STATISTICAL DESIGN AND MONITORING OF THE CAROTENE AND RETINOL EFFICACY TRIAL (CARET)

CARET is a chemoprevention trial of beta-carotene and vitamin A with l ung cancer as the primary outcome. Participants at high risk for lung cancer are drawn from two populations: asbestos-exposed workers and he avy smokers. The intervention is a daily combination of 30 mg beta-car otene and 25,000 IU vitamin A as retinyl palmitate. Nearly 18,000 part icipants will be followed for a mean 6 years, yielding over 100,000 pe rson-years of follow-up. We project that this sample size will have 80 % power to detect a 23% decrease in the incidence of lung cancer cases . The purpose of this paper is to present the values of the key sample size parameters of CARET; our schemes for monitoring CARET for sample size adequacy, incidence of side effects, and efficacy of the study v itamins; an overview of the data collected; and plans for the primary, secondary, and ancillary analyses to be performed at the end of the t rial. These approaches to the design, monitoring, and analysis of CARE T are applicable for many other prevention trials.