concordância entre 2 testes diagnósticos

The diagnosis of latent tuberculosis infection (LTBI), namely in healthcare workers, has been done by tuberculin skin test (TST). Recently, Interferon-λ Release Assays (IGRA tests) have been introduced. Unlike TST, they do not turn positive after BCG vaccine or after most of non-tuberculous mycobacteria infections. In the absence of a gold standard for the diagnosis of LTB, the aim of this study was to analyze the correlation between the two LTB diagnostic techniques, by determining Kappa coefficient and concordance rate between TST and IGRA test, in healthcare workers of a Portuguese university hospital. We carried out a cross-sectional and retrospective study, and we analysed TST and IGRA tests records, that are performed simultaneously (up to 15 days apart) in Occupational Health Department, in 2010 and 2011 (n 137). Most of the sample did BCG and 44.5% did two or more inoculations. The average diameter of PT was 17.5 mm (SD 4.3). Only two subjects reported TST < 10 mm, both showing a negative IGRA test. Only 53 (39.3%) of the 135 participants with positive PT had a positive IGRA test too. The level of agreement between TST and the IGRA test was determined by Kappa coefficient. Respectively to a cut off for PT of 10 mm, 15 mm and 20 mm, the degree of agreement was 0.019 (p = 0.26), 0,19 (p = 0.001) and 0.26 (p = 0.003). Concordance rates were respectively 40%, 54% and 65%. It was found that the concordance between the two methods increased as the cut off for TST also increased. Nevertheless, a higher concordance rate would be expected with cut off of 15 mm, and particularly of 20 mm, since BCG vaccine administered in childhood usually induce smaller reactions. The existence of more than one BCG inoculation in the sample, especially after childhood, can be partly responsible for the low concordance between the two methods. Nevertheless, we cannot also exclude IGRA test's false negatives. In the decision to treat LTB, it is necessary to take into account the limitations of both tests, their level of agreement and weighting individual, occupational and epidemiological factors.publishersversionpublishe