Editor’s Note: Special Section on Estimands, Design and Analysis of Clinical Trials with Time-to-Event Outcomes

Abstract

In several disease areas, such as cardiovascular disease, oncology/cancer or HIV, clinical trials often collect and analyze multiple time-to-event (or survival) outcomes from patients to assess the effects of interventions. Methods for time-to-event outcomes are more complex than for binary or continuous outcomes. The design, monitoring, analysis and reporting of clinical trials with time-to-event outcomes (time-to-event clinical trials) will require considerable care. A common practice in time-to-event clinical trials is to first create a composite endpoint that combines several clinically relevant time-to-event outcomes (e.g., major adverse cardiovascular events (MACE), consisting of death, myocardial infarction, and stroke in cardiovascular disease; progression free survival (PFS) consisting of time-to-progression and overall survival), and then to perform a time-to-first-event analysis for the composite endpoint. The advantages and challenges of using composite endpoints are well known and have been discussed in the statistical and medical literature. Recently, many statisticians have attempted to redefine the estimand(s) of interest to capture the effects of interventions and the corresponding estimators of the estimand(s) (statistical methods) since the implementation of the estimand framework highlighted in the ICH-E9(R1) guideline (ICH 2019). Common survival analysis methods, such as Kaplan-Meier method, log-rank test, or Cox proportional hazards regression, have many strengths and are well accepted in practice. However, there are situations in which they may not be feasible or provide reliable results. The common methods are based