Table 3. Logistic regression model for G-ve infections in our cohort.

Clinical variables available at birth along with TLR variants were investigated by logistic regression for association with G-ve infections. A step-wise elimination was used to discard non-significant variables. Then CVL days (postnatal variable) was entered into the model and stepwise elimination performed till only significant (P < 0.05) risk factors associated with G-ve infections remained. The final model representing variables significantly associated with the outcome is depicted.

Variable	Odds Ratio	95% CI	Alpha
Birth-weight (gm)	0.998	0.996 – 0.999	0.002
TLR4 variant presence	3.1	1.2 – 7.8	0.018
IRAK1 T/C genotype	0.09	0.01 – 0.77	0.028
CVL days (day)	1.03	1.02 – 1.05	<0.001