Abstract WP229: The association between HMG-CoA reductase inhibitor use and the risk of cerebral hemorrhage: a target trial emulated using cardiovascular health study cohort with parametric g-formula

Introduction: HMG-CoA reductase inhibitors are widely used for prevention of ASCVD, yet their effect on hemorrhagic stroke risk remains controversial. Some studies report no increased risk, while others suggest harm or protection. These inconsistencies may reflect short follow-up in RCTs, limited reporting of hemorrhagic stroke events, and inadequate adjustment for co-medications. Emulating a target trial with parametric g-formula can better address these gaps by accounting for time-varying confounding and mimicking clinical treatment strategies. Methods: We used data from the Cardiovascular Health Study, a prospective cohort of U.S. adults aged ≥65 years recruited between 1989–1993 and followed through 2012. Eligibility was restricted to participants meeting ACC ASCVD risk–based criteria for HMG-CoA reductase inhibitor therapy. Baseline was CHS Year 2 (1989–1990), with time zero at Year 3 (1990–1991), when participants were classified as HMG-CoA reductase inhibitor users or non-users. Hemorrhagic stroke was defined as neuroimaging-confirmed hemorrhage, bloody cerebrospinal fluid, stroke-related death within 24 hours, or autopsy evidence. Two strategies were compared: (1) initiate and continue HMG-CoA reductase inhibitors, and (2) remain medication-free. Follow-up continued until hemorrhagic stroke, death, loss, or study end. The parametric g-formula estimated 22-year risks under each strategy, adjusting for time-varying confounders (antithrombotics, anticoagulants, antihypertensives, and non-statin lipid-lowering agents), baseline demographics, and comorbidities. Results: Of 5,888 CHS participants, 5,128 met eligibility criteria and were included in analyses and there were 89 hemorrhagic stroke events. The estimated absolute risk of hemorrhagic stroke under the natural course (no therapy throughout the follow-up) is 2.45 (95% CI: 1.77, 3.10), compared with 1.30 (95% CI: 0.28, 99.71) among HMG-CoA reductase inhibitor users. The risk of hemorrhagic stroke under the hypothetical use of HMG-CoA reductase inhibitors throughout the follow-up time is 0.53 times (95% CI: 0.10,41.09) compared with the risk under the natural course. Conclusions: We found no evidence of an increased risk of hemorrhagic stroke associated with HMG-CoA reductase inhibitor use. This study demonstrates the utility of the parametric g-formula for addressing time-varying confounding in long-term observational data, particularly in the context of complex medication use.