Abstract
BACKGROUND: To appropriately treat tuberculosis (TB) with regimens that combine novel and older drugs, evidence-based, context-specific strategies for drug-susceptibility testing (DST) will be required. METHODS: We created a Markov state-transition model of 100 000 adults with TB receiving a novel, fluoroquinolone (FQ)-containing regimen. We estimated clinical outcomes and resource utilization with no FQ-DST, universal FQ-DST, or FQ-DST only for patients with rifampin-resistant TB ("targeted FQ-DST"). We considered scenarios of stronger (South Africa) and weaker (Southeast Asia) correlation of fluoroquinolone resistance with rifampin resistance, with sensitivity analysis for other setting and regimen characteristics. RESULTS: Relative to no FQ-DST, targeted FQ-DST increased cure of FQ-resistant TB by 7.5% (interquartile range [IQR], 6.7%-9.2%) in South Africa and 1.7% (IQR, 0.7%-2.5%) in Southeast Asia. However, rare FQ resistance among the more prevalent rifampin-susceptible TB accounted for 50% of FQ-resistant TB in South Africa and 83% in Southeast Asia. As a result, universal FQ-DST further increased cure of FQ-resistant TB by 3.4% (IQR, 2.3%-5.4%) in South Africa and 5.8% (IQR, 5.1%-6.3%) in Southeast Asia. With targeted FQ-DST, 1 additional patient was cured per 50 (IQR, 42-70) tests in South Africa and 44 (IQR, 37-51) in Southeast Asia. When expanding from targeted to universal FQ-DST, 1 additional cure required 3500 (IQR, 2300-5500) tests in South Africa and 410 (IQR, 370-450) in Southeast Asia. CONCLUSIONS: FQ-DST improved patient outcomes and was particularly important for high-risk patient groups and less robust regimens. A universal strategy was favored in generalized epidemics of fluoroquinolone resistance.