Antepartum or postpartum isoniazid treatment of latent tuberculosis infection.
OBJECTIVE: To compare health outcomes and costs of different strategies for treatment of latent tuberculosis infection in pregnancy. METHODS: Using a Markov decision-analysis model, the following three strategies were evaluated for treatment of latent tuberculosis infection in pregnancy, defined as positive tuberculin skin reaction of 10 mm or greater and negative chest radiograph: no treatment, antepartum isoniazid administration, in which women were given 300 mg of isoniazid with pyridoxine beginning at 20 weeks' gestation for 6 months; and postpartum isoniazid, in which women were given isoniazid and pyridoxine for 6 months after delivery. Sensitivity analyses were performed for a wide range of probability and cost estimates, and considered discount rates. RESULTS: Under base-case assumptions, the fewest cases of tuberculosis within the cohort occurred with antepartum treatment (1400 per 100,000) compared with no treatment (3300 per 100,000) or postpartum treatment (1800 per 100,000). Antepartum treatment resulted in a marginal increase in life expectancy due to the prevented cases of tuberculosis, despite more cases of isoniazid-related hepatitis and deaths, compared with no treatment or postpartum treatment. Antepartum treatment was the least expensive. Only if the case-fatality rate for tuberculosis was tenfold lower than the base-case and the risk of fatal hepatitis tenfold higher did antepartum treatment become the least advantageous strategy. CONCLUSION: Rather than delaying treatment until postpartum, consideration for antepartum treatment of latent tuberculosis during pregnancy should be given. If isoniazid is not administered antepartum, then efforts to improve postpartum compliance should be instituted, as either antepartum or postpartum treatment is better than no treatment.
Boggess, KA; Myers, ER; Hamilton, CD
Volume / Issue
Start / End Page
Pubmed Central ID
International Standard Serial Number (ISSN)
Digital Object Identifier (DOI)