Neonatal screening for congenital Chagas infection: application of latent class analysis for diagnostic test evaluation

摘要

The present study had the aim of evaluating conventional serum tests that are used in neonatal screening for Chagas disease, with a discussion on the statistical methods available. A random sample among 23,308 newborns who were screened for congenital Chagas disease was studied using the following three tests: enzyme immunoassay, indirect immunofluorescence and indirect hemagglutination. The data were analyzed by different statistical methodologies: latent class analysis, Kappa test and relative sensitivity analysis. Using latent class analysis, enzyme immunoassay had the highest sensitivity (48.6%), followed by indirect immunofluorescence (39.8%) and indirect hemagglutination (23.2%). The kappa value was 0.496. The ratio between the sensitivities of enzyme immunoassays and indirect immunofluorescence tests was 92% [0.74;1.13]. Latent class analysis was not found to be adequate for sensitivity and specificity determination, but it provided important data about the equivalence of the tests, corroborated by relative sensitivity analysis. The results showed that enzyme immunoassaying of dry blood samples can be used as safely as the indirect immunofluorescence test.