TITLE: Impact of exercise type and dose on pain and disability in knee osteoarthritis: A systematic review and meta-regression analysis PURPOSE: To analyze the effect of published exercise therapy programs aimed at reducing pain in patients with osteoarthritis (OA) in the knee, in order to identify the most optimal exercise program characterized by number of supervised sessions, type of exercise, relative exercise intensity and duration of exercise sessions. METHODS: A systematic review and meta-analysis based on published exercise studies identified by a comprehensive literature was performed. Eligible were randomized controlled trials comparing any exercise intervention with a sham or non-intervention control in patients with knee OA, using a least one patient reported outcome on pain. Effect sizes were calculated as standardized mean differences (SMD) estimated from differences in mean changes between intervention- and control group after an exercise program, divided by the pooled standard deviation. SMDs were combined by using a random effects meta-analysis model; incl. both 95% confidence intervals and prediction intervals. Stratified analyses and meta-regression analyses were used to examine study-level covariates. RESULTS: Forty one trials with 3274 patients with knee OA comparing 48 interventions with controls were included in the analyses. The pooled SMD for the reduction of pain was 0.48 [95% CI: 0.36; 0.60] in favor of exercise, although based on a large between study inconsistency (I2= 58%). Interpreting this heterogeneity, the 95% prediction interval (-0.14; 1.10) contains values below 0 and so, although on average the intervention seems effective, it may not always be beneficial in an individual setting. Focusing on pain in pre-specified stratified analyses showed almost similar effects for programs aiming at improving muscle strength, aerobic capacity or functional performance (SMD[pain]: 0.56, 0.67, and 0.49, respectively, P=0.822; pooled SMD= 0.57 [95% CI: 0.44; 0.69], I2= 53%). Whereas combined programs were less effective in reducing pain (SMD [pain]: 0.14 [95%CI: -0.11; 0.40], I2=58%), with a statistically significant difference (P=0.003) between exercise subgroups SMD= 0.43 [95% CI: 0.41; 0.71] in favor of using only one type of exercise. Heterogeneity in aerobic exercise trials could be explained in a meta-regression model using numbers of supervised sessions. The reduction of pain (beta-coefficient = 0.026, [95% CI: 0.01-0.05], I2=23%), was significantly increased with a larger number of supervised exercise sessions. In resistance exercise heterogeneity could not be explained by study-level covariates. Stratified analyses for pain showed larger effect when focusing on quadriceps strengthening only compared to lower limb strengthening or strengthening the whole body (SMD[pain]: 0.72, 0.42 respectively, P=0.04). CONCLUSION: Exercise programs have on average a moderate effect for reducing pain in knee osteoarthritis trials. However, inconsistency across studies suggests that it will not always imply benefit in comparison with a matched control group. Stratified analyses provide evidence that exercise programs aiming at improving strength, aerobic capacity or functional performance are more efficacious than combined exercise programs. While the number of supervised sessions may enhance the benefits of the aerobic exercise, focusing on the quadriceps only, may increase benefits of resistance training.
Osteoarthritis and Cartilage, 2011, Vol 19, Issue Supplement 1