Table 2. Univariate and multivariable regression analysis of MLRs in tuberculosis.

			Univariate analysis		Multiple linear regressiona
		n	MLR	P-value	Coefficient	P-value
			Log mean ± SE		± SE
Sex	Male	176	-0.38 ± 0.020	0.1130
	Female	88	-0.43 ± 0.025
Smear	Positive	67	-0.25 ± 0.034	0.0001	-0.076 ± 0.13	0.555
	Negative	197	-0.44 ± 0.017
Culture	Positive	157	-0.33 ± 0.021	<0.0001	0.042 ± 0.038	0.268
	Negative	107	-0.49 ± 0.022		(0.095 ± 0.029)b	(0.0011)
Site of TB	Pulmonary	103	-0.32 ± 0.028	0.0002	-0.06 ± 0.059	0.345
	Extra-pulmonary	161	-0.44 ± 0.019
HIV	Positive	9	-0.48 ± 0.104	0.3442
	Negative	242	-0.39 ± 0.017
Diabetes	Positive	35	-0.39 ± 0.043	0.8813
	Negative	229	-0.39 ± 0.017
Alcohol problem	Yes	36	-0.38 ± 0.047	0.5369
	No	228	-0.40 ± 0.017
Drug resistance	Yes	28	-0.41 ± -0.059	0.2030
	No	130	-0.34 ± 0.020
BCG scar	Yes	195	-0.39 ± 0.019	0.8778
	No	54	-0.40 ± 0.032
	Linear regression analysis
			R-squared
Age (years)		263c	0.0074	0.1644
White blood count		264	0.0775	<0.0001	0.013 ± 0.005	0.011
					(0.016 ± 0.005)	(0.0013)
Albumin (g/L)		254	0.257	<0.0001	0.092 ± 0.133	0.49
					(-0.015 ± 0.0021)	(<0.00001)
Globulin (g/L)		251	0.0414	0.001187	-0.011 ± 0.014	0.407
Albumin-to-globulin ratio		250	0.1758	<0.0001	-0.018 ± 0.493	0.971
Log CRP (mg/L)		174	0.2709	<0.0001	-0.335 ± 0.274	0.223

^a Data are given for the initial model with the

^b best-fit model in brackets

^c One outlier with a monocyte count of 8 was excluded