Characteristics of the included studies.

Background
Abdominal tuberculosis (TB) is a severe extrapulmonary TB, which can lead to serious complications. Early diagnosis and treatment are very important for the prognosis and the diagnosis of abdominal TB is still difficult.
Methods
We searched PubMed, the Cochrane Library, Embase, China National Knowledge Infrastructure, and the Wanfang database for studies evaluating the diagnostic accuracy of NAATs for abdominal TB until August 2020. Any types of study design with full text were sought and included. The risk of bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies tool. Subgroup analysis, meta-regression analysis and sensitivity analysis were used to explore the sources of heterogeneity. Stata version 15.0 with the midas command packages was used to carry out meta-analyses.
Results
We included a total of 78 independent studies from 53 articles; 64 with CRS as the reference standard, and 14 with culture as the reference standard. The pooled sensitivity, specificity, and the areas under summary receiver operating characteristic (SROC) curves (AUC) were 58% (51%–64%; I2 = 87%), 99% (97%–99%; I2 = 81%), and 0.92 (0.89–0.94) compared with CRS, respectively. The pooled sensitivity, specificity, and the AUC values of the SROC were 80% (66%–90%; I2 = 56%), 96% (92%–98%; I2 = 84%), and 0.97 (0.95–0.98) compared with culture, respectively. The heterogeneity of sensitivity and specificity was significant.
Conclusions
NAATs had excellent efficacy in the diagnosis of abdominal TB regardless of the reference standard and regardless of the subtype of abdominal TB. Multiplex PCR with multiple target genes may improve diagnostic sensitivity, and stool specimens may also be used for the diagnosis of abdominal TB in addition to tissue and ascites.