Background:
Parks are an important attribute of the built environment and are known to be associated with psychological, social and physical health benefits, perhaps as a result of improved opportunities to engage in physical activity (PA). Parks may be more likely to influence PA if they are accessible and if attributes within the park support PA. However, the amount of time spent in parks, and the level of PA engaged in while in a park are difficult to measure. Also, inferences about the relationships between park attributes and PA often assume park use based upon a participant’s residential address, even though parks with different attributes may be visited outside of residential neighborhoods. Thus, the true nature of the relationships between parks and PA remains elusive. Advances in GPS and accelerometer data collection allow us to objectively study physical activity in specific locations.

Objectives:
The primary aim of this study was to better understand the association between time spent in parks and physical activity behavior in adolescents. The secondary aim was to accomplish this by merging accelerometer and GPS data collected from adolescents and preparing it for GIS analysis through the use of a novel system, PALMS, the Physical Activity Location Measurement System.

Methods:
The Teen Environment And Neighborhood Study (TEAN) was conducted in five counties in the Baltimore, MD and in King Country, WA. 400 participants between 12 - 16 years old were recruited from selected neighborhoods that varied by walkability and income. The current study examined a sub-sample of 80 participants from Baltimore, twenty randomly selected from each of four walkability/income quadrants.
Each participant was mailed a packet containing a GPS device (Globalstat: DG-100) and an accelerometer (Actigraph: GT1M) along with instructions on how to wear each device. After seven days, participants mailed the devices to the research center where data were downloaded, cleaned and merged in a multi-step process.

Time stamped data from the GPS and accelerometer was merged using PALMS algorithms that control for inaccuracies and gaps in GPS data collection. Data were cleaned for extraneous points and non-wear time, and age-adjusted cut offs were applied for the detection of moderate to vigorous physical activity (MVPA). PALMS then exported data into a geographic information system (ESRI INC: ArcGIS) containing shape files for all parks in the Baltimore study region. To examine the association between park use and overall physical activity a multiple regression model was built with mean minutes of MVPA per day as the dependent variable. Independent variables included a park use classification (no park use, park use but no MVPA, and park use with MVPA) neighborhood walkability (low vs. high), neighborhood income (low vs. high), age, gender, BMI and ethnicity.

Results:
Seventy-seven adolescents had complete data. The sample was 52% female, with a mean age of 14.6 (SD = 1.5), and 55.7% identified themselves as a non Hispanic-white race/ethnicity. The merging of accelerometer and GPS data files resulted in 949,848 data points representing 15,831 hours of data collection for the sample. Participants had average of 6 valid days (SD = 1.8) of accelerometer wear time and a mean of 67.3 minutes of MVPA per day (SD = 38.6). Approximately 30% of the participants did not visit a park during the data collection period, 41% visited a park but did not engage in MVPA, and the remaining 29% had at least one minute of MVPA within a park. The regression results indicated that participating in MVPA within a park (ß = 0.31), age (ß = -0.33) and gender (ß = 0.33) were significant predictors of mean minutes of MVPA per day. Participating in MVPA in a park was associated with a 26-minute increase in daily MVPA.

Conclusions:
This study served the dual purpose of examining the role of parks on overall physical activity for adolescents and providing a proof-of-concept for the use of PALMS to analyze multiple objectively measured data sources on PA and location. Participants who engaged in MVPA within a park had significantly greater minutes per day of MVPA when compared to participants who had no MVPA in a park and participants who did not spend time in a park. PALMS will next be used to analyze the entire study sample and, if appropriate, subsamples (e.g. by gender, BMI), as well as additional independent variables related to park use (e.g. distance).

Integrating GPS, accelerometer, and GIS data, though challenging from a methodological perspective, is essential if we are to understand the role that parks and park features play in promoting the health of youth. Park and recreation planners can use this knowledge to create parks that optimally support physical activity. Further, assessing behaviors in locations both inside and outside the residential neighborhood may strengthen the evidence for a relationship between the built environment and obesity prevention.

Support:
NIH/NCI Grant 1 U01 CA130771-01 Genes, Environment and Health Initiative (GEI)