Background and objectives: Results of prospective cohort studies can be biased when subjects selectively refuse to participate or be included in follow-up.
Goal of this study: To assess the potential for bias in a longitudinal study of sexual risk behavior among men who have sex with men.
Study design: This was a cross-sectional comparison of clinical data regarding men who have sex with men attending an urban human immunodeficiency virus testing clinic.
Results: Of 3,390 men who have sex with men invited to participate, 2,063 refused, 589 dropped out after completing an initial study questionnaire, and 738 participated in follow-up at 6 months. There were no significant differences in the same-gender sexual behaviors of participants, dropouts, and nonparticipants, with one exception: Nonparticipants were more likely to abstain from receptive oral sex (27%) compared with participants (18%) or dropouts (21%).
Conclusion: The similarities in reported activities among participants, dropouts, and nonparticipants suggest that selection bias may have limited impact on cohort studies of sexual behavior.