Analysis and Design of Clinical Trials Using Additive Hazards Survival Endpoints

Abstract–The additive hazards model provides benefits in interpretation and theoretical properties over the proportional hazards model, particularly in noninferiority trials. We describe techniques for designing and analyzing trials with a score-type test using the additive hazards model, with a focus on noninferiority trials. Accordingly, the techniques are illustrated using two noninferiority trials, the SPORTIF III trial of anti-coagulants for patients with atrial fibrillation, and the CPORT-E trial evaluating percutaneous coronary intervention in different clinical setting. We also derive a powerful test, which can be conducted using standard statistical software to distinguish between the additive and proportional hazards models. This test is shown to be effective in simulated data and applied to data from a study of melanoma as well as a study of colon cancer.