Evaluation of three syphilis algorithms: resolving discordance to ensure diagnostic accuracy

Objectives With the increasing incidence and persistent prevalence of syphilis, it is necessary to reevaluate the clinical utility of three syphilis algorithms. Methods Cases with concurrent Treponema pallidum -electrochemiluminescence immunoassay (TP-ECLIA), T. pallidum particle agglutination (TPPA) and toluidine red unheated serum test (TRUST) results from a large hospital in China were collected. The diagnostic performance of three syphilis algorithms—the traditional algorithm (start with TRUST and confirmation by TPPA), reverse algorithm 1 (start with TP-ECLIA and confirmation by TRUST with TPPA for TRUST-negative cases) and reverse algorithm 2 (start with TP-ECLIA and confirmation by TPPA with TRUST for TPPA-positive cases)—was evaluated using kappa coefficient analysis and Bayesian latent class model (BLCM), along with assessing their capabilities, discrepancies and limitations in identifying discordant results for T. pallidum antibodies. Results High agreement was observed between the two reverse algorithms. BLCM revealed sensitivities of 49.1%, 98.9% and 98.5% and specificities of 99.6%, 99.2% and 100.0% for the traditional algorithm, reverse algorithm 1 and reverse algorithm 2. The traditional algorithm identified 11 biological false-positive cases but missed 1177 TPPA + /TRUST − syphilis diagnoses. Discordant treponemal test results (TP-ECLIA + /TPPA −/± ) were common (18.2%) and associated with significantly lower TP-ECLIA cut-off index values. The reverse algorithm 1 identified 500 treponemal test-discordant cases, while the reverse algorithm 2 identified 517 cases. Besides, 17 discordant cases were identified among untested components of the algorithms. Conclusions The reverse algorithm 2 is recommended, as the confirmation of TP-ECLIA by TPPA reduces the underdiagnosis of previously treated syphilis compared with the traditional algorithm and compensates for the limited specificity of TP-ECLIA, thereby representing an improvement over reverse algorithm 1.