Meta-analysis of the effects of behavioral HIV prevention interventions on the sexual risk behavior of sexually experienced adolescents in controlled studies in the United States.

To estimate the effect of behavioral and social interventions on sexual risk of HIV among sexually experienced adolescents in the United States and to assess factors associated with variation in outcomes, we selected studies from the HIV/AIDS Prevention Research Synthesis project database. Twenty studies published or reported during the years 1988 through 1998 met criteria: 16 presented sufficient data; of these, 15 evaluated behavioral interventions and 1 a social intervention. Summary odds ratios (ORs) and 95% confidence intervals (CIs), weighted by study precision, indicated significantly less sex without condoms (number of studies, k, 13; OR, 0.66; CI, 0.55-0.79) and lower behavioral risk (k, 2; OR, 0.66; CI, 0.50-0.88), but no difference in number of partners (k, 8; OR, 0.89; CI, 0.76-1.05) or STDs (k, 2; OR, 1.18; CI, 0.48-2.86). A composite sexual risk behavior variable (k, 16; 1 outcome per study; preferred order, sex without condoms, number of partners, risk index) was used for heterogeneity and publication bias tests and stratified analyses. Overall, these interventions had a significant protective effect on sexually experienced adolescents (k, 16; OR, 0.65; CI, 0.50 - 0.85), although there was a suggestion of publication bias. Study design and intervention variables did not explain outcome variation. An exploratory finding may merit investigation: interventions tested with single ethnic groups out-of-class (k, 5) had larger effects than in-class interventions with mixed ethnic groups (k, 11), whether the mixed groups were in- (k, 6) or out-of class (k, 5).