Screening for glucose intolerance and development of a lifestyle education programme for prevention of type 2 diabetes in a population with intellectual disabilities: the STOP Diabetes research project

Work package 1

The aims of WP1 were to:

• develop and assess the feasibility of a diabetes screening programme in a community setting for adults with ID (see Chapters 5 and 6)

• determine the prevalence and demographic risk factors for T2DM and IGR in people with mild to profound ID (see Chapters 5 and 6)

• validate the Leicester Self-Assessment diabetes risk score in people with ID (see Chapters 5 and 6)

• determine the cost-effectiveness of lifestyle intervention (see Work package 2), compared with current care (see Chapter 12)

• establish data linkage to Hospital Episode Statistics and the Office for National Statistics (ONS) (see Chapters 5 and 6).

Work package 2

The aims of WP2 were to:

• develop a lifestyle education programme for people with ID and IGR (or high risk of T2DM/CVD based on elevated BMI) (see Chapters 8 and 9)

• assess the feasibility of collecting outcome measures before and 3 months after attendance at lifestyle education (see Chapter 10).

Work package 3

The aim of WP3 was to:

• develop a quality assurance (‘intervention fidelity’) process for the assessment of educators who are delivering the education (see Chapter 11).

Approvals

The University of Leicester acted as sponsor for the programme of research. NHS research ethics approval was obtained from the East of England – Cambridge Central Research Ethics Committee (reference 12/EE/0340). Research and development approval was obtained for the research sites from Leicestershire Partnership NHS Trust (LPT); Leicester City Clinical Commissioning Group (CCG), East Leicestershire & Rutland CCG and West Leicestershire CCG; and the University Hospitals of Leicester NHS Trust.

Adherence to mental capacity legislation

Obtaining consent was the largest ethics consideration for this programme. Strict standard operating procedures needed to be established to ensure that valid consent was obtained in accordance with English capacity legislation,3 while taking into account the heterogeneity in capacity of individuals. More details on the assessment of capacity and taking consent are contained in the methodology section for the screening programme (see Chapter 4). This included providing people with all of the information that was relevant to making the decision on whether or not to participate in the research, and communicating this information in a way that was appropriate to them (such as using simple language and visual aids).

The process for those who lacked capacity involved talking to a ‘consultee’, whose role was to consider the study from the participant’s perspective (see Appendix 1, Figure 26). Regardless of whether or not the person with ID had the capacity to decide on participation, the research was discussed with the individual to help him or her to understand the project as far as he or she had the capacity to do so, and to ascertain any opinion that he or she had on participation. For example, if a person without decision-making capacity appeared even slightly anxious or reluctant to take part, then this was respected, and he or she was not recruited to the study.

Programme steering group

Strategic oversight and direction of the research programme was provided by the programme steering group (Figure 2), which comprised the chief investigator (KK), the lead researcher/project manager (AD) and co-applicants listed in the application, with ad hoc attendance from service users. The meetings were held four times per year and were independently chaired by Dr Colin Greaves, University of Exeter (see Figure 2). The meetings involved discussion of contractual issues, staffing, protocol and ethical amendments, public involvement (a rolling agenda item), recruitment progress, economic analysis, education development, anticipated timelines and progress against project aims.

FIGURE 2.

Governance structure of the STOP Diabetes programme. NETSCC, NIHR Evaluation, Trials and Studies Coordinating Centre.

Operational groups

The research team (researchers, ID research nurses and research administrator) met frequently throughout the programme to plan the individual components of the programme and to discuss progress. The details of these meetings were fed back to the steering group.

The education development team [a multidisciplinary team of health-care professionals (HCPs) and researchers with expertise in the field of both ID and developing diabetes and CVD prevention programmes] met regularly to oversee and facilitate the development of the lifestyle education programme (WP3). Progress and key decisions were fed back at steering group meetings.

Service user groups

A number of service users were involved in the research programme, but two service user self-advocacy groups were particularly influential. The groups met regularly, facilitated by an experienced supporter, and their comments were fed back to the steering group. More information about these and other service users’ involvement is detailed in Chapter 4.

Data protection

A six-digit study code was used to identify all of the study participants. This code was used for all hard and electronic copies of data that were collected for this programme (including questionnaires, anthropometric data and blood samples), which were retained in a secure setting.

The Leicester Clinical Trials Unit (UK Clinical Research Collaboration registration number 43) was responsible for the development of a secure database for the data that were collected as part of this research programme.

Definition of intellectual disability and case identification

Intellectual disability, also known as learning disability, is a lifelong condition with onset before adulthood, characterised by a reduced ability to understand new or complex information and learn new skills, and a reduced ability to cope independently. 4 Severity levels for ID are typically categorised by broad intelligence quotient (IQ), alongside the required deficits in independent living skills, into mild (IQ 50–69), moderate (IQ 35–49), severe (IQ 20–34) and profound (IQ < 20) ID. 5 Acknowledging the wide variation that exists between individuals with ID, typical abilities suggested for each category are outlined in Table 1 [based on World Health Organization (WHO) International Classification of Diseases, Tenth Edition (ICD-10)]. 5 More recently in the UK, the Learning Disabilities Public Health Observatory has offered a ‘working definition’, which includes brief practical guidance to improve the recognition of ID and to assist agencies in targeting services. 6

TABLE 1 - Intellectual disability categories and suggested abilities

Severity of ID	Suggested abilities and skills	IQ level
Mild	Good verbal communication, and basic reading and writing skills Usually independent in self-care and practical domestic tasks Often able to form/maintain good social relationships May have employment
Moderate	Limited language Able to achieve some independence with support, but requirements for support will vary Usually fully mobile
Severe	Uses some words and gestures Activities need to be supervised and ongoing support is necessary May have problems with movement
Profound	Communication is very limited Support is needed for all daily living activities Mobility is usually severely impaired

The aetiology of ID can be broadly divided into problems that occur in the antenatal, perinatal or postnatal periods, or as a result of multiple factors. Common causes of ID include genetic and chromosomal disorders – both non-inherited (e.g. Down syndrome)7 and inherited (e.g. fragile X syndrome) – and non-genetic factors, such as infection and environmental factors. However, in the majority of cases no specific cause is found. 8

A recent meta-analysis of population-based studies suggests that, overall, around 1% of people worldwide have ID, with wide variation dependent on age group and the income of the country (lower, middle and higher); for adults, the proportion is around 0.5%. 9 The evidence from existing ID registers and general practice lists in England suggests that the prevalence of ID is approximately 3–5 per 1000 individuals. 10,11 However, it is thought that the true prevalence could be as high as 2% of the adult population, as people with mild ID are generally under-represented. 11

For the current research programme, cases were identified via (1) records held on adults with ID in general practices and (2) a register of adults attending ID services owned by the local mental health trust (LPT), the Leicestershire Learning Disability Register (LLDR; see below).

General practices in the UK are now incentivised to maintain a register of people with ID. 12,13 Locally, for practices within Leicester City, East Leicestershire and Rutland, and West Leicestershire CCGs, the total number of adults (aged ≥ 18 years) on general practice registers with an identified ID is estimated to be around 4300 (based on figures provided by LPT).

The LLDR comprises adults with ID (aged ≥ 19 years) who live in the unitary authorities of Leicester city, Leicestershire and Rutland. 14 The register was established in 1987 to help facilitate the provision and monitoring of services, and to enable the collection of public health data. It is currently a joint venture between LPT and Leicester City CCG. Enrolment is via a large network of service providers, including specialist ID services, social services and primary care. Currently, there are ≈3900 people with mild to profound ID on the register. However, as the learning disability register is based on service use, some adults, particularly those with mild ID who have little or no support from services, may not currently be identified. This potentially accounts for some of the differences between the number of people identified on this register and the number on local general practice registers.

Type 2 diabetes and impaired glucose regulation

Type 2 diabetes is a serious chronic disease, characterised by prolonged hyperglycaemia. 15 Its symptoms can reduce quality of life and lead to serious health complications, including blindness, renal failure and amputation; 50% of new cases have demonstrable atherosclerosis at diagnosis. 15–17 The prevalence of diabetes in England is estimated to be 6.2%,18 rising to 8.0% [95% confidence interval (CI) 5.7% to 11.7%] when undiagnosed cases are included. 19 T2DM accounts for around 85–90% of diabetes cases; it creates a huge economic burden on NHS resources, at a cost of £8.8B annually (≈10% of total NHS expenditure). 20

Impaired glucose regulation is a condition in which blood glucose concentrations are elevated above the normal range but do not satisfy the criteria for T2DM. 21,22 Approximately 12% of the UK adult population have IGR, of which an estimated 5–12% go on to develop T2DM each year. Observational studies show a consistent and continuous association between glycaemia and CVD risk, whereby people with IGR have a significantly elevated risk of CVD. 23–25 Given the economic burden associated with this condition and its related comorbidities, this group represents an important target for preventative strategies. 26 Other commonly used terms to describe IGR include pre-diabetes, non-diabetic hyperglycaemia and high risk of diabetes; throughout the report, this high-risk group will be referred to as having IGR.

Previously in clinical practice, T2DM and IGR were identified using the ‘gold standard’ oral glucose tolerance test. 22 However, since the publication of updated WHO guidance in 2011 and subsequent National Institute for Health and Care Excellence (NICE) guidance in 2012, there has been a shift away from the use of the oral glucose tolerance test to the glycated haemoglobin (HbA_1c) test. 27,28 The potential benefits of the HbA_1c test include it being a non-fasting blood test, less inter-test variability and its ability to provide an indication of longer-term hyperglycaemia (over 6–8 weeks). 29 A HbA_1c level of ≥ 48 mmol/l (6.5%) is suggestive of T2DM, whereas a level of 42–47 mmol/l (6.0–6.4%) is suggestive of IGR or a high risk of diabetes. 27 Further details on the methods used to identify T2DM and IGR for this programme of research are provided in Chapter 5 (see Outcomes) and Figure 15.

Risk factors for type 2 diabetes and cardiovascular disease in people with intellectual disabilities

In the general population, increasing levels of obesity and sedentary lifestyles have been associated with a rise in non-communicable diseases, including T2DM and CVD. 30–33

Chronic conditions are becoming increasingly important for people with ID as their life expectancy increases. 34 There are a number of risk factors for T2DM that are known to be highly prevalent in people with ID, suggesting that T2DM and CVD may be more prevalent in this group. These risk factors include:

• sedentary behaviour35–38

• a high prevalence of obesity32–34,39

• increased antipsychotic drug use for the management of challenging behaviour40,41 and psychosis,42 which are associated with weight gain, hyperglycaemia and worsening of other metabolic CVD risk factors43–45

• genetic conditions associated with obesity (e.g. Prader–Willi syndrome). 46

Physical inactivity and sedentary behaviour are both common among people with ID, with only a minority (18–33%) achieving the recommended 30 minutes of moderate to vigorous physical activity (MVPA) daily,47,48 and < 15% of people with ID complete the recommended 10,000 steps per day. 49 Furthermore, < 10% of adults with ID who live in supported accommodation have an intake of fruit and vegetables that is sufficient for a balanced diet. 50 The evidence suggests that paid carers know little about public health recommendations on dietary intake. 50

However, little is known about T2DM, CVD and associated risk factors in the population with ID. UK-based data on the prevalence of T2DM are currently unclear. 32 Current estimates for diabetes prevalence in the UK are based on routinely reported data rather than on population-based studies. The suggested prevalence of diagnosed diabetes in people with ID in England is around 6–7%, but estimates are unable to distinguish between T2DM and other forms of diabetes. 13,51 Similarly, the prevalence of CVD among people with ID is reported to be greater than that among the general population, but the overall prevalence is unclear. 52

Further information on the current prevalence of T2DM, CVD and related risk factors in the ID population is presented in the systematic review in Chapter 2.

Diabetes screening

Given the increasing prevalence of diabetes, and the conferred risk of developing CVD, early identification and intervention through screening has been shown to be a useful approach in the general population. 53,54 The value of screening for IGR has also been demonstrated.

It is currently unknown if screening for asymptomatic glucose disorders is viable within UK populations with ID; there is a lack of evidence on feasibility, acceptability, outcomes and benefits. People with ID have been recommended by NICE as being an important group to consider in terms of diabetes prevention strategies, given their supposed high risk of developing diabetes. 27

General practitioners (GPs) in England have been incentivised to provide annual health checks to adults with ID since 2008–9 (for those aged ≥ 14 years since 2014). Recent data suggest that, nationally, the uptake of checks is around 44%. 55 However, the proportion who additionally have bloods taken as part of the health check, including HbA_1c (7%) and cholesterol (30%), is extremely low. 13

Risk scores for the early identification of impaired glucose regulation and type 2 diabetes

A staged approach to screening is recommended by NICE for those at risk of diabetes in the general population. 56 This involves using a risk score to pre-screen for individuals at the greatest risk of T2DM followed by a blood test in those at the highest risk. However, this approach has not been tried with populations with ID.

Risk scores are a non-invasive way of stratifying a population for targeted screening. They use information data from non-invasive risk factors to calculate an individual’s score: a higher score reflects a higher risk. Risk scores can be applied to (1) an individual as a questionnaire (these scores generally require data from only non-invasive risk factors, which would be known by members of the public) or (2) a population (e.g. in primary care, for which software is used to calculate the score using routine data from electronic medical records) and then screening invitations can be sent to those at highest risk. A number of diabetes risk scores have been developed and validated for use in the UK general population. 56–60 One such score is the Leicester Self-Assessment risk score (see Appendix 2), which allows people to easily assess their own risk of having undiagnosed IGR or T2DM and then self-refer for screening with a HCP. 58 The score contains seven questions that ask about age, sex, ethnicity, BMI, waist circumference, family history of diabetes and high blood pressure (BP). The score has been validated for use in a multiethnic UK population58,61 and is specifically recommended by NICE for identifying people who are at risk opportunistically. 56

To date, we are not aware of any risk scores that have been specifically assessed for use in populations with ID. However, it cannot be assumed that a risk score that is developed for a specific population will work well in another;62 for people with ID, there may be different risk factors, or weightings for specific risk factors may change, when compared with the general population. Therefore, this programme of work will seek to validate the Leicester Self-Assessment risk score in a population with ID (this work is presented as part of the screening study; see Chapters 5 and 6).

Diabetes prevention in adults with intellectual disabilities

People with ID experience a disproportionate burden of health inequalities compared with the general population, including poorer mental and physical health and higher rates of mortality. 63–66 Despite their increased health needs, they often find it difficult to access primary care services and to participate in health promotion activities. 67–69

Given the health inequalities among people with ID, and the possible increased risk of developing diabetes, people with ID have the potential to benefit from lifestyle changes (with appropriate support) that are addressed in lifestyle education programmes. However, the evidence base for diabetes prevention relates to the general adult population; literature focusing on ID is scarce. Details of the key literature on lifestyle behaviour change interventions aimed at modifying risk factors for T2DM and CVD in people with ID are presented in the systematic review in Chapter 3.

Current evidence from studies conducted in the general population suggest that intensive multicomponent lifestyle interventions aimed at weight loss, a healthy diet and increased physical activity can successfully reduce the risk of diabetes by 30–60% in those with IGR, and are likely to be cost-effective in the long term. 54,70

Increasing physical activity is fundamental to diabetes prevention initiatives, as research suggests that inactivity may have more impact than increased body weight in the development of insulin resistance. 71

For both obesity management72 and prevention of T2DM,27 NICE recommends that lifestyle interventions should be multicomponent, involving both dietary and physical activity advice, and incorporating behaviour change techniques. However, at present there are no national prevention programmes that suitable for people with ID, despite ongoing recommendations to make ‘reasonable adjustments’ to health-care services to address inequities in provision. 73

Education, exercise and leisure pursuits are often determined or influenced by carers (paid or family carers), who may have a range of competing time demands and a number of people for whom to provide support. For people with limited carer support, difficulties in understanding health risks could also influence motivation to change lifestyles. Therefore, there is the potential for this group to benefit from the development of a lifestyle education programme that is targeted at both people with ID and their carers in order to encourage changes in lifestyle behaviours that could reduce the long-term chances of this high-risk group developing diabetes.

Protocol and registration

This systematic review was registered with the International Prospective Register of Systematic Reviews (PROSPERO CRD42015019048). 77

Eligibility criteria

The review was guided by the population, intervention, comparison, outcomes, study design (PICOS) model. 78 We defined the population as adults (aged ≥ 18 years) with ID (whole study population or a defined subsample). The items of interest were defined as T2DM, CVD and their associated risk factors. Context was defined as population-based studies. We defined the outcomes as prevalence and/or incidence rates (or data to enable this calculation). Study designs included cross-sectional, retrospective and prospective cohort studies (Table 2).

All studies published since 1 January 2000 (until 21 April 2015) and in the English language were eligible. We contacted lead authors for further information when inclusion/exclusion could not be determined.

We chose to limit studies to those published in and after the year 2000 so that the current prevalence of T2DM and CVD could be estimated accurately; it is known that the prevalence of both of these conditions has increased substantially in recent decades.

Information sources

For this review, we searched the databases EMBASE, MEDLINE and PsycINFO. The last date of the search was 21 April 2015. We also searched the reference lists of relevant articles for possible additional studies.

Search

We combined medical subject heading terms and key words for T2DM, CVD, overweight/obesity, hypertension, hyperlipidaemia, elevated glucose level/impaired glucose tolerance, metabolic syndrome and ID (MEDLINE search strategy; Box 1). The search was limited to English-language studies with cohorts of adults aged ≥ 18 years, depending on the database.

Study selection

Full texts were identified after titles and abstracts were read separately by two investigators (TC and AD) who discussed discrepancies in selection at a later meeting. Only full-length articles were included; review articles were removed after being examined for references. Once we had retrieved the full texts of the articles, these were examined separately (by TC and AD) to check their suitability for inclusion.

Data collection process

We designed a data extraction form specifically for this review. Data were extracted by one investigator (TC) and verified for accuracy by another investigator (AD).

Data items

For each study, the first author’s name, title of the paper, year of publication, country of the cohort, study type, sampling method, dates of data collection and inclusion/exclusion criteria were extracted. We also extracted total sample size or subpopulation size, mean ages, proportion of male/female, severity of ID and ethnicity. For each of the outcomes, we also extracted how it was defined, how it was measured and the total number, and proportion, of people for whom it was measured. We extracted data separately for males and females, when reported. When framing the research question and designing the search strategy, we did not consider physical activity/sedentary behaviour, dietary factors or smoking; however, we extracted this information for studies that reported it. We also extracted information on general population data.

Risk of bias in individual studies

We used funnel plots79 and the Egger’s test80 to examine potential publication bias in the literature for the outcomes T2DM, ischaemic heart disease, obesity, hypertension and undefined CVD.

Summary measures

The main outcome measure for the meta-analysis was the prevalence of T2DM and CVD. The secondary outcome measures were the prevalence of:

• overweight/obesity

• hypertension

• hyperlipidaemia

• elevated glucose level/impaired glucose tolerance

• metabolic syndrome.

Synthesis of results

Owing to the variation in reporting of outcomes, we extracted descriptions and definitions of each outcome for analytic purposes and subcategorised for meta-analyses. We subcategorised circulatory disease outcomes as ischaemic heart disease, cerebrovascular disease and undefined CVD. We subcategorised diabetes outcomes as T2DM and pooled diabetes. BMI outcomes were labelled as obese (BMI of > 30 kg/m²) and overweight (BMI of 25–29.9 kg/m²). In some articles, overweight and above (BMI of > 25 kg/m²) was used as an outcome. We combined papers reporting both obese and overweight data to create an overweight and above outcome. The outcome definitions can be seen in Appendix 3 (see Table 60).

Owing to the large amount of variability between studies, we used a random-effects model to pool the point prevalence for each outcome. We conducted a secondary meta-analysis including data from a subset of 10 papers,81–90 which additionally reported general population comparison data (from the same population and time period). We assessed heterogeneity using the I² test. 80

Additional analysis

After the meta-analysis, metaregression was used to determine if study characteristics could explain heterogeneity (as measured using the I² test). These study characteristics were severity of ID, mean age and method of data collection (self-/carer reported, researcher collected, retrospective records/database). We conducted all of the analyses using Stata^® statistical software, version 14 (StataCorp LP, College Station, TX, USA). Significance was set at the 5% level (p < 0.05) and 95% CIs are presented throughout.

Study selection

In total, we identified 4513 articles via the literature searches. After duplicates were removed, 3645 articles remained to be screened. We reviewed the full texts of 148 articles once seven articles from other sources had been added (Figure 3). The authors of seven studies91–97 were contacted for information regarding their studies; five authors replied and two studies94,95 were deemed suitable to be included in the systematic review and meta-analysis. We also included a study93 by one of the authors who did not reply, after we had reread and discussed the article collectively in more depth.

FIGURE 3.

Study selection.

After review, 62 articles50,81–90,93–95,98–145 were included. Four of these articles90,102–104 reported findings from the same study and a further two articles109,125 reported findings from the same study, leaving 58 studies (see Table 3) remaining for the final systematic review and meta-analysis.

Study characteristics

The 58 studies included in the quantitative synthesis presented data on > 47,000 individuals. The characteristics of each of the studies are presented in Table 3.

Ten of the studies81–89,102,103 included in the systematic review also presented as general population comparison data for inclusion in the secondary meta-analysis.

The studies represented 23 countries on five continents. One study109 covered 14 European countries. Most studies were conducted in the USA/Canada (n = 1783,84,86–88,93–95,110,112,114,117–119,133,136,137). The remaining studies were conducted in Europe [the Netherlands (n = 781,85,102,123,139,140,145), the UK (n = 950,82,89,105–107,124,128,135), France (n = 298,130), Norway (n = 1111) and Ireland (n = 2126,127)], Israel (n = 1129), Asia (n = 1089,100,101,113,120–122,142–144), Australia/New Zealand (n = 6108,115,116,132,138,141) and South Africa (n = 2131,134). Primarily, the included studies were cross-sectional observational (n = 3181,82,98,100–102,104,106,109–111,114,116,119,121,124,126–135,137,139,142,143,145). The remaining studies involved retrospective database or medical records data (n = 2283–89,94,95,99,107,113,115,117,118,120,122,123,136,138,140,141) or secondary data analysis (n = 550,93,105,112,144).

All studies were published in the years 2000–15. The average mean age of participants was 42.8 years, with an average mean age range of 23.3–65.5 years. The average mean percentage of male participants was 52.4%. The number of people included in the studies ranged from 25 to 8911, with a mean of 824.

Risk of bias within studies

The funnel plots did not show any obvious asymmetry, and Egger’s test was not statistically significant for any of the outcome measures (specifically T2DM, t = –0.22; p = 0.84; ischaemic heart disease, t = –0.13; p = 0.91; cerebrovascular disease, t = 0.35; p = 0.58) (see Appendices 4–6, Figures 27–29, for funnel plots).

Results of individual studies and synthesis of results

Prevalence of type 2 diabetes

Figure 4 shows the individual studies reporting on the prevalence of T2DM and overall pooled prevalence. The prevalence estimates ranged from 2%84 to 13%. 90 The pooled prevalence of T2DM was 7.6%. The prevalence of any diabetes was 8.7%; this ranged from 2%84 to 11%95,102,117 (data not presented).

FIGURE 4.

Individual studies and pooled prevalence of T2DM. ES, effect size.

Prevalence of cardiovascular disease

Figure 5 shows the individual studies reporting on the prevalence of ischaemic heart disease. The prevalence estimates for ischaemic heart disease ranged from 0%140 to 12%. 126 The pooled prevalence of ischaemic heart disease was 3.7%.

FIGURE 5.

Individual studies and pooled prevalence of ischaemic heart disease. ES, effect size.

Similarly, Figure 6 shows the individual studies reporting on the prevalence of cerebrovascular disease. The estimates were fairly consistently in the < 1–4% range. The pooled prevalence of cerebrovascular disease was 2.2%. The pooled prevalence for undefined CVD was 10.6%, but ranged by individual study from 4%143 to 22%,114 reflecting the diverse case definitions.

FIGURE 6.

Individual studies and pooled prevalence of cerebrovascular disease. ES, effect size.

Risk of bias across studies

We found high heterogeneity in a number of outcomes when prevalence was pooled (see Table 4), as well as in a number of outcomes for the general population comparison (see Table 5). We further explored heterogeneity using metaregression (see below).

Summary of evidence

In this systematic review, we found that the prevalence of T2DM was 8% and of any diabetes was 9%. For CVD, the prevalences of ischaemic heart disease, cerebrovascular disease and undefined CVD were 4%, 2% and 10%, respectively.

The current prevalences of T2DM and CVD and associated risk factors in the population with ID were found to be similar to those in the general population. However, we found that ischaemic heart disease was significantly lower in the ID population. The metaregression showed that the method of data collection had minor effects on pooled diabetes and obesity. Mean age had minor effects on hypertension.

Strengths and limitations

A particular strength of this review is that we used robust methods. We wrote to authors to clarify and obtain additional data rather than excluding the articles. To our knowledge, this is the first systematic review and meta-analysis of prevalence of T2DM, CVD and associated risk factors in adults with ID. In addition, it is the first review of its kind to make comparisons with the general population.

However, we had limited data to separate T2DM from other diabetes, and we were sometimes restricted to unclear or poorly defined outcome measure definitions.

There were also limited data available to make comparisons with the general population. We would have benefited from additional general population data alongside the population with ID data to make more valid, generalisable comparisons.

Findings in relation to other studies

Two recent reviews75,76 have been conducted that have focused on diabetes prevalence among people with ID. The reviews found mean prevalences of 8.7%75 and 8.3%76 for combined gestational, type 1 diabetes mellitus and T2DM, respectively, but the reviews were unable to report on specific types of diabetes. The overall prevalence of CVD among people with ID is unclear. 52 However, our finding that ischaemic heart disease was significantly lower in the population with ID differs somewhat from the literature, which suggests that the prevalence of CVD among people with ID is greater than the general population. 52

Conclusions

Findings from the systematic review and meta-analysis presented in this chapter suggest that T2DM is at least as common in people with ID as in the general population. The findings also identify a gap in knowledge in relation to the prevalence of T2DM as many studies did not report this separately. In addition, none of the studies in our review reported on screen-detected T2DM in the population with ID.

Protocol and registration

The systematic review was registered with the International Prospective Register of Systematic Reviews (PROSPERO 2015: CRD42015020758). 156

Eligibility criteria

The review was guided by the PICOS model. 78 We defined the population as adults (aged ≥ 18 years) with ID (whole study population or a defined subsample). We defined the intervention as any multicomponent lifestyle behaviour change intervention aimed at primary prevention of T2DM or CVD, or a reduction in associated risk factors for people with ID and/or their carers. We included studies with and without comparison groups. We defined outcome measures as changes in anthropometric measures (weight, BMI, waist circumference), BP, lipid levels, glucose levels, physical activity levels, sedentary behaviour and dietary habits. The study design was defined as an experimental study [before-and-after study, randomised controlled trial (RCT) or non-RCT] with a follow-up period of at least 24 weeks or 6 months from baseline (to allow for the initiation and maintenance of medium- and longer-term behaviour change)157 (Table 6).

All studies that had been published on or after 1 January 2000 (until 21 April 2015) and in the English language were eligible. Studies were limited to those published in or after the year 2000, when most large diabetes prevention trials were first published in the general population. 70

We contacted lead authors for further information when inclusion/exclusion could not be determined.

Information sources

We searched the electronic databases EMBASE, MEDLINE, Cumulative Index to Nursing and Allied Health Literature, Cochrane Central Register of Controlled Trials and PsycINFO for this systematic review. The last date of the search was 21 April 2015. We searched the references lists of relevant systematic reviews and included papers within those for additional studies.

Search

We combined medical subject heading terms and key words for multicomponent lifestyle interventions and outcome measures and ID. The search was limited to English-language studies with cohorts of adults aged ≥ 18 years, depending on the database. Box 2 shows the MEDLINE search strategy.

Study selection

Full texts were identified after titles and abstracts were read separately by two investigators (TC and AD), who discussed discrepancies in selection at a later meeting.

After retrieval of the full-text articles, the papers were again examined separately by two investigators (TC and AD) to check their suitability for inclusion.

Data collection process

We created a data extraction form for this review. The data were extracted by one investigator (TC) and verified for accuracy by another investigator (AD).

Data items

For each study, we collected the first author’s name, title of paper, year of publication, country of the cohort, study design, sampling method, intervention details, dates of data collection and the intended recipient of the intervention. For the whole study population (and for each group, if applicable), we extracted data on total sample size or subpopulation size, mean ages, proportion of males/females, severity of ID, ethnicity and withdrawals.

For each reported outcome, we extracted information on how outcomes were defined and measured, the total number measured for each outcome, length of follow-up, mean baseline and post-intervention value, mean between-group change and/or change baseline to follow-up along with a measure of variability [standard deviation (SD), standard error (SE), etc.]. We extracted data separately for males and females, when reported.

Quality assessment

The NICE quality appraisal checklist for quantitative intervention studies158 was used to assess the quality of the selected studies. The checklist included criteria for assessing the internal and external validity of experimental and observational quantitative studies (RCTs, non-RCTs, before-and-after studies) and allowed assignment of an overall quality grade (categories ++, + or –). Studies were assessed by one reviewer (TC) and verified for accuracy by a second reviewer (RS).

Risk of bias in individual studies

Prior to carrying out this systematic review, we anticipated using funnel plots79 and the Egger’s test80 to examine potential publication bias in the literature for the collected outcomes. However, owing to the small number of studies included in this review resulting in low power to detect bias, these methods were not used.

Data synthesis

Data synthesis for this review involved describing the study characteristics (country, population size, age, percentage male, ethnicity, severity of ID, eligibility criteria, outcomes report and follow-up period) of the included articles. We then described the details and behavioural strategies of each of the multicomponent lifestyle behaviour change interventions, including their structure and delivery, and the underlying theory behind each of the interventions. Finally, we described the outcome measures and study findings. Given the low number of studies included, a formal evidence synthesis was not undertaken.

Study selection

The literature searches yielded 3508 articles. After duplicates were removed, 3167 articles remained to be screened. We retrieved and reviewed the full text of 39 articles for 32 studies (Figure 7). Most potentially relevant studies were noted to have small sample size, short follow-up, high attrition rates and/or incomplete data for key outcomes. We contacted the authors of four study protocols159–162 for further information. One of the authors did not reply, and we were informed by the remaining three that their study results were still awaiting publication and could not be included in the review. In total, we identified only four studies163–166 for inclusion in this review.

FIGURE 7.

Study selection.

Study characteristics

The four studies163–166 included in the systematic review presented data on 700 individuals. The characteristics of each of the studies163–166 are presented in Table 7.

The studies163–166 covered three countries (USA, UK and Sweden). Two studies163,164 were single arm, and two studies165,166 had a control group. All of the four studies163–166 reported data for physical activity and/or sedentary behaviour.

The mean age was 42.2 years and the mean percentage of male participants was 42.9%. The mean group size was 174 before dropout and 104 after dropout. Group sizes ranged from 54 to 443 before dropout, and from 44 to 196 after dropout. The majority of participants were white (68% where known); approximately one-quarter (26%) were black and the remaining 7% were from other ethnic groups. Only one study164 provided information on severity of ID, but, based on eligibility criteria, the remaining studies were likely to comprise adults in the mild to moderate ID range. Other descriptive information for each study is presented in Table 7.

Study quality

A breakdown of study quality is presented in Table 8. The studies163–166 were generally of high quality; in particular, all of the studies achieved at least a good quality rating for internal and external validity. However, two163,165 of the four studies failed to account for all of the participants when concluding the study, and three163–165 of the four studies did not report on whether or not the studies were sufficiently powered to detect differences.

Results of individual studies and descriptive data synthesis

Table 9 summarises the multicomponent lifestyle behaviour change interventions evaluated in the individual studies.

Bazzano et al. 163 conducted a single-arm before-and-after intervention in already overweight or obese individuals (BMI of ≥ 25 kg/m²). The intervention involved peer mentoring, one-to-one health education, supervised physical activity and clinical support aimed at reducing weight, improving diet and increasing physical activity.

Melville et al. 164 also conducted a single-arm study in already-obese individuals (BMI of ≥ 30 kg/m²) who had been referred to a dietitian by their GP. Nine lessons, every 2–3 weeks, were provided for participants and their carers. The lessons were aimed at increasing physical activity and better diet, as well as weight loss. Interventions also consisted of personalised diet plans with calorie restrictions [600 kilocalories (kcal) per day].

Bergström et al. 166 conducted a two-armed trial in community residential homes, targeting both people with ID and their carers. The intervention offered a ‘study circle’ for carers and also an appointed health ambassador at each residential home. An educational health course for the residents was also provided. The community residences in the control arm received the option to take part in the intervention after study completion (wait list control). The primary outcome for this trial was increasing physical activity; the secondary outcomes were decreasing weight and BMI.

Finally, McDermott et al. 165 conducted a two-arm RCT. Intervention participants were assigned to eight weekly lessons in nutrition, exercise and changing ways of thinking. The lessons focused on stress management, complications of obesity and behaviour management. The classes emphasised MVPA, healthy eating and BMI reduction. The control group was assigned to eight weekly lessons on safety and hygiene.

Table 10 summarises the components of the individual behaviour change interventions. All of the interventions used both dietary and exercise components.

Of the four included studies,163–166 the two single-arm studies,163,164 with follow-ups of 7 months163 and 24 weeks,164 indicated significantly improved outcomes; reductions in weight, BMI and waist circumference were demonstrated after the implementation of a behaviour change intervention programme aimed at increasing physical activity and improving diet. Additionally, both studies163,164 demonstrated a significant improvement in physical activity outcomes, specifically for ‘minutes per week’ and ‘frequency of sessions’163 and ‘reduction in sedentary behaviour’. 164 Both cohorts163,164 were overweight to obese when they were enrolled into the study. For the further two studies,165,166 for which the cohort was not recruited based on health status, one of the studies166 showed significant positive improvements in waist circumference, BMI and steps per day in those who received the intervention compared with the control subjects; the second study165 did not show any significant differences between control and intervention arms (Table 11). 165

Summary of evidence

This review contributes to the existing knowledge on the effectiveness of multicomponent lifestyle behaviour change interventions in adults with ID. Three of the interventions included in this review led to some reductions in BMI, weight and waist circumference,163,164,166 but inferences are limited owing to small sample sizes, missing data, selected populations and/or lack of control groups.

Strengths and limitations

To our knowledge, this is the first systematic review and meta-analysis focusing on long-term multicomponent behaviour change interventions for people with ID in order to reduce CVD and/or T2DM risk. We used robust methods and sought additional information from authors where relevant. However, only four papers163–166 met our inclusion criteria. Significant findings were observed only for the single-arm studies, which are known to overestimate effect sizes. 168 We were unable to test for publication bias or to carry out meta-analytical work to explore combined effects, particularly as the interventions that were evaluated were so diverse. Similarly, research has shown that improvements in health can be difficult to sustain in the longer term;169 only two of the included studies165,166 had a follow-up period of at least 12 months, and even this may not be enough to indicate long-term sustained benefits.

Findings in relation to other studies

In line with previous systematic reviews in this area,153–155 findings from this systematic review demonstrate a lack of quality evidence on the effectiveness of multicomponent behaviour change interventions in people with ID. In 2010, Jinks et al. 153 focused a systematic review on qualitative studies of behavioural change approaches in people with ID to aid weight loss and health. The review found 12 papers, of which only one was qualitative. The authors noted an overall lack of research on behavioural approaches and using qualitative methods. Similarly, in 2013, Spanos et al. 154 reviewed 22 papers that assessed interventions for weight loss in people with ID. They noted that many of the interventions did not meet the recommended duration in clinical guidelines and were too specific. Brooker et al. 155 also reviewed interventions with a primary focus on physical activity in people with ID. Again, the review noted small sample sizes and invalid measurement tools, and recommended further longer-term intervention studies.

Implications of findings

This systematic review informed the evidence base for the development of the STOP Diabetes educational programme, which is described in Chapters 8 and 9. The studies also revealed a high rate of missing follow-up data for participants who completed the multicomponent lifestyle behaviour change interventions, which helped to inform further development work on feasibility testing (see Chapter 10). The wider implications for research and practice are discussed in Chapter 13.

Selection of service users for involvement

Service users were approached from different sources to encourage a diverse range of views and to minimise burden. Members from the two self-advocacy groups approached the Speaking up for Health Group and the Charnwood Action Group, who agreed to help the team with the study. In addition, the manager and residents of a communal care establishment were approached through the lead nurse’s contacts, and they agreed to help us with the study. Service users who entered the poster competition (see next section) were indirectly involved by providing publicity materials for the team (Figure 8).

FIGURE 8.

Service users involved in the research programme. Charnwood logo from Charnwood Action Group Partnership Board.

Service user involvement in the programme management

A common way of involving service users in research is through representation in steering group meetings,197 and we discussed the potential for this with the service users. We considered tailoring these meetings to make them more accessible, but past experience suggested that they could be lengthy, involving complex discussions about procedures, accelerometer data, health economics and statistical methodologies, and often used conference call facilities. We were concerned that the meetings would be isolating for the service users and their supporters so, instead, we agreed to feed back key points from the meetings and that service users could attend on an ‘ad hoc’ basis.

Service user involvement in promoting the research programme

The study involved a Leicestershire-wide screening programme and it was important to promote the research as widely as possible. Among other considerations, the study logo and publicity materials needed to be suitable and appropriate for the target population. Both of the service user groups that we approached used ‘word police’ cards, which were shown whenever another member of the group or visitor used an acronym or abbreviation that they did not understand. Therefore, the proposed use of an acronym for the programme was not received favourably and, instead, the team opted to call the programme the ‘STOP Diabetes study’. The creative director subsequently devised four corresponding logos (Figure 9) and asked the Charnwood Action Group for their preference, using a feedback form with a scale and pictures. The preferred logo (see Figure 9, option 4) was shown to the Speaking up for Health Group, and there was discussion about ways in which the logo and other publicity materials, such as posters and fliers, could be used to publicise the programme. Service users recommended printing the STOP Diabetes logo on notepads, pens and fridge magnets. When directly asked, they also thought that the logo should be printed on the study documentation, such as information leaflets and consent forms. Members of the group also suggested holding a poster competition as a means of publicising the programme.

FIGURE 9.

Proposed logos for the STOP Diabetes programme. Option 4 was the service users’ preferred logo.

All of the ideas were collated and reported back to the research team for further discussion and to determine if sufficient resources were available to action them. All of the suggestions were taken up by the research team.

Service users were invited to enter the poster competition using brief easy-read information distributed by staff at local day centres, health clinics and other organisations. People who entered the competition were given a certificate and a small award of art and craft materials. Four of the pictures were chosen for the promotional materials (Figure 10); this decision was made by both service users and members of the research team.

FIGURE 10.

Some of the artwork submitted for the poster competition. The four pictures on the left were used for the publicity materials.

The research programme was publicised via the Leicestershire Diabetes Centre website and was also published in the National Institute for Health Research INVOLVE Summer 2013 newsletter,198 both as means of raising awareness about the programme and sharing our experiences of service user involvement (Figure 11). Similarly, having attended one of the programme steering group meetings, one of the co-chairpersons of the Charnwood Action Group contacted the media and was consequently interviewed about the programme (radio and newspaper199), which helped with recruitment.

FIGURE 11.

Service users’ involvement in promoting the research programme. Image of article ‘Volunteers needed for study’ reproduced with permission of Leicester Mercury. Copyright © 2016 Local World. All rights reserved. Image of article ‘STOP Diabetes study’ reproduced with permission of INVOLVE. Copyright © INVOLVE. All rights reserved 2015.

Service user involvement in study documentation and process development

It was important that information about the programme should be available in simple language, free from jargon and acronyms, and using pictures and symbols, so that potential participants had every opportunity to understand what the research team were doing and reach an informed decision about whether or not to take part. Prior to submission to the research ethics committee, the research team, with support from local ID services, drafted an easy-read (symbols and words) information sheet and consent form to partnership boards, local ID services and service user groups for feedback on the symbols, text and whether or not additional communication aids might be necessary.

The service users who we asked to read the easy-read documents did not report any difficulties in understanding them, but recommended additional modes of communication, such as flash cards and story cards. A member of the Speaking up for Health Group assisted the team by taking photographs to illustrate the information leaflet, flash cards and story cards; three service users modelled for the photographs (Figure 12).

FIGURE 12.

Service users’ involvement in assisting with study documentation.

Service user involvement in staff recruitment

Research nurses were integral to successful recruitment to the study, needing to be patient, sensitive and responsive to participants’ needs (e.g. seeing participants outside typical working hours), as well as able to communicate effectively with people with ID. Two members from the Speaking up for Health Group offered to help with the interview process for recruiting nurses into the research programme. Supported by their group facilitator, they created two questions to assess how good the nurses were at communicating with people with ID, and how they might adapt their style of communication if that person did not understand them. On the day that the nurses were interviewed, the service users asked these questions in a separate room, with the facilitator present. They then rated the nurses’ responses on a 4-point scale (Figure 13). Their input helped to reinforce the panel’s decision on who to recruit and was particularly valuable in helping the panel to decide between two similar applicants.

FIGURE 13.

Service users’ rating form for recruitment of nurses.

Service user involvement in training staff and assessing acceptability of measures

Service users at the participating communal care establishment helped to train staff by allowing them to practise communication-based interactions, consent-taking and measurement collection. Service users from one of the self-advocacy groups were invited to attend a follow-on staff training session so that nurses could put their new skills into practice. They gave feedback on nurses’ skills and discussed what they liked and what they did not like, enabling staff to gain confidence and develop competency in various procedures. The mock clinics also helped the research team to determine how long the appointments might take and how many visits may be needed. The service users reported that some of the questionnaires were too lengthy and complex. The team met to discuss the issues raised and made changes to reduce participant burden: these included swapping one of the questionnaires [Psychiatric Assessment Schedules for Adults with Developmental Disabilities (PAS-ADD) mini200 for the PAS-ADD checklist200] and removing two questionnaires (Dietary Instrument for Nutrition Education201 and International Physical Activity Questionnaire202) entirely.

For the research programme, and to help with the design of the anticipated future trial, two members of the Speaking up for Health Group and one member of the Charnwood Action Group wore the activity monitors (both wrist- and waist-worn monitors) and provided feedback on their ease of use.

Inclusion and exclusion criteria

Participant recruitment process

Eligible participants were invited to take part in the screening programme using a four-pronged approach (summarised in Figure 14):

1. approach via general practice registers

2. approach via the LLDR

3. approach via specialist ID psychiatric service clinics

4. direct contact with the research team.

FIGURE 14.

Recruitment pathway to screening programme. LDR, learning disability register.

Informed consent

At the participant’s first appointment, a final face-to-face capacity assessment was undertaken by a trained ID research nurse and informed consent was obtained; appropriate mental capacity legislation (see Chapter 1, Adherence to mental capacity legislation) was followed. Consent was taken only when it had been established that the person understood the consent form and information sheet, and that they had been given the opportunity to ask questions.

People with capacity to consent were asked to sign a consent form. For those who could decide, but were unable to read, the consent form was read to them in the presence of an independent witness. For people who did not have capacity to consent, an appropriate consultee was identified and consulted about the person’s potential participation. The consultee was asked to sign a consultee declaration form confirming that they had been consulted, had their questions answered and had considered the study from the participant’s perspective.

The participant and/or personal/nominated consultee (if appropriate) were asked to confirm that:

• they understood the study and were happy with what taking part would mean for them

• they understood that they could withdraw from the study at any time, without giving a reason (and that this would not affect their care)

• they had been given a chance to discuss their questions with the research team

• they agreed for their GP to be notified about their participation and of their screening results

• the research team could access their medical records or records held at their residential home or day centre for additional information, if unable to obtain from the participant or carer

• relevant sections of their medical notes and/or study data could be looked at by responsible people/regulatory authorities for purposes of auditing the research.

Participants provided their consent for screening to be undertaken (see Informed consent and Box 3), including a blood test (if the participant was willing). Additional optional consent items that participants could chose to agree to, or not, included:

• being contacted to take part in further phases of the study if they screened positive for IGR or high risk of developing T2DM (based on elevated BMI)

• having an additional blood sample taken for storage and future anonymised genetic analyses

• allowing access to medical records for long-term follow-up

• contact details being stored by the research team so that participants could be informed of the study findings and be contacted about future research studies.

At the end of the appointment, the nurse highlighted the office’s telephone number on the participant information sheet, which could be used if the participant decided to withdraw from the study or had any queries.

At the start of any subsequent appointments, participants’ retention and understanding of the study was reconfirmed.

Following the final study appointment, a photocopy of the signed consent form/consultee advice form (as appropriate) was sent to both the participant and the general practice; the copy of the form accompanied the screening results, which were subsequently sent (see Informing of screening results). The original consent and advice forms were retained at the research offices. See Appendices 12–14 for examples of the consent/advice forms.

Data collection

Data collection was usually undertaken over two appointments but could be longer (the maximum was five). The data collection process is summarised in Box 3. All of the data were collected in a standardised way by specially trained research nurses, following study-specific standard operating procedures. Full details of the assessment of outcomes are described below (see Assessment of outcomes).

Anthropometric measurements, BP and demographic and lifestyle data were frequently obtained at the first appointment, after consent was obtained, and usually took between 1 hour and 90 minutes. Questionnaires were completed via interview during the initial screening visit (or at a subsequent appointment) or were given to carers to be completed outside the appointment, as applicable (see Box 3). These typically took between 30 and 60 minutes to complete. Venous blood samples were usually taken during a separate appointment after deciding with participants (and their carers, where relevant) whether a fasting or non-fasting sample would be more appropriate; this decision was based on potential behavioural difficulties and/or cognitive understanding of participants. This appointment lasted about 30 minutes. Medical history and prescribed medication were collected during screening or at a later date from medical records. Other additional information was extracted by a researcher from the LLDR or from records held at residential homes or day centres.

Informing of screening results

All participants were informed of their key biomedical screening results in an easy-read format, supplemented by verbal explanations as appropriate. Anthropometric measures and BP readings were presented to participants on the day that they were taken. Results of blood tests taken to determine diabetes status were provided within 7–10 days.

Participants with normal results were informed by post and given the option to contact the research team and discuss further if they wished. For participants who were screen positive for IGR or T2DM, a research nurse telephoned them to explain their results and answer any questions, prior to a letter being sent in the post. In some cases, this also involved a face-to-face visit by the nurse to support the participant and/or their carer. In accordance with consent taken, these participants were then referred to their general practice for usual care.

As agreed at the time of consent, participants’ GPs were provided with full details of the screening results, including diabetes status. Additionally, for all of the participants who were identified as meeting the criteria for IGR or T2DM, a member of the research team contacted their general practice and informed their GP, prior to any results letters being sent.

See Appendices 15 and 16 for example letters that were used to inform participants and GPs of the results.

Primary and secondary outcomes

The primary outcomes for the screening study were the prevalence of T2DM, IGR and abnormal (T2DM or IGR) blood glucose level.

Diagnosis of T2DM was made following the most recent WHO criteria,28 more specifically a HbA_1c level of ≥ 48 mmol/l or 6.5%. IGR was defined as impaired fasting glucose, following the WHO criteria of a HbA_1c level of 42–47 mmol/l or 6.0–6.4% (Figure 15).

FIGURE 15.

Diagnosis of T2DM and IGR for participants in the screening programme. FPG, fasting plasma glucose.

The secondary outcomes included:

• physical activity levels, including sedentary behaviour, measured by brief questions and accelerometer (for a small subgroup)

• lipid levels [triglycerides, total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol]

• BP (systolic, diastolic)

• cardiovascular risk, as measured by the Framingham Risk Score212,213

• health-related quality of life, as measured by the EuroQol-5 Dimensions (EQ-5D) questionnaire214

• dietary/nutritional intake (food groups and fruit and vegetable intake)

• behavioural disorders, as measured by the Aberrant Behaviour Checklist (ABC)215,216

• psychiatric disorders, as measured by the PAS-ADD checklist217

• depression, as measured using the Glasgow Depression Scale (GDS) and Carer Supplement. 218

Assessment of outcomes

All blood and urine samples were analysed at by the University Hospitals of Leicester NHS Trust laboratory services, using stable methodology standardised to external quality assurance reference values. HbA_1c level was measured using an ARKRAY ADAMS HA-8180T analyser (ARKRAY, Kyoto, Japan). Plasma glucose level (fasting and non-fasting); serum total cholesterol, HDL cholesterol and triglycerides, and urine albumin and creatinine were measured using a Siemens Advia 2400 analyser (Siemens Healthcare Diagnostics, Camberley, UK). The Friedewald equation was used to estimate LDL cholesterol. 219

Resting BP was assessed in a seated position on the brachial artery, using an Omron M5-I automatic BP monitor (Omron Healthcare UK, Milton Keynes, UK); a series of three measurements was recorded, with a mean value calculated from the final two. Waist circumference was measured to the nearest millimetre over minimal clothing, midway between the costal margin and the iliac crest, and in the mid-axillary line; hip circumference was measured to the nearest millimetre at the widest point over the buttocks; a soft tape was used for both anthropometric measures (WM02 Body Tape; Chasmors Ltd, London, UK). Weight was assessed in light clothing and no shoes to the nearest 0.1 kg, using a seca 875 digital floor scale (seca, Birmingham, UK); and height to the nearest centimetre using a Leicester portable height measure (Chasmors Ltd, London, UK) and with head placed in the Frankfurt plane.

Additional data on health-related quality of life (EQ-5D)214 and depression (GDS and Carer Supplement),218 were collected face to face via interview-administered questionnaires at an appointment. To assess problem behaviour (ABC)215,216 and psychiatric disorders (PAS-ADD checklist),217 questionnaires were taken away by carers and self-completed following the appointment. The validated questionnaires used are described in detail in Appendix 17. Deprivation was assessed according to the 2015 Index of Multiple Deprivation. 220

Ambulatory activity and sedentary behaviour were measured for a small subsample of participants; full details of the physical activity substudy undertaken are presented in Chapter 7.

Uptake of screening was measured by recording the number of (1) invitations sent, (2) people responding and refusing at each stage and (3) people attending for screening.

If BP, anthropometric measures and/or bloods were unable to be assessed, then details of the reason were recorded (refused, physical/behavioural difficulty, equipment error, other). For demographic, lifestyle, medical history and prescribed medication, additional details were recorded on how the data were obtained, for example from the volunteer, carer/relative or a combination of both, or if personalised records such as a health action plan4 were used.

Feasibility of diabetes screening in adults with intellectual disability

The feasibility of conducting a diabetes screening programme in a community setting for adults with ID was assessed using a flow diagram of the screening process and summarising the number of dropouts and those for whom data were unobtainable at each step of the screening process. Particular outcomes of interest in terms of the feasibility are the proportions of people who (1) were invited and who complete the screening programme (including the blood tests) and (2) attended the screening session but did not have a blood test. We also assessed the completeness of key data items from the case report form and questionnaire to assess the feasibility of data collection for future research projects in this group.

Characteristics

The characteristics of those who were screened were summarised using means (SDs for continuous variables) and ‘n (%)’ for categorical variables.

Additional analyses were conducted to compare the representativeness of the STOP Diabetes study cohort with the LLDR. 14

Prevalence of type 2 diabetes and impaired glucose regulation

The overall prevalence of IGR, T2DM and any abnormal glucose regulation was calculated with 95% CI.

Cardiovascular risk

Cardiovascular risk was calculated for participants aged 35–75 years with no previous history of CVD. The Framingham CVD risk score212,213 was used to assess risk in white European participants and ETHRISK (a modified version for British black and minority ethnic groups) was used for South Asians. 221 Participants with incomplete data for key variables (total and HDL cholesterol, systolic BP, smoking status) were unable to be included in analyses. The overall mean risk at 10 years and the level of risk (high, intermediate, low), based on thresholds determined by National Cholesterol Education Program,222 were calculated.

Factors associated with abnormal glucose regulation

Logistic regression was used to assess the association between key biomedical and anthropometric characteristics and the outcome – abnormal glucose regulation. ORs and 95% CIs were calculated.

Validation of Leicestershire self-assessment risk score

Our initial analysis plan was to update the Leicester Self-Assessment risk score,58 described in Chapter 1 (see Risk scores for the early identification of impaired glucose regulation and type 2 diabetes), for use in a population with ID. This may have included adding or removing risk factors and updating the relative weighting given to risk factors. However, given the low prevalence of IGR/T2DM that we found in our screening study (see Prevalence of type 2 diabetes and impaired glucose regulation), this was not considered feasible. There are no formal sample size requirements for developing risk scores, although it has been suggested that data sets that are used to develop risk scores should contain between 10 and 20 events for each risk factor being assessed. 223,224 Therefore, our data set would be very underpowered to develop a risk score.

Hence, it was decided that instead of updating the original Leicester Self-Assessment risk score, alternatively, we would assess the risk score’s performance to detect undiagnosed IGR/T2DM. Although this validation would also be underpowered (studies suggest that external validation data sets should have at least 100 events and 100 non-events),225 this analysis should provide some preliminary results to suggest if the score is sensitive and specific in our cohort with ID.

The Leicester Self-Assessment risk score contains seven risk factors (age, sex, ethnicity, family history of diabetes, waist circumference, BMI and high BP). 58 To maximise the number of people included in the analysis, the data were analysed in two ways: (1) complete case basis (including only those with all seven risk factors recorded and the outcome) and (2) imputing missing data for family history of diabetes and high BP. For both a family history of diabetes and high BP, the imputed data set assumed a negative response if these items were missing. In both data sets the sensitivity, specificity, positive predictive value and negative predictive value were calculated for a cut-off point of ≥ 16 points. This is the cut-off point that was used in the general population for invitation to screening. 58

All of the analysis was conducted using Stata; statistical significance related to p < 0.05 and 95% CIs are presented throughout.

Participant recruitment

Initial approach

Participants were recruited to the STOP Diabetes screening study between February 2013 and September 2015. In total, 3201 adults with ID were invited to take part via the four different routes (Figure 16).

FIGURE 16.

Invitation flow chart. LDR, learning disability register.

Fifty-one per cent (n = 73) of all general practices in Leicester City, Leicestershire and Rutland (with adults with ID on their practice list) agreed to be involved with the study. Subsequently, 1736 potentially eligible people were identified and sent an invitation letter by their practice (median 19, range 3–116). People who were invited via this route accounted for the majority of study invitations (54%).

For practices refusing, 1595 people were identified for possible approach via the LLDR. Of these, contact details for 418 people (who had previously agreed to be contacted for research purposes) were passed directly to the research team so that they could be invited (13% of the overall study total). A further 864 (27%) people were invited directly by the principal investigator in LPT.

A much smaller proportion of people were invited via specialist ID psychiatric service clinics or after making direct contact with the research team: 2% (n = 52) and 4% (n = 131), respectively.

Full-stage invitation

From the initial invitation, approximately 30% of people refused, 29% were classed as non-responders and 40% expressed an interest in participating in the study (Figure 17). Following a preliminary assessment of volunteers’ decision-making capacity, 1209 individuals (38% of those initially invited) were then provided with full study information (postal invitation or a face-to-face visit). Subsequently, 984 (31%) proceeded to the screening stage.

FIGURE 17.

Recruitment.

For people who refused (or agreed) at the initial or full invitation stages, details relating to the method of recruitment and reasons for refusal are presented in Table 12.

TABLE 12 - Responses (%) according to recruitment method at initial and full invitation stages

Estimates provided were appropriate for refusals; percentages are rounded, so may not add up to 100%.

Characteristics and method	Stage
	Initial invitation (or chasing non-responders)		Full invitation (or capacity 1 or 2)
	Refuse	Agree	Refuse	Agree
Total number	N = 918	N = 1259	N = 233	N = 984
Male, n	53a	58	62a	58
Age in years, mean	–	43	40a	44
Resided in Leicester City	40a	41	48a	40
Recruitment method, n
GP	50	47	67	41
Learning disability register	33	18	12	19
Previous consent to research	15	19	17	20
Direct invite	1	12	2	14
Psychiatrist clinic	1	5	3	6
Refusal/acceptance method, n
Reply slip	28	28	24	28
Telephone call	14	20	21	20
Chasing person via telephone	55	33	55	28
GP notified team	2	0	0	0
In person	1	18	1	24
Via e-mail	0	1	0	1
Reason for refusing, n
Not known	77	–	72	–
Behavioural issues	7	–	6	–
Carer would not agree consent	7	–	5	–
Health issues	3	–	4	–
Recent health check	3	–	6	–
Too busy	2	–	2	–
Other	1	–	6	–

Screening: consent and data collection

At the consent stage, 930 people (29% of those originally approached) agreed to participate and were recruited to the screening study; 54 people either refused (n = 19) or were ineligible (n = 35) (see Figure 17). Thirty-eight per cent of participants (n = 350) were able to consent for themselves; the others required a nominated (39%) or personal (23%) consultee.

The availability of data for the key screening outcomes is presented in Table 13. The full details regarding the availability of data for all study variables are reported in Appendix 18, Table 61. Anthropometric measures and BP were obtained for most participants: approximately 86% and 89%, respectively. In the majority of cases, the documented reason for not obtaining anthropometric measures was physical or behavioural difficulties; for BP, the main reason was participant refusal.

A high proportion of participants agreed to attend for a blood appointment (n = 825); 700 (75% of those recruited) proceeded to have a blood test, and bloods to allow screening were successfully obtained for 648 (70%). For a few additional participants, when a blood test was refused or a sample was not obtained, recent results were available from their medical records (HbA_1c test, n = 27; for other tests the number varies) (see Figure 17 and Table 13). For a further five participants, HbA_1c results were not included because of potential unreliability in assessing diabetes status (n = 4) and poor kidney function (n = 1, possible haemoglobin variant). Therefore, we were able to assess diabetes status for a total of 675 participants.

Validated questionnaires administered via interview were successfully completed for a high number of participants (EQ-5D ≈94%; GDS or Carer Supplement ≈85%). Carer completion of questionnaires outside the appointment (for the ≈80% of participants who had an identified carer) was less successful; approximately 45% of carers completed the ABC and/or the PAS-ADD.

Demographic characteristics

The mean age of those screened was 43.3 (SD 14.2) years, 58% were male and the majority were of white ethnicity (80%) (see Table 14).

Most participants lived either with family (36%) or in a residential/nursing home (38%), with 6% living alone. A high proportion required 24-hour support (71%) and only 7% reported to be independent.

The majority of individuals was able to access the community to undertake regular daytime activities. Common activities included attending college (18%), voluntary work (16%) or involvement in service user/advocacy meetings (13%). Around half of the participants attended day opportunities/day placements. Only a small number of people were in regular paid employment (8%).

Anthropometric and biomedical measures

Among those screened, the mean waist size was 100.4 (SD 16.5) cm, weight 76.4 (SD 20.8) kg and BMI 28.7 (SD 7.1) kg/m² (see Table 15). Based on their BMI, 31% of participants were classed as overweight and 37% obese. Mean values for systolic and diastolic BP were 121.4 (SD 16.9) mmHg and 78.2 (SD 11.1) mmHg, respectively.

Among participants for whom blood results were available, the mean HbA_1c level was 35.0 mmol/mol (SD 5.1 mmol/mol) (5.3%; SD 1.5%), FPG was 4.7 mmol/mol (SD 0.7 mmol/mol) and non-FPG was 5.3 mmol/mol (SD 1.5 mmol/mol). For lipids, mean total cholesterol was 4.9 mmol/mol (SD 1.0 mmol/mol), HDL cholesterol was 1.3 mmol/mol (SD 0.4 mmol/mol), LDL cholesterol was 2.9 mmol/mol (SD 0.9 mmol/mol) and triglycerides were 1.4 mmol/mol (SD 0.9 mmol/mol).

Current medication and medical history

When details on severity of ID were available (n = 816, 88%), similar proportions of participants were classified as mild, moderate or severe (≈30% each), and 4% were classified as profound ID (see Table 16). Most participants had no confirmed diagnosis or identified cause of their ID (≈70%); when the causes were known, the most common were Down syndrome (n = 133, 14%) and cerebral palsy (n = 58, 6%).

The overall prevalence of existing CVD was 2% (n = 19). A history of stroke was reported for 12 (1.3%) people and coronary heart disease (CHD) for six (0.6%) people, and one person had a history of both conditions.

Congenital heart disease (2%) and other heart problems (2%) were less frequently reported.

Seventy-four participants (8%) had a history of high cholesterol and/or were prescribed a lipid-lowering medication, 85 (9%) had a history of previously diagnosed hypertension and/or were prescribed an antihypertensive drug, and 36 (4%) were prescribed an anti-thrombotic drug. A minority of participants were either current smokers (8%) or ex-smokers (4%).

When known, approximately one-third of participants had a first-degree family history of diabetes. Only one participant reported a previous diagnosis of pre-diabetes and one reported polycystic ovary syndrome. Nine per cent were currently being prescribed a steroid medication (the majority of these medications were inhaled).

Overall, the most commonly reported diagnosed physical health problems were epilepsy (n = 262, 28%), hypothyroidism (n = 93, 10%) and chronic breathing problems (n = 88, 9%). Thirteen per cent of participants had no significant medical history and 19% were not currently prescribed any medication.

For mental health-related problems, 152 participants (16%) had a history of a mood spectrum disorder (ICD-10 codes F30–F39), 35 (4%) a psychotic spectrum disorder (ICD-10 codes F20–F29) and 52 (6%) had a history of both; 143 people (15%) had neurotic, stress-related or somatoform disorders (ICD-10 codes F40–F48). Additionally, 28% of participants were prescribed antipsychotic medication and 32% were prescribed depression- or anxiety-related medication. Other frequently reported problems included autistic spectrum disorders (18%) and a recognised behavioural problem (14%).

When comorbidities (two or more diagnosed health problems) were considered, 121 (13%) participants had co-occurring physical health problems, 182 (20%) had co-occurring mental health problems and 286 (31%) had multiple physical and/or mental health problems.

Lifestyle and well-being

Eighty-five per cent of those screened were able to walk independently (without the help/support of another person), but including 6% who required a walking aid (see Table 17). The data reported directly by participants and/or carers indicated that most people did at least ‘some’ walking on a typical day, but only 25% achieved ‘a lot’ of walking. Additionally, around half of the participants reported spending ‘a lot’ or ‘most/all’ of the day sitting.

Sport/exercise or other physical activities that individuals reported undertaking in a typical week included dance (25%), swimming (20%) or walking (21%). Around half of the participants reported doing housework (such as dusting/hovering) and ≈20% gardening. A small number of people (7%) did regular chair-based exercise.

Problems with eating and drinking were reported for some people: 24% had difficulties in chewing or swallowing and 13% needed help to feed themselves (< 1% were tube fed). For food shopping and preparation, overall ≈35% of participants did their own food shopping (either independently or with some support) and a similar number were able to prepare at least simple hot and cold food (with or without supervision). Reported daily intake of fruit, vegetables or salad indicated that only around 30% of participants were eating the recommended five or more portions a day.

When questionnaire data were available, the proportion of participants identified with possible depression (using a cut-off point of 13) by the GDS or GDS Carer Supplement was 22% and 16%, respectively. 218 For health-related quality of life, the mean EQ-5D descriptive score was 0.8 (SD 0.3) and for the visual analogue scale was 78.1 (SD 19.4). The mean scores for the five problem behaviour subscales measured by the ABC (for participants with carers) were ≈4 for irritability, lethargy and hyperactivity, and ≈1 for stereotyped behaviour and inappropriate speech. The prevalences of mental health problems for organic, affective/neurotic and psychotic disorders (as measured by PAS-ADD checklist) were 6%, 9% and 5%, respectively.

Comparison with the Leicestershire Learning Disability Register

Participant demographic characteristics from this study were compared with adults with ID on the Leicester Learning Disability Register. Comparison of age, sex and ethnicity suggests that the STOP Diabetes cohort is a representative sample of the population with ID, known to services within the Leicester, Leicestershire and Rutland area (Table 18).

Summary of main findings

Utilising a variety of approaches to identify/invite potential volunteers, 930 adults with ID (29% of those approached) participated in the screening programme; 38% were able to consent for themselves, whereas other participants required a consultee. Anthropometric measures (≈86%) and BP (89%) were obtained for most participants. A high proportion of participants agreed to attend for a blood test and, subsequently, prevalence of T2DM/IGR was assessed for 675 participants (73%).

The mean age of participants was 43.3 years, 58% were male and the majority were of white ethnicity (80%). Most lived either with family (36%) or in a residential/nursing home (38%); a high proportion required 24-hour support (71%). Most participants were either overweight or obese; 2% had a history of existing CVD.

Screening results indicated the overall prevalence of undiagnosed T2DM was 1.3% (95% CI 0.5% to 2%) and IGR was 5.2% (95% CI 4% to 7%). Participants of non-white ethnicity were almost four times more likely to have abnormal glucose levels than white European participants; those with a first-degree family history of diabetes were over three times more likely.

Comparison with previous evidence

The prevalence of previously undiagnosed T2DM detected in the screening programme is much lower than previously reported. 90 Combined evidence from other studies, as presented in the meta-analysis in Chapter 2, suggests a prevalence rate of 8% for T2DM in adults with ID. Data to enable comparison of rates for T2DM in the UK population with ID are scarce (it is suggested that 85–90% of diabetes is T2DM20). Current estimated prevalence of diabetes (type not specified) in England, based on combined data reported by partnership boards, is 6.8% (range 6.2–8.4%) for people with ID of any age. 51 Based on current data supplied by 40 (55%) of the general practices that took part in the STOP Diabetes study, the suggested prevalence of diagnosed diabetes (type not specified) locally is 9.5% (n = 148 of 1553 adults aged 18–74 years with ID).

The estimates above are based on previously diagnosed diabetes. Our study aimed to screen adults with ID to identify undiagnosed T2DM. The rates suggested by data supplied by local general practices, alongside the higher recorded uptake of health checks locally (57–66% across the three CCGs)226 than the national average (44%),55 suggests that at a local level the lower rate may simply reflect a successful annual health checks programme. In the general population, estimated prevalence of diabetes rises from 6.2% to 8.0% when including undiagnosed cases. 18 However, it is acknowledged that the proportion of adults with ID who currently have bloods checked, including for diabetes, as part of their annual health check is unclear.

Strengths and limitations

To our knowledge, this is the first diabetes screening study that has been conducted in adults aged 18–74 years with mild to severe/profound ID. The successful integration of a multidisciplinary team, consisting of experienced researchers and ID HCPs, enabled the successful development and conduct of the STOP Diabetes screening programme. This multidisciplinary approach allowed for sharing of knowledge and best practice, and was complemented by service user involvement, particularly in the early stages of developing and trialling study procedures/processes.

The screening programme developed utilised robust methods. All of the data were collected by staff who had undertaken study-specific training and were following standard operating procedures. Minimal exclusion criteria were applied to the study, and reasonable adjustments to facilitate inclusion – such as staged invitation, easy-read documents, flexible appointments and carer involvement – maximised participation. This ensured that as many people as possible participated rather than being arbitrarily excluded. Additionally, we applied a staged approach to invitation and made efforts to contact/chase all people when possible.

It is acknowledged that we were unable to establish any contact with approximately 30% of people who were non-responders. We therefore do not know if they are different in any way from those who were included in the screening programme; evidence suggests that people with mild ID may be at increased risk as a result of unhealthier lifestyles and are less likely to access services. 11 However, similarities in the demographic characteristics (age, sex, ethnicity) between participants in this study and adults with ID on the Leicester Learning Disability Register suggest that the STOP Diabetes cohort is a representative sample of the population with ID that is known to services within the Leicester, Leicestershire and Rutland area.

The validation of the Leicester Self-Assessment risk score in the population with ID was successful despite the limited number of events and wide 95% CI. Estimates suggest that the Leicester Self-Assessment risk score works as well in populations with ID as in the general population: sensitivity 81.8%. Based on this, 140 participants would be referred for unnecessary screening. However, the tool is designed for use in a multistage screening programme and we would rather send more people through the first stage than falsely reassure.

Implications for clinical practice and future research

The screening uptake of those approached, at 29%, was relatively low, but it was favourable compared with two previous screening/prevention studies conducted locally in the general population, for which 22% and 19% of those invited took part. 227 These relatively low rates of uptake might reflect the fact that participants were invited to screening as part of a research project. If the intervention was rolled out in clinical practice, higher rates would be expected; for example, uptake rates to the NHS Health Check Programme,228 which is not a research project, are double those reported in this and other research screening studies. Future research should focus on increasing uptake to screening in all groups.

Bloods to enable diabetes screening were successfully obtained for a high proportion of participants. However, future research may want to consider allowing for separate consent for blood tests so as to not deter people at the initial recruitment stage. Very few people directly expressed ‘the blood test’ as a reason for refusal to participate in the screening study, but anecdotal evidence suggests that this may have deterred some. Alternatively, a staged approach to screening, involving risk stratification as recommended by NICE, might be considered. 27

Our findings suggest that the Leicester Self-Assessment risk score is statistically effective at identifying people with ID who are at a high risk of undetected IGR/T2DM. However, the feasibility of using it in practice with people with ID – given the levels of heterogeneity within the population with ID – needs to be considered. It may not be practical or acceptable for people with ID to calculate their own score, with or without added support from carers. Future research could involve developing an easy-read version (plus a carer supplement) and additional supportive material/communication aids, such as digital audio/visual materials; qualitative research would be needed to supplement this work. Alternatively, a better way may be to integrate the risk score at practice level and incorporate it into the Learning Disability Health Check (the learning disability Annual Health Check scheme).

Participants

Participants who met the eligibility criteria, as outlined below, were asked to wear an accelerometer as part of the main screening component of WP1.

Participant recruitment process

Initial assessment of eligibility to participate in the physical activity substudy commenced during the capacity assessment process (outlined in Chapter 5) and was subsequently confirmed once consent to the main screening study had been obtained. Eligible participants were then approached about wearing an accelerometer. For most people, this was usually at the end of their first screening appointment.

Data collection

Participants were asked to wear an accelerometer for 7 consecutive days, not including the appointment day. The procedure for wearing the accelerometer was explained to the participant and/or carer by an ID research nurse. Participants were also provided with a brief accelerometer information leaflet/diary in an easy-read format, which explained how to use the accelerometer; this diary was also used to log when participants had worn the accelerometer, with a page for each of the 7 days.

After wearing the accelerometer, participants were asked to return it at their next appointment. If a participant was not having another planned appointment, a member of the research team would contact them to arrange a convenient time for the accelerometer to be collected. If an accelerometer was not returned or unsuccessfully collected, the research team made repeated attempts (at least three) to try and retrieve it.

Two different accelerometers were used to collect data. Initially, physical activity data were recorded using a waist-worn accelerometer. Later, it was decided to also trial a wrist-worn accelerometer, given the poor compliance that was emerging with the waist-worn device (see Results) and following discussion with service users (who were assisting with patient and public involvement activities). The wrist-worn accelerometer was anticipated to encourage greater compliance, as it is waterproof and can also be worn when sleeping; therefore, participants could wear it continuously over the 7-day period.

Full details of the assessment of outcomes are described in Assessment of physical activity outcomes.

Outcomes

Physical activity levels were included as one of the secondary outcomes for the main screening study (see Chapter 5). Other anthropometric (BMI, waist circumference), demographic (ethnicity and social deprivation) and biochemical (fasting blood glucose and HbA_1c levels) outcomes assessed are described in Chapter 5.

Sample size

Initially, we aimed to include at least 50 participants who were wearing the waist-worn accelerometer. This was updated to include a comparable number with the wrist-worn device.

Assessment of physical activity outcomes

The participants who were attending screening were offered the option of having their physical activity levels assessed by a waist-worn accelerometer. Once we had achieved our initial aim of at least 50 individuals with data, the remaining cohort were offered an alternative wrist-worn accelerometer. Details of the two accelerometers and analytical methods used are presented below.

Inclusion of physical activity data

Physical activity data were included from a device if there was a minimum of 8 hours’ wear per day for at least 3 days.

Data analysis

Data are presented as means (SDs). Analysis of covariance models were used to compare differences in levels of assessed physical activity between monitors, adjusted for age, sex, social deprivation and wear time.

Feasibility of using accelerometers to assess physical activity in adults with intellectual disabilities

Participants were recruited to take part in the physical activity substudy between October 2013 and August 2015 (Figure 19).

FIGURE 19.

Collection of accelerometer data. (Data validity was based on a minimum of 8 hours’ wear per day for at least 3 days.)

Overall, 203 participants were approached to wear the ActiGraph waist-worn accelerometer. Subsequently, 97 participants (48%) agreed to wear the ActiGraph, and valid data (≥ 8 hours per day for 3 days) were obtained for 55 participants (57%). Reasons for attrition included 14 participants (14%) not returning their accelerometer and 28 participants (29%) not having enough valid days of wear for analysis.

A total of 76 participants were asked to wear the GENEActiv wrist-worn accelerometer and 47 participants (62%) agreed. Valid data were obtained for 39 participants (83%). Two individuals (4%) did not return their accelerometer and six participants did not have enough valid days of wear (13%).

Characteristics of participants in physical activity substudy

The characteristics of those who agreed to wear an accelerometer, on either the wrist or the waist, and those with valid physical activity data stratified by accelerometer type are shown in Table 24. Characteristics were similar between those who had valid physical activity data and those who did not. Characteristics were also similar between those who had valid waist-worn and and those who had wrist-worn accelerometer data. Overall, 54% of participants were male, their mean age was 39.9 (SD 13.0) years and 85% were of white ethnicity. Thirteen per cent lived alone, 42% lived in supported living and 46% lived with family; the majority (88%) had support from a carer for at least some of the time.

Main findings

The estimates of time spent in MVPA, light-intensity physical activity time and sedentary time are presented across the different monitors and physical methods used (Table 25). Estimates for MVPA and sedentary time were significantly higher with the wrist-worn device, whereas estimates of light-intensity physical activity were lower. The total physical activity volume measured by the wrist-worn device was 26.7 mg (SD 8.7 mg).

Recruitment for interviews

To enable a range of views to be captured, a provisional quota was set of conducting up to 25 interviews with the various stakeholders (HCPs and service users with ID).

The recruitment of HCPs commenced in May 2013. A variety of HCPs were identified through ID services at LPT. The invited HCPs all had previous experience of working with adults with ID. Purposive sampling was used to ensure the inclusion of HCPs who could offer a range of perspectives based on their occupation/professional background. Potential interviewees were sent an invitation pack.

For service users (and carers) the recruitment began in January 2014. The eligibility criteria and method of approach are described previously (see Participants).

Data collection and recording

Topic guides were developed to ensure that relevant issues were captured (see Appendix 19 for service user example). The interviews were semistructured and based on open questioning to elicit issues surrounding knowledge, understanding and experience of T2DM and modifiable risk factors, relevance of IGR, perceived barriers to behaviour change and support needs for people with ID. The practical aspects of the delivery of an education programme were also explored, for example whether or not to develop separate interventions for carers (family members and/or key workers) and people with ID, the inclusion of follow-up sessions and the length of the programme.

Questions were asked appropriately depending on who was being interviewed. Additional communication tools were used when interviewing service users, such as prompt cards depicting images of various activities (e.g. swimming, bowling, walking) to help with eliciting contributions.

The interviews with HCPs were conducted between June and August 2013. All of the interviews were conducted by an experienced qualitative researcher at the HCP’s normal place of work.

The interviews with service users were conducted from January to June 2014. There was an initial delay in finding service users who were either eligible to be invited or willing to be approached/interviewed. All of the interviews were conducted by the same qualitative researcher, with assistance from an ID research nurse. The researcher had expertise in developing and modifying diabetes prevention programmes for different populations; prior to commencing the interviews, the researcher had undertaken additional training within the research team to increase his or her knowledge and skills in the area of ID. To suit the needs and preferences of individual participants, the interviews were conducted in a variety of community settings, including at a participant’s family home, a residential/care home, an assisted independent-living flat and a community clinic.

Data analysis

Audio-recordings of interviews were transcribed verbatim and a thematic analysis was conducted using NVivo version 7 (QSR International, Warrington, UK), a qualitative software programme. Subsequently, themes that were relevant to the development of the intervention were identified.

Characteristics of participants

Key points from interviews with service users and carers

The interviews conducted with service users ranged between 9 and 15 minutes in duration.

In a few of the interviews it was possible to explore awareness of diabetes. Service users related this to ‘sugar’; they also spoke about family members who had diabetes and recalled them being on tablets and having injections. Attempts to gauge service users’ knowledge about healthy lifestyles elicited that some were able to describe basic health messages, such as eating vegetables, eating a high-fibre diet and exercising.

The interviews did yield some useful insights into the lives of service users, for example the types of food that they enjoyed and the degree of choice and control they had in relation to foods consumed; discussions about commonly consumed foods ultimately influenced the food images and food models in the dietary sections of the curriculum. For a few participants, the additional use of prompt cards enabled useful discussion around the types of physical activities undertaken; for those who were more independent, walking appeared to be the most preferred and accessible form of physical activity.

It was difficult to explore service users’ preferences towards learning as part of a group or learning on an individual basis. However, the majority of participants spoke about going to some form of group activity sessions, such as sessions held at a local day care centre or a college; activities included arts and crafts, and learning ‘life skills’ to facilitate independence. Further discussion about participants’ preferences for photographs or pictorial images (on educational resources) suggested that most preferred photographs.

More general points arising from the interviews included practical considerations to be taken into account. First, education sessions needed to be held locally (minimal travelling distance/time for participants) in a setting that was familiar and in a venue that was easily accessible via public transport or similar. Second, it was important to include carers in the education sessions to help (1) support participants and make them feel at ease during sessions and (2) facilitate service users in making changes to their diet and physical activity outside the sessions. It was clear from the interviews that both professional and family carers currently fulfilled this role in the daily lives of participants.

Key points from interviews with health-care professionals

All of the HCPs were enthusiastic to share their knowledge and experience of working with people with ID. Most HCPs had previous or current experience of promoting positive behaviour change with people with ID for behaviour management and/or health promotion.

Key points for initial curriculum from the literature

At the start of the curriculum development process described in this chapter, a structured literature review was conducted with a focus on existing lifestyle interventions for adults with ID, aimed at the primary prevention of T2DM and/or CVD or modification of risk factors. This was supplemented by reviewing relevant published guidelines, consensus statements, interventions currently in practice and service evaluations.

Later, this was formalised by conducting a systematic review to consider evidence on the effectiveness of multicomponent lifestyle behaviour change interventions for reducing risk factors for T2DM and/or CVD. The methods and findings of the systematic review are presented in Chapter 3.

Key findings from the literature that directly informed the content, theory and process of the programme are described below. There were only a small number of studies with a focus on people with ID and behaviour change lifestyle interventions. Few of these studies163–166 provided a description of their theoretical underpinning, although they did recommend the use of social cognition models, such as the Theory of Planned Behavior245 and Reasoned Action. 246,247

The Healthy Lifestyle Change Program, which was developed by Bazzano et al. ,163 was the only published intervention at this initial stage in the development of the STOP Diabetes programme that outlined a conceptual model. Thus, the STOP Diabetes theoretical conceptual framework, as shown in Figure 21, was influenced by this approach.

FIGURE 21.

Theoretical framework for the education programme. PA, physical activity.

The STOP Diabetes framework that was developed then informed all aspects of the education programme; the framework highlights the importance of an individual’s beliefs about health, ill health and its consequences, specifically the impact on them as individuals and their life. In terms of ‘attitude’, outcome expectancies were explored, that is to say ‘what would happen if I engage in a particular behaviour and how important is the outcome for me?’. Methods of learning, specifically vicarious, observational and concrete kinaesthetic were also highlighted in the literature, as was the importance of social support and peer norms. 163,167,245–247

Self-efficacy is a key component of behaviour change,167 that is, the person’s belief that they can perform the behaviour. However, there are many real barriers to behaviour change in this population, such as disability and/or a lack of control over the physical environment, for instance not being the person who buys or cooks the food. Therefore, the concept of actual behavioural control was included in the theoretical framework to ensure that these issues were addressed in the programme. The influences of strong intentions and a detailed action plan were also acknowledged. Intrinsic motivation and the power of reinforcing feedback loops were also highlighted via distal and proximal reinforcers, and the positive impact on quality of life and psychological well-being. These specific components of the programme can be viewed below (see Table 28).

Additional key lessons from the literature were the need to (1) maximise carer involvement; (2) recognise that people with ID have extremely heterogeneous needs and any intervention would require a multimodal approach; and (3) acknowledge that pre-group preparation is essential.

Key points from the multidisciplinary development group

The overarching framework, content, process and learning methods for the programme were formed at a large multidisciplinary meeting following a systematic process; this meeting was additional to the regular monthly meetings that were held throughout the development process. The qualitative findings from HCP interviews, relevant literature and core theoretical constructs were presented and debated, and a consensus was formed. This was later supplemented by findings from the qualitative interviews with service users (and carers), once available.

The core multidisciplinary team – which included ID nurses, education team members, a qualitative researcher and the lead study researchers – met monthly throughout all stages of the development, supplemented by more frequent meetings at key points in the process.

Key points agreed included:

• using a concrete kinaesthetic learning style

• ensuring that the resources developed and methods used to convey messages allowed for tailoring to different levels of intellectual ability

• developing a specific carer session to engage and promote involvement

• ensuring that participants were appropriately prepared prior to attendance and at the start of each session

• reflecting on their own levels of risk

• self-monitoring diaries and pedometers

• goal-setting and action planning

• exploring of barriers and individualised solutions.

The specific methods used to ensure that the themes highlighted above were operationalised are detailed below.

Key behavioural goals of the education programme

For the STOP Diabetes programme, the key behavioural goals and lifestyle messages incorporated into the education sessions were based on those of the Let’s Prevent programme (nutritional)240 and PREPARE programme (physical activity). 241,242 Specific goals included losing weight, reducing consumption of total and saturated fat, increasing dietary fibre consumption, and increasing physical activity and/or reducing sedentary behaviour (Table 28). However, the emphasis of the STOP Diabetes programme was on enabling the individual tailoring of goals, based on a participant’s needs and abilities, including potential mobility restrictions, level of independence with food shopping and preparation, potential dietary restrictions, opportunities to access the community, cognitive level, and availability and level of carer support required. Therefore, more generalised behavioural goals were emphasised, rather than setting specified targets (see Table 28).

Participants and recruitment

The inclusion/exclusion criteria for WP2 were described in Chapter 8 (see Participants). Those invited to engage in the pilot phase were service users who had taken part in the screening stage (see Chapter 5), had screened positive for IGR or had a BMI of ≥ 25 kg/m², and at that time consented to being approached to assist with later phases of the research programme. Recruitment followed a similar process to that in the earlier development phase (see Chapter 8, Participants). An initial telephone call was made to potential participants, followed by further information sent in the post or provided at a face-to-face visit.

Delivery of the education

Potential volunteers with ID were approached about attending the education programme, approximately 4–6 weeks prior to the planned programme start date. Carers were invited to an initial session held 1 week before the delivery of the main education sessions. The aim of the carer session was to provide carers with an overview of the education programme, and explore their role in supporting individuals with ID, both within and between the sessions.

Subsequently, the initial curriculum was delivered to a group of individuals with ID. Carers were also invited to attend the sessions to support the service users. Following feedback and refinement of the curriculum (see Refinement), the modified curriculum was then delivered to a second separate group, which, again, was followed by feedback and refinement.

Three educators were involved with delivering the programme at each session: a registered ID nurse, a diabetes specialist with an education background, and an additional ID nurse or health-care assistant in a supporting role. The educator training process for the study is described in Chapter 11.

Data collection

A range of methods was used to evaluate the education sessions and collect feedback. These included observations recorded during the sessions by an experienced researcher, reflections from the educators leading the programme and qualitative interviews with people who received the programme (those with ID and carers). Additionally, subsequent to the first pilot phase, educators were also interviewed to explore their views about the content and style of delivery, experiences from delivering the programme and perceived practical issues. Feedback and reflection on educator training are described in Chapter 11.

Participants were approached to take part in a feedback interview prior to the last session of the education programme. Interviews were held as soon as possible after the final session. Written consent was obtained immediately prior to the interview. Participant interviews took place in July 2014 for the first cycle and December 2014 for the second cycle.

The purpose of these interviews was to explore participant and carer views about the education sessions, to identify whether or not the education sessions resulted in changes to participants’ diet and physical activity, and to inform changes to the next iteration of education sessions based on participant feedback.

Interviews were conducted and analysed by the same qualitative researcher that carried out the previous interviews (see Findings: second pilot phase). Key areas and topics that were explored included experiences of receiving the education programme, ease of understanding, views about the content and style of delivery, usefulness, relevance and practical issues (including duration, provision of support and suggestions for improvement).

Refinement

At the end of each pilot cycle, modifications were made to the curriculum prior to it being used in the next iteration. Modifications and refinements were informed by findings from participant and carer interviews, observations made during the education sessions, and the ongoing reflection and feedback of the educators.

Uptake and attendance at education sessions

The first iteration of the education programme was held in a community resource centre. A total of 21 participants were invited to take part in the education programme. Five participants (four of whom had carers) initially agreed to attend the programme. Following the carers’ session, one person (and their carer) withdrew completely. Subsequently, four participants (and three carers) took part in the main education programme. The overall attendance at the education sessions (7 weeks, one session per week) was very good, with one participant (and carer) attending all of the 7 days and three participants attending 6 days (Figure 22).

FIGURE 22.

Uptake and attendance at first and second testing phases of pilot cycle. a, Participants were supported by various care workers who were present for part (or some) of the session(s).

Characteristics of participants

Of the four participants taking part in the first iteration of the education programme, two (50%) were male, the median age was 35 (range 29–60) years, three (75%) lived in a supported environment with family or carers, and one lived independently. None of the participants was in paid employment, one attended college and did voluntary work, and all four participated in other activities within the community.

Feedback interviews: first pilot cycle

Following on from the first iteration of the education programme, all of the participants (n = 4) and carers (n = 3) agreed to be interviewed. One participant was interviewed independently. For the remaining three interviews, carers and participants were interviewed together; in two of these interviews the carers made major contributions to the interviews.

Interviews with participants

Uptake and attendance at education sessions

The second iteration of the education programme was held in a residential setting. A total of nine participants were invited to take part in the education programme. Several staff (care workers) from the residential home attended the initial carers’ session, and seven participants agreed to take part in the education programme. In general, attendance at the education sessions was good, with three participants attending on all 7 days, one attending on 6 days and the remainder attending on at least 4 days. Care workers from the residential home also attended at various points during the seven sessions (see Figure 22).

Characteristics of participants

Of the seven participants who took part in the second iteration, three (43%) were male; the median age was 43 (range 29–50) years and all seven lived in a residential home supported by carers. One of the participants had paid employment, two did voluntary work and all seven participated in other community activities.

Feedback interviews: second pilot cycle

After the final education session, a total of five participants with ID were interviewed, along with two members of staff (carers) who had attended the education sessions. A care support worker who had attended some of the education sessions provided support to one of the participants during the interview, helping the participant to feel at ease and assisting her to recall and discuss some of the lifestyle changes that she had made. The remaining four participants chose to be interviewed in pairs with their partners. This arrangement worked well in terms of facilitating participants’ recall and support of each other, although it was challenging to ensure that the contribution of both participants to the interview was maximised and balanced.

The presentation of quotations to support the summary of findings includes singular quotations as well as sections of the discussion with the researcher to help contextualise some of the responses to the questions.

Revisions made after the first cycle

The main revisions made after the first cycle included the following.

Revisions made after the second cycle

Study design

Following initial development, testing and refinement of the curriculum (see Chapters 8 and 9), the education programme was delivered to another group of participants to assess the feasibility of collecting pre- and post-intervention outcome measures.

Study setting

The feasibility phase was conducted between January and June 2015.

Participants

Data collection

Outcomes

Particular outcomes of interest in terms of the feasibility included (1) the proportion of people invited who attend the baseline appointment, education programme, individual sessions within the programme and 3-month follow-up; and (2) the completeness of key data items at baseline and 3 months’ follow-up.

Problems encountered during data collection appointments and implementing the education were also considered.

Delivery of intervention

Following on from the pilot phase of testing, evaluation and modification described in Chapter 9, the final modified education programme (see Chapter 9, Outline of the STOP education programme) was delivered to another group of adults with ID, and their carers, at a local community venue.

Sample size

We aimed to conduct the feasibility study with at least one group of participants (four to eight people with ID, plus carers).

Data analysis

The feasibility of recruiting adults with ID to attend for baseline data collection, education sessions and 3-month follow-up data collection was assessed using a flow diagram to summarise dropouts at each stage. Completeness of outcome data collected at each stage was summarised using counts and proportions.

Recruitment and consent

In total, 19 participants were invited to take part in the feasibility phase, of whom five (26%) agreed to attend an initial appointment at the end of February 2015, at which consent and baseline data were obtained. All of the participants had the capacity to give consent for themselves, without the need to involve a consultee.

Feasibility of collecting baseline data

At baseline, all of the participants required measurements to be taken, as it was > 3 months since they had originally taken part in the screening study. Baseline data were obtained for all five participants, with the exception of physical activity data that were collected for only four (80%) participants, as one person refused to wear the accelerometer (Table 34).

Uptake of and attendance at education programme

Following the baseline appointment, all five of the participants took part in the education programme, which was held from March to April 2015. A total of four carers (two paid care workers and two family members) attended at least some of the sessions (three regularly). Overall attendance at the education sessions was good: four (80%) participants attended on ≥ 5 days and two participants attended on all 7 days (Figure 23). However, one participant attended on only 3 days. It is important to note that the most common reason for not attending sessions was other existing commitments such as appointments and holidays.

FIGURE 23.

Uptake of and attendance at data collection and education sessions.

Feasibility of collecting data at 3-month follow-up

All of the participants agreed to attend a 3-month follow-up appointment at the end of June 2015 to obtain repeat measures. Four of the participants (80%) attended the appointment as arranged; one participant needed to have a second appointment arranged because they did not attend the first (this was the same participant who attended only 3 days of the education programme).

Anthropometric measures (weight, BMI, waist circumference) were obtained for all of the participants (see Table 34). BP readings were successfully obtained for four (80%) participants. Accelerometer data were obtained for three of the four participants who wore one at baseline; four participants initially agreed to wear the accelerometer, but one participant later telephoned the research team to inform them that he had changed his mind.

Characteristics of participants

The key demographic characteristics for the five participants are presented in Table 35. Four (80%) participants were male, the median age was 40 (range 20–51) years and all were of white ethnicity. Two participants lived alone, one lived in supported living accommodation and two lived in a family setting; the majority (n = 4, 80%) had support from a carer for at least some of the time. None of the participants was in paid employment, two did voluntary work on a regular basis, and two attended college and other community-related activities,

Baseline data

Biomedical and lifestyle characteristics, from data collected at baseline, are presented in Table 36. The median values for weight, BMI and waist circumference were 110.2 kg, 36.1 kg/m² and 114.3 cm, respectively. For BP, the median systolic value was 123 mmHg and the mean diastolic value was 85 mmHg. Reported daily intake of fruit, vegetables or salad indicated that only two (40%) participants were eating the recommended five or more portions a day (median 5). For physical activity, the median minutes per day for MVPA and light-intensity activity were 107.5 and 93.0, respectively; the median time spent sedentary was 555.0 minutes per day.

Three months’ follow-up data

Data collected at 3 months’ follow-up are also presented in Table 36. The main aim of the work conducted was the feasibility of collecting data at two time points, and not looking for change.

However, on an individual basis, three participants lost weight (range 1.0–4.2 kg), but two participants gained weight. The participant who had gained a lot of weight (17.6 kg) had poor attendance at the education programme (three out of seven sessions); anecdotal evidence at 3 months’ follow-up suggested that significant changes had occurred in this participant’s personal circumstances since baseline, including moving out of the family home to live independently in his/her own flat.

When data were available, participants showed improvements in physical activity levels and sedentary behaviour at 3 months’ follow-up compared with baseline.

Educator training

The educators for delivery of the intervention (outlined in Chapter 10) were a registered ID nurse and a diabetes specialist with an education background, with support from an ID nurse or health-care assistant.

The process of training the educators involved, first, professional development around the DESMOND251 education programme. This stage took place between January and March 2014, and included observations of DESMOND251-based programmes and attendance at a core DESMOND251 training day. This provided an introduction to the theoretical basis, philosophy and core skills of DESMOND. 251

Second, the educators attended an initial study-specific training day in April 2014, when they were introduced to the theoretical component, content, structure and delivery of the STOP Diabetes education programme (see Table 37). This was followed by a second study-specific day in June 2014. A staged approach was used for delivering the educator training to take into account any necessary changes to the programme (curriculum content, delivery and structure) based on the delivery of the early sessions.

The training was delivered by two members of the development team, a consultant clinical health psychologist and diabetes specialist nurse with an education background. Training included detailed information on the content, structure and delivery of the programme, and incorporated additional support and considerations around working with people with ID. Guidance around indirect approaches for educating participants (e.g. role-playing behaviour change techniques) and direct approaches (e.g. presentations) were included.

Educators used a specific training curriculum to equip them to support the delivery of education sessions and were encouraged to use personal reflection and peer review tools to reflect on their delivery. They were also supported with mentorship from trainers attached to the research team.

An outline of the training programme is presented in Table 37. The first training day covered the first five sessions of the education programme (the carers’ session and weeks 1–4).

Day 2 educator training, which covered the curriculum and resources for the final three sessions (weeks 5–7), followed a similar format to the first training day, taking into account feedback relating to the initial training. In addition, educators were able to feed back their experiences on delivery of the first four sessions, and to reflect on their own self-monitoring and behaviour change experiences.

Evaluation of educator training

Subsequent to completion of their training, educators completed an evaluation form (see Appendix 20) for the STOP Diabetes training sessions. Following their first delivery of the education programme, educators involved in delivering the sessions were interviewed by a qualitative researcher to explore their views about the content and style of delivery, experiences from delivering the programme and perceived practical issues; feedback from these interviews is presented as part of the first pilot phase (see Chapter 9).

Preliminary assessment of educator behaviour

The quality development tool used to conduct the preliminary assessment of educator behaviour was based on current assessment tools developed by the DESMOND collaborative;251 which recognises that training of health professionals to deliver education programmes does not always result in appropriate or consistent delivery of the programmes. 256

The tool was developed in 2015 and consists of five ‘global’ categories, each containing specific items to evaluate programme delivery; it is a revision of a previous DESMOND tool. 257 The tool is designed to be used by a separate observer who assesses the educator’s behaviours when they teach/deliver education. Each item in the tool represents a discrete DESMOND251 behaviour, which is paired with an ‘opposite’ item (labelled as a non-DESMOND251 behaviour), which are coded when observed. The observer is asked to rate which behaviour (i.e. DESMOND251 or non-DESMOND251) is most commonly seen during the training session. For example, the first item requires the observer to assess whether or not the educator uses open body language to support engagement of participants (Table 38). DESMOND251 behaviours for this item include nodding and smiling at the participants, and non-DESMOND251 behaviours include avoiding eye contact and the educator turning their back on the participants when asking a question. By noting down these behaviours when they occur, the observer can determine which behaviour was most common during the training session.

Prior to using the tool to assess educator behaviour for the STOP Diabetes programme, three members of the education development team reviewed the existing quality development tool and its application in the STOP Diabetes programme. As a result, only one of the items in the reflective learning category was changed (see Table 38) because it is recognised that people with ID are more amenable to concrete concepts and visual imagery than to more abstract concepts, such as analogies165 (see also Chapters 8 and 9).

Between March and April 2015, a member of the research team attended the final iteration of the education programme using the tool. The researcher was an experienced member of the research team, had attended specific STOP Diabetes educator training and was involved in the development of the education programme. The tool was used to describe how the educators interacted with the group, to identify differences between educators’ behaviours and to assess appropriateness to this client group. The researcher used the quality development tool to conduct assessments during six out of seven of the education sessions. During these sessions the researcher positioned themselves outside the group and completed the tool separately for each educator.

In addition to noting educator behaviour, an assessment of participant–educator interaction was undertaken during one observation visit, using a 10-second event coding to estimate the amount of time during which the educators were speaking. When the beep sounded, the coder indicated on a response sheet who was talking at that point in time (whether an educator or a participant), with other activity classed as ‘miscellaneous’ (indicating silence, laughter or multiple conversations during learning activities). The 10-second event coding is an objective measure and an established method of measuring talk time-to-educator ratio. 258,259

Identifying key important educator behaviours

Once the preliminary assessment had been conducted, members of the research team met to explore the items within the tool, primarily to distinguish between educator behaviours that were seen to be important compared with those that appeared less core or essential when facilitating adult learners with ID.

Educator training

The STOP Diabetes specific training was attended by two diabetes education specialists, four ID HCPs and other members of the research team. It is important to note that not all of these members of the team went on to deliver the education programme.

Based on evaluation forms completed by six of the educators following the two training days, the training was universally well evaluated. Using a standard evaluation self-report measure, all of the educators agreed with the statement that they had learned new skills and believed they could apply these skills in practice. The most helpful aspects of training cited by educators included going through the curriculum and resources; learning about the development and theoretical underpinning of the programme; and group discussions and learning from colleagues.

Preliminary assessment of educator behaviour

This was based on findings from one observation visit that was conducted to assess participant–educator interaction. In general, the observer noted that the diabetes specialist (who was the trained educator) generally displayed more of the facilitation of learning behaviours, such as using open body language, seeking clarification of participants’ contributions and providing overviews of the sessions; the ID nurse displayed more engagement behaviours, prompting exploration of feelings, involvement of quieter participants and facilitating full engagement in interactive tasks. The DESMOND251 behaviour of prompting the group to summarise the key messages was rarely observed.

Context of proposed screening with existing intellectual disability care pathways

Learning Disability Health Check

The Royal College of General Practitioners guidelines state that NHS general practices should identify individuals who are a high priority for health checks, expressing that mild cases of ID are a lesser priority. 260 However, the current 2015–16 NHS General Medical Services contract states that all individuals with ID who are aged > 14 years should be offered a Learning Disability Health Check and we therefore do not exclude those with mild ID from the analysis.

It was assumed that the model starts at the point following attendance of a Learning Disability Health Check by all baseline individuals.

Figure 24 shows how the screening and intervention elements of the STOP Diabetes study would fit into the Learning Disability Health Check and the scope of the economic evaluation.

FIGURE 24.

How screening for suitability for the intervention fits within Learning Disability Health Check. LD, Learning Disability.

There is no need to explicitly model the Learning Disability Health Check process, as the baseline characteristics from the STOP Diabetes study are assumed to reflect characteristics post health check and the resulting baseline risks of CVD and T2DM.

Attending a Learning Disability Health Check obviates the need to participate in the general population NHS Health Check programme for those aged > 40 years.

In 2013–14, nationally, 44% of those on an ID register attended a Learning Disability Health Check; this varies greatly by geographical area, ranging from 10% to 60%. Specifically, 29% missed a check, implying an offer rate of 73% and an uptake rate (among those offered) of 44%/73% = 60%. 261

During the Learning Disability Health Check, BP, weight and height are measured260 [unless it is not possible because of physical disability (meaning that the patient is unable to stand for height and/or weight measurement) or behavioural difficulties] so that their BMI can be calculated. Some blood tests may also be carried out during a Learning Disability Health Check (or in advance of their Health Check, or as a follow-up test recommended by patients in their Health Action Plan). There is also scope to screen for hyperglycaemia (and dyslipidaemia) and assess cardiovascular risk as part of routine care, although there is much variation in practice as to whether or not such blood tests are deemed to be necessary or a priority within a Learning Disability Health Check, or indeed uptake of blood tests by individuals. For the purpose of the modelling, we assume that these blood tests would be carried out as part of the Learning Disability Health Check.

The model starting point is an individual having been offered, and attending, a Learning Disability Health Check, that is, one of the 44% of individuals noted above.

Overview of approach

This section provides an overview of the economic work before more details are described in subsequent sections.

The first phase of work involved obtaining all necessary parameter inputs and assumptions for the economic model, specifically:

1. data from the STOP Diabetes study to provide the baseline patient characteristics

2. data collected, and quotations and input from the study team to provide the details required to calculate the cost per participant of the STOP Diabetes lifestyle intervention

3. assumptions on uncertain inputs such as rate of uptake of the intervention and the durability of the benefits of the intervention through discussion with the clinical team.

The modelling itself comprised:

1. Model development An existing economic model of cardiovascular and diabetes risk, driven by characteristics such as BMI, total cholesterol, HDL cholesterol and HbA_1c level, was adapted to incorporate the relationship between changes in physical activity (steps) and changes in the above risk factors.

2. Modelling the screening process Determining those individuals from the STOP Diabetes study who are suitable for, and take up, the intervention. The modelling takes account of individuals’ BMI and capacity to participate in the intervention and rates of uptake of the intervention.

Modelling the screening process involved an adaptation of the prevention ‘Sheffield School for Public Health Research (SPHR) Type 2 Diabetes Model’. The model is used to simulate the lifetime patient clinical pathways, incidence of complications and associated cost, and health utility impacts arising from an intervention compared with routine care. The assumptions for the modelling are described in detail later in the report, but the key ones are listed here:

1. The STOP Diabetes intervention was estimated to cost £1097 for the initial intervention and eight maintenance sessions, delivered within a 1-year time frame.

2. The benefits of increases in physical activity could be mapped to changes in BMI, systolic BP, and total and HDL cholesterol, using a relationship identified in the literature.

3. The durability of the intervention effects was uncertain so, for the modelling, two scenarios were adopted with the effects lasting (but decreasing linearly) for 3 and 5 years (from the start of the intervention), respectively.

The usual approach to economic evaluation is to estimate the incremental lifetime discounted costs and quality-adjusted life-years (QALYs) of an intervention compared with usual (routine) care. From this, the incremental cost-effectiveness ratio (ICER) can be calculated and compared with usual acceptability thresholds (£20,000–30,000 per QALY). However, because the clinical effectiveness of the STOP Diabetes intervention is not known, the economic modelling was primarily based on threshold analysis. Under this approach, the requisite clinical effect size needed for the intervention to be just cost-effective (given the cost of the intervention) was estimated. Because the STOP Diabetes intervention promotes physical activity and dietary change, the output of the threshold analyses was not a single effect size, but various permutations of these that would be adequate to make the intervention cost-effective.

Uncertainty around the results was analysed primarily using one-way sensitivity analysis. Owing to the computational demands of undertaking threshold analyses, probabilistic sensitivity analysis (PSA) was restricted to one scenario to give an indication of the degree of uncertainty around such an intervention.

Data sources: STOP Diabetes study