Meta-analysis of the effect of exercise interventions on fitness outcomes among adults with type 1 and type 2 diabetes.

This review applied meta-analytic procedures to integrate primary research that examined exercise fitness outcomes among adults with type 1 or type 2 diabetes. We used extensive literature searching strategies to locate published and unpublished intervention studies that tested exercise interventions. We conducted meta-analytic procedures using both fixed- and random-effects models. Analyses were conducted separately for type 1 and type 2 diabetes and separately for two-group and single-group data under two assumptions about pre-post fitness outcomes association: no association and a strong positive association (0.80). Codable data were extracted from 35 reports with 1074 subjects. The type 1 diabetes overall weighted mean effect size for two-group comparisons was 0.65. For treatment group pretest versus post-test, the effect was 0.62. The effect size among adults with type 2 diabetes for two-group comparisons was 0.35. The treatment group pretest versus post-test comparison effect was 0.37. Effect sizes among control group participants were not statistically different from 0. For type 2 diabetes two-group comparisons, larger effect sizes were associated with exercise prescription, fitness testing, supervised exercise, group sessions, and recommendations for longer duration exercise sessions. HbA1c, age, body mass index, and gender distribution were unrelated to fitness outcomes.