Estimand-based design strategies for cardiovascular recurrent event trials of combination therapies

Cardiovascular trials increasingly evaluate combination therapies in settings where patients may experience recurrent events such as myocardial infarction, heart failure hospitalization, or stroke. Traditional single-event analyses fail to capture the cumulative disease burden and temporal dynamics, motivating the integration of recurrent event methods within an estimand framework. This manuscript reviews estimand strategies for recurrent event analyses in cardiovascular trials of combination therapies, emphasizing alignment between clinical questions, intercurrent events, and causal interpretation. We compare commonly used models, including the Andersen-Gill, Prentice-Williams-Peterson, and Negative Binomial approaches, highlighting their suitability for different estimands. We distinguish each model's conditional or marginal estimand, its identifiability assumptions, and the difference between sensitivity analyses that vary assumptions under the same estimand and supplementary analyses that target alternative estimands with different intercurrent event strategies. The impact of intercurrent events such as treatment discontinuation, rescue therapy, and mortality on estimand interpretation is explored. Novel methodological considerations specific to combination-therapy trials, including component-specific intercurrent events (ICE), drug-drug interaction estimands, mixed intercurrent event (ICE) strategies, and principal-stratum estimands conditional on dual adherence, are formally defined and discussed with work examples. We further discuss how the choice of estimand influences causal inference and model selection. Finally, we provide a regulatory perspective on estimand-based approaches in cardiovascular trials. This manuscript serves as a methodological guide for designing and analyzing cardiovascular trials with recurrent events for combination therapies to ensure clarity of clinical interpretation, robustness of causal conclusions, and regulatory alignment.