Abstract
BACKGROUND: Programs, such as Medicaid, the Supplemental Nutrition Assistance Program, and the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), provide access to vital medical and nutrition services. Few studies have investigated whether demographic, social, and economic determinants of health, including length of time spent on these programs, are associated with diet quality and weight status in early childhood. OBJECTIVES: Classification and regression tree analysis, a machine learning method, was used to determine health predictors to identify patterns of children with higher compared with lower diet quality and higher compared with lower weight status. METHODS: Using the WIC infant and toddler feeding practices study-2 (unweighted N = 3051; weighted N = 413,211), classification and regression tree identified the sequence of binary splits that best differentiated the sample on Healthy Eating Index-2020 (HEI-2020; range 0-100) and HEI-2020 subscales (adequacy and moderation), and body mass index (BMI in kg/m(2)) z-score at 2-5 y. Predictors, including maternal BMI, child birthweight, sociodemographics, and length of time spent on safety net programs, were considered. RESULTS: Higher HEI-2020 scores were primarily predicted by race and ethnicity (e.g., Hispanic, Spanish, or non-Hispanic White), and longer WIC and shorter Supplemental Nutrition Assistance Program duration. In examining HEI-2020 subscales, higher HEI adequacy was primarily predicted by higher education, older maternal age, longer WIC duration, and race and ethnicity. Higher HEI moderation was primarily predicted by race and ethnicity and longer WIC duration. Higher BMI z-score was primarily predicted by higher birth weight. CONCLUSIONS: Child diet quality and weight status were associated with different social determinants of health, which included maternal weight status, race and ethnicity, and food assistance program participation, particularly WIC.