Bayesian latent class analysis produced diagnostic accuracy estimates that were more interpretable than composite reference standards for extrapulmonary tuberculosis tests

Table 5 Observed counts, expected counts, and TB meningitis probability by test result pattern

Culture	Xpert	Smear	ADA	No. observed	No. expected by LCA (95% CrI)	Probability of TB meningitis by LCA (95% CrI)	Probability of TB meningitis by CRS
Culture	Xpert	Smear	ADA	No. observed	No. expected by LCA (95% CrI)	Probability of TB meningitis by LCA (95% CrI)	CRS1+	CRS2+	CRS3+	CRS4+
−	−	−	−	170	166 (158,173)	0.01 (0, 0.04)	0	0	0	0
−	−	−	+	27	29 (22,37)	0.33 (0.08, 0.76)	1	0	0	0
−	−	+	−	2	2 (1,6)	0.004 (0, 0.07)	1	0	0	0
−	+	−	+	3	2 (1,4)	0.96 (0.06, 0.99)	1	1	0	0
+	−	−	−	1	2 (1,6)	0.30 (0.03, 0.93)	1	0	0	0
+	−	−	+	6	4 (2,7)	0.97 (0.08, 1)	1	1	0	0
+	+	−	+	7	6 (4,10)	1.0 (0.99, 1)	1	1	1	0
+	+	+	−	2	1 (0,3)	1.0 (0.99, 1)	1	1	1	0
+	+	+	+	6	6 (4,9)	1.0 (0.99, 1)	1	1	1	1
−	−	−	NA	6	NE	0.03 (0, 0.08)	NE	0	0	0

Tests were all performed in CSF samples (n = 230). A “+” indicates positive test result and “−” indicates negative test result. CRS 1+ indicates any one positive test in CRS versus all four parameters being negative. CRS 2+: any two tests positive in CRS versus all four parameters being negative. CRS 3+: any three tests positive in CRS versus all four parameters being negative. CRS 4+: all four tests positive in CRS versus all four parameters being negative
ADA adenosine deaminase, CrI credible interval, CSF cerebral spinal fluid, LCA latent class analysis, NA value not available, NE result not estimable, No. number of, Smear smear microscopy, TB tuberculosis

ISSN: 2397-7523