Thirteen RCTs were included (368 patients; range 12 to 58). The mean PEDro score was 7 out of 10 (range 6 to 8).
For the overall analysis, the pooled standardised mean difference was 0.86 (95% CI 0.74 to 0.98; 13 trials; 150 comparisons). There was evidence of statistically significant heterogeneity, but no evidence of publication bias.
By type of programme the standardised mean differences were 1.04 (95% CI 0.70 to 1.39; 25 comparisons) for strength with aerobic exercise, 1.17 (95% CI 0.75 to 1.60; 18 comparisons) for electrical stimulation, and 1.11 (95% CI 0.93 to 1.28; 107 comparisons) for strengthening exercise.
The effect estimates were similar for short-duration interventions (less than eight weeks) compared with long-duration interventions (eight to 12 weeks).
Further results, including over 20 meta-analyses for individual outcomes, were presented. Heterogeneity results were presented only for the overall analysis. There was no evidence of publication bias.