Diagnostic accuracy of the Xpert MTB/RIF assay for abdominal tuberculosis: a systematic review and meta-analysis

Background We performed a meta-analysis to determine diagnostic accuracy of Xpert MTB/RIF for diagnosis of abdominal (intestinal or peritoneal) tuberculosis (TB) in various tissues (intestinal, omental/peritoneal tissue or ascitic fluid). Methods Electronic databases were searched for observational studies on use of Xpert MTB/RIF in ascitic fluid, peritoneal, or omental tissue for diagnosis of peritoneal and intestinal TB. We calculated the pooled sensitivity, specificity and diagnostic odds ratio of Xpert MTB/RIF for diagnosis of peritoneal TB in comparison to composite reference standard (CRS) and culture, and in comparison to CRS for intestinal TB. Results Twenty-five observational studies were included. The pooled sensitivity and specificity as assessed with peritoneal culture from ascites as an Index test was 64% (95% Confidence Interval [C.I.] 49-76%) and 97% (95% C.I., 95-99%) respectively and with peritoneal CRS was 30% (95% C.I., 22-40%) and 100% (95% C.I., 98-100%) respectively. In the intestinal group, the pooled sensitivity and specificity of Xpert MTB/RIF was 23% (95% C.I., 16-32%) and 100% (95% C.I., 52-100%). The AUC of peritoneal culture and intestinal tissue was 0.935 and 0.499. Conclusion Xpert MTB/RIF has modest sensitivity for diagnosis of peritoneal and intestinal tuberculosis but has a good specificity. Prospero registration CRD42020140545