Easily applicable multiple testing procedures to improve the interpretation of clinical trials with composite endpoints

BACKGROUND: Cardiology trials often consider composite endpoints as primary efficacy outcomes thereby combining several time-to-event variables in a single time-to-first-event measure. The main motivation to use a composite endpoint is to increase the number of expected events thereby reducing the required sample size. However, interpretation may be difficult as the effect observed for the composite endpoint does not necessarily reflect the effects for the single components. To improve interpretation, it is therefore a current standard to analyze the individual components in a descriptive way. However, a descriptive analysis does not allow a statistical proof of concept. Therefore the gain in information is limited.

METHODS: This paper systematically explores multiple testing procedures aimed at improving the interpretation of composite endpoints by confirmatory tests of the components. A simulation study demonstrates, on the basis of a real cardiology clinical trial example, the benefit of these easily applicable multiple testing procedures.

RESULTS: By applying adequate multiple testing strategies to assess the components of a composite endpoint there is a high chance to get additional confirmatory evidence on the components without the need to increase sample size. With a moderate increase in sample size, a gain in evidence can often also be ensured with a predefined power.

CONCLUSION: The interpretation of composite endpoints can be improved by applying multiple testing procedures that assess the components. The methods discussed here are easy to apply and provide a substantial benefit for clinical interpretation of study results.