Setting: Standard treatment for latent tuberculosis infection (LTBI) is 9 months daily isoniazid (9INH). An alternative is 4 months daily rifampin (4RMP), associated with better completion and less toxicity; however, its efficacy remains uncertain.
Objectives: To assess the cost-effectiveness of these regimens for treating LTBI in human immunodeficiency virus negative persons, using results from a recent clinical trial, plus different scenarios for 4RMP efficacy, and to estimate the costs of an adequately powered noninferiority trial and resulting savings from substitution with 4RMP.
Design: A decision-analysis model tracked TB contacts and lower-risk tuberculin reactors receiving 9INH, 4RMP or no treatment. For different 4RMP efficacy scenarios, we estimated the cost-effectiveness, sample size and cost of non-inferiority trials, and potential cost savings substituting 4RMP for 9INH for 10 years in Canada.
Results: With an assumed 4RMP efficacy of 60%, 9INH was more effective but slightly more expensive. Above a threshold efficacy of 69%, 4RMP was cheaper and more effective than 9INH. If the true efficacy of 4RMP is ≥75%, a trial powered to detect non-inferiority with a lower limit of 60% estimated efficacy (~20 000 subjects) may lead to cost savings within 10 years, even with the extreme assumption that Canada bears the entire cost.
Conclusion: 4RMP may be a reasonable alternative to 9INH. Costs of a large-scale non-inferiority trial may be offset by subsequent savings.