117 Using composite area-level measures as a proxy for self-report family income

Abstract

Abstract Background Socioeconomic status (SES) is a well-established social determinant of child health. When reliable self-report family income is unavailable, area-level measures, such as median neighbourhood income, are commonly used as a proxy. However, median neighbourhood income is not a good proxy for self-report family income. Newer area-level measures, such as the Neighbourhood Equity Score (NES) and the Child and Family Inequities Score (CFIS) are composite scores comprised of indicators of well-being such as income, parental education and physical surroundings. Objectives The primary objective was to evaluate the agreement between self-report family income and three area-level measures: median neighbourhood income, NES, and CFIS. The secondary objective was to examine the association between self-report family income, NES, and CFIS with two health indicators associated with SES: overweight/obesity (BMI z-score>1) and short breastfeeding duration (<6 months). Design/Methods We conducted a cross-sectional study using data from a healthy urban Canadian cohort of young children (0-5 years) attending a scheduled health supervision visit in primary care. Parents completed a questionnaire including family income, postal code and breastfeeding duration. Research assistants measured height and weight (to calculate body mass index). Postal code was used to determine each area-level measure. Agreement between self-report family income and area-level measures was evaluated using kappa coefficients. The percentage of families accurately classified by area-level measures compared with self-report family income was calculated. Multivariable logistic regression was used to evaluate the association between self-report family income, NES, and CFIS (quintiles) with the two health indicators (present/absent). Results 5149 children were included (mean age 21 months). Agreement between self-report family income and both NES and CFIS was ‘fair’ (weighted k=0.29 for both), and agreement with median neighbourhood income was ‘poor’ (weighted k=0.09). Accurate classification between self-report family income and the three measures were: median neighbourhood income (5.6%), NES (32.2%), CFIS (32.4%). For children in the lowest vs. highest quintile, the odds (95% CI) of overweight/obesity were: self-report family income OR=2.75 (1.60-4.72), NES OR=2.02 (1.23-3.30), CFIS OR=2.04 (1.25-3.33); and having short breastfeeding duration: self-report family income OR=1.61 (1.32-1.97), NES OR=1.84 (1.50-2.25), CFIS OR=1.84 (1.46-2.31). Conclusion Agreement and accurate classification between self-report family income was strongest for composite area-level measures (NES and CFIS), compared with median neighbourhood income. Both self-report family income and composite area-level measures supported the same conclusion that lower SES was associated with poorer health outcomes. These newer measures may be more appropriate than median neighbourhood income when self-report family income is unavailable.