imputation

Not discounted.

Multiple imputation	Intervention (SE)	Control (SE)	Difference
EQ-5D (reported + mapped)
Baseline	0.778 (0.010)	0.819 (0.009)	–0.041 (p = 0.002)
6 months	0.796 (0.010)	0.814 (0.009)	–0.018 (p = 0.164)
12 months	0.801 (0.009)	0.815 (0.009)	–0.013 (p = 0.308)
24 months	0.776 (0.012)	0.802 (0.010)	–0.027 (p = 0.095)
36 months	0.774 (0.013)	0.779 (0.012)	–0.005 (p = 0.788)
Total QALYsa	2.356 (0.029)	2.414 (0.025)	–0.0585 (p = 0.130)

FIGURE 11.

Complete-case mean utility scores for merged EQ-5D utility (reported + mapped) by intervention and control group at different follow-up time points. Error bars indicate CIs. Ctrl, control; Int, intervention.

FIGURE 12.

Mean utility scores after multiple imputation for merged EQ-5D utility (reported + mapped) by intervention and control group at different time points. Error bars indicate CIs. Ctrl, control; Int, intervention. Reproduced from Leal et al. ,218 figure 1, Cost-effectiveness of a pragmatic structured education intervention for the prevention of type 2 diabetes: economic evaluation of data from the Let’s Prevent Diabetes cluster-randomised controlled trial. BMJ Open 2017;7:e013592. This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

FIGURE 13.

Difference in mean utility scores between intervention and control groups at difference time points for different instruments. Positive values indicate higher utility in intervention group. MI reports the mean utility scores following multiple imputation.

Cost-effectiveness analysis

The results of the cost-effectiveness analysis post multiple imputation are reported in Table 37, with a complete-case analysis included for comparison. In our analysis, the costs of the intervention group were found to be £168 higher than the control group over 3 years after accounting for clustering. The majority of this difference is accounted for by the intervention cost.

TABLE 37 - Cost-effectiveness analysis

Complete-case analysis: non-inpatient costs not included. SD reported in brackets for absolute values in complete-case analysis. SE reported in brackets for multiple imputation results and difference values for complete-case analysis.

Discounted.

Corrected for difference between groups in utility at baseline.

Note

Reproduced from Leal et al. ,218 table 2, Cost-effectiveness of a pragmatic structured education intervention for the prevention of type 2 diabetes: economic evaluation of data from the Let’s Prevent Diabetes cluster-randomised controlled trial. BMJ Open 2017;7:e013592. This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Outcome	Complete (N = 225)a		Imputed (N = 880)
	Intervention (n = 102)	Control (n = 123)	Intervention (n = 447)	Control (n = 433)
Costs, £b	Medication, intervention		Non-inpatient, medication, intervention
Year 1	192 (9.3)	109 (8.4)	656 (49)	561 (93)
Year 2	154 (9.5)	106 (8.0)	603 (47)	542 (89)
Year 3	143 (8.9)	112 (7.7)	560 (36)	536 (34)
QALYsb
Year 1	0.835 (0.015)	0.830 (0.013)	0.793 (0.009)	0.816 (0.008)
Year 2	0.800 (0.014)	0.796 (0.013)	0.762 (0.010)	0.781 (0.008)
Year 3	0.759 (0.015)	0.760 (0.014)	0.723 (0.011)	0.738 (0.010)
Total costs, £	489 (26)	326 (21)	1818 (114)	1639 (192)
Total QALYs	2.394 (0.042)	2.386 (0.038)	2.278 (0.028)	2.334 (0.024)
Difference: QALY (unadjusted); 95% CI	0.0087 (0.057); –0.103 to 0.120		–0.0567 (0.04); –0.130 to 0.016
Difference: QALY (adjusted for baseline); 95% CIc	0.0349 (0.031) (–0.0257 to 0.0955)		0.0389 (0.02); –0.004 to 0.082
Difference: costs; 95% CI (£)	162 (33); 97 to 228		179 (239); –296.52 to 655.03
Cluster adjusted
Difference: QALYc (adjusted); 95% CI	0.0349 (0.031); –0.0250 to 0.0948		0.0461 (0.03); –0.0171 to 0.109
Difference: costs; 95% CI (£)	171 (41); 91 to 252		168 (285); –395.36 to 731.63
ICER (£)	4906		3643
Probability cost-effective at £20,000	0.80		0.86

The costs as reported in the complete-case analysis (n = 225) are lower, as these do not include non-inpatient costs which were omitted owing to the fact that the high level of missing data would significantly reduce the number of cases available for analysis. However, the difference in costs after accounting for clustering is similar (£171) to the imputation analysis.

A cost-effectiveness acceptability curve is reported in Figure 14, showing the probability that the intervention is the most cost-effective option at difference values of willingness to pay per QALY. The Let’s Prevent intervention was found to result in a net gain of 0.046 QALYs over 3 years at an overall cost of £168 per patient, with an ICER of £3643 and a probability of 0.86 of being cost-effective at a willingness-to-pay threshold of £20,000.

FIGURE 14.

Cost-effectiveness acceptability curve showing the probability that the Let’s Prevent intervention is cost-effective for different ceilings of willingness to pay. Reproduced from Leal et al. ,218 figure 2, Cost-effectiveness of a pragmatic structured education intervention for the prevention of type 2 diabetes: economic evaluation of data from the Let’s Prevent Diabetes cluster-randomised controlled trial. BMJ Open 2017;7:e013592. This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Discussion

We have estimated the cost-effectiveness of the Let’s Prevent intervention, a structured lifestyle modification programme, using QALYs as our outcome measure. We have found that the intervention is likely to be cost-effective at a willingness-to-pay threshold of £20,000 per QALY gained.

The Let’s Prevent study is the first structured education- and lifestyle-based diabetes mellitus prevention RCT in participants with PDM states. Therefore, there are no directly comparable studies. However, previous studies have examined the cost-effectiveness of structured education and lifestyle interventions in patients with T2DM. The most relevant to our study is the DESMOND intervention, which delivered a structured education and lifestyle intervention (on which the Let’s Prevent intervention is based) and which was found to be cost-effective49 in its study population of patients with established T2DM, with an ICER of £5387 per QALY gained. Notably, the ‘real world’ costs of DESMOND are reported to be considerably less expensive, resulting in a ‘real world’ ICER of £2092. Although not included in this report, the ‘real word’ costs for the Let’s Prevent intervention would be significantly lower than the within-trial costs reported here. In practice, there would be economies of scale in terms of both man power and resources required. For example, the one-off costs incurred (educator training, food models, venue costs for training and training materials) would be used over a much longer period than in the trial, bringing the cost per patient down. In addition, in the trial educators were paid to run courses; in practice, the programme would be a commissioned service and existing members of staff might work in a different way to deliver the intervention. Under most such scenarios it could be anticipated that the intervention would be even more cost-effective when implemented in clinical practice than in a trial setting.

The strengths of the present analysis are that it is based on randomised data from a broad-based (43 GP practices) multiethnic patient population with a good length of follow-up (3 years) and relatively low patient drop-out rate. Data were collected on both quality of life and resource use simultaneously and at different time points.

The major limitations of this cost-effectiveness analysis relate to the degree of missing data and the reliability and validity of the data available. We attempted to compensate for these limitations by making several assumptions in our primary analysis and then exploring these assumptions in our sensitivity analysis. For such a cost-effectiveness analysis, the ideal data set would have consisted of EQ-5D utility scores at all time points, hospital use data recorded through case report forms and GP records of medications and appointments. In this study we had access to EQ-5D data at later time points, patient-reported yearly medications at all time points, and patient-reported hospital and health-care use at later time points.

In this study the EQ-5D instrument was introduced through the economic questionnaire part way through the study and, therefore, the majority of our quality-of-life data were in the form of the 15D instrument, although a significant number of EQ-5D data had been collected at later time points. Given the preference for EQ-5D in health economics, our goal was to use all reported EQ-5D data available and to estimate EQ-5D scores using 15D data if available, to reduce the number of data requiring multiple imputation. In our primary analysis we used a published regression equation to achieve this. As demonstrated in our sensitivity analysis, irrespective of the instrument or composite instrument used to estimate QALYs, we consistently demonstrated a greater increase in QALYs in the intervention than in the control group, once differences in baseline utility had been accounted for. Although the 15D instrument is not as well validated as the EQ-5D instrument, in this study it formed our most reliable and complete single data source on quality of life. Using only data from the 15D instrument, the net gain in QALYs in the intervention group was similar to our primary analysis using composite EQ-5D-based data. The difference noted in baseline utility may relate to the cluster randomised nature of this trial at the level of GP practices. Furthermore, we noted a significantly higher IMD score in the intervention group than in the standard care group at baseline.

Medication use data in this study were collected through the main trial questionnaire and the data set was comparatively complete. Although several assumptions were made in estimating medication costs, these were applied to both intervention and control groups, and the impact on difference in costs between the groups is likely to be minimal, although the accuracy of the absolute values for each group are likely to be less robust. We found no statistically significant difference between the intervention and control groups in medication cost. We did not include hospital data in our primary analysis owing to the limited number of data available, which were of questionable validity. When included in the multiple imputation models as part of the sensitivity analysis, inpatient admission data exerted a disproportionate effect on the ICER, with the net effect being markedly skewed depending on whether or not clustering was accounted for. This was thought to be a reflection of the relatively small sample of the inpatient data available, the imprecision with which it was recorded by patients and the high cost of inpatient admissions, such that the degrees of random variation that would be expected when handling data of a smaller sample would exert a dramatic effect on overall cost.

The volume of missing data was significant, ranging from 20% to 56% in the data that were used for the primary analysis. Some of the missing data relate to the introduction of the economic questionnaire half-way through the study. The remainder of the missing data is accounted for by attrition in patient response to questionnaires over time. To account for this level of missing data, we used multiple imputation, a validated technique224 for handling missing data in RCTs, which is superior to complete-case analysis when there are significant degrees of missing data. We were unable to use multiple imputation on 12-month non-inpatient data, as the degree of missing data (> 95%) did not allow for a workable imputation model. Therefore, we used 24-month estimates as a ‘best guess’ for 12-month data. Given the similarity in non-inpatient data, both between intervention and control groups, and between time points, the effects of this assumption on the analysis were minimal.

In order to account for the clustering in this trial we used a mixed-effects model to analyse data post multiple imputation. We also included the cluster code in the imputation itself as a baseline explanatory variable, although this does not correct the data for clustering but rather uses GP practice as one of the variables (in addition to age, sex, BMI, and quality of life and economic indices) to generate replacement values for missing data, without exerting any effect on complete data. Analysing cluster-randomised data in the presence of missing data is not straightforward and future research might fruitfully explore the performance of different methods of doing so.

A further limitation of this study is that the time horizon of the analysis was restricted to 3 years. A longer duration of follow-up might be expected to yield greater quality-of-life benefits for the intervention arm, thus further improving the cost-effectiveness of the intervention. Longer-term economic modelling was originally proposed in the study design, but not undertaken for two reasons. First, the primary within-trial analysis showed that the intervention was cost-effective (£3643 per QALY gained) over the 3-year follow-up period. We considered it highly likely that any extrapolation would simply make the intervention even more cost-effective and would not qualitatively change the conclusion. Second, the original intention to conduct some extrapolation was based on the assumption that the trial might produce differences in numbers with diagnosed diabetes mellitus, and the long-term outcomes of these differences could then be propagated by modelling. In fact, the trial did not find a statistically significant difference in the proportion of patients developing diabetes mellitus, in any between-group differences for CVD or CHD 10-year risk, in the proportion with a CVD risk > 20%, or in any risk factor other than a very small 0.06% difference in HbA_1c. The outcomes were driven by the quality-of-life differences, which would not have been propagated by either of these models. However, additional follow-up of patients in the study would be very desirable to assess whether or not quality-of-life differences are maintained and whether any significant differences emerge in risk factors or in the development of diabetes mellitus.

Summary

This study has provided a reliable estimate of the costs of the Let’s Prevent intervention and broad-based estimates of the outcomes of the intervention (QALYs and health-care costs). Even taking into account the issues caused by missing data, the results appear robust across different assumptions, thereby indicating that the Let’s Prevent intervention is cost-effective using quality of life as our outcome measure. Owing to the modest number of cases of T2DM prevented, this study has not attempted to quantify the additional long-term benefits that this would provide.

Implementation pathways, resources and processes

Our programme grant coincided with the first round of NIHR Collaboration for Leadership in Applied Health Research and Care (CLAHRC) funding. Several investigators on the programme grant were centrally involved in the successful Leicestershire, Northamptonshire and Rutland CLAHRC, particularly in the organisation and leadership of the prevention and early detection themes. These twin pillars of NIHR-funded infrastructure afforded by CLAHRC and our programme grant supported a dedicated programme of work that was focused on developing clinically effective and feasible prevention pathways within primary care.

The initiation of the NHS Health Check programme in 2009 provided the first clear national-level policy for the prevention of vascular disease within primary care. 7 The prevention of T2DM through the identification of at-risk individuals and the provision of lifestyle change programmes was at the heart of the original business case and economic modelling that supported the creation of the NHS Health Check programme and remains a fundamental target of the policy. 225 However, there was no clear framework or mechanism for how diabetes mellitus prevention pathways could be co-opted into current primary care systems. We therefore set out to overcome this limitation by working with commissioners, GPs, practice nurses and members of the public.

After an initial scoping exercise, we centred this work around the DESMOND collaborative (see www.desmond-project.org.uk/). Since 2003, NICE has recommended the use of structured education in the management of diabetes mellitus;149 all individuals with T2DM are recommended to receive some form of structured education to aid their self-management and to promote lifestyle change at the time of diagnosis. DESMOND is one of the most prominent structured education programmes available to commissioning organisations nationally and the only programme to undergo a multicentre cluster RCT to quantify clinical effectiveness and cost-effectiveness. 48,49 As such, it is the most widely commissioned T2DM education programme in the UK, with over half of all CCGs (formally PCTs) having commissioned it for use within their diabetes mellitus care pathways. With the advent of the NHS Health Check programme, proactive regions expressed an interest in working with us to develop the systems needed to extend the delivery of structured education through DESMOND into the prevention of T2DM.

In 2010 work began with three early implementation sites to develop and pilot resources and pathways for extending structured education into the prevention of T2DM. At the time, the ‘Walking Away from Diabetes’ programme was the only structured education programme aimed at the prevention of T2DM to be underpinned by RCT evidence of efficacy. 50,51,226 Walking Away (originally called the PREPARE programme) acted as a precursor to, and informed the content of, the Let’s Prevent programme (see Chapter 4). Given that in 2010 Walking Away was already fully developed through CLAHRC, and considering the underpinning evidence of efficacy, it was chosen as the initial test-case structured education programme to inform our national implementation pilot work. However, the developed pathways are not specific to any one programme and can be utilised by other programmes, including Let’s Prevent.

Implementation pilot sites

The three initial pilot sites were Cumbria, Brighton and County Cork, Republic of Ireland. Cumbria Partnership NHS Foundation Trust had successfully embedded DESMOND into their diabetes mellitus care pathway and were planning on using opportunistic screening through the NHS Health Check programme to identify and refer at-risk individuals into a diabetes mellitus prevention programme. Brighton PCT had recently commissioned DESMOND and had a coexisting cohort of individuals within primary care who were identified and coded with IFG or IGT. There was local commitment for supporting and developing a pathway for these participants with referral into a lifestyle intervention. Finally, County Cork, Ireland, had secured funding to run a community-based model of prevention, which included the identification of at-risk individuals through the promotion of the Leicester self-assessment risk score and community campaigning. High-risk individuals were then able to request or seek referral into a lifestyle intervention. The sites provided three different test cases which would enable generalisability of developed approaches to future sites.

All three sites were visited by our team for initiation visits with relevant stakeholders. A detailed plan of implementation was than drawn up for each site. A cohort of DESMOND educators from each site were provided with one full day of vocational training to enable them to deliver Walking Away in addition to their DESMOND training. Training and subsequent quality-assurance models were based on those developed by the programme grant (see Chapter 4). The trained cohort of DESMOND educators and commissioners then met with our team several months later to discuss progress, to share successes and failures in terms of referral pathways and attendance rates, to highlight planned revisions to their pathways and to provide feedback on how the programme content and delivery could be improved. Finally, our team conducted a final site visit, which included undertaking focus groups with participants who had been through the prevention pathway from identification to referral through to attendance on the Walking Away programme. The focus groups were used to understand user-level perceptions of the referral pathway and programme content. This work provided a blueprint for the implementation of structured education in the prevention of T2DM which rapidly expanded to include differential models of training for DESMOND or non-DESMOND sites and registered or non-registered health-care professionals.

To date, our pathway has been commissioned in over 15 regional locations in the UK, from Edinburgh to Essex; in two regional locations in the Republic of Ireland; and in Gibraltar, with ongoing plans for implementation in Australia. We have worked with our implementation partners to evaluate specific elements of the pathway. For example, the implementation in Cumbria was estimated to cost as little as £30 per patient referred. Up until mid-2014, over 3000 individuals have been referred to and attended the Walking Away programme in the Cumbria site alone. A recent evaluation of the pathway in the Wigston and Oadby area of Leicestershire reported highly positive feedback from participants and demonstrated significant changes to behaviour 6 months after attending the Walking Away programme to levels that were consistent to changes reported in the original RCT (for report see Appendix 33).

Links with and dissemination through industrial partners

The work around the development of risk scores and risk identification algorithms supported by the programme grant (see Chapter 3) enabled significant collaboration and partnership with charity and industry partners, resulting in substantially increased reach and dissemination to the general population. In 2009 we were commissioned by the charity Diabetes UK to develop a self-assessment risk score for use by the general population. This work supported the development and validation of a scoring protocol tailored to information gained through self-assessment rather than practice records. 138 The self-assessment score and practice risk score contain similar variables and used the same development and validation analysis plan, but have been developed with their specific use in mind. The practice risk score is calculated by the LPRS software and therefore has a more sophisticated equation for calculating the score which gives greater discrimination; the self-assessment score is calculated by hand and is therefore a crude score with integer points given to categorised variables. The self-assessment additionally asks about waist circumference; at the time of developing the practice risk score this was not routinely collected in primary care and, therefore, has not been included in the practice risk score. Developing risk scores with a specific use in mind and therefore tailored to the requirements of the potential future use is recommended to those developing risk scores but has rarely been considered in the development of other risk scores. 132 This has been highlighted as a possible reason for the limited uptake and use in clinical practice of the many risk scores developed internationally for detecting those at risk of diabetes mellitus. 132

Both scores are recommended by NICE for the identification of those at risk. 126 Below is an outline of the implementation of both scores which has been achieved partly through the collaboration with charitable partners.

The practice risk score is freely available for download to be used in practices across the entire country (http://leicesterdiabetescentre.org.uk/The-Leicester-Diabetes-Risk-Score). This was highlighted in the September 2014 newsletter of the Royal College of General Physicians which is distributed to over 50,000 GPs nationally. We are also working with Diabetes UK and medical record software providers to incorporate the practice risk score into one or more of the most commonly used electronic medical records software providers. This will enable GPs to use the risk score without the need for downloading and installing it.

The self-assessment score has been completed by over 650,000 members of the public on the Diabetes UK website (http://riskscore.diabetes.org.uk/2013). The self-assessment score is also available in all Boots (Boots UK, Ltd, Nottingham, UK) and Tesco (Tesco PLC, Welwyn Garden City, UK) pharmacies across the country. We have also worked with Boots to integrate HbA_1c point-of-care testing alongside the risk score. This is currently being piloted in local stores. The risk score is also available on the Tesco (www.diabetes.org.uk/Tesco/Know-your-risk/Diabetes-UK-Online-Risk-Score/), Boots (www.boots.com/en/Pharmacy-Health/Health-information/Type-2-Diabetes-Find-out-your-risk/), NHS Direct (www.nhs.uk/Tools/Pages/Diabetes.aspx) and BUPA (the British United Provident Association Ltd, London, UK; www.bupa.co.uk/individuals/health-information/tools-calculators/hi-hra-diabetes) websites.

Diabetes UK has used the self-assessment score at local road shows since 2011. In 2012 (latest available figures) they carried out 20,911 risk assessments and referred 10,945 visitors, of whom 3700 were at high risk of having or developing T2DM in the next 10 years. In a survey conducted in 2011, of those who were in these higher-risk categories, 69% had been to their general practice or intended to go. Recall of the risk factors for T2DM and how to reduce risk was high among visitors 2 months after attending the road show. Evaluation of the use of the self-assessment score within Diabetes UK activities showed that after being risk assessed, 41% had started to eat more healthily and a further 44% intended to; 33% had increased their physical activity levels and a further 43% intended to; and 44% of those referred to their GP had been to their GP to seek a test. Bridget Turner, Diabetes UK, 2013, personal communication, says:

The diabetes risk score plays an important part in encouraging more people to take greater notice of their health and their lifestyles – helping to tackle the growing public health challenge of Type 2 diabetes and working towards earlier diagnosis and prevention.

Diabetes UK has embarked on a programme of awareness-raising within BME communities by recruiting ‘Community Champions’ – volunteers who are trained to deliver healthy lifestyle and diabetes mellitus awareness messages through information stands and talks at community centres, places of worship and at festivals and events. Over 100 champions have been trained in London, and the programme is being rolled out in four other towns across England. Some champions have been trained to undertake the risk assessments.

The self-assessment score was featured in the Embarrassing Bodies television programme,227 reaching around 2 million viewers. The risk scores were awarded gold at the national Quality in Care Diabetes Awards 2011 for the best early detection/screening initiative.

National policy and guidance

The work undertaken through the programme grant and associated work has significantly informed national policy and guidance around the prevention of T2DM. The principal investigator (Professor Melanie Davies) and a coinvestigator (Professor Kamlesh Khunti) were involved as a panel member and chair, respectively, in the NICE Programme Development Group that developed new guidance for the prevention of T2DM in high-risk populations. The published risk scores developed through the programme grant were included in the guidance and expert testimony involving unpublished data from the screening and structured education phases of the programme grant, including the importance of standardising educator training and quality-assurance pathways, substantially informed guidance on the content and structure of lifestyle intervention programmes. The finalised NICE guidance was published in 2012. 13

In addition to NICE, work supported through the programme grant contributed to The Handbook of Vascular Risk Assessment, Risk Reduction and Risk Management (both the original 2008 version and updated 2012 version228) and our work in South Asian populations significantly informed guidance issued by the South Asian Health Foundation for diabetes mellitus research priorities in British South Asians. 229

Summary

The programme grant has supported a host of associated activity that significantly adds value to the primary research objectives. By developing a referral pathway and freely available materials and tools, supporting the commissioning of our various prevention programmes, working with charity and industry partners and through contributing to NICE guidance we have ensured that the research supported by the programme grant has significantly advanced methods of prevention within primary care. We feel that we have improved individual patient health within the lifetime of the 5-year funding window.

Main findings and outputs from this programme grant

• We developed and validated a risk score for detecting undiagnosed PDM/T2DM in a multiethnic UK population using data from existing medical records. We have also developed a freely available piece of software which allows GPs nationally to use this risk score for screening for PDM/T2DM.

• We developed a group education programme for the prevention of diabetes mellitus (Let’s Prevent) in those with PDM suitable for roll-out within a NHS setting. The programme consists of an initial 6-hour course followed by yearly refresher sessions with motivational telephone calls throughout. The programme was extensively piloted and then tested in a large-scale cluster RCT.

• We identified 17,972 people at high risk of PDM/T2DM from 44 general practices using the risk score and invited them for screening. In total, 3449 participants were screened, of whom 30% had PDM/T2DM. The rates of glucose abnormalities detected were significantly higher than those previously reported when using a population-based approach. The 880 people identified with PDM were included in the RCT.

• We conducted a cluster randomised trial; 44 GP practices were randomly assigned to receive either the Let’s Prevent structured education programme or standard care. Participants were followed up for 3 years and the primary outcome was progression to T2DM. We found a non-significant 25% reduction in the progression to T2DM in those receiving the education compared with the standard care group; this was increased to 35% when excluding those who did not attend. Positive findings were also seen for HbA_1c, LDL cholesterol, psychosocial well-being, sedentary time and step count up to 3 years post the initial education programme.

• Intervention effectiveness increased in a linear manner with increasing face-to-face contact; those who attended Let’s Prevent and both group-based follow-on support sessions (at 12 and 24 months) had a statistically significant 88% reduction in their risk of T2DM.

• We conducted a cost-effectiveness analysis and showed that the Let’s Prevent intervention is highly cost-effective. The costs for the intervention group were found to be £168 higher over 3 years after accounting for clustering than for the standard care group; the majority of this difference is accounted for by the intervention cost. The Let’s Prevent intervention was found to result in a net gain of 0.046 QALYs over 3 years at an overall cost of £168 per patient, with an ICER of £3643 and a probability of being cost-effective of 0.86 at a willingness-to-pay threshold of £20,000.

• Alongside this work we have developed a referral pathway and freely available materials and tools, supporting the commissioning of our various prevention programmes. Through working with charity and industry partners and contributing to NICE guidance, we have ensured that the research supported by the programme grant has significantly advanced methods of prevention within primary care.

Implications for practice

The evidence presented here suggests that a two-stage screening programme utilising a non-invasive risk score followed by a blood test is feasible in primary care and identifies people with previously undiagnosed PDM/T2DM with a high level of discrimination and calibration. Those found to have PDM can then be enrolled into a prevention programme. The evidence here suggests that Let’s Prevent offers a clinically effective and cost-effective option for use in the NHS. Let’s Prevent is based on the DESMOND programme, which has been successfully commissioned across the UK. This gives an existing infrastructure which could be used to roll out this programme with scale and at pace. Team members were actively involved in the development of the NICE guidelines on the identification of those at high risk of diabetes mellitus. The programme of work presented here meets those guidelines and gives an infrastructure for how they can be met. The risk score is already available for use and is being used around the country. We aim to make the self-management structured education intervention also available quickly for CCGs to commission.

Research recommendations

A number of areas for further research emerge from this large body of work.

• The risk score was developed using a large data set from Leicester and Leicestershire. The ethnic make-up of this area means that the increased risk weighting for being from a BME group was dominated by those from a South Asian background. In other areas, other ethnic groups are predominant. Further research is warranted to assess whether or not the current BME coefficient within the risk score is valid for other ethnic groups.

• Only 19% of those invited to be screened actually attended. This may reflect the fact that participants were also being asked to take part in a research study. The NHS Health Check programme sees uptake rates of around 40%, which, although still low, are double those seen in research. Research is needed into methods for increasing the uptake to screening. This could include providing screening services outside primary care (i.e. pharmacy, community events), enhancing risk communication to patients in the screening invitation, increasing opportunistic referral from GPs through the use of prompting tools and using GP- and patient-level incentives to increase uptake.

• The Let’s Prevent structured education intervention was offered as either a 6-hour programme held on 1 day or as two 3-hour sessions held on different days. Around 23% of those in practices randomised to the intervention arm did not attend the initial education session, with only 29% of those in the intervention group attending offered face-to-face contacts (i.e. initial care session plus both refresher sessions). Offering prevention programmes in alternative modalities may increase attendance, particularly for follow-on support after initial face-to-face approaches. With near-universal mobile phone ownership and three-quarters of the population now having daily access to the internet, eHealth and mHealth interventions are increasingly being used to promote lifestyle intervention. 230,231 The integration of such technological platforms into traditional behaviour change interventions may help to maximise engagement and better reflect changing societal norms around communication and feedback. Further research is therefore needed to investigate the efficacy of integrating mHealth and eHealth approaches with traditional interventions such as Let’s Prevent.

• Pedometers were a key part of Let’s Prevent; they were identified as important in the development of the programme and have previously been shown to be an effective interventional tool in the prevention of T2DM. 232 Research is needed to investigate whether or not the provision of a pedometer with brief counselling and promotional materials could be used to promote physical activity behaviour change in those with PDM who are unable or unwilling to attend Let’s Prevent and to investigate the acceptability of pedometer use in different ethnic populations.

• The Let’s Prevent structured education intervention was adapted for people from a South Asian background. The programme was delivered via translators and the food models and games were tailored to be culturally appropriate. Using the same model of adaptation, the programme could be adapted for other minority groups, for example for those from Chinese or Afro-Caribbean backgrounds.

• The RCT followed up participants for 3 years. A natural extension to the programme of work would be to follow up these participants over a longer period, such as 10 years. Such a follow-up could be used to see if the event rate difference between the two groups in terms of T2DM increases over a longer duration, and we could also assess other outcomes such as CVD. Finally, Chapter 8 assessed the costs over the duration of the study; using modelling approaches, projected future costs savings could be assessed over a 10-year time horizon.

Dissemination activities and plans

In consultation with the Leicester Diabetes Focus Group, The Leicester Centre for Ethnic Health and our public and patient involvement lead (Kenneth Jones), we shall disseminate the findings of the Let’s Prevent study though a public event. All participants and GPs who took part in the study will be invited. At the event the main findings will be announced and there will be an opportunity for people to ask questions and speak to members of the study team. For those who are not able to attend, we shall also produce a newsletter.

In terms of academic dissemination, many of the work streams reported here have already been published in high-impact journals. 52,113,145,146,201,210 The cost-effectiveness analysis will also be published and presented both locally, nationally and internationally. Locally, this will be achieved through our existing collaborations with NIHR CLARHC East Midlands and the East Midlands Academic Health Sciences Network. We presented the Leicester Diabetes Prevention Pathway at the Academic Health Sciences Network Diabetes Innovation Exchange in January 2015. National and international dissemination was achieved by presenting the results at key conferences, such as Diabetes UK and the International Diabetes Federation Congress. 233–235

Summary

The results of this programme of work show that it is feasible to identify those with PDM in a primary care setting and to intervene to reduce their incidence of T2DM and promote healthy lifestyle options, using a group programme that is both clinically effective and cost-effective. The programme of work completed mirrors NICE guidelines and achieves their recommendations to meet a previously unmet need in the UK.

Contributions of authors

Professor Melanie J Davies (Professor of Diabetes Medicine) was the principal investigator for the trial, contributed methodological and practical advice to all components of the research programme and contributed to publications arising from the research.

Dr Laura J Gray (Senior Lecturer of Population and Public Health Sciences) conducted the statistical analyses and authored chapters of the manuscript.

Dr Dariush Ahrabian (Foundation Year 2 Doctor, Public Health rotation) conducted the review of health economics and cost-effectiveness and coauthored a chapter in the manuscript.

Dr Marian Carey (Director, DESMOND Research) was a coapplicant, assisted with the development of the education programme and reviewed the final manuscript.

Professor Azhar Farooqi (GP and Visiting Professor to Diabetes Research Centre) was a coapplicant and provided practical advice, particularly with regard to implementation in primary care.

Professor Alastair Gray (Professor of Health Economics) conducted the review of health economics and cost-effectiveness and coauthored a chapter in the manuscript.

Miss Stephanie Goldby (Senior Trial Manager) was responsible for the management of the trial and reviewed the final manuscript.

Mrs Sian Hill (Trial Manager) was the co-ordinator for the study.

Dr Kenneth Jones (GP and public and patient involvement representative) was a coapplicant and reviewed chapters of final report.

Dr Jose Leal (Senior Researcher) conducted the review of health economics and cost-effectiveness and coauthored a chapter in the manuscript.

Miss Kathryn Realf (Research Associate) authored chapters of the manuscript and prepared the final manuscript for publication.

Professor Timothy Skinner (Professor of Psychological and Clinical Sciences) was a coapplicant and contributed to the initial design of the study and to the education programme.

Mrs Bernie Stribling (Director, DESMOND programme) was a coapplicant and provided practical advice as regards the implementation of structured education into primary care.

Mrs Jacqui Troughton (Dietitian, Advanced Practitioner in Diabetes) led on the development of the educational intervention and training programmes for educators, the quality assurance of educational intervention and the piloting of the intervention.

Dr Thomas Yates (Senior Lecturer of Physical Activity, Sedentary Behaviour and Health) contributed to the development of the physical activity component of the educational intervention and authored chapters of the manuscript.

Professor Kamlesh Khunti (Professor of Primary Care Diabetes and Vascular Medicine) was a coapplicant, contributed methodological and practical advice to all components of the research programme and contributed to publications arising from the research.

The Let’s Prevent Diabetes team

Keith Abrams, University of Leicester, Leicester, UK.

Sayjal Amin, University Hospitals of Leicester, Leicester, UK.

Mary Bancroft, Hockley Farm Medical Practice, Leicester, UK.

Janette Barnett, University Hospitals of Leicester, Leicester, UK.

Hannah Berkeley, University Hospitals of Leicester, Leicester, UK.

Danielle Bodicoat, University of Leicester, Leicester, UK.

Michael Bonar, University Hospitals of Leicester, Leicester, UK.

Louise Boyles, University Hospitals of Leicester, Leicester, UK.

Paul Bray, University Hospitals of Leicester, Leicester, UK.

Nichola Cairns, University Hospitals of Leicester, Leicester, UK.

Sandra Campbell, University Hospitals of Leicester, Leicester, UK.

Patrice Carter, University of Leicester, Leicester, UK.

Sudesna Chatterjee, University Hospitals of Leicester, Leicester, UK.

Pauline Cowling, Sheffield Teaching Hospitals NHS Foundation Trust, Sheffield, UK.

Carolyn Currie, University Hospitals of Leicester, Leicester, UK.

Heather Daly, University Hospitals of Leicester, Leicester, UK.

Alison Dunkley, University of Leicester, Leicester, UK.

Sue Enright, University Hospitals of Leicester, Leicester, UK.

Geri Gray, University Hospitals of Leicester, Leicester, UK.

Colin Greaves, University of Exeter Medical School, Exeter, UK.

Joe Henson, University Hospitals of Leicester, Leicester, UK.

Stephen Hiles, University Hospitals of Leicester, Leicester, UK.

Jayne Hill, University Hospitals of Leicester, Leicester, UK.

Hannah Holdsworth, University Hospitals of Leicester, Leicester, UK.

Rosie Horne, University of Leicester, Leicester, UK.

Zin Zin Htike, University Hospitals of Leicester, Leicester, UK.

Shenaz Jamal, University Hospitals of Leicester, Leicester, UK.

Janet Jarvis, University Hospitals of Leicester, Leicester, UK.

Carolyn Johnson, University Hospitals of Leicester, Leicester, UK.

Janet Jones, University Hospitals of Leicester, Leicester, UK.

Sabera Khan, University Hospitals of Leicester, Leicester, UK.

Anita Khulpateea, University Hospitals of Leicester, Leicester, UK.

Judith Leonard, University Hospitals of Leicester, Leicester, UK.

Hamidreza Mani, University Hospitals of Leicester, Leicester, UK.

Lorraine Martin-Stacey, University Hospitals of Leicester, Leicester, UK.

Val Morgan, University Hospitals of Leicester, Leicester, UK.

Frances Morris, University Hospitals of Leicester, Leicester, UK.

Samiul Mostafa, University Hospitals of Leicester, Leicester, UK.

Alison Northern, University Hospitals of Leicester, Leicester, UK.

Kayleigh O’Brien, University Hospitals of Leicester, Leicester, UK.

Hersha Patel, University Hospitals of Leicester, Leicester, UK.

Naina Patel, University of Leicester, Leicester, UK.

Rachel Plummer, University Hospitals of Leicester, Leicester, UK.

Sheila Porter, University Hospitals of Leicester, Leicester, UK.

Mo Radia, University Hospitals of Leicester, Leicester, UK.

Dean Richmond, University Hospitals of Leicester, Leicester, UK.

Clare Russell, University of Leicester, Leicester, UK.

Rebecca Saker, University Hospitals of Leicester, Leicester, UK.

Jane Sennet, University Hospitals of Leicester, Leicester, UK.

David Sheppard, Saffron Group Practice, Leicester, UK.

Rebecca Spong, University of Leicester, Leicester, UK.

Margaret Stone, University of Leicester, Leicester, UK.

Nick Taub, University of Leicester, Leicester, UK.

David Webb, University of Leicester, Leicester, UK.

Emma Wilmott, University Hospitals of Leicester, Leicester, UK.

Carolina Wilson, University Hospitals of Leicester, Leicester, UK.

Panna Wilson, University Hospitals of Leicester, Leicester, UK.

Publications

Gray LJ, Davies MJ, Hiles S, Taub NA, Webb DR, Srinivasan BT, et al. Detection of impaired glucose regulation and/or type 2 diabetes mellitus, using primary care electronic data, in a multiethnic UK community setting. Diabetologia 2012;55:959–66.

Gray LJ, Khunti K, Edwardson C, Goldby S, Henson J, Morris DH, et al. Implementation of the automated Leicester Practice Risk Score in two diabetes prevention trials provides a high yield of people with abnormal glucose tolerance. Diabetologia 2012;55:3238–44.

Gray LJ, Khunti K, Williams S, Goldby S, Troughton J, Yates T, et al. Let’s Prevent Diabetes: study protocol for a cluster randomised controlled trial of an educational intervention in a multi-ethnic UK population with screen detected impaired glucose regulation. Cardiovasc Diabetol 2012;11:56.

Carter P, Gray LJ, Talbot D, Morris DH, Khunti K, Davies MJ. Fruit and vegetable intake and the association with glucose parameters: a cross-sectional analysis of the Let’s Prevent Diabetes Study. Eur J Clin Nutr 2013;67:12–17.

Yates T, Henson J, Khunti K, Morris DH, Edwardson C, Brady E, Davies MJ. Effect of physical activity measurement type on the association between walking activity and glucose regulation in a high-risk population recruited from primary care. Int J Epidemiol 2013;42:533–40.

Dunkley A, Bodicoat DH, Greaves CJ, Russell C, Yates T, Davies MJ, et al. Diabetes prevention in the real world: effectiveness of pragmatic lifestyle interventions for the prevention of type 2 diabetes and of the impact of adherence to guideline recommendations. A systematic review and meta-analysis. Diabetes Care 2014;37:922–33.

Gray LJ, Khunti K, Wilmot EG, Yates T, Davies MJ. External validation of two diabetes risk scores in a young UK South Asian population. Diabetes Res Clin Pract 2014;104:451–8.

Troughton J, Chatterjee S, Hill SE, Daly H, Martin Stacy L, Stone MA, et al. Development of a lifestyle intervention using the MRC framework for diabetes prevention in people with impaired glucose regulation. J Public Health 2015:1–9.

Davies MJ, Gray LJ, Troughton J, Gray A, Tuomilehto J, Farooqi A, et al. A community based primary prevention programme for type 2 diabetes integrating identification and lifestyle intervention for prevention: the Let’s Prevent Diabetes cluster randomised controlled trial. Prevent Med 2016;84:48–56.

Leal J, Ahrabian D, Davies MJ, Gray LJ, Khunti K, Yates T, Gray AM. Cost-effectiveness of a pragmatic structured education intervention for the prevention of type 2 diabetes: economic evaluation of data from the Let’s Prevent Diabetes cluster-randomised controlled trial. BMJ Open 2017;7:e013592.

Data sharing statement

Data sharing requests should be made to the corresponding author.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, CCF, NETSCC, PGfAR or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the PGfAR programme or the Department of Health.