Four rapid tests for the serologic diagnosis of leptospirosis were evaluated, and the performance of each was compared with that of the current standard, the microscopic agglutination test (MAT). The four rapid tests were a microplate immunoglobulin M (IgM)-enzyme-linked immunosorbent assay (ELISA), an indirect hemagglutination assay (IHA), an IgM dipstick assay (LDS), and an IgM dot-ELISA dipstick test (DST). A panel of 276 sera from 133 cases of leptospirosis from four different geographic locations was tested as well as 642 sera from normal individuals or individuals with other infectious or autoimmune diseases. Acute-phase sera from cases (n = 148) were collected or=15 days after onset (median = 29.1). By a traditional method (two-by-two contingency table), the sensitivities for detection of leptospirosis cases were 93.2% by LDS, 92.5% by DST, 86.5% by ELISA, and 79.0% by IHA. Specificity was 98.8% by DST, 97% by ELISA and MAT, 95.8% by IHA, and 89.6% by LDS. With a latent class analysis (LCA) model that included all the rapid tests and the clinical case definition, sensitivity was 95.5% by DST, 94.5% by LDS, 89.9% by ELISA, and 81.1% by IHA. The sensitivity and specificity estimated by the traditional methods were quite close to the LCA estimates. However, LCA allowed estimation of the sensitivity of the MAT (98.2%), which traditional methods do not allow. For acute-phase sera, sensitivity was 52.7% by LDS, 50.0% by DST, 48.7% by MAT and ELISA, and 38.5% by IHA. The sensitivity for convalescent-phase sera was 93.8% by MAT, 84.4% by DST, 83.6% by LDS, 75.0% by ELISA, and 67.2% by IHA. A good overall correlation with the MAT was obtained for each of the assays, with the highest concordance being with the DST (kappa value, 0.85; 95% confidence interval [CI], 0.8 to 0.90). The best correlation was between ELISA and DST (kappa value, 0.86; 95% CI, 0.81 to 0.91). False-positive LDS results were frequent (>or=20%) in sera from individuals with Epstein-Barr virus, human immunodeficiency virus, and periodontal disease and from healthy volunteers. The ease of use and significantly high sensitivity and specificity of DST and ELISA make these good choices for diagnostic testing.
