Final model of multiple logistic regression of factors associated with CMV seroprevalence in children 3-5 years of age in the USA.

Table 4: Final model of multiple logistic regression of factors associated with CMV seroprevalence in children 3-5 years of age in the USA.

Variable	Adjusted OR (95% CI)	P Value
Breastfeeding (any vs none)	3.34 (1.85-6.05)	< 0.001
Birthweight in gm (less 1500 vs 1500+)	3.93 (0.256-60.43)	0.373
Age of child in years (4-5 vs 3)	1.96 (1.03-3.70)	0.042
Race (non-White vs White)	1.54 (0.96-2.45)	0.059
Cotinine Levels in ng/mL (0.05 vs 0.05+)	0.67 (0.0.44-1.02)	0.062

In this final model, breastfeeding was the most significant predictor of CMV seroprevalence. Breastfed children at any time post-delivery were 2+ times more likely to be CMV sero +ve as compared to their non-breastfed counterparts. As measured by cotinine levels, children not exposed to tobacco/E-cigarettes were 0.7 times less likely to be CMV sero +ve as compared to their unexposed counterparts (P = 0.062).