Sexual health interventions: a meta-analysis

In the second of two companion papers, we conducted a meta-analysis of sexual health interventions in three domains. The interventions chosen for the meta-analysis were a subset of studies presented in a narrative review (the first of the two companion papers); these in turn were selected on the basis of fit to principles derived from the World Health Organization (WHO) and other definitions of sexual health. Studies (n=20) were drawn from Medline and PsycINFO databases (English language, adult populations, 1996-2011) and fell into three domains: knowledge, attitudes, and sexual behaviors. We estimated intervention effects via Hedges' g, using the random-effects approach. Initial estimates revealed a large effect for knowledge, g=1.32 (95% CI=0.51-2.14), and smaller effects for attitude change, g=0.17 (0.11-0.24) and behavior, g=0.21 (0.13-0.29). After removing outliers to produce more precise estimates, the final effect sizes for knowledge, attitudes, and sexual behavior were, respectively, 0.25 (0.03-0.48), 0.18 (0.12-0.24), and 0.18 (0.11-0.24). Interventions yielded positive effects across populations and in all the domains studied.