Concordance between individual vs. Area-level socioeconomic measures in an Urban setting

Area-level socioeconomic status (SES) measures have been used as a proxy in child health research when individual SES measures are lacking, yet little is known about their validity in an urban setting. We assessed agreement between census block-group and individual-level SES measures obtained from a caregiver telephone survey in Jackson County, Missouri. Associations with prevalence of childhood overweight (OW), low birth weight (LBW), and household smoking exposure were examined using logistic regression models. Seven hundred eighty-one households were surveyed: 49% male, 76% White, mean child age 9.4 years. We found misclassification rates of 20-35% between individual vs. arealevel measures of education and income; Kappa indices ranged from 0.26-0.36 indicating poor agreement. Both SES measures showed an inverse association with LBW and smoking exposure. Area-level SES measures may reflect a construct inclusive of neighborhood resources; routine substitution of these measures should be interpreted with caution, despite similar correlations with health outcomes.