Xpert MTB/RIF test for tuberculosis - The Lancet

Correspondence

We would like to congratulate Catharina Boehme and colleagues (April 30, p 1495)1 on their study, the results of which show the excellent performance of Xpert MTB/RIF in health facilities in high-burden, resource-poor settings. The sensitivity of Xpert MTB/RIF was not reduced in individuals infected with HIV compared with those who were uninfected and convincingly outperformed smear microscopy. Thus, Xpert MTB/RIF represents a fundamental advance in the diagnosis of tuberculosis. We have shown a diminished negative predictive value (NPV) and negative likelihood ratio (LR) of Xpert MTB/RIF for the detection of tuberculosis in individuals infected with HIV versus those who were uninfected,2 suggesting that substantial numbers of patients infected with HIV who are negative by Xpert MTB/RIF actually have active tuberculosis. The likelihood of this false-negativity occurring increased with advanced immunosuppression. The conclusiveness of our findings was limited by our retrospective study design and small sample size. However, these data have now been prospectively confirmed in 6648 patients by Boehme and colleagues. In their study, the NPV was 91·3% (389/426; 95% CI 88·3–93·6) in patients infected with HIV versus 96·0% (748/779; 94·4–97·2; p