Using latent class analysis to profile risk behaviors among sexual minority students
Yongwen Jiang, Jan Mermin, Tara Cooper, Rosemary Reilly-Chammat, Samara Viner-Brown
Introduction: The purposes of this study were (1) to summarize multiple related health indicators within a health domain using latent class analysis (LCA) to identify the patterns of risky behaviors and health conditions; (2) to profile high-risk behaviors and health conditions among sexual minority students. Methods: The Rhode Island Youth Risk Behavior Survey is a population-based high school survey conducted in odd years between 2007 and 2015 with a total sample size of 14,718. Thirty-four indicators of five domains were identified: safety and being bullied, depression and suicide, substance use, sexual behavior, and health conditions. LCA was used to categorize students into high- and low-risk classes based on similar patterns of indicators, then these two classes were cross-tabbed with sexual minority categories. The analysis was conducted in 2017. Results: Bisexual, lesbian and gay, and unsure (of sexual orientation identity) students had higher health-related risks and conditions than their heterosexual peers. Students who reported sexual contact were more likely to have engaged in risky behaviors than those who reported they never had sexual contact. Those who identified as heterosexual, but who had sexual contact with the same sex or both sexes, were more likely to have engaged in risky behaviors than those who identified as heterosexual, but who did not have sexual contact with the same or both sexes. Conclusions: The LCA method can be applied to identify and define risk behavior patterns among subgroups, which can improve the specification of high-risk populations and allocation of prevention resources.