Review of nucleic acid amplification tests and clinical prediction rules for diagnosis of tuberculosis in acute care facilities.

Tuberculosis (TB) remains an important cause of hospitalization and mortality in the United States. Prevention of TB transmission in acute care facilities relies on prompt identification and implementation of airborne isolation, rapid diagnosis, and treatment of presumptive pulmonary TB patients. In areas with low TB burden, this strategy may result in inefficient utilization of airborne infection isolation rooms (AIIRs). We reviewed TB epidemiology and diagnostic approaches to inform optimal TB detection in low-burden settings. Published clinical prediction rules for individual studies have a sensitivity ranging from 81% to 100% and specificity ranging from 14% to 63% for detection of culture-positive pulmonary TB patients admitted to acute care facilities. Nucleic acid amplification tests (NAATs) have a specificity of >98%, and the sensitivity of NAATs varies by acid-fast bacilli sputum smear status (positive smear, ≥95%; negative smear, 50%-70%). We propose an infection prevention strategy using a clinical prediction rule to identify patients who warrant diagnostic evaluation for TB in an AIIR with an NAAT. Future studies are needed to evaluate whether use of clinical prediction rules and NAATs results in optimized utilization of AIIRs and improved detection and treatment of presumptive pulmonary TB patients.