Twenty-five trials were included (n=1,390, range 11 to 156). Study quality was A in one trial, B in seven and C in 17.
Compared with controls people assigned a psychological intervention (12 trials) had lower mean percentage glycated haemoglobin with a pooled SMD of -0.32 (95% confidence interval (CI): -0.57, -0.07) or an absolute difference of 0.76% (95% CI: -1.34, -0.18). There was significant heterogeneity present (p=0.04). Removing two studies (in which the control was a less intensive psychological therapy) the pooled SMD was -0.44 (95% CI: -0.67, -0.22) or a decrease in glycated haemoglobin of 1.06% (95% CI: -1.61, -0.51). A one percent decrease in glycated haemoglobin was considered to be sufficient to reduce the risk of development and progression of microvascular complications. Under this analysis, heterogeneity was not significant and there was no evidence for publication bias. There was no significant difference for blood glucose or body weight.
Compared with controls psychological distress (five trials) was significantly lower in the intervention groups: SMD -0.58 (95% CI: -0.95, -0.20); heterogeneity was not significant.