Table 4.

Predictors of tuberculosis risk, multivariate analysis by Cox proportional hazard model of PLWHA at Nekemte referral hospital and Nekemte Health center followed from September 2009-September 2012.

Variables (n=600)	Category	Unadjusted HR(95%CI	P-value	Adjusted HR(95%CI	P-value
Age	<=30	Referent		Referent
	31-40	0.47(0.23-0.95)	0.035∗	0.47(0.23-1.05)	0.06#
	41-50	1.05(0.46-2.38)	0.92	1.05(0.46-2.38)	0.072#
	>50	2.9(1.13-7.5)	0.027∗	2.9(1.13-7.5)	0.08#
CD4 Count	>500	Referent
	351-500	1.25(0.314-5.16)	0.749	1.25(0.89 -5.16)	0.68#
	201-350	4.9(1.6-14.6)	0.005∗	4.9(1.6-14.6)	0.007
	<=200	17.5 (6.15-50)	<0.001∗	17.5 (6.15-50)	<0.001
Sex	Male	Referent
	Female	1.095(0.83-	0.525
BMI	>=18.5	Referent		referent
	<18.5	3.13(1.77-5.53)	<0.001∗	3.13(1.77-5.53)	0.042
Weight	>50	Referent		Referent
	<=50	1.72(1.28 -2.28)	0.001∗	1.72(1.28 -2.28)	0.041
WHO Stage	I/II	Referent		Referent
	III/IV	3.22(0.96-4.56)	0.08∗	3.22(1.96-4.56)	0.037
Functional Status	Working	Referent		Referent
	Ambulatory/Bedridden	3.38(1.9-5.96)	<0.001∗	3.38(1.9-5.96)	0.022
Cotrimoxazole Treatment	Yes	Referent		Referent
	No	4.21(2.42-7.4)	<0.001∗	4.21(2.42-7.4)	<0.001
Previous TB	No	Referent
	Yes	1.43(0.56-3.58)	0.45
Opportunistic Infection	No	Referent		Referent
	Yes	1.02(0.51-2.02)	0.06∗	1.02(0.51-2.02)	0.061#
Cohort Group	Yes	Referent		Referent
	No	2.5(1.3-4.9)	.007∗	2.5(1.3-4.9)	0.03

^∗All candidate variables for multivariate analysis.

^#Lost statistical significance in multivariate analysis.