Abstract P2018: Re-evaluating the Effect of Pravastatin on Mortality and Cardiovascular Events Using G-methods to Adjust for Adherence in the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial – Lipid Lowering Trial

Abstract

Introduction: The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack – Lipid-Lowering Trial (ALLHAT-LLT) was a large-scale, pragmatic clinical trial comparing pravastatin to usual care. In contrast to placebo-controlled statin trials, ALLHAT-LLT did not find an effect on all-cause mortality when data were analyzed under the intention-to-treat (ITT) principle. Objective: We replicated the ITT analysis and estimated the per-protocol effect of pravastatin on the 5-year risk of all-cause mortality and major adverse cardiovascular events (MACE). Methods: To replicate the ITT analysis, we used discrete time methods (pooled logistic regression). To estimate the per-protocol effect, we censored individuals when they deviated from protocol, i.e., in the pravastatin arm when they discontinued the study drug in the absence of adverse events. In the usual care arm, following the trial’s protocol, we considered the initiation of lipid-lowering therapy to be clinically indicated and not a deviation from protocol. We adjusted for baseline and time-varying factors associated with adherence using inverse probability weighting. Our outcomes of interest were all-cause mortality and a composite endpoint of MACE (myocardial infarction, stroke, heart failure, and cardiovascular mortality). Results: Of 10,355 individuals randomized, we included 9,741 individuals who had complete baseline data. This restriction did not result in an imbalance between the randomization groups, nor substantial changes in the distribution of characteristics compared to the original trial. The ITT findings were consistent with the original trial, (hazard ratio 1.00; 95% confidence interval (CI) 0.88, 1.13). Over one-third (35%) of individuals in the pravastatin arm were nonadherent to protocol at year 5. After adjusting for this, we found a 5-year risk difference of -2.99 (-4.94, -1.46) and risk ratio of 0.79 (0.68, 0.89) for all-cause mortality and -3.56 (-7.06, 0.18) and 0.82 (0.66, 1.01) for MACE, comparing pravastatin to usual care. Findings were consistent across sensitivity analyses modifying definitions of adherence and missing data methods. Conclusion: We found that use of pravastatin was protective against all-cause mortality and cardiovascular events after adjusting for adherence to protocol. Results support the use of g-methods to adjust for adherence when interested in understanding the effect of dynamic, sustained treatments in pragmatic trials.