Abstract:
|
Many studies have examined the link between neighborhood food store availability and obesity. Classic regression models that use food store counts within buffers of pre-specified size around participants' residences can yield biased or inefficient health effect estimates. We use longitudinal data from MESA obtained during 2000-2010 across six U.S. cities and distributed lag models (DLMs) to examine: (1) associations between favorable food store availability and body mass index (BMI) as a function of distance from participants' residences; (2) differences in these associations by participants' characteristics; and (3) cumulative associations of favorable food stores and BMI within 1 and 2 miles from participants' residences estimated from the DLM. We found that associations differ by city. Further, within New York, the association was modified by participant's age and distance: the association within 1 mile of participants' residence was not significant for younger participants, but was significant for older ones (-0.017; 95% CI -0.033,-0.002); within 2 miles, the associations were significant for both (younger: -0.009; 95% CI -0.016,-0.003; older: -0.018; 95% CI -0.028,-0.009).
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.