Background: U.S. guidelines recommend respiratory isolation during evaluation of suspected pulmonary tuberculosis (TB) until demonstration of serial negative acid”fast bacilli(AFB) sputa. We evaluated the strategies of 1 vs. 2 rapid nucleic acid GeneXpert MTB/RIF(Xpert) tests in comparison to AFB smear for the initial diagnostic evaluation, including in persons with HIV, for whom prompt TB diagnosis is a priority.
Methods: Patients undergoing pulmonary TB evaluation had 2 sputa tested by Xpert(G4 cartridges) and compared to 2 sputum AFB smears; a subset had 3 AFB smears available. TB culture status was determined by 2 sputa samples, each cultured on both liquid and solid media. Those with M. tuberculosis on any of 4 cultures were classified as TB+. All had HIV testing. Exact McNemar’s test was used for comparisons.
Results: 633 participants had 2 AFB smear results: median age 49 years, 69% male, 78% inpatient at time of evaluation, 38% HIV+. 91(14%) were TB+; 10(11%) of TB+ were HIV+. A subset of 361(57%) had 3 AFB smears.
The initial Xpert identified 75 of 88 (85.2%) TB+ cases, compared to 69.3% for 2 AFB smears (p=0.001) (Table). Two Xperts identified 82/90 (91.1%)TB+ cases. Initial Xpert identified 59/61(96.7%) AFB+/TB+ and two Xperts identified all 62 AFB+/TB+. For the subset with 3 AFB, 1 Xpert and 2 Xperts identified 41/50 (82.0%) and 46/52 (88.5%) of TB+ cases compared to 60.4% for 3 AFB smears, and 1 Xpert and 2 Xperts identified 30/31 (96.8%) and 32/32 (100%) of AFB+/TB+, respectively.
Specificity was 98.7% for the first Xpert and 98.4% for 2 Xperts, compared to 94.8% for 3 AFB smears (p =0.008 and 0.019, respectively). Of 18 AFB+/TB-, both 1 and 2 Xperts had 1 false positive result. Of 524 AFB-/TB-, 1 Xpert yielded 3 false positive results; 2 Xpert had 5. The negative predictive value(NPV) of 1 Xpert was 97.6%, NPV of 2 Xperts 98.5%, compared to the NPV of 2 smears of 94.8%(14% TB prevalence). The NPV of 3 AFB was 93.3%(4% TB prevalence). 1 Xpert had similar performance (p>0.05) in HIV+ vs. HIV- with sensitivity 87.5% vs. 85.0% and specificity 99.6% vs. 99.0%.
Conclusions: A strategy of 2 Xperts missed no AFB+/TB+ patients, identifying all TB patients requiring respiratory isolation on the basis of AFB smear positive sputum. 1 and 2 Xperts were each significantly more sensitive and specific than 3 AFB smears for identifying culture-positive patients. These data support consideration of a 2 Xpert strategy to discontinue respiratory isolation of U.S.TB suspects, regardless of HIV status.