Spatial Clustering of Objectively Measured Physical Activity in Massachusetts Adults: Preliminary Findings

Background: Accelerometers and global positioning system (GPS) units to monitor participants’ activity allow for a dynamic spatial examination of the locations where physical activity occurs. The application of spatial clustering analysis to these geographically linked physical activity data may provide a better understanding of how moderate-to-vigorous physical activity (MVPA) concentrates in certain locations among individuals and eventually may provide insights into important environmental determinants.

Purpose: To detect and describe patterns of spatial clusters of objectively measured MVPA in a sample of Massachusetts adults.

Methods: This study involved spatial analysis of minute-by-minute physical activity data obtained via accelerometry and linked to geographic coordinates via GPS monitoring. The data were collected from 144 Massachusetts adults (age = 44±13 yrs) recruited at five trails in Massachusetts. Participants wore an ActiGraph accelerometer and small GPS unit for four days in 2004-2005. Minutes with an activity count ≥ 760 were classified as MVPA. A spatial scan statistic was used to detect statistically significant (p <.05) spatial clusters of high rates of MVPA. A relative risk (RR) is generated for each cluster, which indicates that minutes inside the cluster have high rates of MVPA relative to minutes outside the cluster.

Results: Out of a total of 61,600 monitoring minutes with geographic coordinates, 24,866 were classifi ed as MVPA (40%) and 36,734 were inactive or light intensity. Seventy-fi ve spatial clusters of MVPA were identifi ed with RRs ranging from 1.34 to 2.48. Fifty-six percent of MVPA minutes were inside clusters (n = 13,982). The number of participants in clusters ranged from 1-37. The mean distances from participants’ homes to the cluster centroids ranged from 0.33 to 50.3 km; and the cluster radii were all less than 5 km. Twenty-eight clusters (37%) were located inside the city of Boston, 25 (33%) intersected the five study trails, and 30 (40%) occurred near participants’ homes.

Conclusions: Spatial cluster methods detected areas with a greater concentration of objectively measured MVPA. As expected, some clusters were located on study trails. However, a substantial number of clusters were found off trails and near participants’ homes