Table 4

Multivariable analysis for estimation of prevalence of MDR-TB in the new cases^* (n=1554) and retreatment cases^† (n=418)

Variable	New cases (n=1554)			Retreatment cases (n=418)
Variable	% of MDR-TB	AOR (95% CI)	p Multilevel	% of MDR-TB	AOR (95% CI)	p
Sex
Men	7.9	1.45 (0.91 to 2.31)	0.114	28.1	0.72 (0.43 to 1.21)	0.213
Women	5.5	1 (reference)	0.114	36.1	1 (reference)	0.213
Age
<45 years	8.5	1.72 (1.13 to 2.61)	0.012	36.9	1.44 (0.91 to 2.30)	0.122
≥45 years	5.5	1 (reference)	0.012	25.6	1 (reference)	0.122
Not working
Yes	5.0	0.71 (0.43 to 1.18)	0.184	27.4	0.68 (0.39 to 1.16)	0.158
No	8.2	1 (reference)	0.184	31.5	1 (reference)	0.158
Delay in initiating TB treatment
>60 days	6.8	1.09 (0.71 to 1.66)	0.692	38.6	2.64 (1.61 to 4.30)	<0.001
≤60 days	7.4	1 (reference)	0.692	21.4	1 (reference)	<0.001
Cavitary disease
Yes	6.9	1.05 (0.68 to 1.61)	0.835	34.3	1.14 (0.72 to 1.83)	0.575
No	7.4	1 (reference)	0.835	27.8	1 (reference)	0.575
TB in organs other than lungs
Yes	13.6	2.40 (0.66 to 8.76)	0.183	45.5	1.03 (0.26 to 4.05)	0.966
No	7.1	1 (reference)	0.183	30.1	1 (reference)	0.966
Diabetes mellitus
Yes	6.7	1.39 (0.48 to 4.03)	0.544	31.6	1.45 (0.49 to 4.32)	0.507
No	7.2	1 (reference)	0.544	30.3	1 (reference)	0.507
Provider of last treatment
TB dispensary/TB hospital				35.6	1.84 (1.14 to 2.97)	0.012
General hospital/others				25.6	1 (reference)	0.012
Length of treatment with rifampin and isoniazid
>180 days				56.0	4.82 (2.97 to 7.81)	<0.001
≤180 days				20.7	1 (reference)
ICC		0.113			0.023

The associations between multidrug resistant tuberculosis (MDR-TB) and independent variables were expressed as adjusted OR (AOR) with 95% CI.
↵* Two-level multilevel model was fitted using SAS PROC GLIMMIXED.
↵† Binominal regression model was fitted using PROC LOGISTIC.
ICC, intra-class correlation coefficient.