Table 2.

Diagnostic performance of Xpert, Xpert Ultra, and MGIT culture for the diagnosis of tuberculous meningitis

	Number	Sensitivity vs composite microbiological reference standard*	p value†	Sensitivity vs uniform case definition	p value‡	Negative predictive value vs uniform case definition	Specificity vs uniform case definition
Xpert Ultra	204	92·9% (80·5–98·5); 39/42	..	76·5% (62·5–87·2); 39/51	..	92·7% (87·6–96·2); 153/165	100% (97·6–100); 153/153
Xpert	166	65·8% (48·6–80·4); 25/38	0·0063	55·6% (44·0–70·4); 25/45	0·0010	85·8% (78·9–91·1); 121/141	100% (97·0–100); 121/121
MGIT culture	142	72·2% (55·9–86·2); 27/37	0·092	61·4% (45·5–75·6); 27/44	0·020	85·2% (77·4–91·1); 98/115	100% (96·3–100);98/98

Values are percentage (95% CI); numerator/denominator unless stated otherwise. Xpert=Xpert MTB/RIF. MGIT=mycobacterial growth indicator tube.

^*The composite microbiological reference standard included a positive CSF test on any of Ziehl–Neelsen stain microscopy, Xpert, Xpert Ultra, and MGIT culture. Specificity (and the positive predictive value) versus the composite endpoint is by definition 100% as the index test is included in the reference standard of definite tuberculous meningitis. If the Xpert Ultra result is excluded when assigning the case definition, the specificity of Xpert Ultra is 96% (95% CI 91–98; 153 of 160 patients) and the positive predictive value is 82% (66–93; 32 of 39).

^†McNemar's test comparing the sensitivity of Xpert or MGIT culture with that of Xpert Ultra.

^‡Cochran–Mantel–Haenszel test comparing the distribution of Xpert or MGIT results with that of Xpert Ultra results against the uniform clinical standard of definite or probable tuberculous meningitis.