Eight RCTs (1,656 patients), one RCT analysed as a pre-intervention post-intervention study for both arms (64 patients), and three pre-intervention post-intervention studies (236 patients) were included.
The sample sizes ranged from 5 to 596. Methodological limitations included the weakness of studies with a pre-test post-test design, high attrition rates, and studies with small sample size that lacked statistical power. In addition, only one study analysed data on an intention-to-treat basis.
RCTs (8 studies).
Glycaemic control: 5 of the 8 RCTs found that, compared with the control, the intervention improved glycaemic control by the end of the intervention. However, one RCT had a high (50%) attrition rate and another tested statistical significance using a 1-tailed test. Two of the 3 RCTs reporting 6-month follow-up results found no significant different between the treatments. The third RCT found that improved control was partially sustained at 6 months.
Quality of life: 3 of the 4 RCTs reporting quality of life outcomes found no significant difference between the treatments in quality of life scores at the end of the intervention. The fourth RCT (11 patients in the intervention arms and 13 patients in the control arm) found that a monthly support group over 18 months significantly improved quality of life scores compared with no support (78 points versus 71 points, P<0.05). The trialists reported no significant difference between the treatment groups at baseline, but the reviewers commented that baseline values were neither provided, nor adjusted for in the analysis.
Uncontrolled studies (4 studies).
Three studies found that glycaemic control significantly improved after the intervention (2 studies immediately after the intervention and a third study 1 month after the intervention). The fourth study found no change in glycated haemeoglobin at the end of the intervention. This study found improved quality of life post-intervention and at the 6-month follow-up.
Characteristics of the 10 studies that showed a positive effect for the intervention included: poor glycaemic control at baseline (glycated haemoglobin greater than 11%); interventions aimed at changing behaviour as opposed to using traditional didactic methods (all 10 studies); interventions culturally tailored (4 studies); an intervention modified for older people (1 study); group counselling (8 studies); one-to-one counselling (3 studies); nurse educators (8 studies); nutritionalist educators (8 studies); and the involvement of the family (6 studies).