Twenty-three placebo-controlled RCTs (median sample size of 55; range: 35 to 63) with 57 treatment groups were included in the review. The majority of the studies used a parallel-group design (52 treatment groups).
Forty treatment groups had less than 15% attrition. Further details about the quality of the trials was not reported.
Across all interventions, the pooled reduction in CRP in comparison with placebo was 28% (95% confidence interval, CI: 26, 30). However, significant statistical heterogeneity was detected (p<0.0001) and the effect size was less with the fixed-effect model (16%) than with the random-effects model. Significantly greater reductions in CRP were reported for statin-ezetimibe combinations and statin-only therapies versus other LDL-lowering therapies, and for 80 mg/day statin versus lower statin doses. A dose relationship with greater LDL reduction was also observed with increasing statin dosages. Study type, duration of the intervention and study sample size had no impact on the change in CRP after adjusting for change in LDL. The meta-regression model showed a significant correlation between change in CRP and change in LDL (correlation, r=0.80, p<0.001). Overall, 89 to 98% of the reduction in CRP was related to the reduction in LDL, and 2 to 11% of the reduction in CRP was related to the effects of the statin, independent of the reduction in LDL.
The funnel plot and Begg's analysis showed a significant risk of publication bias. However, stratification for median reduction in LDL showed no evidence of bias, suggesting that the overall result was due to differences in LDL reduction across the various studies, rather than directly related to publication bias.