Statisticians have proposed a new method for estimating treatment effects in clinical trials when intercurrent events that themselves take time to occur complicate the analysis. The approach, described in BMC Medical Research Methodology, unifies the hypothetical strategy for handling intercurrent events with causal mediation analysis under a semi-competing risks framework, and it targets what is known as the controlled direct effect.
Intercurrent events—occurrences after treatment initiation, such as a treatment-related death, that complicate or preclude observation of the outcome of interest—remain an unresolved methodological issue in time-to-event settings. The ICH E9 (R1) addendum proposed five strategies for handling such events, one of which is the hypothetical strategy: estimating the effect that would have been observed in a hypothetical scenario in which the intercurrent event does not occur. The authors note that hypothetical estimands align with the controlled direct effect in causal mediation, but that no current study had shown how to estimate this effect when the intercurrent event is a time-to-event outcome.
To address this gap, the research team defined the effect of treatment under a hypothetical scenario in which the intercurrent event does not occur as the controlled direct effect within a semi-competing risks framework, a structure suited to settings where one event type can censor or modify the risk of another. They then developed a generalized Nelson–Aalen estimator and evaluated its performance on three scales: the cumulative hazard, survival, and restricted mean survival time (RMST) scales. The evaluation was carried out in a structured simulation study based on a cardiovascular trial setting.
The simulation results, drawn from 760 replicates, showed small bias across prespecified follow-up times. Relative bias ranged from −4.54% to 2.83% on the cumulative-hazard scale, from −5.13% to 1.60% on the survival scale, and from −8.84% to −2.79% on the RMST scale. The hypothetical, treatment-policy, and composite-variable estimators each closely tracked the true values of their corresponding estimands, with coverage probabilities ranging from 93.42% to 95.79% on the cumulative-hazard scale.
A key contribution of the study lies in quantifying how alternative strategies diverge from the hypothetical target. The researchers summarized the differences between the treatment-policy or composite-variable estimates and the controlled direct effect estimand as estimand discrepancies. Their simulations showed that the treatment-policy and composite-variable estimands displayed different degrees of discrepancy from the controlled direct effect over the follow-up period, meaning that the choice of strategy can materially change the interpretation of trial results when time-to-event intercurrent events are present.
The method was also illustrated in a case study involving a comparison between ALL and AML groups, where the three estimand strategies produced different estimates of the contrast in treatment-related death between the groups. This practical example underscores the authors’ conclusion that the treatment-policy and composite-variable strategies can yield discrepant answers relative to the hypothetical controlled direct effect, so analysts must be explicit about which question they intend to answer.
The authors conclude that the proposed estimator showed unbiased estimation for the hypothetical controlled direct effect under the simulation settings considered. Because the performance was assessed in simulations calibrated to a cardiovascular trial context and in a single case study, the findings apply to the settings examined; the work is presented as a methodological contribution to trial design and analysis rather than as evidence about the efficacy of any particular treatment.
Subject of Research: A new statistical estimator for handling time-to-event intercurrent events in clinical trials using controlled direct effects under a semi-competing risks framework.
Article Title: Estimating hypothetical estimands with time-to-event intercurrent events using controlled direct effect under semi-competing risk framework
Article References: Wu, K., Feng, S., Wang, T., & Chen, W. (2026). Estimating hypothetical estimands with time-to-event intercurrent events using controlled direct effect under semi-competing risk framework. BMC Medical Research Methodology. https://doi.org/10.1186/s12874-026-03039-0
Image Credits: AI Generated
DOI: 10.1186/s12874-026-03039-0
Keywords: clinical trials, estimands, intercurrent events, hypothetical strategy, controlled direct effect, causal mediation, semi-competing risks, Nelson–Aalen estimator, restricted mean survival time, ICH E9 (R1), simulation study, biostatistics
News Source: Ophelia Keating. (October 29, 2026). New Estimator Targets Hypothetical Trial Effects When Time-to-Event Events Intervene. Scienmag.
