Predicting adherence to a dietary regimen in a cancer prevention trial

Abstract

In light of the importance of preventive measures to effectively reduce the mortality rate associated with colon cancer, and the importance of adherence to treatment in the study of cancer prevention, the aim of this study was to identify variables that explain adherence to treatment in a long term, colon cancer prevention clinical trial. The Wheat Bran Fiber study is a double-blinded, placebo-controlled trial evaluating the efficacy of daily wheat bran fiber supplementation on polyp recurrence in 1,400 randomized persons with adenomatous colon polyp history. Adherence was defined as consumption of the cereal regimen during the specific evaluation period. It was evaluated through the count of returned unopened cereal boxes, and self-monitoring. Data were collected prospectively and discretely during clinic visits that took place approximately every three months. The Life-Style Questionnaire was constructed as a theory driven instrument with three major components: past adherence behavior, social network and self-motivation. I was given a baseline to 910 participants. The evaluation of potential predictors of attrition was conducted using survival analysis techniques. Specifically, a Weibull distribution was used to model the data. Demographic characteristics, physiological changes, and psychological variables constituted the explanatory variables of the model. The two most important predictors of attrition in this study, for both males and females, were exposure to high fiber and experiencing side effects, whether or not study related. Other significant predictors were self-motivation and adherence at baseline for males, and self-motivation, smoking history, and study expectations for females. One of the most interesting findings is the difference in risk between males and females. The higher hazard for females when toxicities were present suggests that women have more difficulties managing side effects. Also, smoking history was not a significant predictor of the hazard for males. Although the reason of such difference is unknown, it can be speculated that the motivations for and the circumstances under which males and females smoke might be completely different. Overall, these results confirm the multifactorial determination of attrition as a time-to-event phenomenon.