Can peer educators influence healthy eating in people with diabetes? Results of a randomized controlled trial


Professor Janet Cade, Nutritional Epidemiology Group, Centre for Epidemiology and Biostatistics, University of Leeds, Level 8, Worsley Building, Leeds, LS2 9JT. E-mail:


Aims  To assess whether the Expert Patient Programme (EPP), adapted for people with Type 2 diabetes, can be used to promote healthy eating to improve glycaemic control.

Methods  Adults with Type 2 diabetes (n = 317) were randomized to receive either a diabetes-specific EPP (n = 162) or individual one-off appointments with a dietitian (control group) (n = 155). The diabetes-specific EPP followed the standard National Health Service programme although all participants in the group had diabetes only, rather than a mix of chronic conditions. Participants attended a group session for 2 h once per week for 6 weeks. In addition, a final seventh-week 2-h session was included that was specific to issues concerning diabetes. Outcomes were assessed at baseline, 6 and 12 months.

Results  There were no statistically significant differences between the control and the intervention group in any of the clinical outcomes measured. There was no significant difference between the groups in any dietary outcome. There was a higher starch intake in the EPP group, although this did not reach statistical significance (effect size for starch adjusted for baseline values 8.8 g; 95% CI −1.3 to 18.9). There was some loss of participants between baseline measurement and randomization, although this did not appear to have had an important impact on baseline balance.

Conclusions  In this study of people with Type 2 diabetes, the EPP approach was not effective in changing measures of diabetes control or diet.


Audit of Diabetes-Dependent Quality of Life


Diabetes Empowerment Scale


Expert Patient Programme


general practitioner


glycated haemoglobin


National Health Service


With an aging population and increasing prevalence of obesity, the prevalence of Type 2 diabetes is expected to continue to rise. In addition, the number of people with diabetes is 1.6 times greater in areas where there are high levels of poverty and where there are higher proportions of people of South Asian origin [1]. As the incidence of diabetes in the population rises, the need has emerged for effective methods of helping people with diabetes improve their management of the disease and improve their long-term health prospects.

In common with people suffering from other long-term problems, who are managed in the community, [for example by themselves and their general practitioner (GP)], < 50% of people with diabetes follow their treatment correctly. Less than 10% of adults with diabetes manage to follow all the steps required to achieve good glycaemic control [2]. People with diabetes often do not know enough about their condition and few have been helped to help themselves [2]. The concept of self-management has emerged as one promising route to improved health and well-being for those with chronic illness, in particular in the format of the Expert Patient Programme (EPP), in which participants learn skills such as problem solving and relaxation from peer educators who themselves have a chronic condition (i.e. lay people who are trained to give advice to other people; the terms lay tutor and peer educator are interchangeable) [3,4]. A review of research studies focusing on self-management found that such approaches could improve people’s health in the short term [5]. For example, disabled people have been trained to provide health messages to others with disabilities—the success of this project is partly as a result of the understanding between them about their situation [5,6]. In the UK, the ‘Challenging Arthritis’ programme has led the way in successful peer education as has the ‘Living with Long-Term Illness’ project [7]. Consequently, in 2002 the Department of Health (DH) introduced generic self-management courses nationally, in the form of the Expert Patients Programme (EPP) [8]. The standard EPP consists of six, weekly sessions, each lasting around 2 h. Whether the EPP can lead to improvement in diet of people with diabetes, through better self-management, in not known. In order to provide a more focused course, an additional diabetes-specific session was added to the EPP in this study at week 7, developed by the University of Leeds but delivered by the peer educators.

The aim of this study was to train people with diabetes to deliver the EPP to other people with diabetes. This analysis aims to answer the key question: can peer educators promote healthy eating to improve control in people with Type 2 diabetes?

Participants and methods


Participants with Type 2 diabetes were recruited from GP practices in the Burnley, Pendle and Rossendale area of Lancashire, following ethical approval from East Lancashire Local Research Ethics Committee. There were two phases to recruitment. First, participants were approached to become ‘peer educators’. These were people with diabetes, living in the community, who were willing to be trained in chronic disease self-management and to deliver group sessions on chronic disease self-management to other people with diabetes. Five people volunteered to become peer tutors. Of these, four attended a residential training course for this purpose, provided by the National Health Service (NHS) EPP within East Lancashire. One participant withdrew as a result of unavoidable delays in setting up the residential course. Subsequently, two tutors who had been trained withdrew through ill health and were replaced by an additional two tutors; therefore, four tutors were available to run courses for the trial. In addition, a senior trainer who had worked for the NHS EPP was paid to support the project for 6 months. It was a requirement of the EPP to have a more experienced tutor work with less experienced tutors. The senior trainer worked alongside volunteer peer tutors to deliver the programme, thereby providing five lay tutors in total. All tutors in this study were given additional training to deliver the diabetes-specific elements of the course (week 7).

Participants for the intervention phase were also recruited from GP practices in the Burnley, Pendle and Rossendale area between October 2003 and October 2004. Eligibility criteria were adults with Type 2 diabetes (defined as onset over age 30 years, not on insulin within the first year); registered with GP practices selected from socially deprived catchment areas (using Jarman scores).

Following informed consent, potential participants were randomized into either the intervention or the control arm of the study before being invited for baseline measurements at the start of the intervention. The decision to randomize prior to baseline measures was a pragmatic one, designed to reduce the burden on control participants who would otherwise have to attend an additional appointment for the collection of baseline measures. Initially, the gap between randomization and baseline measures was minimal. However, as recruitment progressed, delays were encountered that were out of the control of the research project team, which led to a greater delay between randomization and collection of baseline measures. This resulted in some dropout between randomization and collection of baseline measures with subsequent potential effect on baseline balance between intervention and control groups.


The EPP is being delivered by the NHS using a highly structured and standardized delivery format to people with a range of chronic health problems. Our intervention used this programme. It differed from the standard NHS practice in that each group included only people with diabetes. One extra session relating to aspects of diabetes was also added. Subjects attended a 2-h session, once a week, for 7 weeks. The first six sessions covered aspects of learning to cope with a long-term health problem, and improved eating, relaxation and exercise patterns. A new seventh session was also delivered, specifically about diabetes. This session covered the following: identifying common problems for people with diabetes; monitoring diabetes; self-managing diabetes in terms of food intake, physical activity, blood glucose and blood pressure; goal setting.


The control group received ‘standard care’, comprising an individual appointment with a dietitian, lasting approximately 15–30 min, depending on the duration of time since the patient was diagnosed with diabetes. At this session, the dietitian provided standard dietary advice, as would be delivered within the NHS setting.


The primary clinical outcome under study was a change in glycated haemoglobin (HbA1c). Subjects were followed up at 6 and 12 months following the EPP course. Group follow-up sessions were arranged for the EPP group and individual follow-up was arranged for the control group. Extensive measures were put in place to maximize response to the 12-month follow-up, including two reminders to the initial follow-up postal contact; telephone contact (at which minimal data were collected if participants would not return questionnaires or attend a face-to-face appointment) and a medical note search (for which additional ethical approval was received). In addition, £5 gift vouchers were provided to participants for each questionnaire returned. Subjects who did not attend the 12-month follow-up were contacted by postal questionnaire and asked for permission to obtain clinical data from GP notes search.

Other outcomes of clinical relevance were weight, body mass index (BMI), waist circumference, lipid profile and blood pressure. Dietary changes were measured through the completion of repeated 3-day food diaries and questionnaires. Psychological measures were collected through questionnaires. These were the short-form Diabetes Empowerment Scale (DES), adapted for use in UK populations [9] and the Audit of Diabetes-Dependent Quality of Life (ADDQoL) [10]. The DES has three subscales: managing the psychosocial aspects of diabetes (nine statements), assessing dissatisfaction and readiness to change (nine statements) and setting and achieving diabetes goals (10 statements). Scores for each subscale range from a minimum of 1 to a maximum of 5, with 5 showing the greatest empowerment. The 18 questions in the ADDQoL were related to the freedom to eat and drink and enjoyment of food with diabetes. The combined score for impact on life ranged from −9 (maximum negative impact of diabetes) to +9 (maximum positive impact of diabetes).

Sample size

The sample size for the study was based on 90% power to detect a standardized difference of 0.4 in any of the main outcome measures. This required 135 subjects in each group (270 in total). This sample size would give adequate power to detect a difference between the groups of 2% in mean percentage of energy from fat (e.g. 37 vs. 35%) or a difference of 200 kcal/day in total calorie intake or a difference of 3/4 of a portion of fruit and vegetables per day (400 g/week). To allow for the effect of losses to follow-up we aimed to randomize an additional 10% of participants (300 in total).

Statistical analysis

Analyses were performed using Statistical Package for the Social Sciences (spss for Windows, version 11.5; SPSS, Chicago, IL, USA). Independent sample t-tests (or non-parametric tests where appropriate) and χ2-tests were used to investigate differences in baseline characteristics and response rates between the two groups. However, a poor response rate was seen in the intervention group, which meant that the original balance obtained through randomization could not be guaranteed and a full intention-to-treat (ITT) analysis could not be performed. Therefore, analysis of covariance was performed, adjusting for baseline values. Sensitivity analyses were also performed to explore the effect of obtaining data from the medical note search, attendance at four or more sessions of the EPP and attendance at the diabetes-specific session of the EPP (week 7).


Response rates

Progress through the study is given in Fig. 1. We targeted 1726 subjects as potential participants in the study. Of these, 319 subjects consented to take part in the study. Following randomization, baseline measurements were available from 112 EPP and 127 control subjects (total 239 subjects). However, at 6 months there was a considerable dropout rate among the EPP group, with 6-month follow-up dietary data from 61 (55%) of the EPP participants. Follow-up of the control group at 6 months was much higher with 98 (77%) providing some data.

Figure 1.

 Participant flow. EPP, Expert Patient Programme; HbA1c, glycated haemoglobin.

Attendance rates for the EPP group at baseline found that, for 110 subjects for whom data were available, 18 attended only one session and 22 attended all seven sessions; 63 subjects attended the final diabetes specific session.

Attempts to improve follow-up at 12 months were successful. At 12 months, clinical data (HbA1c) was available on 86 (77%) of the intervention participants and 118 (93%) control participants, with dietary data available on 74 (66%) of intervention participants and 101 (80%) of control participants. Although less than originally intended, this maintained almost 80% power for the same effect size as the study was originally powered for and easily achieves more than 80% power for standardized differences of 0.5. Twelve-month data were collected on some participants for whom there were no baseline measures (23 control and 21 intervention participants with 12-month but no baseline HbA1c). Baseline and 12-month clinical measures were available on 63 (56%) intervention and 91 (72%) control participants, with dietary data on 74 (66%) intervention and 103 (81%) control participants. For 12-month data, where baseline values were not always present, but response rates were higher, unadjusted figures are presented.

Characteristics of participants at baseline

The mean age of participants at baseline in the EPP group was 65.4 ± 11.6 years (sd) and in the control group was 66.2 ± 11.5 years. Of the EPP participants at baseline, 43% were female and 57% were male and, of the control subjects, 40% were female and 60% were male. Over 95% of each group characterized themselves as of white European origin. In terms of employment, 8% of the EPP and 15% of the control group classified themselves as working full-time; 65% of the EPP and 63% of the control group were classified as retired. For education, 21% of the EPP and 20% of the control group had a degree level education. Similar percentages in each group (36% of the EPP and 37% of the control group) were living on their own. These results suggest that the groups were largely comparable at baseline.

As dropout was an issue, we also present baseline results for participants at 12 months, there were some small differences, particularly for the EPP group compared with the full baseline sample. Mean age was 65.8 ± 11.0 years in the EPP group and 66.6 ± 11.0 years in the control group. Of the 12-month participants in the EPP group, 38% were female and 62% were male, while, in the control group, 42% were females and 58% were males. Over 95% of each group characterized themselves as of white European origin. In terms of employment, 10% of the EPP and 14% of the control group classified themselves as working full-time, 62% of the EPP and 65% of the control group were classified as retired, 20% of the EPP and 23% of the control group had a degree level education and 29% of the EPP and 34% of the control group lived on their own.

Changes in primary clinical outcomes at 12 months

The key variables at baseline, 6 and 12 months for all participants are given in Table 1. In subjects for whom there were both baseline and 12 months’ data, there was no significant difference in any clinical or dietary measure. The numerical difference in unadjusted values for starch was not statistically significant, even after adjusting for baseline values (see Table 2). Using all 12-month data collected (including the medical note search), there was also no significant difference in any primary clinical or lifestyle outcome. There was no significant difference in any dietary outcome, apart from a borderline difference in starch intake (EPP–control difference = 11.2 g; 95% CI 0.2, 22.2; P < 0.05). Qualitative exploration of the sources of starch from food diaries failed to find any particular differences in food sources of starch between the groups.

Table 1.   Comparison of key variables: baseline, 6 and 12 months (all participants)
Baseline6 months12 monthsBaseline6 months12 months
Mean ± sdnMean ± sdnMean ± sdnMean ± sdnMean ± sdnMean ± sdn
  1. BMI, body mass index; BP, blood pressure; CHO, carbohydrate; Chol:HDL ratio, cholesterol:high-density lipoprotein ratio; EPP, Expert Patient Programme; HbA1c, glycated haemoglobin; NSP, non-starch polysaccharides; sd, standard deviation.

 Weight (kg)86.2 ± 17.613385.3 ± 16.310785.5 ± 16.812384.9 ± 19.311882.2 ± 17.27684.7 ± 18.796
 BMI (kg/m2)30.8 ± 5.613330.4 ± 5.010730.6 ± 5.211930.7 ± 6.011729.8 ± 5.57430.4 ± 5.793
 HbA1c (%)7.5 ± 1.41187.5 ± 1.4917.6 ± 1.31087.3 ± 1.3897.5 ± 1.2737.6 ± 1.286
 Cholesterol (mmol/l)4.6 ± 1.11214.4 ± 1.0934.3 ± 1.11065.0 ± 1.1904.6 ± 1.0694.5 ± 1.087
 Chol:HDL ratio4.5 ± 1.11154.1 ± 1.0883.8 ± 1.01014.7 ± 1.4904.5 ± 1.3614.2 ± 1.183
 Systolic BP (mmHg)152 ± 21133148 ± 20115152 ± 21125154 ± 24115147 ± 2087149 ± 2198
 Diastolic BP (mmHg)82 ± 1213381 ± 1211580 ± 1512583 ± 129079 ± 118680 ± 1398
 Energy (kcal)1728 ± 4731221542 ± 883901563 ± 4731031886 ± 4441021829 ± 706611668 ± 44373
 Fat (g)64 ± 24.512268 ± 39.79057 ± 24.910370 ± 22.810265 ± 28.56158 ± 19.873
 Saturated fat (g)22 ± 9.112224 ± 15.09020 ± 9.810325 ± 10.110222 ± 10.26120 ± 8.973
 CHO (g)211 ± 60.5122220 ± 87.890197 ± 68.0103227 ± 53.8102225 ± 82.561211 ± 56.873
 Protein (g)84 ± 21.812295 ± 41.39076 ± 20.010389 ± 21.810284 ± 22.86177 ± 18.373
 NSP (Englyst g)16 ± 5.912217 ± 8.49015 ± 5.610316 ± 5.710217 ± 6.66116 ± 5.173
Table 2.   Unadjusted means and effect sizes adjusted for baseline values, for key clinical and dietary variables at 12 months
Difference in means adjusting for baseline 95% CI P
  1. *Weight, BMI n = 118; HbA1cn = 95; cholesterol n = 97.

  2. †Weight n = 94; HbA1cn = 65; cholesterol n = 67.

  3. BMI, body mass index; BP, blood pressure; CHO, carbohydrate; CI, confidence interval; EPP, Expert Patient Programme; HbA1c, glycated haemoglobin.

Clinical measures(n = 120*)(n = 93†)   
Weight (kg)85.984.2−0.3−1.4, 0.80.6
BMI (kg/m2)30.630.4−0.04−0.4, 0.30.8
HbA1c (%)−0.3, 0.40.8
Cholesterol (mmol)−0.2, 0.30.9
Systolic BP (mmHg)152149−3.8−8.7, 1.10.1
Diastolic BP (mmHg)8180−1.0−4.4, 2.40.6
Dietary measures(n = 87)(n = 64)   
Energy (kcal)1545164821.7−108.7, 152.10.7
Fat (g)56.055.9−2.7−9.0, 3.70.4
Protein (g)75.377.80.4−5.4, 6.20.9
CHO (g)194.5209.24.1−15.0, 23.20.7
Starch (g)107.6130.48.8−1.3, 18.90.09
Englyst fibre (g)14.815.60.8−0.7, 2.30.3
Fruit and vegetable intake (g)364.2368.719.4−39.2, 78.10.5

Sensitivity analyses

Participants for whom 12-month data were extracted from clinical notes only (i.e. non-responders to the 12-month follow-up session who consented to having their GP notes searched for relevant information), were compared with participants who provided data at 12 months’ follow-up. There were no significant differences for any clinical measure between the two groups. Comparison was also made between individuals who attended four or more EPP sessions and control subjects and individuals who attended the diabetes-specific session (week 7 of the course) and control subjects. Again, there were no significant differences between the groups in clinical or dietary measures. There were also no significant differences in psychological measures (data not presented).


This study failed to show any significant difference in outcome between intervention and control subjects at 12 months. However, a high dropout, particularly from the intervention group, may have compromised the results of the trial. The sensitivity analyses performed also failed to elicit any differences between groups.

A major weakness of this research was the dropout between randomization and baseline measurement. The possible effect of this was to reduce the benefit of randomization leading to potential selection bias and possible imbalance in baseline covariates. Nevertheless, imbalance was slight at worst and results, adjusting for baseline covariates, were broadly the same as for unadjusted figures.

This problem was caused by unexpected delays in setting up the EPP which were beyond the control of the research project team. These included delays in implementation of the scheme at national and local level, including the development of generic course materials, setting up of training for tutors, lack of availability of senior trainers and changes in procedures associated with EPP delivery, which required all lay tutors to be supervized while delivering their first course.

A further problem was the relatively low response rate to follow-up. However, the medical note search did go some way to address the differences in response rate between intervention and control participants. The lack of any intervention effect is disappointing, given the significant investment (at least £18 m) in the EPP by the NHS. However, it is interesting to note that these results are similar to those found in a small trial (n = 83) comparing peer advisors with health professionals who delivered a self-management training programme to people with Type 2 diabetes. This study concluded that, because there was no difference in outcomes between the peer advisors and health professionals, this showed the success of the peer advisors. In fact, there was no difference in either group between baseline and follow-up in measures of glycaemic control (HbA1c), implying that neither group was successful in improving outcomes [11].

The mean glycated haemoglobin at baseline was 7.3–7.5%. This is lower than in other studies which have recruited participants with poor glycaemic control and who may therefore be more likely to experience a positive outcome [12]. Diet was recorded using a 3-day food diary. The food diary technique is often considered to be a gold standard, although it can be prone to under-reporting. Energy intakes reported in our study are similar to those found in the low income diet and nutrition survey [13], which were estimated to be about 80% of the estimated average requirements.

Although in this study peer educators do not appear to have been effective in promoting healthy eating and improving glycaemic control, other group-based, education programmes have been successful [14]. The X-PERT Programme, which is health professional led, was found at 14-month follow-up to improve glycaemic control, body mass index, reduced requirement for glucose-lowering medication and increased consumption of fruit and vegetables [15]. That project had excellent attendance rates [16]. This suggests people may be more committed to a programme if it is led by health professionals. A recent review of therapeutic patient education for people with diabetes found that most education was undertaken by a multi-professional team according to best-practice guidance [17].

It is possible that the intervention, i.e. the EPP, was not appropriate for the population under study. The evidence base for the EPP primarily derives from the USA [18], where a different system of health care is in place. Self-management is a more familiar concept there than in the UK and this may raise issues around its transferability. People who agreed to take part in this study were not representative of the local population; for example, there were fewer people of South Asian origin (5%) taking part than live in Burnley, Pendle and Rossendale (10%). However, this would not affect the outcome of the study, only its potential generalizability.

To explore our findings in more detail, we undertook qualitative interviews with participants from the intervention and control groups, along with tutors and trainers involved in delivering the EPP. The results of that qualitative study will be reported in detail elsewhere. In brief, there was considerable expression of disappointment regarding this course, although there were some participants who felt they had benefited. Positive aspects were around exchanging ideas with others who had diabetes and the diabetes-specific week 7 material. Nevertheless, these benefits do not appear to have translated into measurable improvements in health and few changes in knowledge, attitude or behaviour could be attributed to the course. Most participants who had attended any session of the EPP course expressed a wish that there had been a higher proportion of diabetes-specific information and less instruction in techniques of self-management. A frequently expressed view was that concrete, relevant information applicable to their situation as people with diabetes would be more helpful than self-management skills that are broadly applicable in any situation. Participants who dropped out early in the course felt the most strongly about this.

The lack of impact of EPP in this trial may be in part because of a lack of receptivity to its basic concepts in a local culture where many see self-management as no more than the common sense they already possess. Future EPP courses may benefit from screening participants to determine potential likelihood of benefiting from the course. The other clear recommendation emerging from this study, however, is for a brief, intensive, disease-specific course such as the X-PERT Programme, containing both medical information and self-management techniques, to be offered to all diabetes patients within the first 2 months of diagnosis.

This trial was conducted in the early stages of delivery of the EPP; now that the programme has become more established, repeating this evaluation may yield a different result. It could be that softer outcomes, associated with other factors such as improved self-esteem or greater confidence may show greater differences between the two groups. While no differences were seen between groups in the psychological measures that were collected, these were mainly diabetes-specific outcomes and more generic measures may show greater differences. The generic EPP is now running in a number of areas of the country. A UK-wide assessment of the pilot phase of the national EPP programme has been conducted by the National Primary Care Research and Development Centre (NPCRDC). A randomized controlled trial of the standard EPP course found moderate gains in self-efficacy for people attending courses. They note that caution should be applied to this finding as those results are only pertinent to people who volunteer to go on such a course and not those with long-term conditions more generally [19]. Diet, as measured in that study by one item, was not different between the EPP and the waiting list control groups.

In conclusion, this randomized controlled trial of an EPP targeted for people with diabetes found that the EPP approach was not effective in changing measures of diabetes control or diet.

Competing interests

Nothing to declare.


Thanks to Clare Witham for working with the control group and James Thomas for database support. This study was commissioned by the Food Standards Agency.