Recent PEPFAR guidance seeks to diminish HIV testing volume and focus testing on higher positivity populations. We sought to define testing criteria to reduce total tests performed and increase positivity in Botswana.
We analyzed data collected October 2018-August 2019 as part of routine HIV testing provided in 134 Botswana Ministry of Health facilities supported by the International Training and Education Center for Health (I-TECH). We randomly split the data into prediction and validation datasets of equal size and used multivariate logistic regression to identify demographic characteristics, testing strategies, and testing sectors (e.g., antenatal) associated with HIV positivity; factors with significant adjusted odds ratios (aOR) ≥1.5 were included in the testing criteria. Testing strategies and sectors where cessation of testing was deemed unacceptable a priori (TB, STD, VCT, partner services, antenatal, labor & delivery, pediatrics and gynecology) were excluded from model development and included in the testing criteria. We applied the new testing criteria to the validation dataset to determine the number of tests performed, test positivity, proportion of positives missed, and costs averted. Costs were derived based on total budgets allocated to I-TECH to support HIV testing and estimated costs of test procurement.
The analysis included 262,230 tests of which 4.3% were HIV positive. Model derivation analysis identified ages 23-29, 30-39, and 40+, non-citizenship, and emergency department testing as significantly associated with positivity. Among 131,115 tests in the validation analysis, 5,580 (4.3%) were HIV positive. Restricting testing to persons age >30 years and other defined criteria would reduce testing volume by 23% and increase positivity to 4.9%; 649 (2.1%) of the 30,178 persons who would not be tested were HIV positive representing 11.6% of all positive tests in the validation dataset. Positives missed by the criteria had a median age of 25 years and were mostly female (67%) and tested in the general outpatient department (86%). Assuming no changes in staffing, implementing the new testing criteria would decrease total HIV testing costs by 13%, a savings of $18 per positive test missed.
In Botswana, a targeted approach to HIV testing could reduce testing volumes by 23% and modestly increase HIV test positivity while missing 11.6% of positive tests. Cost saving would be modest unless implementation was accompanied by changes in staff costs.