What does IGRA testing add to the diagnosis of ocular tuberculosis? A Bayesian latent class analysis

BMC Ophthalmology

Table 3 Parameters from latent class model

	Estimated effect	95% CI
Equation estimating the probability of ocular TB
Asian ethnicity^a	1.33	0.94 to 1.89
African ethnicity^a	1.13	0.76 to 1.68
Female^a	0.89	0.64 to 1.24
Age (10 year increment, centred on 40)^a	1.08	0.87 to 1.37
Equations predicting observed signs and symptoms based on the probability of ocular TB
Mean QFT4 if TB+	1.19	1.12 to 1.26
Mean QFT if TB+	5.15	4.30 to 6.19
SD of QFT4 if TB+	0.44	0.39 to 0.49
SD of QFT4 if TB-	0.06	0.05 to 0.07
Crossover between TB+ and TB- QFT distribution (not transformed to QFT4)	0.05	0.04 to 0.07
Effect of estimated probability of ocular TB on anterior uveitis symptoms^b	0.69	0.45 to 1.02
Effect of estimated probability of ocular TB on intermediate uveitis symptoms^b	0.55	0.36 to 0.82
Effect of estimated probability of ocular TB on posterior uveitis symptoms^b	1.54	1.09 to 2.18
Effect of estimated probability of ocular TB on choroiditis^b	21,246	0.97 to 3.10
Effect of estimated probability of ocular TB on bilateral symptoms^b	1.04	0.75 to 1.46
Effect of estimated probability of ocular TB on ACE^c	3.99	−0.40 to 8.23

TB Tuberculosis, QFT QuantiFERON Gold In-Tube, ATT Anti-tubercular therapy, SD Standard deviation, CI Confidence interval
^aThis can be interpreted as an odds ratio for this characteristic in patients with tuberculous uveitis compared to those without
^bThis can be interpreted as the odds ratio for having a particular sign or symptom if the patient has tuberculous uveitis (predicted probability > 50%), compared to not
^cThis can be interpreted as an additional odds ratio that is applied to the prediction of treatment failure

ISSN: 1471-2415