Background
In 2020, Tanzania introduced a 16-question screening tool to determine adult eligibility for HIV testing services (HTS). This validation study compares the characteristics of the current tool to a) an optimized tool and b) HIV self-testing (HIVST).
Methods
From March to June 2024, adults aged ≥15 years with unknown HIV status seeking care were enrolled at 32 health facilities in Tanzania. The current national screening tool was administered, and all participants were tested for HIV regardless of eligibility screening outcome. Demographic information, screening responses, and HIV test results were collected. Eligibility for HTS was defined as ≥1 positive response among 16 screening questions. An optimized screening tool was developed using regularized logistic regression to identify the most predictive of the 16 screening questions and optimal weighting (1 to 3 points), with a total score ≥2 as the threshold for HTS eligibility. Sensitivity, specificity, positive and negative predictive values (PPV, NPV), and 95% confidence intervals were generated for both screening tools. PPV and NPV were calculated based on manufacturer’s reported sensitivity and specificity for HIVST.
Results
We enrolled 23,444 participants; median age was 32 years (interquartile range: 24-43 years), and 13,715 (58.5%) were women. HIV was diagnosed in 356 (1.5%) participants, with higher prevalence among women (1.7%) compared to men (1.2%) and persons aged ≥35 years (2.0%) compared to persons aged <35 years (1.1%). Of participants, 22,228 (94.8%) screened eligible for HTS using the national tool (97.8% sensitivity, 5.2% specificity). Slightly fewer (22,097; 94.3%) screened eligible for HTS using the optimized tool, which had similar characteristics to the national screening tool (97.8% sensitivity, specificity 5.9%). Using HIVST for screening yielded 402 (1.7%) participants eligible for HTS, including all in whom HIV was diagnosed.
Conclusions
The national and optimized tools had high sensitivity but low specificity resulting in almost all participants screening eligible for HTS. Both tools misclassified some persons as ineligible for HTS who were later diagnosed with HIV. HIVST was more efficient than the two screening tools, resulting in fewer individuals screening eligible for HTS but identifying all persons in whom HIV was diagnosed. Further econometric and qualitative analyses are needed to determine the implications of these results on HTS eligibility screening policy.
Table or Figure
