Assessing diagnostic accuracy of onchocerciasis rapid diagnostic tests using Bayesian latent class models

Abstract

Abstract Background Onchocerciasis elimination programs increasingly rely on rapid diagnostic tests (RDTs) that detect antibodies to the Ov16 antigen, yet the performance of currently available rapid tests remains uncertain. The commercially available SD BIOLINE Ov16 RDT has shown inconsistent sensitivity when used on whole blood and does not consistently meet international performance thresholds for mapping or decisions to stop mass drug administration (MDA). Two novel onchocerciasis RDTs have been developed, DDTD RDT and GADx RDT, but their diagnostic accuracy has not been fully evaluated using methods that account for the absence of a true reference standard. Methodology/Principal Findings We pooled participant-level data from three field studies conducted in Mozambique, Ghana, and Benin in 2023 to evaluate two novel onchocerciasis RDTs and the commercially available SD BIOLINE Ov16 RDT. A Bayesian latent class model was used to estimate diagnostic sensitivity and specificity without assuming any test was a perfect reference standard. Across all model specifications, the GADx rapid test showed the highest sensitivity, with posterior median estimates (92.0%–92.8%) consistently above the 89% threshold recommended for decisions to stop mass drug administration (MDA). Posterior median sensitivity for the SD BIOLINE test was above this threshold in most models but below it under the field-based prior (83.2%). The DDTD RDT demonstrated lower sensitivity overall (86.6%–88.4%). Specificity estimates were consistently high for all tests but remained below the recommended threshold of 99.8% (highest median: 98.8%). Alternative definitions of a positive result for the multi-antigen DDTD RDT did not meaningfully alter its estimated performance. Conclusions/Significance In the study communities sampled, the two novel onchocerciasis RDTs generally yielded higher sensitivity estimates than the commercially available SD BIOLINE Ov16 RDT, with the GADx assay having the highest estimated sensitivity. However, posterior estimates for none of the evaluated tests reached the specificity threshold recommended to support decisions to stop MDA on their own. These findings highlight the need for improved diagnostic tools or complementary testing strategies to support elimination decision-making. Author Summary In this study, we evaluated three RDTs used to detect exposure to the parasite that causes river blindness. These tests are important tools for national programs that are working to eliminate the disease and determine when preventive treatment can be safely stopped. One challenge is that there is no perfect reference test against which new diagnostic tests can be evaluated. To address this, we pooled participant-level data from studies conducted in Mozambique, Ghana, and Benin and used a Bayesian latent class model to estimate test accuracy without assuming that any single test was perfect. We found that one of the novel tests, the GADx RDT, identified people with exposure more accurately than the other evaluated tests, meeting the established WHO threshold for stopping MDA. The commercially available SD BIOLINE test and the DDTD test showed lower estimated accuracy than the GADx test. Although all three tests correctly identified most people who were not exposed, none achieved the very high level of specificity recommended to confidently support decisions to stop preventive treatment. As such, stop-MDA decisions should not rely on any of these RDTs alone. Further improvements, developments, and robust assessment are needed first. Our findings suggest that while newer onchocerciasis RDTs are improving, additional advances are needed before these tests can independently support decisions to stop preventive treatment. These results provide evidence to inform the evaluation and selection of diagnostic tools as national programs progress from disease control toward onchocerciasis elimination.

The authors' abstract, as published at the source. PLoS neglected tropical diseases, 2026 · DOI ↗