P5.089 Confirmation of High Specificity of an Automated ELISA Test For Serological Diagnosis of Syphilis - Results from Confirmatory Testing After Syphilis Screening and Sensitivity Analysis in the Absence of a Gold Standard

Abstract

Background In clinical microbiology laboratories, serological diagnostic assays are usually implemented after evaluation using a selected sample collection. We have previously evaluated the performance of the Bioelisa Syphilis 3.0 compared with the Treponema pallidum Particle Agglutination (TPPA) in a selected collection of serum samples (syphilis positivity rate 44%) and found a sensitivity and specificity of both 100%. In the current study we have compared the specificity of Bioelisa Syphilis 3.0 after clinical implementation as a syphilis screening test with the specificity found in the previous evaluation to assess whether the high specificity would stand up in clinical practise. Methods We included 14,622 sera (positivity rate 0.9%) sent to the laboratory for syphilis serology in the period between October 2007 and February 2010. Results We confirm the initially reported specificity and further narrow down its confidence interval (specificity 99.5%, 95% CI 99.4–99.6%), and show that this high specificity is valid across diverse patient categories. Here we demonstrate that differences in positive predictive values between patient categories reflect the prevalence of syphilis in these categories, and are not due to differences in specificity. In addition, in a sensitivity analysis we show that these conclusions are robust for several assumptions. Conclusion Our analysis shows that the high specificity found in the initial study, stands up after implementation in a population with a low syphilis prevalence (0.9%). Using a selected serum sample collection is therefore a valid manner in the evaluation of syphilis serological diagnostic assays. Confirmatory syphilis testing remains mandatory in low prevalence populations, even when the screening test has a very high specificity.