Performance of Xpert MTB/RIF and Alternative Specimen Collection Methods for the Diagnosis of Tuberculosis in HIV-Infected Children

Background. The diagnosis of tuberculosis in human immunodeficiency virus (HIV)-infected children is challenging. We assessed the performance of alternative specimen collection methods for tuberculosis diagnosis in HIV-infected children using Xpert MTB/RIF (Xpert). Methods. HIV-infected children aged <= 13 years with suspected intrathoracic tuberculosis were enrolled in 8 hospitals in Burkina Faso, Cambodia, Cameroon, and Vietnam. Gastric aspirates were taken for children aged < 10 years and expectorated sputum samples were taken for children aged >= 10 years (standard samples); nasopharyngeal aspirate and stool were taken for all children, and a string test was performed if the child was aged >= 4 years (alternative samples). All samples were tested with Xpert. The diagnostic accuracy of Xpert for culture-confirmed tuberculosis was analyzed in intention-to-diagnose and per-protocol approaches. Results. Of 281 children enrolled, 272 (96.8%) had >= 1 specimen tested with Xpert (intention-to-diagnose population), and 179 (63.5%) had all samples tested with Xpert (per-protocol population). Tuberculosis was culture-confirmed in 29/272 (10.7%) children. Intention-to-diagnose sensitivities of Xpert performed on all, standard, and alternative samples were 79.3% (95% confidence interval [CI], 60.3-92.0), 72.4% (95% CI, 52.8-87.3), and 75.9% (95% CI, 56.5-89.7), respectively. Specificities were >= 97.5%. Xpert combined on nasopharyngeal aspirate and stool had intention-to-diagnose and per-protocol sensitivities of 75.9% (95% CI, 56.5-89.7) and 75.0% (95% CI, 47.6-92.7), respectively. Conclusions. The combination of nasopharyngeal aspirate and stool sample is a promising alternative to methods usually recommended by national programs. Xpert performed on respiratory and stools samples enables rapid confirmation of tuberculosis diagnosis in HIV-infected children.