What is the effect of peer support on diabetes outcomes in adults? A systematic review


Jeremy Dale. E-mail: jeremy.dale@warwick.ac.uk


Aim  There is increasing interest in the role that peers may play to support positive health behaviours in diabetes, but there is limited evidence to inform policy and practice. The aim of this study was to systematically review evidence of the impact and effectiveness of peer support in adults living with diabetes.

Methods  We searched the Cochrane Library, MEDLINE, PubMed, EMBASE and CINHAL for the period 1966–2011, together with reference lists of articles for eligible studies. Data were synthesized in a narrative review.

Results  Twenty-five studies, including fourteen randomized, controlled or comparative trials, met the inclusion criteria. There was considerable heterogeneity in the design, setting, outcomes and measurement tools. Peer support was associated with statistically significant improvements in glycaemic control (three out of 14 trials), blood pressure (one out of four trials), cholesterol (one out of six trials), BMI/weight (two out of seven trials), physical activity (two out of five trials), self-efficacy (two out of three trials), depression (four out of six trials) and perceived social support (two out of two trials). No consistent pattern of effect related to any model of peer support emerged.

Conclusions  Peer support appears to benefit some adults living with diabetes, but the evidence is too limited and inconsistent to support firm recommendations. There remains a need for further well-designed evaluations of its effectiveness and impact. Key questions remain over its suitability to the needs of particular individuals, populations and settings, how best to implement its specific components and the sustainability of its effects.


Diabetes is a growing worldwide health problem, with between 300 and 350 million people anticipated as having diabetes by 2025 [1]. It is associated with considerable human, social and economic costs, and places great demands on healthcare resources. Changes in lifestyle behaviours, such as diet, exercise, self-monitoring and adherence to medication regimens, are key to improving outcomes in diabetes [2,3]. While specialist nurses and diabetes educators are being used to promote self-management, peer support offers an approach that is increasingly being considered [4]. However, it is unclear how best to harness its potential.

Peer support has been defined as ‘support from a person who has experiential knowledge of a specific behaviour or stressor and similar characteristics as the target population’ [5]. Shared characteristics include age, gender, disease status, socio-economic status, religion, ethnicity, culture, etc. Psychosocial processes that may be important in peer support include social support, experiential knowledge, and those described by social learning theory, social comparison theory and the helper–therapy principle [6]. In so doing, peer support may enable exploration of feelings, social support, problem solving, goal setting, self-efficacy and hence self-management [7,8]. The reciprocal relationship that occurs through the sharing of life experiences may also benefit the peer supporter, such as by achieving an increased sense of interpersonal competence, gaining new personally relevant knowledge and receiving social approval from the person they help [9].

A broad range of models of peer support have been described in the context of diabetes, including face-to-face management programmes, peer coaching, telephone-based peer support, and web- and email-based support [10]. These models vary in the extent to which they offer one-to-one or group support. They also differ in their focus, how they build on the shared knowledge and experience that peers can offer each other, and in the ways that they provide one or more of the following [5]:

  • 1 Emotional support, including expressions of care, encouragement, active listening, reflection, reassurance and usually the absence of criticism.
  • 2 Appraisal support, including communication of information that is relevant to self-evaluation and the appropriateness of emotions, cognitions and behaviours; for example, motivation and encouragement to persist in problem solving.
  • 3 Informational support, including provision of knowledge relevant to problem solving.

There has been relatively little research or systematic appraisal of the evidence relating to peer support with which to guide diabetes health policy, service developments and delivery [11]. A recent Cochrane review looked at telephone peer support interventions and found limited evidence of impact and effectiveness related to diabetes [12]. Simmons et al. [4] conducted a systematic review of the development of peer support initiatives for diabetes, but its scope was limited to New Zealand. Other reviews of peer support models in diabetes have lacked systematic appraisal of clinical outcomes [9,13]. A recent review by Tang et al. [14] reported on 12 volunteer peer support intervention studies, including seven randomized controlled trials (RCTs). It found inconsistent evidence of impact and concluded that the preliminary evidence for volunteer-based peer support interventions in diabetes was promising, although limited.

The aim of this study therefore was to address the need for rigorous systematic appraisal of the published evidence related to the impact and effectiveness of interventions that use peer support to improve the outcomes of adult patients with diabetes.


The review was undertaken in accordance with the 27-item checklist of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [15].

Search protocol

An extensive literature search was conducted using the Cochrane Library, MEDLINE, PubMed, EMBASE and CINAHL for the period 1966–December 2011, using the terms diabetes, intervention, healthcare professional, peer support, peer, peer and support, support groups, group psychosocial support, self-help, education, lay health educators, lay workers, lay health advisor, coach and community health worker (see also Supporting Information, Appendix S1). Reference lists of articles were also searched.

Paper inclusion and exclusion criteria

The inclusion criteria were: (1) published in English; (2) described a specific programme which included peers providing support to adults with diabetes; (3) all subjects (patients) were diagnosed and being treated for diabetes; (4) study designs included randomized or quasi-randomized controlled trials, controlled clinical trials, before-and-after studies, interrupted time series, descriptive studies or case studies; (5) interventions were aimed at improving the care or management of diabetes.

Paper selection

SMW examined the titles and abstracts for their applicability according to the inclusion and exclusion criteria. All papers that were eligible were read by SMW and VB, who independently examined each according to the selection criteria. All authors discussed papers considered ‘borderline’ for inclusion until consensus was reached.

Paper classification, data extraction and synthesis

Selected papers were classified independently by SMW and VB into predetermined categories according to their design. Data were extracted by SMW and VB from eligible papers on the intervention (either delivered by healthcare professional or peer), mode of delivery (e.g. face-to-face, group etc), diabetes type of participating patients, the nature of peer support, where the intervention was conducted (home, hospital, primary care, community setting), details about the research (e.g. sample size, control or comparison group, data collected measures and findings and significance levels). JRD confirmed the data extraction.

Data were synthesized in a narrative review because of the heterogeneity of the included studies. The risk of bias in the studies that were RCTs was assessed using the Cochrane method [12]. In addition, a checklist (see also Supporting Information, Appendix S2) adapted from previous reviews [16,17] was used to assess the quality of all selected papers in terms of whether key information about the intervention and its delivery and receipt was included (scores were recorded as Yes or No for each element included; maximum score: 30), in addition to an overall rating of the research quality (maximum score: 13). A random sample of 50% of the papers was scored by two authors (VB and SMW) to allow inter-rater agreement to be calculated using the following formula [12]:



Studies reviewed

Figure 1 shows the results of the search strategy. Of 569 papers identified, 474 were screened by reading abstracts and were excluded. Full texts for the remaining 95 were read and 70 papers not meeting the inclusion criteria were eliminated. As a result, 25 papers were included. Seventeen (68%) studies were conducted in the USA, four (16%) in the UK, one (4%) in Ireland, one (4%) in Australia, one (4%) in the Netherlands and one (4%) in Canada.

Figure 1.

 Literature search results.

There were ten (38%) randomized controlled trials [18–27], four (15%) randomized comparative trials [28–31], one (4%) non-randomized comparative study [32], six (23%) studies with a before-and-after design [33–38], two (8%) descriptive studies [39,40], one (4%) feasibility study [41] and one (4%) case study [42].

Sample size varied widely, from 8 to 761 adults (mean 192), and most of the studies examined Type 2 diabetes exclusively. A minority described peers’ characteristics; 9 (36%) reported on their age and gender.

Intervention objectives and content

As shown in Table 1 and in the Supporting Information (Table S1), there was considerable heterogeneity in the format and scope of the interventions, and in the ways in which peers were recruited and trained for their role. Of the 14 RCTs and comparative trials, eight were based entirely on a group face-to-face format [18,21–23,26,27,29,30] and two included such elements as peer telephone calls in addition to a group support [24,25], one study involved peer phone calls alone [20], one a paired face-to-face and telephone format [28] and two involved peer interaction over the Internet [19,31].

Table 1. Summary of setting, design, objectives, participants and intervention characteristics for the randomized controlled studies
Author, year, location, designPrimary objectivesParticipantsPeer support intervention
Smith et al., 2011 [18]
Randomized controlled trial
Group face to face
To test the effectiveness of peer support for patients with Type 2 diabetes395 patients (192 in the intervention group; 203 in the control group) and 29 peer supporters with Type 2 diabetesThe intervention ran over 2 years. There were nine peer-support sessions. Each meeting was facilitated by the peer supporter with a suggested theme and a small structured component. The emphasis was on social support. Content included basics of diabetes, lifestyle and medication issues. The nine sessions covered: (1) introduction; (2) heart and vascular disease; (3) blood sugar levels; (4) healthy eating; (5) medication; (6) exercise; (7) foot care; (8) eye and kidney complications; (9) living with diabetes
Lorig et al., 2010 [19]
Randomized controlled trial
To evaluate an online diabetes self-management program and an additional peer reinforcement group761 people with Type 2 diabetes (including 110 American Indians/Alaska Natives were randomly assigned to (1) the program (n = 259), (2) the program with peer-group reinforcement (n = 232) or (3) usual-care controls (n = 270)This was a 6-week programme. Each week participants logged on to topics that were available, and were asked to reply to a question such as ‘What problems do you have because of your diabetes?’ The questions and answers were posted on bulletin boards that could be seen by all participants. Tools were available, such as exercise and medication logs, audio relaxation exercises, meal planning, etc. Two peers facilitated each programme, and assisted participants by reminding them to log on, modelling action planning and problem solving, offering encouragement and posting to the bulletin boards. All facilitation took place online
Dale et al., 2009 [20]
Randomized controlled trial
Telephone calls
To evaluate the impact on self-efficacy and clinical outcome of a peer telephone intervention231 patients with Type 2 diabetes and raised HbA1c levels (90 in telephone peer support group; 97 in routine care group; 44 in Diabetes Specialist Nurse group)Telecare support was intended to supplement routine care by motivating adherence to the advice provided by the general practitioner or practice nurse at the time of change (medication and/or lifestyle) in the patient’s diabetes care. The first telecare call was made 3–5 days later and then at 7–10, 14–18, 28–35, 56–70 and 120–150 days
Cade et al., 2009 [21]
Randomized controlled trial
Group face to face
To assess whether peer educators can influence healthy eating in people with Type 2 diabetes317 patients with Type 2 diabetes randomized to either a diabetes-specific (n = 162) or individual one-off appointment with a dietician (n = 155)Subjects attended a 2-h session each 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. The final session covered identifying common problems for people with diabetes; monitoring diabetes, self-managing diabetes in terms of food intake, physical activity, blood glucose and blood pressure, and goal setting
Lorig et al., 2009 [22]
Randomized controlled trial
Group face-to-face
To determine the effectiveness of a community-based peer-led diabetes self-management programme345 adults with Type 2 diabetes (186 in peer-led diabetes self-management programme; 159 in usual-care group) This was a 6-week community-based peer-led diabetes self-management programme, offered as 2.5 h weekly by peer leaders in community settings. Programme content included all areas of the American Association of Diabetes Education Standards with two exceptions (glucose monitoring and insulin injection). Topics covered included: overview of self-management and diabetes, action plans, nutrition/healthy eating, problem solving, preventing low blood glucose, preventing complications, fitness/exercise, stress management, relaxation techniques, difficult emotions, monitoring blood glucose, depression, positive thinking, communication, medications, healthcare professionals, healthcare system, sick days, skin and foot care, future plans
Murrock et al., 2009 [23]
Randomized controlled trial
Group face to face
To test a dance intervention and explore the role of peer support to improve diabetes outcomes46 African American women with Type 2 diabetes (n = 24 in dance group; n = 22 in usual-care group)There were 24 dance classes over 12 weeks, each dance class was 60 min. The low-impact dance intervention was taught by an experienced African American dance instructor. The peer component was during the first week of dance classes; each woman chose a personal goal for improving one diabetes outcome and shared it with the group. After each class there was sharing of progress and tips
Lorig et al., 2008 [24]
Randomized controlled trial
Group face to face and reinforcement telephone calls
To determine whether a peer-led programme would lead to improvement in health status, health behaviours and self-efficacy in diabetes patients and whether receiving monthly automated telephone reinforcements would maintain improvement at 18 months567 Spanish-speaking adults with Type 2 diabetes randomized to usual-care control group or intervention6-week community-based peer-led Spanish Diabetes Self Management Programme. The programme was offered for 2.5 h per week by two peer leaders. Spanish Diabetes Self Management Programme participants were also re-randomized to receive 15 months of automated telephone messages or no reinforcement
Anderson-Loftin et al., 2005 [25]
Randomized controlled trial
Group face to face and telephone
To test the effectiveness of a culturally competent, dietary self-management intervention on physiological outcomes and dietary behaviours for African Americans with Type 2 diabetes97 adult African Americans with Type 2 diabetesThe intervention consisted of four sessions at weekly intervals in dietary strategies, five peer-professional group discussion session at monthly intervals and weekly telephone follow-up by a nurse case manager. The peer-professional 1 hour discussion groups were facilitated by a nurse case manager. The peer-professional approach facilitated cultural translation of content, culturally competent learning methods and emotional support from peers and family
Keyserling et al., 2002 [26]
Randomized controlled trial
Group face to face
One-to-one phone calls
To determine whether a culturally appropriate clinic and community-based intervention for African American women with Type 2 diabetes will increase moderate-intensity physical activity200 African American women with Type 2 diabetes (67 in group A; 66 in group B; 67 in group C)Participants were randomized to either: clinic and community (group A); clinic only (group B); minimal intervention (group C). In addition to the clinic-based intervention with a nutritionist received by participants in groups A and B, participants in group A also received three group sessions and 12 monthly calls from a peer counsellor. The peer supporter’s role was to (1) provide social support and feedback to study participants; (2) reinforce diet and activity goals through monthly telephone calls and (3) to assist with group sessions
Pratt et al., 1987 [27]
Randomized controlled trial
Group face to face
To assess the impact of nutrition education and peer support on weight change and glycaemic control in older patients with Type 2 diabetes79 patients with Type 2 diabetesPeer support groups (10 60-min long group sessions). The purpose of the peer support groups was to provide participants with support for changing health behaviours that related to diabetes. The sessions immediately followed the nutrition education classes led by a registered dietician. The facilitator’s primary role was to foster peer support so he/she focused on the interaction of the group, asking members to contribute ideas to one another
Heisler et al., 2010 [28]
Randomized comparative trial
Telephone with optional group face to face
To compare a reciprocal peer-support programme with nurse case management244 male adults (125 in the reciprocal peer-support group; 119 in the nurse case management group) with HbA1c levels greater than 58 mmol/mol (7.5%) during the previous 6 monthsPatients in the reciprocal peer-support group were matched with another age-matched peer patient and received peer communication-skills training. Peers were encouraged to talk at least once a week. Participants could also attend optional group sessions at 1, 3 and 6 months. These were completely participant-focused, where they were encouraged to share concerns, questions, strategies and progress on their action plans
Baksi et al., 2008 [29]
Randomized comparative trial
Group face to face
To assess the effectiveness and acceptability of peer advisers in diabetes in delivering a programme of training on self-management for people with diabetes83 adults aged 18–75 years with diabetes randomized to intervention delivered by peer advisers in diabetes (n = 40) or by the specialist health professional (n = 43)An education programme was delivered by trained peer advisers or by specialist health professionals. The curriculum was considered to be suitable for both Type 1 and Type 2 diabetes. Each course consisted of six sessions held at weekly intervals
McKay et al., 2001 [30]
Randomized comparative trial
Group face to face
To evaluate the short-term benefits of an Internet-based supplement to usual care that focused on providing support for sedentary patients with Type 2 diabetes to increase their physical activity levels78 adults with Type 2 diabetesParticipants in the intervention group could communicate with other members in the intervention group via the ‘Active Lives Support Group’ online conference area. This allowed group members to share information and provide emotional encouragement and support for engaging in their physical activity programme by posting messages. A separate peer-interaction area was available for focus topics (e.g. ‘Physical Activity Barrier Busters’) and participants were encouraged to post comments and share their thoughts on the topic
Glasgow et al., 2003 [31]
Randomized comparative trial
To evaluate the effects of adding tailored self-management training and peer support maintenance components to a basic information-based Internet nutrition intervention320 adults with Type 2 diabetes who were relatively novice Internet usersIndividuals in the peer support arm of the study participated in several activities that allowed for exchanges in diabetes-related information, coping strategies and emotional support. The main activity area was a peer-directed forum for participants to interact with one another in a safe, supportive setting. There was a more structured conference area where, periodically, research staff introduced specific diabetes-related topics to stimulate group discussion. There were real-time live chats and e-newsletters

Ten of the RCT studies reported that peers were trained [18–22,24,26–29], with the duration of training ranging from 2 evenings to 4 days. Some form of supervision was reported in five studies [18,21,28,29,31].

Of the 11 non-RCT studies, six were based entirely on a group face-to-face format [32–35,38,39], two (13%) involved peer phone calls alone [36,42], one peer interaction over the Internet [37], one a paired face-to-face and telephone format [38] and one the use of an information kiosk [40]. Six studies reported that peers were trained [32,34,38,39,41,42], with the duration of training ranging from 2 h to over 5 days. Some form of supervision was reported in three studies [33,37,41].

Peer-led education

Topics reported as being covered in RCT peer-led sessions included introduction to diabetes and its complications [18,19,21,22,25,26,29,31], the structure of the body related to functions [29], behaviour change/goal setting [19,21–23,27], the role of diet/nutrition [18,19,22,24,26,29], lifestyle issues [18], exercise [18–22,24,26,29,30] and BMI [18], medication [18,19,22,29], the importance of self-monitoring blood glucose [18,19,22,29], blood pressure [18,19,21], relaxation techniques (including for stress management [19,21,22,24], depression [19], coping strategies [31], emotional support [25,31], communication skills [28], examining feet/foot care [24], sick days [19,22], relationships with family [19], communication [22], working with the healthcare system [19,22], future plans/goal setting [19,22,27], problem solving [22,33], positive reinforcement (when a member tried something new or did something successfully the group verbally rewarded that person) [27], modelling (group members described exactly how they would or had responded to problems brought up by other members) [27], preventing complications [22], difficult emotions [22], depression [22], positive thinking [22] sharing and encouragement of sharing [27], normalization of thoughts and behaviours that appeared to be negative or dysfunctional reactions to the stress of chronic illness [27], coping with living with diabetes [18,19,21,31], ‘working with your healthcare professional’ [22], the structure of the National Health Service (NHS) and organizational issues relevant to the intervention [29].

The topics covered in the non-randomized controlled trial studies’ peer-led sessions included diabetes and its complications/information [32,36,37,40], behaviour change/goal setting [33,34,36,38,41,42], problem solving [34], the role of diet/nutrition [32,33,36,41], exercise/weight control [32,35,41,42], medication [32,33,41] sexual health information [33] the importance of self-monitoring blood glucose [32,35], identifying social supports [33], coping strategies [37] and emotional support [37].

Other intervention elements

Encouragement to subjects to attend their healthcare provider for follow-up visits was an element in one randomized controlled trial study [24] and a non-randomized comparative study [32], and the provision of social support was described as an element in two randomized controlled trials [26,31] and one case study [42].

Quality of studies

A mean agreement of 89% (range 76–100%) was attained for ratings of the description of the intervention and the target population, and a mean agreement of 92% (range 85–100%) for ratings of research quality.

There was considerable variation in the quality of the design and the reporting of the studies reviewed. Given the risks of bias and limitations evident in the studies reviewed, their findings should be viewed cautiously. The risk of bias (see also Supporting Information, Table S2) was considerable across most of the randomized controlled trials, in part reflecting issues around blinding that are implicit to a complex intervention such as peer support, as well as specific limitations in the design and reporting of the randomized controlled trials included. The specific quality scores (Table 2, and see also Supporting Information, Table S3) for the majority of the studies were fair to good, with the randomized controlled trials gaining the highest scores.


The findings reported by the RCTs and by studies with non-RCT designs are summarized in Table 2 and in the Supporting Information (Table S3), respectively. A broad range of clinical, behavioural, knowledge, empowerment and satisfaction outcome measures were reported on, reflecting the diverse objectives of the peer interventions and the associated studies.

Table 2. Summary of key findings from RCTs
Author, year, location, designMeasuresFollow-upKey findingsAuthor’s conclusionsQuality rating (score)
for the intervention (out of maximum score of 30)for research design (out of maximum score of 13)
Smith et al., 2011 [18]HbA1c Cholesterol Systolic blood pressure Wellbeing score24 monthsWhile there was a trend towards decreases in the proportion of patients with poorly controlled risk factors at follow-up; these changes were not statistically significant.A group-based peer support intervention is feasible in general practice settings but the intervention was not effective when targeted at all patients with Type 2 diabetes. The results do not support the widespread adoption of peer support.20 Fair13 Good
Lorig et al., 2010 [19]HbA1c Health distress Activity limitation Patient Health Questionnaire – depression Patient Activation Measure Self efficacy Exercise Physician visits6 monthsAt 6 months HbA1c, patient activation and self efficacy were improved for program participants compared with controls (P < 0.05). The American Indians/Alaska Natives programme participants showed improvements in health distress (P < 0.01) and activity limitation (P < 0.05) compared with controls. At 18 months self-efficacy (P < 0.05) and patient activation (P < 0.01) were improved for programme participants. Reinforcement showed no improvement.An online diabetes self-management program is acceptable for people with Type 2 diabetes. The program may have beneficial effects in reducing HbA1c and American Indians/Alaska Natives populations can be engaged in and benefit from online interventions. The follow-up reinforcement appeared to have no value.18 Fair13 Good
Dale et al., 2009 [20]Self efficacy HbA1c Cholesterol BMI Diabetes distress Satisfaction and experience6 monthsThere were no significant differences in self efficacy scores HbA1c or other secondary outcome measures. There was a high level of acceptability but peer support was less highly valued than support from Diabetes Specialist Nurse.Further consideration needs to be given to the targeting of the telecare peer support, its intensity, the training and ongoing supervision of peer supporters and the extent to which information and advice should be incorporated.21 Good13 Good
Cade et al., 2009 [21]HbA1c Weight BMI Waist circumference Lipid profile Blood pressure Diabetes Empowerment Scale (DES) Audit of Diabetes Dependent Quality of Life (ADDQoL)6 and 12 monthsThere were no significant differences between groups in any of the clinical or dietary outcomes measured.The peer intervention was not effective in changing measures of diabetes control or diet.18 Fair13 Good
Lorig et al., 2009 [22]HbA1c BMI Depression Glucose monitoring Symptoms of hypoglycaemia Communication with physicians Healthy eating Reading food labels6 and 12 monthsAt 6 and 12 months the intervention group showed no significant differences in HbA1c. Participants did have significant improvements in depression, symptoms of hypoglycaemia, communication with physicians, healthy eating and reading food labels (P < 0.01). At 12 months Diabetes Self Management Program participants continued to demonstrate improvements in depression, communication with physicians, healthy eating, patient activation and self-efficacy (P < 0.01).The findings suggest that people with diabetes without elevated HbA1c can benefit from a community-based peer-led diabetes programme.18 Fair13 Good
Murrock et al., 2009 [23]HbA1c Weight Body fat Blood pressure12 weeksResults showed a significant group mean difference in systolic blood (P < 0.01) pressure and body fat (P < 0.05).Dancing twice per week for 12 weeks produced significant differences in systolic blood pressure and body fat. Dancing in a supportive environment with peers may be an effective strategy for diabetes educators to help those with diabetes to become more physically active and improve diabetes outcomes and overall health.16 Fair13 Good
Lorig et al., 2008 [24]HbA1c Health status Health behaviours Health care utilization Self efficacy Symptoms of hypo and hyperglycaemia6 and 18 monthsAt 6 months Spanish Diabetes Self Management Program (SDSMP) participants compared with control subjects showed improvements in HbA1c (P < 0.05). There were also improvements in health distress, symptoms of hypo and hyperglycaemia and self-efficacy (P < 0.05). At 18 months all improvements were maintained (P < 0.05). SDSMP participants also demonstrated improvements in self-rated health and communication with physicians, had fewer emergency room visits (-0.18 visits in 6 months, P < 0.05) and trended toward fewer visits to physicians. At 18 months reinforced (i.e. received telephone calls) showed increased glucose monitoring (P < 0.001).The SDSMP demonstrated effectiveness in lowering HbA1c and improving health status. Telephone reinforcement did not add to its effectiveness.16 Fair13 Good
Anderson-Loftin et al., 2005 [25]BMI Dietary fat behaviours Weight HbA1c Lipids6 monthsBMI and dietary fat were significantly lowered in the experimental group (both P < 0.01). The experimental group reduced high fat dietary habits to moderate while high-fat dietary habits in the control group remained essentially unchanged. Trends towards reduced HbA1c and lipids were observed.Results suggest the effectiveness of a culturally competent dietary self-management intervention in improving health outcomes for southern African Americans.13 Fair13 Good
Keyserling et al., 2002 [26]Physical activity and energy expenditure Dietary intake HbA1c Lipids Diabetes knowledge Mental wellbeing Social wellbeing6 and 12 monthsGroup A (P < 0.01) and Group B (P < 0.05) increased physical activity compared with Group C. There was enhanced diabetes knowledge for Groups A, B and C at 6 and 12 months (P < 0.05). High levels of satisfaction were reported for both interventions.The intervention was associated with a modest enhancement of physical activity and was acceptable to participants.13 Fair13 Good
Pratt et al., 1987 [27] Weight HbA1c Social and psychological variables Peer support levels Satisfaction Ratings of classesWeeks 8 and 16There was a significant reduction in weight within the peer support group after 8 weeks (P < 0.05).No significant changes were observed in HbA1c levels.This finding should encourage further serious investigation of peer support as a facilitator of weight reduction for older adult patients with diabetes.16 Fair11 Good
Heisler et al., 2010 [28]HbA1c Insulin therapy Blood pressure Self-reported medication adherence Diabetes-specific distress Diabetes-specific social support6 monthsMean HbA1c level decreased in the reciprocal peer support (RPS) group, with a difference in HbA1c change between groups of6mmol/mol (0.58%) (P < 0.01). Eight patients in the RPS group started insulin therapy compared with 1 patient in the nurse care management group (P < 0.05).Reciprocal peer support holds promise as a method for diabetes care managementGood (21)Good (13)
Baksi et al., 2008 [29]Knowledge was primary outcome: what is diabetes, nutrition, exercise, monitoring, medications. HbA1c6 monthsKnowledge scores improved in both groups but there were no significant differences between groups for any of the five knowledge domains or HbA1c levels.Trained patients are as effective in imparting knowledge to their peers as specialist health professionals. Both are also acceptable to patients as trainers. However, lay tutors require training specific to the education programme they would be delivering.Fair (18)Good (12)
McKay et al., 2001 [30]Physical activity per week Depressive symptoms8 weeksThere was an overall moderate improvement in physical activity levels within both the intervention and control conditions, but there were no significant differences between groups. Those who used the site more regularly derived significantly greater benefits reporting greater overall satisfaction on a six point scale (p < 0.05).Internet-based self-management interventions for physical activity and other regimen areas have great potential to enhance care of diabetes. Greater attention should be focused on methods to sustain involvement with Internet-based intervention health promotion programmes over time.Fair (19)Good (13)
Glasgow et al., 2003 [31]Dietary outcomes Behavioural outcomes including physical activity HbA1c and lipid ratios Depression Social support Usage of internet intervention Number of barriers10 monthsParticipant’s website usage decreased over time. There were improvements in total cholesterol (p < 0.0001), low density cholesterol (p< 0.0001), dietary behaviour and psycho-social outcomes (p< 0 .0001).The basic D-Net intervention showed improvements across a variety of patients, interventionists and clinics. There were difficulties in maintaining usage over time and additions of tailored self-management and peer support components generally did not significantly improve results.Fair (17)Good (10)

Randomized controlled trial studies

Clinical outcomes

HbA1c:  All but one [30] of the 14 RCTs reported HbA1c as an outcome measure. Of these, three found peer support to have a statistically significant beneficial impact [19,24,28]. The latter studies had a mean number of participants of 524 (range 244–761), compared with 184 (range 46–395) for those that found no statistically significant difference.

Blood pressure:  One RCT found statistically significant reduction in systolic blood pressure in the peer support group [23]; three found no significant differences [18,21,28].

Cholesterol:  Of the six RCTs measuring cholesterol, one found significant differences at follow-up [31], while the others found no differences [18,20,21,25,26].

Symptoms of hypo- and hyperglycaemia:  Self-reported symptoms of hypo- and hyperglycaemia were an outcome measure in two RCTs [22,24], and in both there were statistically fewer symptoms in the intervention groups at follow-up.

BMI/weight/body fat:  Weight and/or BMI was an outcome measure in seven RCTs. Two reported statistically significant improvement at follow-up [25,27], and there was a significant difference in weight for both the intervention group and usual care group between baseline and follow-up for another [23]. The latter also reported a significant reduction in body fat in the intervention group [23]. There were no significant differences for the others [19–22].

Fatigue:  Fatigue was an outcome measure in one RCT, and was reported to have marginally improved for the intervention group [22].

Health behavioural outcomes

Physical activity/fitness:  Physical activity was an outcome measure in five RCTs, of which two showed statistically significant improvement in levels of physical activity as measured by accelerometer [26] and self-reported exercise [22], and three showed no improvement in self-reported exercise [19,29,31].

Glucose monitoring:  One RCT used glucose monitoring as an outcome measure and found a significant increase in self-reported monitoring [24]. Another found knowledge about monitoring as evaluated in a written test was not significantly different at follow-up [29].

Diet:  Four RCTs used self-reported diet and eating habits as an outcome measure, and three reported statistically significant improvements in terms of healthier eating [22,25,31]. One found no significant differences between groups for dietary behaviour [30].

Insulin therapy:  One study [28] used initiation of insulin therapy as an outcome measure and reported that patients in the reciprocal peer support group were more likely to start insulin therapy than those in the nurse case management group.

Clinic and communication visits:  Three RCTs used clinic visits as an outcome measure. One [22] found that self-reported communication with physician significantly improved, and another [24] reported improvements in self-rated communication with physicians, fewer emergency room visits and a trend towards fewer visits to physicians. No significant differences in the number of physician visits were reported in the third [19].

Empowerment outcomes

Self-efficacy:  Three RCTs used self-efficacy as an outcome measure. Two found significant improvement in self-efficacy, which was maintained at 12-month follow-up [22,24], while no significant differences for self-efficacy were reported in the third study [20].

Perceived barriers:  One RCT [31] used perceived barriers to self-managing diabetes as an outcome measure and reported that at follow-up there was a significant improvement in the number of barriers scored as ‘moderate’ or more.

Knowledge outcomes:  Two RCTs reported significant improvements in knowledge scores for several knowledge domains as an outcome at follow-up [26,29].

Psychological outcomes

Depression/health distress:  Four of the RCTs that used health distress or depression as an outcome measure [19,22,24,31] found statistically significant improvements at follow-up, while three found no significant differences [18,20,28].

Perceived social support:  Two studies used social support as an outcome measure [28,31]. Both showed significant improvement at follow-up, with one showing significant improvement in diabetes-related social support when compared with nurse care management [28].

Acceptability:  Four studies [20,26,29,30] reported on acceptability and three found high levels of acceptability for peer support. However, two compared peer-led and health professional-led interventions and both [20,29] reported that the intervention was more highly valued when delivered by a health professional. One study [30] reported that only 35% found peer-to-peer support group helpful.

Comparison between peers and healthcare professionals

Three RCTs compared a peer-led intervention with a healthcare professional-led intervention [20,28,29]. Of these, one reported that peers may be as effective as specialist health care in promoting self-efficacy [20]. Another study [28] reported that the peer-led intervention was superior to a healthcare professional, resulting in a significant difference between groups for HbA1c and a greater number of patients in the peer support group starting insulin therapy compared with those in the nurse care management group. In a third study, peers were found to be as effective at imparting knowledge to their peers as specialist healthcare professionals [29].

Intervention fidelity

Of the four RCTs reporting on intervention fidelity, one [22] reported observation of sessions that confirmed that the peer-leaders were careful in maintaining fidelity to the structured program. Another [20] used telephone record sheets to ensure consistency of documentation regarding call content, goal setting and achievement and length of calls. The third [25] reported supervision of peer group classes by a nurse case manager who was a certified diabetes educator. The other study reported that research associates completed a checklist of key areas covered and communication skills used [28].

Other study designs

Clinical outcomes

The non-randomized comparative study reported [32] statistically significant improvements in HbA1c, and that both the nurse case management and peer education/empowerment group had statistically significant improvements in diastolic blood pressure. It also reported significant improvements in cholesterol levels in both groups.

In one before-and-after study [34] there was a significant improvement in blood pressure, weight and waist girth at follow-up. In another before-and-after study [35] there was a significant improvement in weight at follow-up.

Health behavioural outcomes

In two before-and-after studies, one found significant improvement in self-reported physical activity at follow-up [35] and the other showed significant improvement in pedometer-determined physical activity [34]. Another before-and-after study reported an overall moderate improvement in self-reported physical activity levels for both groups, but no between-group differences [37]. In another before-and-after study, 73% of participants reported that their peer partner helped them do things to stay healthy, such as exercising more [36], while another reported improvements in self-reported days per week of exercise [33]. In a before-and-after study it was reported that there was significant improvement in a dietary assessment [35].

Empowerment outcomes

In one before-and-after study there was also improvement in self-efficacy and use of supportive resources [35]. Another before-and-after study reported that participants gained significantly greater self-efficacy scores [36]. A non-randomized comparative study reported significant improvements in knowledge of diabetes and cultural-based beliefs [32].

Psychological outcomes

Three before-and-after studies [33,36,38] reported that the peer intervention was a positive and beneficial experience for most participants. In a feasibility study, the majority of patients were satisfied with the individual sessions [41]. In the descriptive study [39], all participants reported that they would recommend the peer intervention to others and had learned a lot about their diabetes and how to manage it better.


Peer support shows some potential to improve outcomes in adults living with diabetes, but as identified in this review the evidence base is inconsistent and too limited to support firm recommendations about wider implementation. In a minority of studies, peer support was associated with improvements in clinical and behavioural outcome measures. These included statistically significant improvements in glycaemic control (three out of 14 trials), blood pressure (one out of four trials), cholesterol (one out of six trials), BMI/weight (two out of seven trials), physical activity (two out of five trials), self-efficacy (two out of three trials), depression (four out of six trials) and perceived social support (two out of two trials). Studies that were RCTs tended to report less evidence of beneficial impact than those that had weaker methodological designs. However, several of the RCTs appeared to be underpowered and there was a significant risk of bias evident in most of them. Furthermore, data collection for many of the studies appeared to have been undertaken at an early stage of implementation when ‘teething problems’ might still have been present. No pattern emerged of elements of peer support that appear most important to achieving specific diabetes outcomes, and hence the review does not point to the superiority of any particular model of peer support.

The interventions varied in the extent to which peer support was intended as an adjunct to routine clinical care, or were less formal, user-initiated interventions that patients might self-select or volunteer into. The importance of establishing a strong theoretical understanding of how a complex intervention, such as peer support, causes change is recognized as a prerequisite to optimizing its design and implementation [44,45]. This appeared to be a weakness in many of the studies and most appeared to have adopted a pragmatic approach to intervention development. There may be considerable scope for increasing the effectiveness of peer support through strengthening its theoretical foundations and linking this to the processes involved in all aspects of its implementation. Many components may contribute to the take-up and effectiveness of peer support and each of these requires careful consideration during intervention design. This includes: the target population and intended aims and purpose of the peer support; the educational, counselling or behavioural approaches that are intended; the process through which peers are recruited, trained and supervised; the information given to those receiving support, and their understanding and expectations of peer support; the location, frequency, duration and flexibility of contact; and the extent to which the peer support is integrated with diabetes and other health services.

It is inherent to peer support that it is culturally located in time, place and population, and that it is unlikely that any standardized model will be widely applicable [45]. For example, several of the studies were directed at disadvantaged populations, such as African American women, Hispanics and Bangladeshis in the UK, who may have especially limited access to routine health care. Although the shared experiential knowledge that is implicit in peer support may have generic value to such groups, the model of peer support that is most suited to each group may be distinctive.

Some difficulties in recruiting and retaining peers were described in the papers reviewed, raising questions about the applicability and sustainability of some interventions outside the contexts studied. The peers were often specific to particular communities; hence, their enthusiasm, dedication and motivation may be difficult to replicate more widely. It is likely that those who chose to participate in these studies, whether as patients or peer supporters, may have been more favourably disposed towards, and so more likely to gain benefit from, the intervention than others who had lower expectations about its relevance to their needs. Joseph et al. [38], for example, reported difficulty in finding appropriate peer coaches who had changed their behaviour and appreciated the struggle that occurs with such change. They stated that the detailed matching of peers and patients was time-consuming and may not be practical or possible in everyday practice. Furthermore, with a larger population of peers it might be difficult to maintain the fidelity of the intervention. Indeed, few of the studies had investigated intervention fidelity, and ways of establishing and maintaining such fidelity across large-scale implementation of peer support need to be investigated. There is also a need to be aware of how peer support may impact on the peers themselves. Although none of the studies reported adverse effects on those providing the peer support, there is a need to consider the potential harmful consequences that might occur.

The trials comparing models of peer-led and healthcare professional-led interventions were designed to demonstrate superiority rather than equivalence or non-inferiority. Hence, the failure to identify differences should not be taken as evidence of equivalence, particularly given the risk of false negative results (type 2 errors) as a result of inadequate sample size and power. Although there may be scope for substitution of activities between healthcare professionals and trained peers, further research of how peer support interventions can most effectively complement and extend routine clinical services is needed before this can be recommended.

Three studies used the Internet as the mode of contact between peers and reported encouraging results [19,31,37]. There was evidence that this could lead to improved HbA1c and self-efficacy [19], improved dietary behaviour [31,37] and other psychosocial and biological measures [31]. However, there were difficulties in maintaining usage over time. Ways of optimizing the use and sustaining the effectiveness of social networking and online media for peer support in diabetes needs further investigation.

Finally, there are a number of important gaps in the literature reviewed that should be addressed in future research. This includes a lack of data related to cost-effectiveness and little research aimed at understanding the clinical and psychosocial benefits gained as a result of providing peer support to another individual with diabetes. Subgroup analyses to identify the characteristics of individuals who may particularly benefit from peer support are also needed. In addition, longer-term follow-up studies are needed to determine the sustainability of peer support impact on behaviour change and clinical outcomes.


Peer support for adults living with diabetes has been delivered through diverse formats to widely differing populations, but the quality of the evidence is insufficient to determine which models and elements of peer support may be most applicable and effective in relation to the needs of an individual patient, specific population or setting. There are many unanswered questions about its effectiveness and impact. As recommended by the World Health Organization Consultation on Peer Support, further well-designed evaluations are needed before peer support interventions can be recommended as a policy option for diabetes [11]. These should investigate all aspects associated with the design and implementation of different models of peer support, including cost-effectiveness, in order to better understand the efficacy of different formats, the characteristics of individuals who are most suited to becoming peers, the target populations who may benefit most and the sustainability of effects [46]. The underlying theories that inform the design of specific peer interventions, including the associated recruitment, training and supervision strategies, need more attention.

In the meantime, there is an opportunity for shared learning about models of peer support in diabetes that are being developed and implemented. Peers for Progress, a global initiative of the American Academy of Family Physicians Foundation, developed out of the World Health Organization Consultation on Peer Support Programmes in Diabetes, is a strategic initiative to promoting best practice in peer support [6]; its website provides guidance on how to evaluate peer support programmes in managing diabetes and may be useful for future reference [47].

Funding sources


Competing interests

Nothing to declare.