Background and Purpose
Research findings on the association between the built environment and energy balance (e.g., physical activity and body mass index (BMI)) are inconsistent. One reason for discordant findings across studies may be the inconsistency in how studies operationalize features of the built environment and neighborhood boundaries. Previous built environment research has relied mainly on pre-defined areas, or buffers, around a geocoded home address to characterize a participant’s exposure to the built environment. Buffer definitions are based primarily on the transportation literature, which describes the upper limit of the distance that individuals will walk. However, there is no uniform buffer type used across studies in public health, and there is little published data on the effect of buffer choice on study results.

Objectives
We aimed to examine how choice of buffer size and shape influence associations between different measures and features of the built environment, physical activity, and BMI.

Methods
We conducted cross-sectional analyses using linear regression to examine the relationship between various measures of the built environment and physical activity and BMI. Data for this analysis came from adult female participants in the Nurses’ Health Study II living in Texas in 2009 with a geocoded home address and data on BMI (n=6,255). Home addresses were based on questionnaire mailing addresses and were limited to those with a geocode match at the street level. Using these addresses, we used geographic information systems (GIS) to create line-based network and radial buffers of 400m, 800m, 1200m, and 1600m. Within these buffers, we ascertained business and intersection counts and calculated densities.  We then examined relationships between these measures and three energy balance-related variables: 1) Total physical activity in metabolic equivalent hours per week (MET hrs/wk), 2) Walking in MET hrs/wk, and 3) BMI. Energy balance outcomes were based on self-reports. Data on street networks came from the commercially available Topologically Integrated Geographic Encoding and Referencing (TIGER) 2010-based road network resource (Streetmap USA, ESRI) and business data came from the commercially available InfoUSA 2009 database. Data was analyzed using ArcGIS 10 and UNIX SAS (Cary, NC).

Results
Participants were all white and ranged in age from 44-64 years old. The mean BMI was 28.2 [standard deviation (SD) 6.8], the mean total physical activity was 21.0 MET hrs/wk (SD 30.6), and the mean MET hrs/wk from walking was 5.5 (SD 8.1). Findings differed greatly depending on the built environment measure. Buffer size, buffer shape, and the specific factor measured within the buffer affected the direction, magnitude, and statistical precision of estimates. Generally, line-based network buffers had stronger relationships with both outcomes compared with radial buffers. Relationships varied across buffer size for some measures, while for others, buffer size had no effect on the associations observed. The most consistent finding was that increased business count and density was associated with lower BMI, higher walking, and higher total physical activity, although the strength of association differed across buffer size and shape. Results varied for intersection count and density.

Conclusions
Results indicate that the scale and type of built environment measures can have a dramatic impact on study results and may partly explain inconsistent findings from past studies of the built environment and energy balance. These findings underscore the issue of “geographic context,” an emerging key concern for studies of associations between environment and behavior, and may help guide the design and interpretation of future studies utilizing built environment measures. Our results highlight the importance of this concern for active living research.

Support / Funding Source
This research was funded by the Transdisciplinary Research in Energetics and Cancer (TREC) Center funded by NIH (1U54CA155626-01).