Navarese et al recently reported a meta-analysis of randomized controlled trials concerning the effect of various types and doses of statins on new-onset diabetes mellitus (DM). The authors concluded that pravastatin 40 mg/day was associated with the lowest risk for new-onset DM compared with placebo. Carter et al reported a population-based cohort study concerning the effect of various types and doses of statins on new-onset DM by setting pravastatin as a control. Although the clinical evidence was weaker than that in randomized controlled trials, these authors concluded that higher-potency statins might be contributed to the increase of new-onset DM.

I have 2 concerns regarding Navarese et al’s study outcome. First, there were no significant differences in odds ratios versus placebo using a random-effects model, and the authors speculated as to what the potential effects in a powered head-to-head comparison would be. In their final paragraph of the Discussion, they cited a meta-analysis by Sattar et al and mentioned their study limitations. Sattar et al reported that subtotal odds ratio of pravastatin on new-onset DM did not reach significance. If there is no significance in odds ratios, the potential effect of statins on DM cannot be assumed. Mizuno et al reported on the complexity of the relationship between pravastatin therapy and new-onset DM, and their study was cited by Sattar et al. The meta-analysis conducted by compiling each study should be considered with caution, although a mathematical procedure can be applied. Second, the contribution rate (weight %) of each study on the total meta-analysis cohort is informative, and it is possible for the network meta-analysis to show the contribution rate. Further studies are needed for a high-quality meta-analysis to be possible.