Using biomarker data in combination with self-reported risky sexual behaviors in HIV prevention research among Kenya orphan adolescents

Background: There is a lack of consensus about the relative validity of measures of self-reported behaviors and incident STD biological markers. We present baseline data from an ongoing HIV prevention trial in Kenya. We also examine inconsistency between self-reported sexual risky behaviors and biomarker data. Methods: A cluster randomized-controlled trial (c-RCT) to test whether school support reduces HIV, HSV-2 and sexual risk behaviors, is ongoing in western Kenya. Primary schools (N=26) were randomized to condition. Intervention participants receive financial support for school fees, uniforms and medical needs. Baseline and annual surveys were conducted in 2011 and 2012, respectively. The primary outcomes for this paper are HIV and HSV-2 infection (biomarker), as well as sexual debut, marriage and pregnancy (self-report). Results: In the baseline survey, 837 orphan adolescents participated and had HIV and HSV-2 sero-status established (average age 15 years, 48% female). HIV and HSV-2 prevalence was 1% and 3%, respectively. HSV-2 testing in low prevalence area results in low positive predictive value, thus only 33% may be true positive. Twenty-two percent reported having had sex and 9% reported being married. There were, however, inconsistencies in the data. Thirty-six percent among those who were HSV-2 positive and 70% among those who were married reported never having had sex. Multivariate analysis revealed that relative to Grade 8 participants in Grade 7 had significantly higher odds of inconsistency. Conclusions: Using biomarkers with combination of self-reported survey is helpful. However, more careful assessment of the validity of measures from multiple sources is required.