Assessing the relationship between BMI and geographic information systems data for people with mobility impairments using a multi-level linear model

People with mobility impairments are at higher risks of obesity. Ecological models of health promotion incorporate individual, social, organizational, and community domains. Multi-level linear models can be used to explain the relationship between health promotion variables, like body mass index (BMI), and the different ecological domains. The purpose of this study was to predict BMI using a multi-level linear model which incorporates individual, social, and community domains. A convenience sample of 195 people with mobility impairments completed telephone surveys on health promotion behaviors, the quality of their social relationships and environment, and demographic variables. Data on transportation, grocery stores, fitness centers, land use mix, street connectivity, and residential density was gathered to represent the community domain. Identifying as African American significantly predicted BMI (β=7.14, p<.01). The quality of a person's social relationships was a marginally significant predictor of BMI (β=-.748, p<.10). The time a person spent sleeping or lying down were significant predictors of BMI (β=0.539, p<.01). Distance to a fitness center was a significant predictor of BMI (β=-.00226, p<.05). Land use mix was a marginal predictor of BMI (β=-7.79, p<.10). Dietary behaviors were not significant predictors of BMI, but the presence of grocery stores was a marginal predictor of fat consumption (β=-13.01, p=0.10). These surprising results appear to indicate that environmental ecological domains mediate the interaction between health behaviors and BMI in people with mobility impairments. Future studies should use multi-level models to further examine the mediating effects of environmental data on health promotion behaviors and BMI.