Is it appropriate to routinely use a nucleic acid amplification test for the diagnosis of tuberculosis?

The purpose of this study was to compare the usefulness of the nucleic acid amplification (NAA) test against conventional tests under normal laboratory operational conditions. The NAA test was performed on the first sputum specimen of all patients. Liquid media culture, solid media culture, and Ziehl-Neelsen stain for an acid-fast bacilli (AFB) smear were performed on three sputum specimens. The results were calculated using the gold standard of either the culture results or the clinical diagnosis. Of the 593 patients tested, 151 (25.5%) were diagnosed with pulmonary tuberculosis. The sensitivity of the first specimen only was 64% for the NAA test, 54% for the AFB smear, 77% for BACTEC MGIT 960 culture, 40% for Lowestain-Jensen (LJ) culture, and 25% for 7H11 culture. The sensitivity when using all three specimens increased to 63% for AFB smear, 87% for BACTEC MGIT 960 culture, 51% for LJ culture, and 40% for 7H11 culture. The specificity was 100% for all culture tests, 99% for the AFB smear, and 99.5% for NAA test. The mean turnaround time was 1.34 days for NAA, 0.59 days for AFB smear, 11 days for BACTEC MGIT 960 culture, 23 days for LJ culture, and 20 days for 7H11 culture. We conclude that the sensitivity of NAA is still far from ideal, and the test is not cost effective. Thus, the COBAS AMPLICOR PCR system is not suitable for routine use in microbiology laboratories.