Assessment of a screening tool to aid home-based identification of adolescents

The HPTN071 (PopART) for Youth (P-ART-Y) study evaluated the acceptability and uptake of a community-level combination HIV prevention package including universal testing and treatment (UTT) among young people in Zambia and South Africa. We determined whether a four-question primary care level screening tool, validated for use in clinical settings, could enhance community (door-to-door) identification of undiagnosed HIV-positive younger adolescents (aged 10-14) who are frequently left out of HIV interventions. Community HIV-care Providers (CHiPs) contacted and consented adolescents in their homes and offered them participation in the PopART intervention. CHiPs used a four question-screening tool, which included: history of hospital admission; recurring skin problems; poor health in last 3 months; and death of at least one parent. A "yes" response to one or more questions was classified as being "at risk" of being HIV-positive. Rapid HIV tests were offered to all children. Data were captured through an electronic data capture device from August 2016 to December 2017. The sensitivity, specificity, positive predictive value and negative predictive value were estimated for the screening tool, using the rapid HIV test result as the gold standard. In our 14 study sites, 33,710 adolescents aged 10-14 in Zambia and 8,610 in South Africa participated in the study. About 1.3% (427/33,710) and 1.2% (106/8,610) self-reported to be HIV positive. Excluding the self-reported HIV-positive, we classified 11.3% (3,746/33,283) of adolescents in Zambia and 17.5% (1,491/8,504) in South Africa as "at risk". In Zambia the estimated sensitivity was 35.3% (95% CI 27.3%-44.2%) and estimated specificity was 88.9% (88.5%-89.2%). In South Africa the sensitivity was 72.3% (26.8%-94.9%) and specificity was 82.5% (81.6-83.4%). The sensitivity of the screening tool in a community setting in Zambia was low, so this tool should not be considered a substitute for universal testing where that is possible. In South Africa the sensitivity was higher, but with a wide confidence interval. Where universal testing is not possible the tool may help direct resources to adolescents more likely to be living with undiagnosed HIV.