Among 70,130 Medicare beneficiaries aged 65+ hospitalized for myocardial or cerebral infarction, initiating atorvastatin (81% at high intensity) was associated with a sub-hazard ratio of 2.20 (95% CI 1.35–3.58) for acute hemorrhagic cerebrovascular disease versus initiating a different cardiovascular medication — translating to a number needed to harm of 351. Acute hepatic failure showed a sub-hazard ratio of 1.72 (NNH 468). Intermediate-strength signals emerged for biliary tract disease and musculoskeletal injuries; weaker signals appeared for cardiac valve disorders and sensory symptoms. Findings were confirmed via quantitative bias analysis and multiplicity correction across 14 nested daily trial emulations.
Statins remain among the most prescribed cardiovascular medications globally, and their net benefit post-infarction is well established. However, this preprint — not yet peer-reviewed — raises clinically meaningful questions about hemorrhagic risk, particularly in patients on concurrent antithrombotic therapy, where the valve disorder signal was notably amplified. The study's target trial emulation framework is a methodological strength, minimizing immortal-time bias common in claims analyses. Key limitations include its observational design (residual confounding remains plausible), restriction to fee-for-service Medicare (limiting generalizability to younger or privately insured populations), and the inherent constraints of claims-based outcome ascertainment. Clinicians should not discontinue statins based on a single unreviewed analysis, but these signals — particularly hemorrhagic stroke in anticoagulated post-stroke patients — warrant prospective validation. This is an incremental but pharmacovigilance-important contribution.