Seventeen RCTs (n=823) were included in the review .
The included studies scored a mean of 3.1 (± 1.1) (range: 1 to 5) out of a maximum of 5 on the Jadad scale, and a mean of 10.8 (± 2.9) (range: 5 to 14) out of a maximum of 17 on the Oxford Pain Validity Scale.
There was little or no indication of statistical heterogeneity in the analyses (I2 ranged from 0 to 38%).
Outcomes at 3 to 4 months (16 RCTs).
There was a statistically significant difference between the groups, favouring the intervention group over the placebo group, for patient-assessed pain (SMD -0.26, 95% confidence interval, CI: -0.49, -0.03, p=0.03; 13 RCTs, n=501), morning stiffness (SMD -0.43, 95% CI: -0.72, -0.15, p=0.003; 8 RCTs, n=306), the number of painful and/or tender joints (SMD -0.29, 95% CI: -0.48, -0.10, p=0.003; 10 RCTs, n=425) and NSAID consumption (SMD -0.40, 95% CI: -0.72, -0.08, p=0.01; 6 RCTs, n=156). No statistically significant difference between the groups was found when studies were pooled for the outcomes of physician-assessed pain (3 RCTs, n=123) and Ritchie Articular Index (4 RCTs, n=135).
Outcomes at 5 months or longer (6 RCTs).
The reviewers reported a statistically significant difference between the groups, favouring the intervention group, for physician-assessed pain (SMD -0.50, 95% CI: -0.98, -0.01, p=0.05; 2 RCTs n=68) and the number of painful and/or tender joints (SMD -0.51, 95% CI: -1.0, -0.02, p=0.04; 2 RCTs, n=68).There was no statistically significant difference between the groups for other outcomes measures, though sample numbers were small.
Outcomes at 2 months or less (6 RCTs).
There was no statistically significant difference between the groups for any outcome measures.
Sensitivity analyses excluding studies with a Jadad score of 2 or less did not affect the statistical significance of the results.
Subgroup analyses were also reported.