Head-to-head comparison of the diagnostic accuracy of Xpert MTB/RIF and Xpert MTB/RIF Ultra for tuberculosis: a meta-analysis

Abstract

Background

Tuberculosis (TB) diagnosis has significantly improved since the introduction of the automated molecular test Xpert MTB/RIF (Xpert) and the new version Xpert MTB/RIF Ultra (Ultra) that detect Mycobacterium tuberculosis. Due to the rapidly widespread use of Xpert and Ultra, we conducted a meta-analysis to compare the performances of Xpert and Ultra in diagnosing TB and discuss the advantages and limitations of these two tests.

Methods

Web of Science, Medline (via PubMed), Embase (via OvidSP), the Cochrane Central Register of Controlled Trials and Google Scholar (up to April 2020) were searched for relevant studies. The diagnostic performance of Xpert and Ultra for TB was determined using a bivariate random-effects regression model. The sources of heterogeneity were explored via meta-regression and subgroup analyses.

Results

Of 19 studies that examined a total of 5855 samples, the pooled sensitivity and specificity of Xpert in TB diagnosis were 0.69 (95% CI: 0.57–0.78) and 0.99 (95% CI: 0.98–0.99), respectively. However, the pooled sensitivity and specificity of Ultra in TB diagnosis were 0.84 (95% CI: 0.76–0.90) and 0.97 (95% CI: 0.96–0.98), respectively. Regardless of whether the comparisons were indirect or direct, Ultra was consistently found to be more sensitive, but with slightly lower specificity than Xpert in diagnosing TB.

Conclusions

Compared with Xpert, Ultra had higher sensitivity but slightly lower specificity for the diagnosis of TB disease. The excellent upgrade in sensitivity of the Ultra test was particularly relevant in subjects with paucibacillary TB including tuberculous pleurisy, tuberculous meningitis and paediatric TB.

Keywords:

• Xpert MTB/RIF
• Xpert MTB/RIF Ultra
• diagnosis
• tuberculosis

Acknowledgments

We thank all members of the Department of Respiratory and Critical Care Medicine, and the First and Second Affiliated Hospital of Anhui Medical University for their strong support of this study.

Disclosure statement

The authors declare that they have no competing interests.

Author contributions

Conceptualization: JJ and ST. Methodology: JJ and NZ. Software: JJ and JY. Validation: JJ and JY. Formal analysis: JJ, YJ and JY. Investigation: JJ and YS. Resources: JJ and GS. Data curation: JJ and YL. Writing (original draft preparation): JJ and GS. Writing (review and editing): JJ and GS. Visualization: JJ and GS. Supervision: JJ and NZ. Project administration: JJ and GS. Funding acquisition: GS.

Additional information

Funding

This study was supported by two National Natural Science Foundation of China grants (81370170 and 81770081).