Statin-related side effects: the recent Lancet publication is biased toward false-negatives

Product monographs do present the absolute rates of AEs recorded in clinical trials, for both trial arms. This is helpful. The less helpful portion of the PM is the “postmarket AEs” section- a long list of AEs based on spontaneous postmarket reports, for which little evidence of causality might exist. The Lancet authors seem to be suggesting that causality for listed postmarket AEs needs to be reassessed now that clinical trials have accrued sufficient numbers of these AEs.

An important caveat: between-arm AE rate comparisons using clinical trial databases will not be sufficient for detecting ALL possible drug-related harms. Specifically, idiosyncratic harms can still occur for many drugs on the market (e.g., idiosyncratic drug-induced liver injury). These AEs will usually be so rare as to be undetectable in typically-sized clinical trial databases. If an AE case report is very well documented and contains important historical details like positive dechallenge and/or positive rechallenge, we can still make inferences of causality at the level of individual patients. AEs like this should be flagged in PMs, but their mention should include appropriate caveats about their extreme rarity and (in most cases) unpredictability.