Eighteen RCTs (2,580 participants) were included in the review. Fifteen other studies were excluded because of a PEDro quality score of less than 6. The included studies scored from 6 to 8 points. All the included studies reported randomisation. Allocation concealment was reported in 11 studies. Assessors were blinded in 15 studies. Thirteen studies reported drop-outs of less than 15%. Intention-to-treat analyses were used in 13 studies. Follow-up periods ranged from five weeks to 36 months.
Physical exercise therapy compared to no exercise: There were statistically significant benefits with physical exercise therapy compared to no exercise in mobility based on final values (SMD 0.18, 95% CI 0.05 to 0.30, Ι²=9%; six RCTs, 956 participants) and change scores (SMD 0.82, 95% CI 0.54 to 1.10; one RCT, 215 participants) and physical functioning based on final values (SMD 0.27, 95% CI 0.08 to 0.46; one RCT, 424 participants) and change values (SMD 2.93, 95% CI 2.50 to 3.36; two RCTs, 287 participants).
No significant differences were found for physical activity (three RCTs) and quality of life (four studies). There were no differences in effectiveness between groups that used short interventions compared to longer interventions.
High intensity physical exercise therapy compared to low intensity exercise therapy: Statistically significant benefits were observed in quality of life with higher intensity exercise (SMD 0.30, 95% CI 0.04 to 0.56, Ι²=12%; two RCTs, 232 participants). No significant differences were observed for mobility (six studies) and physical activity (one RCT). There were trends towards improvements in physical functioning but the pooled estimate from four RCTs did not reach statistical significance.
Individual interventions compared to group interventions: When physical therapy was compared with no exercise, individual interventions were associated with a non-significant effect on mobility in one study (SMD 0.22, 95% CI -0.24 to 0.68 based on final values) and a significant effect in another study (SMD 0.82. 95% CI 0.54 to 1.10 based on change values). Group interventions were associated with no significant differences (SMD 0.18, 95% CI 0.01 to 0.35, Ι²=27%). Significant differences between groups were found for individual interventions on physical functioning (SMD 2.93, 95% CI 2.50 to 3.36, Ι²=27%) and group interventions (SMD 0.27, 95% CI 0.08 to 0.46). There was insufficient data for analyses of physical activity and quality of life.
Intervention type: The authors reported that the interventions with the largest effect sizes used strength training components as part of the exercise therapy programmes.
Long-term effects: Five studies used follow-up measures post intervention and one study found significant benefits of individualised exercise therapy six months after conclusion of the intervention.
There was no evidence of publication bias in the funnel plots.