### Abstract

© The Author 2017. Although composite endpoints (CE) are common in clinical trials, the impact of the relationship between the components of a binary CE on the sample size requirement (SSR) has not been addressed. We performed a computational study considering 2 treatments and a CE with 2 components: the relevant endpoint (RE) and the additional endpoint (AE). We assessed the strength of the components' interrelation by the degree of relative overlap between them, which was stratified into 5 groups. Within each stratum, SSR was computed for multiple scenarios by varying the events proportion and the effect of the therapy. A lower SSR using CE was defined as the best scenario for using the CE. In 25 of 66 scenarios the degree of relative overlap determined the benefit of using CE instead of the RE. Adding an AE with greater effect than the RE leads to lower SSR using the CE regardless of the AE proportion and the relative overlap. The influence of overlapping decreases when the effect on RE increases. Adding an AE with lower effect than the RE constitutes the most uncertain situation. In summary, the interrelationship between CE components, assessed by the relative overlap, can help to define the SSR in specific situations and it should be considered for SSR computation.

Original language | English |
---|---|

Pages (from-to) | 832-841 |

Journal | American Journal of Epidemiology |

Volume | 185 |

Issue number | 9 |

DOIs | |

Publication status | Published - 1 May 2017 |

### Keywords

- Association measures
- Binary endpoints
- Composite endpoint
- Correlated endpoints
- Outcome assessment
- Sample size

## Fingerprint Dive into the research topics of 'The use of a binary composite endpoint and sample size requirement: Influence of endpoints overlap'. Together they form a unique fingerprint.

## Cite this

Marsal, J. R., Ferreira-González, I., Bertran, S., Ribera, A., Permanyer-Miralda, G., García-Dorado, D., & Gómez, G. (2017). The use of a binary composite endpoint and sample size requirement: Influence of endpoints overlap.

*American Journal of Epidemiology*,*185*(9), 832-841. https://doi.org/10.1093/aje/kww105