114 Distributional Decomposition: A Novel Method for Understanding Inequities in Child Growth, Behavior and Development
Journal Articles
Overview
Research
Identity
Additional Document Info
View All
Overview
abstract
AbstractBackgroundIncome related inequities in child health are well-established, with children from lower income households showing increased risk of obesity, behavior problems, and delayed development. To facilitate clinical diagnosis, outcomes are conventionally measured in dichotomous terms. However, inequities may exist along the entire range of distribution, with implications for population health.ObjectivesOur primary objective was to examine differences in the distribution of three measures of child health by income: body mass index (BMI), behavior difficulties and development.Design/MethodsThis was a cross sectional study of children enrolled in a primary care practice-based research cohort. Our study included generally healthy children recruited from age 0-5 years. Dependent variables were 1) BMI z-score (zBMI) at 5 years; 2) behavior: total score on the Strengths and Difficulties Questionnaire (SDQ), measured at 3-5 years; 3) development: total score on the Infant Toddler Checklist (ITC), measured at 18-24 months. Independent variable was parent-reported annual household income (< $100,000 vs ≥ $100000). We then used distributional decomposition, which uses mathematical re-weighting to construct a counterfactual distribution that describes the distribution of the lower income group based on the predictor profile (child age, sex, birthweight, prematurity, breastfeeding duration; maternal age, education, immigration status, ethnicity) of the higher income group.ResultsOur study samples consisted of 1649 (zBMI), 764 (SDQ) and 761 (ITC) children. Mean BMI z-score was 0.16, median total difficulties score was 7, median ITC score was 48. Comparing distributions graphically (Figure 1), children with low income have a higher risk distribution for all outcomes; for example, children with low income were more likely to have BMI z-scores in the underweight and obese ranges. For each outcome, the counterfactual curve lower income children with the predictor profile of their higher income counterparts reduced inequities somewhat, particularly in the normal or low risk range, but not in the high-risk range. However, there were notable unexplained portions of inequity remaining.ConclusionIn a cohort of generally healthy children, we found evidence of meaningful income-related inequities in the distribution of child zBMI, behavior difficulties, and development. Population health interventions aiming to mitigate these inequities by addressing common predictors may improve outcomes in the normal range; however targeted clinical interventions are likely required for those in the high-risk range.
