Test characteristics and potential impact of the urine LAM lateral flow assay in HIV-infected outpatients under investigation for TB and able to self-expectorate sputum for diagnostic testing

Table 4 Multivariate analysis showing predictors of all-cause mortality, lost-to-follow-up/mortality and baseline TB-related morbidity in HIV-infected patients overall and restricted to culture-positive TB patients only^a

	P-value	AOR	Lower 95 % CI	Upper 95 % CI
6-month all-cause mortality
All HIV-infected (n^* = 173)
Age, years	0.051	0.934	0.867	0.997
CD4 cell count, cells/mm³	0.038	0.996	0.991	0.999
TB culture positive (pos/neg)	0.007	9.207	2.069	53.4
Baseline TB score	0.050	1.349	1.012	1.86
LAM strip result^b	0.170	2.762	0.613	11.72
TB culture positive (n = 117)
CD4 cell count, cells/mm³	0.069	0.9951	0.989	0.999
Baseline TB score	0.007	1.683	1.19	2.555
LAM strip result	0.252	2.59	0.495	13.72
Lost-to-follow-up or deceased
All HIV-infected (n = 268)
Age, years	0.048	0.966	0.932	0.999
Zambia	<0.0001	12.52	4.077	45.15
Durban, South Africa	0.052	3.701	1.027	14.82
CD4 cell count, cells/mm³	0.058	0.998	0.997	1
LAM strip result	0.002	4.382	1.834	11.58
TB culture positive (n = 170)
Age, years	0.010	0.944	0.903	0.985
Zambia	0.002	5.897	2.055	18.61
Baseline TB score	0.006	1.301	1.085	1.583
LAM strip result	0.007	4.717	1.638	15.85
Baseline TB score (n = 146)
TB culture positive
CD4 cell count, cells/mm³	0.02716	0.9977	0.9958	0.9997
LAM strip result (pos/neg)	0.1771	1.886	0.7542	4.717

Abbreviations: ARV, anti-retroviral; CI, confidence interval; AOR, adjusted odds ratio
^aAdditional co-variates included all multivariate models as potential confounders include: study arm (Xpert MTB/RIF or smear), study site, sex, age. Additional co-variates evaluated as additional predictors of outcomes included: positive initial TB result (Xpert MTB/RIF or smear), ARV treatment status, previous TB and CXR suggestive of TB. LAM strip test is always shown while only significant co-variates in each multivariate regression are shown
^*n = number included in each of the multivariate models depending on outcome and sub-group considered
^bLAM strip results are considered as a binary variable (positive/negative) using the grade 2 cut-point. Results of the multivariate analysis do not change if LAM strip results are considered as continuous variable from grade 0-5

ISSN: 1471-2334