Assessing the risk of tuberculosis infection among healthcare workers: the Melbourne Mantoux Study

Objective: To determine the potential prevalence of nosocomial infection with Mycobacterium tuberculosis among hospital employees in teaching hospitals in Melbourne. Design: Cross-sectional survey of positive tuberculin skin test (Mantoux) responses among employees in 14 public hospitals in Melbourne, January 1996 to April 1999. Participants: All consenting employees in participating hospitals (4070 healthcare and 4298 non-healthcare workers; participation rates, 13%-66%). Outcome measures: Prevalence of positive responses to tuberculin skin tests among healthcare and non-healthcare workers and association with employee and hospital characteristics. Results: Healthcare workers were significantly more likely to have a positive tuberculin response than non-healthcare workers (19.3% versus 13.7%; odds ratio, 1.5; 95% CI, 1.3-1.7; P<0.001). Multivariable analysis revealed that age, country of birth (high versus low tuberculosis [TB] prevalence), history of BCG (bacille Calmette-Guerin) vaccination, years since last BCG, occupation (healthcare versus non-healthcare worker) and years of hospital employment were all significantly associated with a positive response. Rates of positive responses among employees varied greatly between hospitals (6%-35%). These differences were not explained by employee characteristics, hospital TB patient load (number of admissions or bed-days) or percentage of hospital patients from countries with high TB prevalence. The hospital with the highest rate of positive responses was notable for its lack of negative-pressure isolation rooms for TB patients, Conclusions: Positive tuberculin responses are relatively common among hospital employees in Melbourne, with rates varying between hospitals and being higher among healthcare than non-healthcare workers. Employee characteristics, such as age, country of birth and past BCG status, explain little of this variation. More emphasis on TB infection control measures and regular staff screening may be needed.