Factors predicting overweight in US kindergartners

Background: Childhood overweight is a substantial public-health problem, but little is known about predictors of early childhood overweight. Objective: We aimed to identify factors - alone and in combination - that predict kindergarten overweight. Design: We analyzed nationally representative data from the Early Childhood Longitudinal Study-Birth Cohort, a longitudinal cohort study of 6800 children followed from birth through kindergarten. Multivariable logistic regression and recursive partitioning analysis (RPA) were performed to identify individual and clusters of parental, prenatal/pregnancy, infant, and toddler factors predicting kindergarten overweight. The main outcome was kindergarten overweight [body mass index (BMI) ≥85th percentile, which includes obesity]. Results: The prevalence of kindergarten overweight was 32%. By using combinations (derived from 131 factors) of a weight-for-length or BMI ≥85th percentile at earlier ages, race/ethnicity, a maternal gestational diabetes history, birth weight, and ages at solid-food introduction and the child pulling to a stand, the RPA identified 6 groups with a particularly high prevalence of kindergarten overweight (56-100%) and 2 groups with a particularly low prevalence (11-15%). An especially high prevalence was noted for children with a ≥85th BMI percentile at preschool age (77%) and in children with a ≥85th BMI percentile at 2 y old, for white children whose mother had gestational diabetes (100%), and for minority children with a birth weight <2695.5 g and who pulled themselves to a stand at <7.5 mo old (89%). Conclusion: Clusters of parental, prenatal/pregnancy, infant, and toddler factors can be used to predict which children are at particularly high and low risk of becoming overweight kindergartners.