Abstract
INTRODUCTION: The diagnosis of latent tuberculosis infection (LTBI) relies largely on the tuberculin skin test (TST) or, more recently, on interferon-gamma release assays (IGRA). Knowledge regarding these tests is essential to improve their usefulness in combating the tuberculosis epidemic. OBJECTIVES: To characterize the agreement between the IGRA and TST tests by determining the kappa coefficient (K) and agreement rate between these two tests in patients with active tuberculosis (TB). METHODS: Retrospective cohort study conducted with data from active TB patients notified in the Portuguese Tuberculosis Surveillance System (SVIG-TB), from 2008 to 2015. TST results were interpreted using a 5 mm (TST-5 mm) and 10 mm (TST-10 mm) cutoff. Kappa coefficient and agreement rate were calculated in order to evaluate the agreement between IGRA and TST (both cutoffs) test results. RESULTS: A total of 727 patients with results for both tests were included in the study, of which 3.4% (n = 25) had HIV infection, 5.6% (n = 41) diabetes, 5.0% (n = 36) oncological diseases and 4.4% (n = 32) inflammatory diseases. Of the 727 patients, 16.5% (n = 120) presented different outcomes between IGRA and TST-5 mm, and 20.5% (n = 149) presented different outcomes between IGRA and TST-10 mm. Kappa coefficient between IGRA and TST-5 mm was 0.402 (p < 0.001) with an agreement rate of 83.5%. Between IGRA and TST-10 mm, the kappa coefficient was 0.351 (p < 0.001), with an agreement rate of 79.5%. Patients with HIV infection, diabetes, oncologic diseases and inflammatory diseases presented a substantial agreement between IGRA and TST-5 mm, while inflammatory diseases was the only variable that presented a substantial agreement between IGRA and TST-10 mm. CONCLUSION: As both tests can present false-negative results, the low level of agreement between the tests can potentially help identify more cases of LTBI if the two tests are used in parallel, with infections not detected by IGRA possibly being detected by the TST and vice versa.