Peer Support Interventions for Adults With Diabetes: A Meta-Analysis of Hemoglobin A1c Outcomes

PURPOSE Peer support intervention trials have shown varying effects on glycemic control. We aimed to estimate the effect of peer support interventions delivered by people affected by diabetes (those with the disease or a caregiver) on hemoglobin A(1c) (HbA(1c)) levels in adults. METHODS We searched multiple databases from 1960 to November 2015, including Ovid MEDLINE, the Cochrane Central Register of Controlled Trials, CINAHL, and Scopus. We included randomized controlled trials (RCTs) of adults with diabetes receiving peer support interventions compared with otherwise similar care. Seventeen of 205 retrieved studies were eligible for inclusion. Quality was assessed with the Cochrane risk of bias tool. We calculated the standardized mean difference (SMD) of change in HbA(1c) level from baseline between groups using a random effects model. Subgroup analyses were predefined. RESULTS Seventeen studies (3 cluster RCTs, 14 RCTs) with 4,715 participants showed an improvement in pooled HbA(1c) level with an SMD of 0.121 (95% CI, 0.026-0.217; P =.01; I-2 = 60.66%) in the peer support intervention group compared with the control group; this difference translated to an improvement in HbA(1c) level of 0.24% (95% CI, 0.05%-0.43%). Peer support interventions showed an HbA(1c) improvement of 0.48% (95% CI, 0.25%-0.70%; P < .001; I-2 = 17.12%) in the subset of studies with predominantly Hispanic participants and 0.53% (95% CI, 0.32%-0.73%; P < .001; I-2 = 9.24%) in the subset of studies with predominantly minority participants; both were clinically relevant. In sensitivity analysis excluding cluster RCTs, the overall effect size changed little. CONCLUSIONS Peer support interventions for diabetes overall achieved a statistically significant but minor improvement in HbA(1c) levels. These interventions may, however, be particularly effective in improving glycemic control for people from minority groups, especially those of Hispanic ethnicity.