Validation of a National Pediatric HIV Risk Screening Tool at 31 Health Facilities in Zambia

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Background: To improve testing efficiency, Zambia introduced a national pediatric human immunodeficiency virus (HIV) Risk Screening Tool (HRST) in 2019. However, concerns about its sensitivity led to its suspension in 2021 pending formal evaluation. We therefore undertook a formal validation of this tool to assess its diagnostic performance. Methods: We conducted a cross-sectional evaluation from February 2021 to March 2022 in 31 health facilities across Lusaka, Southern and Western Provinces. Children aged 2–14 years were screened using the HRST, and all subsequently underwent HIV testing to estimate the tool’s sensitivity and specificity. L1-regularized logistic regression was used to identify the most predictive screening questions to develop an optimized tool. Focus group discussions with 85 healthcare providers explored implementation experiences. We also estimated unit costs for HRST-based screening and universal testing and compared these with published cost estimates for provider-assisted HIV self-testing. Results: Among 15,431 children tested, 90 (0.58%) were HIV-positive. The 6-question HRST demonstrated 80.0% sensitivity and 71.0% specificity. An optimized 4-question tool showed similar performance (77.8% sensitivity, 73.2% specificity). Providers reported that the HRST supported targeted testing and workflow efficiency, but highlighted challenges related to consent, disclosure and caregiver knowledge. Estimated unit costs per child tested were $2.25 for universal testing, $0.39 for HRST-based screening and $3.08 for HIV self-testing. Conclusion: Although the HRST improves testing efficiency, its moderate sensitivity underscores the risk of missed diagnoses when used alone. Cost and implementation findings support a differentiated pediatric HIV testing strategy that deploys universal testing, risk-based screening, or HIV self-testing based on local capacity and resource constraints.