Using the Time-Varying Effect Model to Examine Changes in Predictors of Multiple Sexual Partners from Adolescence through Young Adulthood

Rates of STIs are relatively high in adolescence and young adulthood, and having multiple sexual partners increases STI risk.  A number of predictors of multiple sexual partners have been documented, including substance use and depressive symptoms.  However, little is known about how these predictors change over time as individuals move through adolescence and into young adulthood.  We apply an innovative analytic method, the time-varying effect model (TVEM) to examine how associations between multiple sexual partners and substance use and depressive symptoms change over developmental time.

Methods: Data are from the contractual sample of the National Longitudinal Study of Adolescent Health (Add Health), using data from all four waves ranging from adolescence to young adulthood, including all waves when participants were ages 14 to 32 and unmarried.  Our resulting sample includes 11,963 individuals (52.2% female, M age at Wave 1=16.1 years).  We examined how odds of multiple sexual partners in the past year were differentially predicted by substance use (past year heavy episodic drinking, past month smoking, past month marijuana use) and depressive symptoms, and how these associations varied by gender.  Analyses were performed using TVEM, a flexible, semi-parametric model that estimates associations between a predictor and outcome in near-continuous time.

Results: The proportion of participants having multiple partners increased over time, leveling off at around 30% after age 20. Significant positive associations between substance use and multiple partners were strongest early in adolescence, and decreased sharply by around age 18.  For example, male participants who smoked marijuana had five times greater odds of multiple partners at age 14, compared to three times greater odds at age 20 and two times greater odds at age 26.  Associations were similar for male and female participants, apart from a short period in the late teens, during which the effect of substance use on odds of multiple partners was stronger for female adolescents compared to male adolescents.  The significant positive association between depression and sexual behavior decreased over time, remaining significant in young adulthood for women, but not men.

Conclusions: These findings suggest the strength of predictors of sexual behavior changes from early adolescence through young adulthood.   Prevention programs aimed at substance use and sexual behavior may be effective earlier in adolescence, whereas other types of interventions may be more effective in young adulthood.  TVEM may be useful in helping prevention scientists to understand what predictors are especially salient at different ages, and what periods may be particularly important times to intervene.