Behavioral change in longitudinal studies: Adoption of condom use by homosexual/bisexual men

Abstract

We compared reporting serial cross-sectional prevalence of sexual behavior over time, to reporting individual patterns of behavioral change in a cohort of homosexual men at a six-month interval. Aggregate prevalence rates underestimated the magnitude of change to safer practices, and failed to provide information on relapse to less safe practices. We conclude that it is important to report data based on individual fluctuations in behavior for the evaluation of change over time.