RESUMO
Rationale: Microbiological confirmation of pulmonary tuberculosis in children is desirable.Objectives: To investigate the diagnostic accuracy and incremental yield of Xpert MTB/RIF Ultra (Ultra; Cepheid), a new rapid test, on repeated induced sputum, nasopharyngeal aspirates, and combinations of specimens.Methods: Consecutive South African children hospitalized with suspected pulmonary tuberculosis were enrolled.Measurements and Main Results: Induced sputum (IS) and nasopharyngeal aspirates (NPAs) were obtained. NPAs were frozen; IS underwent liquid culture, and an aliquot was frozen. Ultra was performed on thawed NPAs and IS specimens individually. Children were categorized as confirmed, unconfirmed, or unlikely tuberculosis according to NIH consensus case definitions. The diagnostic accuracy of Ultra was compared with liquid culture on IS. In total, 195 children (median age: 23.3 mo; 32 [16.4%] HIV-infected) had one IS and NPA, and 130 had two NPAs. There were 40 (20.5%) culture-confirmed cases. Ultra was positive on NPAs in 26 (13.3%) and on IS in 31 (15.9%). Sensitivity and specificity of Ultra on one NPA were 46% and 98%, respectively, and similar by HIV status. Sensitivity and specificity of Ultra on one IS were 74.3% and 96.9% respectively. Combining one NPA and one IS increased sensitivity to 80%. Sensitivity using Ultra on two NPAs was 54.2%, increasing to 87.5% with an IS Ultra.Conclusions: IS provides a better specimen than repeated NPA for rapid diagnosis using Ultra. However, Ultra testing of combinations of specimens provides a novel strategy that can be adapted to identify most children with confirmed pulmonary tuberculosis.