Abstract
BACKGROUND: Diagnosis of pulmonary tuberculosis remains grim, especially in resource-limited settings. Low quality of sputum, particularly among seriously ill, HIV/AIDS, and pediatric patients might result in missing the diagnosis. This study evaluated the performance of GeneXpert MTB/RIF for the detection of pulmonary tuberculosis on stool specimens as an alternative to respiratory specimens. METHODS: A cross-sectional study design was used to evaluate the performance of GeneXpert MTB/RIF to detect TB in stool specimens from presumptive TB patients. Sputum culture on Lowenstein-Jensen media was used as the gold standard. Recruitment of patients into the study was conducted in 12 selected health facilities in Tanzania. Two sputa and a stool specimen were collected from each study participant. Both sputa and stool samples were tested at their respective study sites of collection using GeneXpert, and their respective portions shipped to the Central Tuberculosis Reference Laboratory for testing by stool GeneXpert and sputum culture in the LJ media. Statistical analysis was performed using STATA software version 14.1. RESULTS: A total of 590 presumptive tuberculosis patients were enrolled in this study. Their median age was 35 years (IQR = 21-47 years). More than half (57.5%, n = 339) of the study participants, were males. Children aged below 15 years constituted 17.6% (n = 104) of the study participants. A total of 75 tuberculosis cases were detected by sputum culture. The sensitivity and specificity of Stool GeneXpert conducted at CTRL was 84% (95% CI: 81.0-87.0%), and 93.4% (CI: 98.5-99.9%) respectively. The overall sensitivity and specificity of stool GeneXpert at the peripheral laboratories was 63.0% (95% CI: 47.8-76.1) and 76.7% (95% CI: 72.1-81.4), respectively. CONCLUSION: Findings from this study suggest that stool is a potential alternative to respiratory specimen for use in routine diagnosis of tuberculosis, especially when obtaining a respiratory specimen is challenging.