Prevalence and Predictors of Non-Exclusive Breastfeeding at Hospital Discharge in Uruguay

BACKGROUND: The prescription of infant formula during postpartum hospitalization is one of several factors that influence breastfeeding. RESEARCH AIMS: To analyze the prevalence of non-exclusive breastfeeding at hospital discharge in Uruguay, a Latin American country, and to identify the predictors of non-exclusive breastfeeding at hospital discharge. METHOD: This cross-sectional study draws on data from all births between 2008 and 2020 registered in the Uruguayan perinatal information system. XGBoost, a machine-learning algorithm, was used to identify the predictors of infant formula at hospital discharge. A binary variable indicating non-exclusive breastfeeding (1 = yes, 0 = no) was used as dependent variable, and another 35 variables were used as predictors, including mother, child, and hospital characteristics. RESULTS: The prevalence of non-exclusive breastfeeding at hospital discharge descriptively increased in the period, from 7.0% in 2008 to 27.9% in 2020. The model had a moderately adequate prediction performance (ROC AUC = .781, 95% CI [.778, .785], Brier score = .188, 95% CI [.187, .189], PR AUC = .424, 95% CI [.187, .431]). Hospital type was the most important variable in predicting non-exclusive breastfeeding, followed by birth weight and type of birth. Predicted probability of non-exclusive breastfeeding was highest in public hospitals in the capital (Montevideo), for infants with birth weight lower than 3000 g and higher than 3900 g and born through Cesarean delivery. CONCLUSION: Results suggest the need to address the structural and procedural factors that may contribute to the prescription of infant formula to improve breastfeeding practices in the country. Additional studies are needed to identify the specific maternity practices that promote the prescription of infant formula at hospital discharge.