Healthy Kids Obesity Risk Assessment Demonstrates Predictive Validity in Sample of Low‐income Children

Abstract

Medical providers would benefit from a risk assessment tool to supplement a physical exam to predict a BMI trajectory in young children, 3‐5 years. Our objective is to estimate predictive validity of a pictorial assessment tool based on BMI collected 23 months later. Low‐income parents (n=103) provided data about themselves and their young children. Longitudinal data collection included the 43‐item Healthy Kids (HK) and child height/weight for BMI percentile and was repeated 23 months later. Of the 43 potential behavioral items, 14 were selected for the final model using random forests analysis and stepwise regression to predict BMI percentile. The R2 was .74 for the final model that included the HK items and BMI percentile compared to an R2 of .56 for BMI percentile alone. The partial F‐test (F = 4.01, p<.0001) comparing a model with BMI percentile and HK to model with BMI percentile only supports the predictive value of HK tool. Summing the estimates indicates a negative relationship; the lower the HK score, the higher the child's BMI percentile 23 months later. Healthy Kids tool for low‐literate parents adds predictive value to the medical providers' knowledge of the child's BMI.Supported by National Research Initiative #2009‐55215‐05019 & Agriculture and food Research Initiative #2010‐85215‐20658 from the USDA National Institute of Food and Agriculture, Human Nutrition and Obesity 93330.