Examining the components of population-level sexual behavior trends from 1993 to 2007 in an open ugandan cohort.

INTRODUCTION: Sexual behavior changes are widely cited as contributing factors to sexually transmitted disease trends. We explore a rarely examined aspect of behavior trends in an open cohort--the relative impact of individuals' changing reported behavior versus new responses due to a changing respondent base. METHODS: Respondents from an open cohort in Uganda annually answer questions on sexual behavior. We describe the impacts on behavior trends of: respondents' changing reported behavior, migration, mortality, changing eligibility for indicator inclusion, changing survey participation, and misreporting. We report contributions to trends on the following factors: condom use, ever had sex, age at first sex, and number of sexual partners and casual partners. RESULTS: Main trend contributions varied by indicator. Condom use trends were influenced by individuals' changing responses and by increasing condom use among in-migrants and newly interviewed people. Sexual partners were driven by fewer partners among newly interviewed people, although increase of partners in 1999, 2004, and 2006 stemmed mainly from people changing answers. Thirty-nine percent of responses to age at first sex among 17- to 20-year-olds were inconsistent--different ages in different years. Early trends in the factor "ever had sex" among 15- to 19-year-olds were driven by people changing their answers--including ever to never, an impossible sequence. Comparing behavior in one year to mortality in the next, we found little evidence of higher mortality among higher risk takers. DISCUSSION: In an open cohort, various factors contribute to sexual behavior trends. When reporting sexual behavior trends, researchers should acknowledge the contributing factors and attempt to separate the role of interindividual versus intraindividual changes.