Identifying Pregnant Women at High Risk of HIV Acquisition: An External Validation Assessment of an HIV Risk Assessment Tool for Pregnant Women.

BACKGROUND

An HIV risk assessment tool for pregnant and lactating women was developed in Kenya. We evaluated the performance of this tool in Malawi.

METHODS

We conducted a case-control study among pregnant women in Lilongwe, Malawi (2017-2020) to externally validate the Kenyan HIV risk assessment tool. Pregnant women with a recent HIV infection at their first antenatal care (ANC) visit served as cases (n = 44), whereas those who tested HIV-negative at their first ANC visit served as controls (n = 349). Recent HIV infections were identified using a validated, laboratory-based, recent infection testing algorithm with a mean duration of infection of 130 days. We used the Kenyan tool to assign each participant a risk score according to their syphilis status, number of lifetime sex partners, and partner's HIV status. The predictive performance of the risk score was assessed using the area under the receiver-operator characteristic curve (AUC).

RESULTS

Risk scores were somewhat higher among cases (median 5.6, IQR: 3.0-8.1) than controls (median: 3.0, IQR: 2.0-7.0). The AUC of the risk score was 0.67 (95% CI: 0.58 to 0.76). When the risk score was dichotomized at the recommended threshold of >6, 48% of cases and 26% of controls were classified as high risk. Lowering this threshold improved sensitivity but decreased specificity.

DISCUSSION: The Kenyan HIV risk assessment tool was modestly predictive of recent HIV infection in our setting. Use of this tool should consider the benefits and shortcomings of different strategies within the context of local programmatic priorities and resource availability.