Food, Health & Choices (FHC): School Demographics and Neighborhood Food Environment Explains Variations in BMI Across Schools

Abstract

To determine how school-level demographics and neighborhood food environment explain the variation in BMI z-score among twenty urban, low-income high-minority schools. Cross-sectional, multilevel analysis of fifth grade students (n=952) from 20 schools. Heights, weights, gender, and age were collected for each student and BMI z-scores were calculated according to CDC recommendations. BMI z-scores showed a moderate intraclass correlation coefficient of 0.031. School-level demographics were obtained from publicly available school records. Counts of fast food and healthy retail outlets (supermarkets, farmers markets, and mobile produce vendors) were obtained from the commercial vendor ReferenceUSA® for the schools’ attendance zone, and density was calculated per square kilometer. We conducted multilevel linear regression models, to determine variance explained of between-school BMI z-scores, in three stages: MODEL 1 included gender (50.8% female) and age (10.6±0.6 years), MODEL 2 added school-level demographic [percent Black (29.6±22.0%), Hispanic (59.0±23.2%), eligible for free/reduced price lunch (86.0±19.3%) and English-language learners (16.6±10.6%)]; and MODEL 3 added fast food restaurants (7.9±5.3; density: 18.31±11.70/km2) and healthy retail options (8.7±6.4; density: 20.10±14.66/km2). Proportion of between-school variation (R-squared of level-two) was calculated. The models explained increasing amounts of BMI z-score between-school variation: MODEL 1 accounted for 0.5%, MODEL 2 accounted for 83.6%, and MODEL 3 accounted for 98.6%. Childhood obesity is of great concern to health practitioners, educators, parents, and policy makers. Understanding the large degree in which school-level demographics and neighborhood food environments explain between-school BMI z-score variance can help inform policies and practices that can lead to decreased childhood obesity rates.