Retention of Black and White Populations in the Selenium and Vitamin E Cancer Prevention Trial

Abstract

Abstract Background: Adequate retention of Black (African-American) participants in long-term, randomized trials is important for achieving broadly applicable results. Purpose: To determine the incidence of retention failures and the individual and study site factors associated with retention failure among White and Black participants from the Selenium and Vitamin E Cancer Prevention Trial (SELECT), a phase III study of selenium and vitamin E for prevention of prostate cancer. Methods: SELECT randomized 35,533 participants from 427 study sites. Age eligibility was >55 years for Whites, >50 years for Blacks; race was determined by self-report. The primary analysis included 28,118 (87%) White and 4,322 (13%) Black men. Time to retention failure was defined as days to the earliest of either (1) the second consecutive missed in-person or phone visit or (2) refusal of future contact with study staff. Covariates included participant demographic and clinical characteristics, reasons for participating in SELECT and study site characteristics. Results: In SELECT, Blacks had a higher age-adjusted risk of disengagement than Whites [HR = 1.9; 95% CI 1.7–2.0; p-value < .0001]. Younger Black participants, age 50–54, representing 32% of Black participants, were at three times the risk of disengagement than those age 65+ [HR = 3.34, 95%CI 2.24,4.94, p < .0001]. Blacks age 65+ had 1.6 times the risk of disengagement than Whites age 65+ [HR = 1.58, 95%CI 1.37, 1.83, p < .0001]. By six years post-randomization, 83% of Whites were still retained on the study compared to 68% of Blacks. Among Whites, participant risk factors for disengagement were Hispanic ethnicity, graduate degree, living alone or current smoking; among Blacks, younger age and current smoking were risk factors. The site characteristic associated with disengagement among both racial groups was staff missing SELECT training sessions. Sites with a low staff attendance at training sessions had a high percentage of disengaged participants of either race. Both Whites and Blacks at sites receiving SELECT retention and adherence grants had lower risks for retention failure. Conclusion: These results may be useful in identifying personal and study site characteristics to guide recruitment and enhance retention of both Black and White men in long-term prevention trials.