Causal interpretation of the hazard ratio in randomized clinical trials.

ABSTRACT:

Background

Although the hazard ratio has no straightforward causal interpretation, clinical trialists commonly use it as a measure of treatment effect.

Methods

We review the definition and examples of causal estimands. We discuss the causal interpretation of the hazard ratio from a two-arm randomized clinical trial, and the implications of proportional hazards assumptions in the context of potential outcomes. We illustrate the application of these concepts in a synthetic model and in a model of the time-varying effects of COVID-19 vaccination.

Results

We define causal estimands as having either an individual-level or population-level interpretation. Difference-in-expectation estimands are both individual-level and population-level estimands, whereas with

SUBMITTER: Fay MP 

PROVIDER: S-EPMC11502288 | biostudies-literature | 2024 Oct

REPOSITORIES: biostudies-literature