Accuracy and performance of the WHO 2015 and WHO 2019 HIV testing strategies across epidemic settings

Abstract

Background: WHO 2019 HIV testing guidelines recommended a standard HIV testing strategy consisting of three consecutively HIV-reactive test results on serology assays to diagnose HIV infection. National HIV programmes in high prevalence settings currently using the strategy consisting of only two consecutive HIV-reactive tests should consider when to implement the new guideline recommendations. Methods and Findings: We implemented a probability model to simulate outcomes of WHO 2019 and the two strategies recommended by WHO 2015 guidelines on HIV testing services. Each assay in the strategy was assumed independently 99% sensitivity and 98% specificity, the minimal thresholds required for WHO prequalification. For each strategy and positivity ranging 20% to 0.2%, we calculated the number of false-negative, false-positive, and inconclusive results; positive and negative predictive value (PPV, NPV); number of each assay used, and testing programme costs. We found that the NPV was above 99.9% for all scenarios modelled. Under the WHO 2015 two-test strategy, the PPV was below the 99% target threshold when positivity fell below 5%. For the WHO 2019 strategy, the PPV was above 99% for all positivity levels. The number reported 'inconclusive' was higher under the WHO 2019 strategy. Implementing the WHO 2019 testing strategy in Malawi, would require around 89,000 A3 tests in 2021, compared to 175,000 A2 tests and over 4.5 million A1 tests per year. The incremental cost of the WHO 2019 strategy was less than 2% in 2021 and declined to 0.9% in 2025. Conclusions: As positivity among persons testing for HIV reduces below 5% in nearly all settings, implementation of the WHO 2019 testing strategy will ensure that positive predictive value remains above the 99% target threshold, averting misdiagnoses and ART initiations among HIV uninfected people. The incremental cost of implementing the WHO 2019 HIV testing strategy compared to the two-test strategy is negligible because the third assay accounts for a small and diminishing share of total HIV tests.

Country adoption of WHO 2019 guidance on HIV testing strategies and algorithms: a policy review across the WHO African region

“…A retrospective study in Nigeria using household survey data found that the performance of the two-test strategy in a low prevalence setting of about 1.4% was poor with a PPV of 94% and a false-positive rate of 5.5% 22. This positive predictive value is similar to a modelling study assessing the accuracy of the 2015 and 2019 WHO HIV testing algorithms, which was 95.4% using a two-assay testing strategy 19…”

“…However, some countries have yet to prioritise adopting the three-test strategy because they were focused on higher yield testing approaches alone,17 18 or delayed the transition due to costs concerns. While countries will need to develop a plan and identify the optimal time for the transition, in terms of costs, a modelling study found that the total cost of the three-test strategy would have a very limited impact on costs 19. Further, in Ghana after transitioning to a three-test strategy they found that while there were lessons learnt about logistics and training, the new strategy was found to be feasible, improved testing quality and reduced costs due to preventing misdiagnosis 20…”

“…While countries will need to develop a plan and identify the optimal time for the transition, in terms of costs, a modelling study found that the total cost of the three-test strategy would have a very limited impact on costs. 19 Further, in Ghana after transitioning to a three-test strategy they found that while there were lessons learnt about logistics and training, the new strategy was found to be feasible, improved testing quality and reduced costs due to preventing misdiagnosis. 20 We found that among low-prevalence countries (n=34), a significant proportion (41%) still use a suboptimal testing strategy with two assays to make an HIV-positive diagnosis, which will increase the likelihood of falsepositive diagnosis.…”

“…20 We found that among low-prevalence countries (n=34), a significant proportion (41%) still use a suboptimal testing strategy with two assays to make an HIV-positive diagnosis, which will increase the likelihood of falsepositive diagnosis. [19][20][21] A retrospective study in Nigeria using household survey data found that the performance of the two-test strategy in a low prevalence setting of about 1.4% was poor with a PPV of 94% and a false-positive rate of 5.5%. 22 This positive predictive value is similar to a modelling study assessing the accuracy of the 2015 and 2019 WHO HIV testing algorithms, which was 95.4% using a two-assay testing strategy.…”

“…22 This positive predictive value is similar to a modelling study assessing the accuracy of the 2015 and 2019 WHO HIV testing algorithms, which was 95.4% using a two-assay testing strategy. 19 Considering that the recommendation to use dual HIV/syphilis RDTs in ANC was released only in 2019, 6 23 its adoption by 21 countries has been encouraging. The number of countries adopting dual HIV/syphilis RDTs among pregnant women continues to increase and may be in practice higher considering that preliminary National Commitments and Policy Instrument (NCPI) survey data reported that a total of 26 countries in the WHO African region adopted the use of dual HIV/syphilis RDTs in ANC.…”