Sixteen RCTs with 22 comparisons were included in the review. Only two studies reported blinded outcome assessment. Only four reported adequate allocation concealment. Seven studies had attrition rates higher than 20 per cent.
Insulin sensitisers versus placebo (nine RCTs). There was a small but statistically significant effect of insulin sensitisers (WMD -1.5, 95%CI: -2.8, -0.2) .
Insulin sensitisers versus OCPs (five RCTs). There was no statistically significant difference between the groups.
Metformin versus androgens (three RCTs). There was a statistically significant difference in favour of anti-androgens (WMD - 3.7, 95% CI: -6.8, -0.6).
Metformin combined with other treatments (five RCTs). Metformin plus flutamide was statistically significantly more effective than metformin alone (WMD -4.6, 95% CI: -7.9, -1.3; two RCTs). No significant differences were found between metformin combined with flutamide versus OCP (one RCT) or between metformin plus flutamide versus flutamide alone (two RCTs).
High or moderate levels of statistical heterogeneity were found in all analyses. Results of subgroup analyses were reported extensively. Results for studies using patient self-assessment were also reported.