Skip to content


Estimands for Clinical Trial

Chair: Huiman Barnhart (Duke University) and Laine Thomas (Duke University)

Speaker: Chao Cheng, PhD (Yale University)
Title: Multiply Robust Estimation for Causal Survival Analysis with Treatment Noncompliance
Comparative effectiveness research frequently addresses a time-to-event outcome and can require unique considerations in the presence of treatment noncompliance. Motivated by the challenges in addressing noncompliance in the ADAPTABLE pragmatic trial, we develop a multiply robust estimator to estimate the principal survival causal effects under the principal ignorability and monotonicity assumption. The multiply robust estimator involves several working models including that for the treatment assignment, the compliance strata, censoring, and time-to-event of interest. The proposed estimator is consistent even if one, and sometimes two, of the working models are misspecified. We apply the multiply robust method in the ADAPTABLE trial to evaluate the effect of low- versus high-dose aspirin assignment on patients’ death and hospitalization from cardiovascular diseases. We find that, comparing to low-dose assignment, assignment to the high-dose leads to differential effects among always high-dose takers, compliers, and always low-dose takers. Such treatment effect heterogeneity contributes to the null intention-to-treatment effect, and suggests that policy makers should design personalized strategies based on potential compliance patterns to maximize treatment benefits to the entire study population. We further perform a formal sensitivity analysis for investigating the robustness of our causal conclusions under violation of two identification assumptions specific to noncompliance.

Speaker: Bo Liu (Duke University)
Title: Principal Stratification Analysis of Noncompliance with Time-to-Event Outcomes
Post-randomization events, also known as intercurrent events, such as treatment noncompliance and censoring due to a terminal event, are common in clinical trials. Principal stratification is a framework for causal inference in the presence of intercurrent events. The existing literature on principal stratification lacks generally applicable and accessible methods for time-to-event outcomes. In this paper, we focus on the noncompliance setting. We specify two causal estimands for time-to-event outcomes in principal stratification and provide nonparametric identification formula. For estimation, we adopt the latent mixture modeling approach and illustrate the general strategy with a mixture of Bayesian parametric Weibull-Cox proportional model for the outcome. We utilize the Stan programming language to obtain automatic posterior sampling of the model parameters. We provide analytical forms of the causal estimands as functions of the model parameters and an alternative numerical method when analytical forms are not available. We apply the proposed method to the ADAPTABLE trial to evaluate the causal effect of taking 81 mg versus 325 mg aspirin on the risk of major adverse cardiovascular events. We develop the corresponding R package PStrata.

Speaker: Hillary Mulder (Duke University)
Title: Estimation of the Hypothetical Effect of Aspirin Dosing with Perfect Adherence; Re-analysis of the ADAPTABLE Clinical Trial
Abstract: The ADAPTABLE (Aspirin Dosing: A Patient-Centric Trial Assessing Benefits and Long-term Effectiveness) trial was a large, multi-center, pragmatic, open-label randomized controlled trial, comparing aspirin 81mg vs 325mg for the secondary prevention of cardiovascular events in patients with established ASCVD.  Interpretation of the randomized comparison, showing non-significant differences, was complicated by substantial differential dose switching (6.1% in the 81mg and 33.5% in the 325mg groups) and discontinuation of aspirin (10.4% in the 81mg and 15.8% in the 325mg groups). These intercurrent events may have attenuated the effect of dosing by making the actual treatment received more similar between the groups. The reasons for non-adherence were likely preference related and not related to clinical efficacy. Therefore it remains relevant to estimate whether significant differences would be observed if all subjects adhered to their randomized assignment.  We re-analyzed the ADAPTABLE study using inverse probability of censoring weights (IPCW) to estimate the effect of 81mg versus 325mg dosing under perfect adherence, while accounting for measured differences between those who adhere and those who don’t.  Specifically, this approach censors patients at the time of discontinuation or switching, when their observed data becomes inconsistent with the question of interest.  The probability of censoring, for discontinuation or switching, is modeled as a function of baseline clinical risk factors and incorporated into an inverse censoring weight.  This approach is valid under a missing at random (MAR) type assumption, whereby non-adherence is unrelated to outcome given the baseline clinical risk factors. We discuss the plausibility of this assumption in ADAPTABLE and interpret the findings.