Confirmation of high specificity of an automated enzyme immunoassay test for serological diagnosis of syphilis: retrospective evaluation versus results after implementation.

BACKGROUND: The optimal algorithm for serological syphilis screening is still a matter of debate. We have previously evaluated the performance of the Bioelisa Syphilis 3.0, using a selection of archived sera, and in this study compare these results with the Bioelisa results after clinical implementation. METHODS: All Bioelisa Syphilis 3.0 results obtained since clinical implementation were analyzed. Bioelisa-positive or borderline samples were retested using Treponema pallidum particle agglutination, rapid plasma reagin test, fluorescent treponemal antibody-absorption test, and/or immunoblot. On sera sent in together with cerebrospinal fluid, occasionally both the T. pallidum particle agglutination and Bioelisa were performed. RESULTS: The Bioelisa was performed on 14,622 sera. Bioelisa-positive samples, which were not retested by the previously described assays, were withdrawn from the database (n = 36). In 1.3% of the samples (187/14,586), the Bioelisa was positive or borderline and, ultimately, 115 sera were considered true positive (prevalence 0.8%). The specificity of the Bioelisa was 99.5%. CONCLUSIONS: Based on the results of all performed diagnostic assays, the specificity of the Bioelisa of 99.5% is very consistent with that found in the initial study (100%; 95% confidence interval was 98.0%-100%). Interpreting (positive) test results is difficult in the absence of a gold standard, especially when the disease prevalence is low. Results should be viewed in the light of the patients' characteristics.