Sixty trials (8,218 participants) were included in the review. Randomisation was adequate in 60% of the trials, allocation concealment in 42% of the trials, 52% masked outcome assessors, 18% of the trials were at high risk of bias from incomplete data and 88% of the trials had a low risk of selective outcome reporting bias.
Trial sequential analysis showed that as of 2002 enough evidence had been accrued to show significant benefit of exercise intervention over no exercise control for both pain and functional improvement.
There were 13 types of exercise comparisons in the network meta-analysis of the outcome of pain. No comparison had undue influence on the effect estimates in the network. There was no evidence of inconsistency between direct and indirect evidence in the network for the outcomes of pain or function.
The network meta-analysis showed no significant differences in effect estimates between different types of exercise interventions. For pain relief, strengthening, flexibility plus strengthening, flexibility plus strengthening plus aerobic, aquatic strengthening and aquatic strengthening were all significantly more effective than no exercise control. The probability of being the most effective intervention for pain relief was 81% for aquatic strengthening plus aerobic flexibility, 76% for strengthening exercise only and 73% for aquatic strengthening plus aerobic exercise. A combined intervention of strengthening, flexibility and aerobic exercise was also significantly more effective than no exercise for improving limitation in function (SMD -0.63, 95% CrI -1.16 to -0.10). This combination of three types of exercise and aquatic strengthening plus aerobic exercises had the highest probability of being the best overall treatment for function (both 71%).
Further sensitivity analyses were reported.