Background:
In recent years there has been a growing body of evidence demonstrating associations between attributes of the built environment, such as land use mix, density, and street connectivity, and participation in physical activity (PA). Despite improved methods of measuring exposures to the built environment via geographic information systems (GIS) technology and objective assessment of PA with accelerometers, the evidence base is still emerging. In recent correlational studies using objective measures, there appears to be an implicit assumption that all PA comprising the outcome has taken place within an area around a geocoded residential address. With methods used to date, the amount of PA occurring within that buffer is in fact indeterminate. To address this limitation, we monitored a small sample of adults with both accelerometers and portable GPS units.
Objectives:
The objectives of this exploratory study were: 1) to quantify PA of various intensities in specific areas around residential and work locations; and 2) to examine associations between objective built environment variables and location-based PA.
Methods:
Study participants (≥18 years) were recruited from five community trails in Massachusetts during 2004-2005. Participants were instructed to wear an Actigraph accelerometer during waking hours and a lightweight portable GPS unit anytime they were outdoors for four consecutive days (2 weekdays, 2 weekend days). The GeoStats Wearable GeoLoggerTM includes a passive GPS data logger that records second-by-second position and speed. Participants wore the accelerometer on their right hip and the GPS unit in a small backpack. GPS and accelerometer data files were merged using their respective time stamps and converted into a minute-by-minute dataset. Using GPS longitude and latitude fields, GIS procedures were used to create dichotomous (yes/no) location variables that characterized where a participant was during each minute of monitoring; specifically within a 50 meter buffer around home and work and within a 1 km network buffer around home and work. Two processing steps were used to create location-based PA variables. First, established cut-points were used to classify each minute of accelerometer monitoring (e.g., inactive, light, moderate, vigorous). Second, location-based PA variables were created for the 50 m and 1 km buffers using the GPS-derived location variables and activity intensity variables for each minute. We excluded days with less than 600 minutes of accelerometer wear time. GIS was also used to create five measures of the built and natural environment previously shown to be related to PA in adults; intersection density (i.e., street connectivity), land use mix, residential population density, housing unit density, and a vegetation index. Multiple linear regression was used to examine associations between built environment variables and moderate PA outcomes (1952-5724 cts/min), controlling for age, gender and race. We first examined associations with total moderate PA, irrespective of location. In subsequent models, we examined associations between built environment variables and location-based PA outcomes for the 1 km home and work buffers.
Results:
Data were analyzed on 160 adults who had at least one day of valid accelerometer data (out of 178 participants). Their average age was 43.5 ± 12.8 years. Twenty-eight percent were non-white and 81% had a college degree or higher. Eighty-eight percent of participants (n = 141) had 3-4 days of valid accelerometer data; the average monitoring wear time was 14.4 ± 1.6 hrs/d. About 88-89% of the total activity within the 50 m home buffer and 1 km home network buffer was classified as inactive or light intensity and the remaining 11-12% was moderate-to-vigorous intensity. Total moderate PA activity, irrespective of location, averaged 50.9 ± 27.9 min/d, whereas moderate activity within the 1 km home buffer averaged 13.8 ± 15.5 min/d. We found statistically significant positive associations between intersection density, land use mix, population density, and housing unit density and moderate PA within the 1 km home network buffer. In addition, we found a statistically significant negative association between a vegetation index and location-based moderate PA. The variance in PA explained by these models ranged from 8% to 14%. We found no statistically significant relationships between these variables and total moderate PA. Alternatively, intersection density, land use mix and vegetation index within a 1 km work buffer were significantly associated with total moderate PA. Only land use mix within the work buffer showed a significant association with location-based moderate PA.
Conclusions:
Connectivity, land use mix, and density within a 1 km residential buffer were positively related to moderate PA, though these associations were only found for a location-based PA outcome. In contrast several of the same built environment variables within a work buffer were related to overall moderate PA - but not to location-based moderate PA. Simultaneously monitoring individuals with GPS units and accelerometers may provide new insights into the effects of the built environment on PA.
Support:
This research was supported by the Active Living Research Program, the Robert Wood Johnson Foundation, and the College of Liberal Arts, Purdue University.
