Designing a large prevention trial: statistical issues

Abstract

Recent research in Alzheimer's disease (AD) is centred about the early detection and prevention of this disease. Several recent moderate size clinical trials targeted at high risk cohorts have been designed along this theme. There have been few attempts to design a large trial to prevent this disease in elderly individuals at low risk for the disease. The purpose of this paper is to suggest a framework for designing a simple, large AD prevention trial. This framework uses a discrete time hazard model for decreasing the incidence of AD when participants are randomly assigned to one or more active prevention agents or placebo. This design allows for differential incidence among participants due to age, family history, genetic disposition, and ethnicity. It takes into account the length of the follow-up period, participant mortality, drop-outs, drop-ins, and loss to follow-up. This framework is illustrated by PREADVISE, a recently initiated large add-on prevention trial investigating the use of anti-oxidants for preventing AD among men enrolled in a even larger prostate cancer prevention study, SELECT.