Decision tool and Sample Size Calculator for Composite Endpoints

Summary points:

This article considers the combination of two binary or two time-to-event endpoints to form the primary composite endpoint for leading a trial.
It discusses the relative efficiency of choosing a composite endpoint over one of its components in terms of: the frequencies of observing each component; the relative treatment effect of the tested therapy; and the association between both components.
We highlight the very important role of the association between components in choosing the most efficient endpoint to use as primary.
For better grounded future trials, we recommend trialists to always reporting the association between components of the composite endpoint.
Common fallacies to note when using composite endpoints: i) composite endpoints always imply higher power; ii) treatment effect on the composite endpoint is similar to the average effects of its components; and iii) the probability of observing the primary endpoint increases significantly.

Summary points:

This article considers the combination of two binary or two time-to-event endpoints to form the primary composite endpoint for leading a trial.
It discusses the relative efficiency of choosing a composite endpoint over one of its components in terms of: the frequencies of observing each component; the relative treatment effect of the tested therapy; and the association between both components.
We highlight the very important role of the association between components in choosing the most efficient endpoint to use as primary.
For better grounded future trials, we recommend trialists to always reporting the association between components of the composite endpoint.
Common fallacies to note when using composite endpoints: i) composite endpoints always imply higher power; ii) treatment effect on the composite endpoint is similar to the average effects of its components; and iii) the probability of observing the primary endpoint increases significantly.

文献摘要