Health outcomes of non-heterosexual compared to heterosexual adolescents and young adults from a national survey in viet nam: Analysis using propensity score matching

Methodological background: A challenge in LGBT health disparities research is lack of comparable LGBT and non-LGBT samples. Even with population samples (preferred over separately recruited LGBT and non-LGBT samples), confounding remains problematic due to small LGBT percentages. Geographical background: No prior research examined LGBT health disparities in Viet Nam. Methods: We analyzed data from a national survey of adolescents and young adults in Viet Nam. 160 male and 155 female participants reported being attracted to the same or both genders. These non-heterosexual individuals were matched to heterosexual (attracted to the opposite gender) individuals on age, minority/majority ethnicity, urban/rural location, regions, education, work status, family assets, and parents' education. Variable ratio matching was used. Matching combined propensity score with caliper, Mahalanobis distance on continuous variables, and exact matching on important categorical variables, using the R-package MatchIt. We matched in several steps to acquire matches for all non-heterosexual individuals. Risk ratios of various health outcomes were evaluated using matched samples. Results: 1482 male and 622 female heterosexual persons were selected as matches, with excellent balance of propensity score and all characteristics. Non-heterosexual females are more likely than heterosexual females to smoke (RR=20, 95% CI 1.25-318) and to have been injured by family members (RR=4.08, CI 1.79-9.30). Non-heterosexual males are more likely than heterosexual males to have thought about suicide (RR=3.78, CI 1.75-8.13). Both non-heterosexual males and females are more likely than heterosexuals to have been very depressed (RR=1.40, CI 1.08-1.82, and RR=1.36, CI 1.11-1.65, respectively).