Question: In patients with type 2 diabetes mellitus, does exercise training improve glycemic control (reduce glycosylated hemoglobin [HbA1c] levels) and reduce body mass?
Data sources: Studies in all languages were identified by searching MEDLINE (1966 to 2000), EMBASE/Excerpta Medica (1980 to 2000), SPORT DISCUSS (1949 to 2000), HealthSTAR (1975 to 2000), Dissertation Abstracts (1861 to 2000), and the Cochrane Controlled Trials Register; by checking the reference lists of major textbooks, review articles, and relevant articles from the search; and by contacting experts.
Study selection: Studies were selected if they were randomized controlled trials (RCTs) or controlled clinical trials (CCTs) of exercise interventions lasting ≥ 8 weeks in patients with type 2 diabetes in whom compliance could be verified by direct supervision or diaries.
Data extraction: 2 investigators independently extracted data on sample size, baseline and post-treatment means and standard deviations of the intervention and control groups for HbA1c levels and body mass, exercise-program characteristics (type, frequency, duration, intensity, and energy cost), and study quality (randomization, blinding, withdrawals, and concealment).
Main results: 14 trials (11 RCTs) were included (504 patients, mean age 55 y, 50% men, mean duration of diabetes 4.6 y). The duration of the intervention ranged from 8 to 52 weeks (mean 18 wk). Study quality was moderate to low (mean score 1.6 out of 5 points). 12 trials that compared exercise with no exercise showed reduced HbA1c levels with the exercise program (P < 0.001) (Table). 2 trials that compared diet plus exercise with no diet and no exercise also showed reduced HbA1c levels with the intervention (P = 0.008) (Table). A sensitivity analysis showed no difference between results from RCTs and CCTs (Table). The results did not differ by type of exercise (aerobic vs resistance training). Exercise alone (12 trials, P = 0.60) or combined with diet (2 trials, P = 0.24) did not reduce body mass (Table). The results did not differ when RCTs, CCTs, aerobic training, or resistance training were analyzed separately. Meta-regression analysis showed that the reduction in HbA1c levels could not be predicted by change in body mass, exercise intensity, or total amount of exercise.
Conclusion: In patients with type 2 diabetes mellitus, exercise training reduces HbA1c levels but does not reduce body mass.
Exercise vs no exercise for type 2 diabetes at mean 18 weeks*
| Outcomes | Comparisons | Number of comparisons | Weighted mean difference (95% CI) |
| HbA1c level (%) | Exercise vs no exercise | 11 | 0.66 (0.34 to 0.98) |
| Exercise + diet vs no exercise and no diet | 3 | 0.76 (0.29 to 1.32) |
| RCTs only (exercise and diet) | 9 | 0.63 (0.25 to 1.01) |
| | | Standardized mean difference (CI) |
| Body mass (kg) | Exercise vs no exercise | 13 | 0.06 (−0.15 to 0.26) |
| Exercise + diet vs no exercise and no diet | 3 | 0.20 (−0.54 to 0.14) |