Abstract
BACKGROUND: In recent years, studies on the diagnostic accuracy of in-house real-time PCR (hRT-PCR) assay for the detection of Mycobacterium tuberculosis (Mtb) have been reported with unignorable discrepancies. To assess the overall accuracy of the hRT-PCR assay for Mtb diagnosis in different samples for individuals with active pulmonary and extra-pulmonary Mtb infection, a systematic review and meta-analysis were performed. METHODS: The PUBMED, EMBASE, Web of Science, and Cochrane databases were searched up to June 2017 for eligible studies that estimated diagnostic sensitivity and specificity with the hRT-PCR assay in respiratory and non-respiratory samples in pulmonary and extra-pulmonary Mtb infection patients, with Mtb culture as the reference standard. Bivariate random effect models were used to provide pooled estimation of diagnostic accuracy. Further, subgroup and meta-regression analyses were performed to explore sources of heterogeneity. The risk of bias was assessed by the QUADAS-2 tool. RESULTS: Of the 3589 candidate studies, 18 eligible studies met our inclusion criteria. Compared to Mtb culture data, the pooled sensitivity and specificity were 0.96 and 0.92, respectively. The diagnostic odds ratio (DOR) was 192.96 (95% CI 68.46, 543.90), and the area under the summary ROC curve (AUC) was 0.9791. There was significant heterogeneity in sensitivity and specificity among the enrolled studies (p < 0.001). The studies with high-quality assessment and application of respiratory specimen were associated with better accuracy. CONCLUSIONS: In low-income/high-burden settings, our results suggested that the hRT-PCR assay could be a useful test for the diagnosis of TB with high sensitivity and specificity.