Exposure to tuberculosis among newborns in a nursery: decision analysis for initiation of prophylaxis

Newborns in a hospital nursery were exposed to a mother whose sputum was direct-smear negative for acid-fast bacilli but culture positive for Mycobacterium tuberculosis. Given the low risk for exposure, the high susceptibility of infants to M. tuberculosis infection, and the possibility of hepatotoxicity due to isoniazid therapy, a decision analysis model was used to determine whether administration of isoniazid prophylaxis against tuberculosis is preferable to no administration of prophylaxis.

A decision analysis tree was constructed with software, using probabilities from the literature and costs from local health facilities. The expected values for each strategy were obtained, and sensitivity analyses were performed.

For the strategy in which prophylaxis was administered under direct observation (DO), the probability for survival was 0.999980. For the strategy in which no prophylaxis was administered, the probability of survival was 0.999950, which corresponds to 3 more deaths per 100,000 patients than with the DO prophylaxis strategy. The incremental cost-effectiveness of the DO prophylaxis strategy was 21,710,000 US dollars per death prevented. Sensitivity analysis for survival showed that the DO prophylaxis strategy was preferable to the strategy in which no prophylaxis is given if the probability of infection was >0.0002, the probability of tuberculous disease in an infected infant who did not receive prophylaxis was greater than 0.12, the probability of dying from tuberculosis was greater than 0.025, the probability of hepatotoxicity was less than 0.004, and the probability of dying from hepatotoxicity was less than 0.04. For the strategy in which prophylaxis was administered under non-DO conditions (ie, by parents), the incremental cost-effectiveness was 929,500 US dollars per death prevented, which is approximately 5% of the incremental cost-effectiveness of the DO prophylaxis strategy.

This model provides a structure for determining the preferable prophylaxis strategies for different risks of exposure to tuberculosis in a nursery. Administration of prophylaxis is preferable to no administration of prophylaxis, unless the probability of infection is extremely low.