[Birth weight as a factor influencing infant feeding in Polish maternity wards].

Data on 11784 newborns collected by staff in 427 maternity hospitals in 1995 were analysed. In all 97.2% of newborns were breastfed, but only 72.5% with birth weight less than 2500 g. Univariate logistic regression model with birth weight as a continuous variable was applied to calculate the probability of breastfeeding initiation. Using bivariate logistic regression models, different groups of newborns were compared to explain the additional influence of 4 dichotomous variable: type of delivery (normal birth and caesarean section), neonatal problems (jaundice, birth asphyxia, disorders of adaptation after birth, infections, congenital malformations) and two hospital practices: rooming-in and skin-to-skin contact. Multivariate conditional logistic regression model showed that the independent prognostic factors for breastfeeding initiation were: birth weight (continuous), rooming-in (OR=5.5; CI(OR)=3.9-7.9) skin-to-skin contact after birth (OR=3.4; CI(OR)=2.6-4.5) and lack of neonatal problems (OR=2.1; CI(OR)=1.6-2.7). Logistic regression curves showed that influence of skin-to-skin contact and rooming-in after birth on breastfeeding initiation was more evident for newborn with lower birth weight.