As the number of individuals eligible for statin therapy continues to increase, researchers conducted this study to systematically review and synthesise the totality of the RCT evidence on different statins and determine their comparativetolerability and harms across a range of populations eligible for statin therapy.
Overall, 55 two-armed placebo-controlled and 80 two- or multiarmed active-comparator trials in 246 955 individuals with and without cardiovascular disease were included in the analysis.
The following findings were reported:
• Individual statins were not different than control in terms of myalgia, creatine kinase elevation, cancer, and discontinuations because of adverse events.
• Statins as a class resulted in significantly higher odds of diabetes mellitus (odds ratio, 1.09; 95% CI, 1.02– 1.16) and transaminase elevations (1.51; 1.24–1.84) compared with control.
• When individual statins were compared in network meta-analyses, there were numerous statistically detectable differences, favouring simvastatin and pravastatin.
• According to dose-level comparisons, individual statins resulted in higher odds of discontinuations with higher doses of atorvastatin and rosuvastatin. Similarly, higher doses of atorvasatin, fluvastatin, lovastatin, and simvastatin were associated with higher odds of transaminase elevations. Simvastatin at its highest doses was associated with creatine kinase elevations (odds ratio, 4.14; 95% credible interval, 1.08–16.24).
• Meta-regression analyses adjusting for study-level age at baseline, LDL-cholesterol, and publication year did not explain heterogeneity. There was no detectable inconsistency in the network.
The authors advise caution in interpreting these findings in light of the following limitations:
1. The quality of included trials was moderate, with older trials being more prone to bias than newer trials.
2. Given the large volume of available studies in the literature, the meta-analysis did not use individual patient-level data, which would have advantages when exploring potential differences across relative treatment effect modifiers.
3. Although there was no evidence of small-study effects, there was an apparent asymmetry in the evidence network where specific interventions seem to be avoided (e.g. fluvastatin).
4. There was considerable heterogeneity across various pairwise meta-analyses of statins versus control, particularly for hepatic transaminase elevations.
In spite of these limitations, the authors suggest their study has important methodological strengths such as the size of the study, the inclusion of a comprehensive list of trials, irrespective of placebo or active controls, including all clinically used statins, and evaluation of the dose comparative harms of individual statins. They conclude that their study provides strong evidence that statins as a class are generally safe with uncommon side effects; and in addition, there is consistently strong evidence on the comparatively favourable side effect profile of simvastatin and pravastatin, particularly at low-to-moderate doses, which they believe should be favoured in clinical practice.