An assessment of nucleic acid amplification testing for active mycobacterial infection

Population
Patients with clinical signs and symptoms of active TB who have a specimen suitable for AFB microscopy and culture, and who have had < 3 days of anti-TB treatment
Intervention	NAAT with or without AFB microscopy for the detection of MTB-complex DNA and genetic mutations associated with anti-TB drug resistance
Comparator	AFB microscopy
Reference standard	Culture ± DST
Outcomes^a	Sensitivity Specificity Positive/negative predictive value Level of agreement (concordance of data) Diagnostic yield
Publication type	All study designs listed in the ‘Diagnostic accuracy’ column of Table
Search period	2005 – June 2014
Language	Non-English language articles were excluded unless they provided a higher level of evidence than the English language articles identified

^a Due to the large volume of studies, included studies were limited to those that provided 2x2 data suitable for meta-analysis of sensitivity, specificity and likelihood ratios

Pre-specified subgroups for analysis included patients with a high pre-test probability of active TB (e.g. those from a country with high rates of TB) versus those with a low pre-test probability of TB. Although studies conducted in countries with a high incidence of TB were a good surrogate for patients with a high pre-test probability of having TB, there was no good surrogate for patients with a low pre-test probability of having TB. Many of the patients in those studies conducted in countries with a low incidence of TB were most likely recent immigrants from high-incidence countries. Those studies that used extrapulmonary specimens were a more appropriate surrogate for patients with a low pre-test probability of having TB, as the incidence of TB was lower in these patients.

A total of 79 studies provided data to assess the diagnostic accuracy of NAAT and AFB microscopy compared with culture in mixed pulmonary and/or extrapulmonary specimens from patients suspected of having an MTB infection. Culture methods included standard diagnostic laboratory procedures such as L-J or Ogawa solid media and/or liquid BACTEC media. Of these 79 studies, 20 (10 using an in-house NAAT and 10 using the commercial Xpert NAAT) provided data using mixed pulmonary and extrapulmonary specimens, 34 (21 in-house NAAT and 13 Xpert) using sputum specimens and 40 (29 in-house NAAT and 11 Xpert) using non-sputum specimens. Eight studies only provided data for the accuracy of NAAT compared with culture in patients with AFB-negative specimens. Eleven studies (1 in-house NAAT 10 Xpert) assessed the diagnostic accuracy of NAAT compared with culture to identify patients with drug-resistant MTB infections; 3 of the included studies only provided data for this outcome. The study profiles, patient characteristics and quality appraisal of these studies are listed in Table (Appendix ) and the extracted 2x2 data are presented in Appendix (Table to Table ). An overall summary of the body of evidence is presented in Table .

Table Body of evidence matrix for studies reporting on the accuracy of AFB and NAAT compared with culture in diagnosing MTB infections