In-hospital contact investigation among health care workers after exposure to smear-negative tuberculosis

Abstract

Abstract Background Smear-negative pulmonary tuberculosis (TB) accounts for a considerable proportion of TB transmission, which especially endangers health care workers (HCW). Novel Mycobacterium-tuberculosis-specific interferon-γ release assays (IGRAs) may offer the chance to define the burden of TB in HCW more accurately than the Mantoux tuberculin skin test (TST), but the data that is available regarding their performance in tracing smear-negative TB in the low-incidence, in-hospital setting, is limited. We conducted a large-scale, in-hospital contact investigation among HCW of a German university hospital after exposure to a single case of extensive smear-negative, culture-positive TB with pulmonary involvement. The objective of the present study was to evaluate an IGRA in comparison to the TST and to identify risk factors for test positivity. Methods Contacts were prospectively enrolled, evaluated using a standardized questionnaire, the IGRA QuantiFERON®-TB Gold in Tube (QFT-GIT) and the TST, and followed-up for two years. Active TB was ruled out by chest x-ray in QFT-GIT-positive subjects. Independent predictors of test positivity were established through the use of logistic regression analysis. Results Out of the 143 subjects analyzed, 82 (57.3%) had close contact, but only four (2.8%) experienced cumulative exposure to the index case >40 hours. QFT-GIT results were positive in 13 subjects (9.1%), while TST results were positive in 40 subjects (28.0%) at an induration >5 mm. Overall agreement was poor between both tests (kappa = 0.15). Age was the only predictor of QFT-GIT-positivity (Odds ratio 2.7, 95% confidence interval 1.32–5.46), while TST-positivity was significantly related to Bacillus Calmette-Guérin vaccination and foreign origin. Logistic regression analysis showed no relation between test results and exposure. No secondary cases of active TB were detected over an observational period of two years. Conclusion Our findings suggest a low contagiosity of the particular index case. The frequency of positive QFT-GIT results may in fact reflect the pre-existing prevalence of latent TB infection among the study population. TB transmission seems unlikely and contact tracing not generally warranted after cumulative exposure