RESUMEN
Background: Microscopic examination of acid-fast-stained sputum smears is the current standard of care in the United States to determine airborne infection isolation (AII) of inpatients with presumptive pulmonary tuberculosis (PTB). However, nucleic acid amplification testing (NAAT) with the Xpert MTB/RIF assay (Xpert) may be more efficient and less costly. Methods: This prospective observational cohort study enrolled a consecutive sample of 318 AII-eligible inpatients from a public hospital in Seattle, Washington, from March 2012 to October 2013. Sputum samples were collected from each inpatient and analyzed using smear microscopy, culture, drug susceptibility testing, and NAAT. The performance, clinical utility (AII duration and survival), and cost-effectiveness from an institutional perspective were compared for 5 testing strategies. Results: Among the 318 admissions with presumptive PTB, 20 (6.3%) were culture-positive for Mycobacterium tuberculosis. The sensitivity of 1 Xpert, 2 Xperts, 2 smears, or 3 smears compared to culture was 0.85 (95% confidence interval [CI], .61.96), 0.95 (95% CI, .731.0), 0.70 (95% CI, .46.88), and 0.80 (95% CI, .56.93), respectively. A cost-effectiveness analysis of the study results demonstrated that an Xpert test on 1 unconcentrated sputum sample (assuming equivalent results for unconcentrated and concentrated sputum samples) is the most cost-effective strategy (99.9% preferred at willingness-to-pay of US$50000) and on average would save 51.5 patient-hours in AII and up to $11466 relative to microscopy without a compromise in sensitivity. Conclusions: In hospitalized patients with presumptive PTB in a low-burden setting, NAAT can reduce AII and is comparably sensitive, more specific, and more cost-effective than smear microscopy.