Performance of nucleic acid amplification tests for diagnosis of tuberculosis in a large urban setting.

Background: A diagnosis of tuberculosis (TB) relies on acid-fast bacilli (AFB) smear and culture results. Two rapid tests that use nucleic acid amplification (NAA) have been approved by the US Food and Drug Administration for the diagnosis of TB based on detection of Mycobacterium tuberculosis from specimens obtained from the respiratory tract. We evaluated the performance of NAA testing under field conditions in a large urban setting with moderate TB prevalence.

Methods: The medical records of patients with suspected TB during 2000-2004 were reviewed. Analysis was restricted to the performance of NAA on specimens collected within 7 days after the initiation of treatment for TB. The assay's sensitivity, specificity, and positive and negative predictive values (PPV and NPV, respectively) were evaluated.

Results: The proportion of patients with confirmed or suspected TB whose respiratory tract specimens were tested by use of NAA increased from 429 (12.9%) of 3334 patients in 2000 to 527 (15.6%) of 3386 patients in 2004; NAA testing among patients whose respiratory tract specimens tested positive for AFB increased from 415 (43.6%) of 952 patients in 2000 to 487 (55.5%) of 877 patients in 2004 (P < .001 for both trends). Of the 16,511 patients being evaluated for pulmonary TB, 4642 (28.1%) had specimens that tested positive for AFB on smear. Of those 4642 patients, 2241 (48.3%) had NAA performed on their specimens. Of those 2241 patients, 1279 (57.1%) had positive test results. Of those 1279 patients, 1262 (98.7%) were confirmed to have TB. For 1861 (40.1%) of the 4642 patients whose specimens tested positive for AFB on smear, the NAA test had a sensitivity of 96.0%, a specificity of 95.3%, a PPV of 98.0%, and an NPV of 90.9%. For 158 patients whose specimens tested negative for AFB on smear, the NAA test had a sensitivity of 79.3%, a specificity of 80.3%, a PPV of 83.1%, and an NPV of 76.0%, respectively. For the 215 specimens that tested positive for AFB by smear, we found a sensitivity, specificity, PPV, and NPV of 97.5%, 93.6%, 95.1%, and 96.8%, respectively. A high-grade smear was associated with a better test performance.

Conclusion: NAA testing was helpful for determining whether patients whose specimens tested positive for AFB on smear had TB or not. This conclusion supports the use of this test for early diagnosis of pulmonary and extrapulmonary TB.