An unexpected result in an evaluation of a serological test to detect syphilis

Traditional tests for detection of syphilis are labour intensive and costly. Enzyme immunoassays (EIAs) are readily automated and cost effective if large numbers of tests are performed. Four experiments were devised to evaluate a syphilis EIA test kit where resources are limited: (1) testing antenatal patients; (2) testing refugees; (3) testing a high prevalence population; and (4) testing "problem sera" (containing autoantibodies or antibodies to other infective agents). Forty-one available syphilitic sera from antenatal patients were tested to evaluate sensitivity. Specificity was determined through testing sera determined to be nonreactive with rapid plasma reagin and Treponema pallidum hemagglutination tests, calculating the sample size (456) on the confidence interval (CI) required. Two runs were performed on antenatal sera, giving sensitivities of 32% (95% CI: 20%, 47%) and 37% (95% CI: 24%, 52%) and specificities of 92% (95% CI: 89%, 94%) and 90% (95% CI 87%, 92%), respectively. We present a method to evaluate a serological test where resources are limited. Unexpectedly, the test kit performed poorly as a screening test. New serological tests need to be evaluated in-house prior to adoption.