Comparison of the performance and costs of testing algorithms using rapid diagnostic tests for detection and treatment of syphilis among pregnant individuals and men who have sex with men: a modelling study.

Abstract

Syphilis remains a global public health concern, particularly among pregnant individuals and men who have sex with men (MSM). Rapid diagnostic tests (RDTs) enable same-day diagnosis and treatment, but the optimal testing algorithm depends on local epidemiology. We aim to evaluate the performance and cost-effectiveness of three RDT-based algorithms across varying syphilis prevalence in antenatal and MSM populations. We developed an open-source decision-analytic model to compare: (1) a treponemal-only RDT (T-RDT); (2) a standard dual treponemal and non-treponemal RDT (T/NT-RDT; for which treatment is recommended if both components are positive); and (3) a modified T/NT-RDT, for which anyone positive on either component receives treatment, reflecting real-world practice when a person's syphilis history is unclear. The primary outcomes were missed and overtreated cases, evaluated in cohorts of 100 000 individuals. We also assessed cost per person tested (in 2025 US dollars). Cost inputs reflected settings typical of low-income and middle-income countries. Analyses were stratified by population: among pregnant individuals, we compared the T-RDT and modified T/NT-RDT, and for MSM, all three algorithms were evaluated. The time horizon was a single testing episode with syphilis prevalence ranging from 0·5% to 20%. Sensitivity analyses explored uncertainty in test performance, costs, and alternative algorithm options. Among pregnant individuals, the modified T/NT-RDT missed at least 80% fewer active cases than the T-RDT but had more overtreated cases. For example, at a syphilis prevalence of 0·5%, with a RDT sensitivity of 85% and specificity of 90%, the modified T/NT-RDT missed 10 active cases and overtreated 18 948 cases per 100 000 individuals, compared with 68 missed cases and 9993 overtreated cases with the T-RDT. The T-RDT ($1·09-$1·51 per person tested) cost less than the modified T/NT-RDT ($3·59-$4·12). Among MSM, the modified T/NT-RDT missed the fewest cases but increased overtreatment. The standard T/NT-RDT (costing $3·59-$3·87 per person tested) had the lowest rates of overtreatment but missed more cases than either alternative (ie, in the same scenario with a syphilis prevalence of 0·5%, RDT sensitivity of 85%, and specificity of 90%, the standard T/NT-RDT missed 125 active cases and overtreated 999 cases per 100 000 individuals). RDT costs were the main cost driver in all three algorithms. In sensitivity analyses, a sequential algorithm consisting of T-RDT followed by the modified T/NT-RDT reduced missed cases compared with the standard T/NT-RDT, and substantially reduced overtreatment relative to the T-RDT and modified T/NT-RDT. No single algorithm was optimal across all settings. T-RDTs remain suitable for antenatal screening, but the standard or modified T/NT-RDT might be preferable for populations at high risk for syphilis, such as MSM. Australian National Health and Medical Research Council Emerging Leadership Investigator Grants and an Australian Government Research Training Program scholarship.