Abstract P148: Developing a Childhood Obesity Risk Score System Based on Perinatal and Postnatal Risk Factors

Abstract

Introduction: Few studies have examined the relationship between perinatal and postnatal birth factors with later childhood obesity. One notable exception found only infant weight-related factors (such as adiposity rebound) related to weight at age 7, but not other potential risk factors (such as breastfeeding, parity, maternal age, etc., Reilly et al 2005) Our goal with the current study was to expand earlier work, utilizing an Appalachian sample, and to develop a risk score based on an expanded set of risk factors. Methods: The study used longitudinally linked data from three cross-sectional datasets in West Virginia (N=22136). Risk score development followed the tutorial by Sullivan et al (2004). The outcome was defined as childhood obesity at age 10 and calculated as children with 95th percentile BMI for their age and gender. The perinatal predictors included sex, race, health insurance status, family history of cholesterol and cardiovascular disease, smoking during pregnancy, maternal age at birth, breastfeeding intention at birth, and birth weight of the infant. ROC analysis was then performed to establish a cut-off that would maximize the sensitivity and specificity of the perinatal risk factor scores in predicting childhood obesity. Results: Table 1 shows the significant risk factors and associated points. A score of 17 maximized sensitivity (50%) and specificity (60%) using the combined risk factor system for predicting childhood obesity at age 10. Nearly 43% (N=4054 of 9494) of the children had a high perinatal risk score; of those, 34.24% (N=1388) were obese (compared to 26.53% of those with a low perinatal risk score). The odds of being obese in fifth grade were 1.44 (95%CI: 1.32, 1.58) times among those who had a high perinatal score compared to those who had a low score. Discussion: The results of the study can help in identifying infants at birth who are at higher risk of developing childhood obesity. Future research should include other predictors to increase sensitivity and specificity of the risk score system.