O3-S1.06 Diagnostic accuracy of rapid point-of-care tests to detect syphilis: a meta-analyses

Background The World Health Organization estimates that in 2006, there were 12 million new cases of syphilis. In developing countries, there is a lack of proper screening due to limited laboratory services and long distances from clinics. In developed countries, there is limited access to care among hard-to-reach populations. In this context of disconnect with the health care system, point of care (POC) tests have proven to be an invaluable resource, yet their accuracy needs to be established in order to justify their use. Method Electronic databases were searched from 1 January 1980 to 24 September 2010 for articles evaluating syphilis POC tests. Data were extracted and a second reviewer independently reviewed a subset of the articles. Subgroups were made according to the index test, the sample tested, and reference standard employed. Pooled sensitivity and specificity were calculated using Hierarchical Summary Receiver Operating Characteristic Curve. Adjustments were made to account for imperfect reference standards. Results 30 (47%) from 64 full text articles assessed articles were included in the meta-analysis. The most common kits evaluated were Determine, Bioline, Syphicheck, and Visitect in whole blood and sera samples. Using a Treponema Pallidum (TP) specific reference standard, in sera, the Determine test was the most accurate with a pooled sensitivity of 98.43% (96.03, 99.94) and a specificity of 97.74% (96.38, 98.92). In whole blood, Bioline was the most accurate with a sensitivity of 87.70% (84.78, 90.58) and a specificity of 99.07% (98.50, 99.59). The sensitivity of Determine and Visitect were lower when using whole blood than when using serum. When we adjusted for imperfect reference standards, the pooled parameters of accuracy improved when compared to pooled accuracy under the assumption of a perfect reference standard. Conclusions Determine with high sensitivity and Bioline with high specificity appeared to perform the best of the tests studied. Higher accuracy in serum warrants the use of serum rather than whole blood wherever possible. Confirmation with non-TP specific reference standard are required to confirm whether the infection is active or treated.