RESUMO
BACKGROUND: Diagnosis and timely treatment of tuberculosis in children is hampered by the absence of fast and reliable tests, especially in the era of human immunodeficiency virus (HIV). The aim of this study was to evaluate the diagnostic performance of the Xpert MTB/RIF assay (Xpert) in children with suspected tuberculosis in a high tuberculosis/HIV-burden setting. METHODS: In a prospective study with a minimum follow-up of 12 months, 164 children with suspected tuberculosis were assigned to predefined diagnostic subgroups, based on microbiological and clinical findings. Results of smear microscopy and culture were compared against diagnostic performance of Xpert. RESULTS: Twenty-eight of 164 children (17.1%) had confirmed tuberculosis. Xpert detected 100% (95% confidence interval [CI], 59.0%-100%) of smear-positive cases and 66.6% (95% CI, 43.0%-85.4%) of culture-positive but smear-negative cases. In the per-sample analysis, Xpert displayed a similar sensitivity (54.7% [95% CI, 42.7%-66.2%]) compared with culture methods. Xpert detected 3-fold more confirmed tuberculosis cases than smear microscopy but with equal rapidity. Four additional cases (8.5%) with clinical tuberculosis but negative culture were diagnosed by Xpert. Testing second and third samples increased sensitivity by 20% and an additional 16%, respectively. When tuberculosis was reliably excluded, Xpert's specificity was 100%. HIV infection did not affect diagnostic accuracy of Xpert. CONCLUSIONS: Xpert was easy to perform and displayed similar diagnostic accuracy as culture methods in children with suspected tuberculosis. Rapid turnaround times should reduce treatment delay and improve patient outcome, although sensitivity remains suboptimal and access is dependent on local laboratory infrastructure.