Generating comparative data on clinical benefits and harms of statins to inform prescribing decisions: evidence from network meta-analyses

摘要

Background and Importance: comparative evidence generated using systematic reviews and meta-analyses can form the basis of high quality prescribing decisions in clinical practice. Such evidence is imperative when choosing a first-line treatment among multiple alternatives, particularly in the United States where there is no single national authority responsible for providing practice guidelines for prescribers.udObjective: Using cholesterol-lowering statins as a case study, this thesis set out to evaluate the comparative clinical benefits and harms of statins for the prevention of coronary heart disease.udNovelty and Empirical Contribution: The empirical work presented in this thesis was based on a systematic review and network meta-analysis, for the first time combining the placebo-controlled and active-comparator trials of statins. Using 184 randomized trials including 260,630 individuals with or without cardiovascular disease, this thesis makes four major contributions to the literature on the comparative effectiveness and safety of statins, showing the following:(1) cholesterol-lowering effects of statins are less pronounced than suggested by the previous reviews; (2) statins potentially differ in terms of their comparative effects on clinically meaningful benefit outcomes, which are not fully explained by their cholesterol-lowering effects;(3) harms associated with statins are rare; still, some statins are safer than others; and (4) unlike previous findings in the literature, there is no evidence of industry sponsorship bias affecting the trials of statins.udImplications for Clinical Practice: Although there are statistically detectable and clinically relevant differences among individual statins, the empirical work presented in this thesis does not conclusively identify a clear winner among statins that should be favored in clinical practice.udFuture&Research Directions: The potential mechanisms underlying the observed differences between individual statins should be investigated in future studies.udPolicy Relevance: The findings presented in this thesis suggest that statin prescribing patterns over the past decade – and in particular atorvastatin’s exceptional sales performance despite its equivalence to simvastatin – are not supported by the current best evidence. A proposed policy option is to raise the bar for market entry of new drugs by requiring comparative evidence at the time of approval decisions. Network meta-analysis methods can be used at the United States Food and Drug Administration setting, thereby making comparative evidence available before prescribing patterns are established.