Behavioral versus biological outcomes on sexual risk: A comparison of intervention effect sizes

BACKGROUND: Sexual risk-behavior measures can be biological using laboratory assays or behavioral using self-report. Accurate assessment of sexual risk behavior is necessary to make a correct evaluation of sexually transmitted interventions (STIs). The main goal of this work is to compare biological and behavioral outcomes in behavioral interventions of STI/HIV studies. METHODS: Datasets from two published comprehensive meta-analyses were supplemented by additional Diffusion of Effective Behavioral Interventions (DEBIs). Studies were located using the reference section of the two meta-analyses and the Centers for Disease Control and Prevention (CDC). Fifty-seven studies with 276 separate interventions were included. Weighted effect sizes for biological measures, STIs, including or not HIV, and behavioral measure (self-report) for condom use were calculated under random-effects assumptions. Finally, specificity and sensitivity of the behavioral measure were obtained. Positive effect sizes indicated increased condom use and fewer incident STIs, including HIV. RESULTS. Only 35% of the interventions had an increase of condom use with an equivalent reduction in STI and HIV incidences. However, the self-report had a specificity of 73%. This pattern was similar regardless the biological measure and it was more pronounced when the self-report was related to casual sex. CONCLUSIONS. Self-report lacks the power to detect positive effects of behavioral interventions that are captured by a biological test. However, both biological and behavioral test's agreement increases when the intervention effects are negative. Consideration should be given to the increased use of biological measures or elaborate more sensitive self-reports because it could improve the evaluation of behavioral interventions.