Abstract
SETTING
Although it is now widely used for tuberculosis (TB) diagnosis, Xpert® MTB/RIF availability remains inadequate in low-resource settings. Moreover, its accuracy in testing stored samples from non-expectorating patients has not been evaluated.
OBJECTIVE
To assess the performance of Xpert in frozen samples of induced sputum (IS) and sputum from string test (ST) from non-expectorating individuals with presumed TB.
DESIGN
This was a laboratory-based study of 377 ST and IS samples collected between March 2010 and March 2013 at a referral hospital in Uganda. Samples were decontaminated, centrifuged and cultured, and the resultant samples were frozen at −20°C before Xpert evaluation.
RESULTS
TB was detected in ST and IS samples from 19/163 (11.7%) children and 63/201 (29.4%) adults using culture. Xpert sensitivity in frozen sediments from children was 37.5% (95%CI 8.5–75.5) in ST and 41.7% (95%CI 15.2–72.3) in IS samples, with specificities of respectively 100% (95%CI 94.9–100) and 98.6% (95%CI 92.7–100). In adults, sensitivity was respectively 50% (95%CI 31.3–68.7) and 48.5% (95%CI 30.8–66.4) in ST and IS samples, with specificities of respectively 100% (95%CI 95.5–100) and 98.6% (95%CI 92.4–100).
CONCLUSION
Given these results, and particularly the high specificity, the use of Xpert on frozen ST and IS sediment samples from both children and adults is promising.
