Diagnostic Accuracy of Molecular Testing on Saliva and Oral Swabs for Pulmonary Tuberculosis

Background Rapid, accurate, nonsputum tests are needed to close gaps in tuberculosis (TB) detection. We evaluated the diagnostic performance of molecular testing on saliva and oral swabs. Methods We conducted a nested case-control study with 1:1 incidence-density sampling within a prospective cohort of adults and children undergoing evaluation for pulmonary TB at primary care centers in Colombia (July 2023-August 2024). Participants provided a sputum sample for liquid mycobacterial culture and paired saliva and nylon-flocked oral swabs for storage at -80°C. A microbiologist blinded to clinical and culture data performed Xpert MTB/RIF Ultra on thawed saliva and on swab eluate, each mixed 1:1 with sample reagent. We calculated the sensitivity and specificity of saliva and swab against sputum culture and compared them using McNemar's test. Results Among 648 enrolled participants, we tested saliva and swabs from all 95 individuals with culture-confirmed TB and 95 matched culture-negative controls (n = 190). Saliva sensitivity was 90.5% (95% confidence interval [CI], 82.8-95.6), and specificity was 95.8% (95% CI, 89.6-98.8). Swab sensitivity was 71.6% (95% CI, 61.4-80.4), and specificity was 99% (95% CI, 94.3-100). Saliva sensitivity exceeded that of swab by an absolute difference of 18.9% (95% CI, +10.0 to +27.9, P Conclusions Both saliva and swabs were highly sensitive and specific for culture-confirmed pulmonary TB. Saliva sensitivity exceeded the World Health Organization's ≥80% target for a low-complexity, nonsputum TB diagnostic test.