Six RCTs, reporting seven data sets (n=1,533 patients), were included in the review. Five of the six trials were rated as adequate for sequence generation and allocation concealment. Four trials were classified as having adequate blinding. Two trials were judged to have adequately reported incomplete outcome data. No criterion was judged inadequate for any trial.
The pooled estimate of effect size (SMD) for pain reduction with diacerein was -0.24 (95% CI -0.39 to -0.08) compared with placebo, based on seven data sets with high heterogeneity (I2=56.3%).
The pooled estimate of effect size (SMD) for disability with diacerein on the Lequesne impairment index was -0.14 (95% CI -0.25 to -0.03) compared with placebo, based on seven data sets with low heterogeneity (I2=11.4%).
There was a significantly increased risk of diarrhoea for diacerein compared with placebo (RR 3.51, 95% CI 2.55 to 4.83; NNH 4, 95% CI 3 to 7), based on seven data sets with moderate heterogeneity (I2=38.7%).
Subgroup analyses showed no significant differences by adequacy of concealment or joint type.
A funnel plot showed some evidence of publication bias.