Background
Hispanic-American preschool children have higher rates of obesity and Hispanic-American youth may be at higher risk for sedentary lifestyles and metabolic syndrome. Both physical and social environmental factors influence young children’s physical activity. The ability to simultaneously capture objective measurement of physical activity and location of young Hispanic-American children will help provide more reliable and valid findings to better inform environmental interventions and policy changes to promote physical activity among young Hispanic-Americans.

Objectives
This pilot study aimed to 1) assess the usefulness and feasibility of simultaneously objectively measuring the location and activity of Hispanic-American preschool children; 2) evaluate the correspondence between parental diary data of location and travel behavior and Global Positioning System (GPS) data on preschool children’s location and travel behavior (presence, duration, and mode of travel) processed using the Physical Activity Location Measurement System (PALMS); and 3) examine the associations of travel mode (no travel, walking, and motorized travel) and location (child’s home, others home, store, restaurant, church, community centre, park, and other locations) with objectively measured physical activity (defined as accelerometer counts per 30 seconds).

Methods
Fifteen 3-5 year-old Hispanic-American children residing in Houston, Texas were recruited. Children simultaneously wore QStarz BT100X GPS data loggers and Actigraph GT3X accelerometers for a 24-36 hour period on a day they were with a parent the whole day, including a 12 hour run in time. The parent completed a location and travel behavior diary for 12 hours (8 AM-8PM), at 5 minute intervals. Physical activity and GPS data were processed using PALMS. For each participant, the levels of observation-by-observation correspondence between GPS- and diary-assessed locations and travel behavior were estimated with Kappa statistic and percent agreement. Whole-sample point estimates and relative 95% confidence intervals of Kappa statistics and percent agreement were computed. Between-method differences in total minutes spent on specific travel modes and location, aggregated at the person level, were estimated using paired t-tests and Wilcoxon signed-ranks tests. Between-method associations in total minutes spent on specific travel modes and location, aggregated at the person level, were estimated using Kendall’s tau correlation coefficient. Associations of travel mode and activity location with objectively measured physical activity were estimated using multilevel models.

Results
The average age of the children was 4.6 (sd 0.9), 60% were girls. Of the participating Hispanic-American families, 53% reported family income less than $50,000/year; and 53% lived in single family homes, 34% in apartments, and 13% in trailer homes. Eighty percent of the children had valid physical activity data with mean valid activity data of 10.0 hours (sd 2.7). GPS- and diary-based estimates of minutes at each location were positively correlated (Kendall’s tau >0 .69) and mean values were not significantly different. The average correspondence between GPS- and diary-based estimates of types of location ranged from 91.9% to 99.9% agreement and Kappa statistics ranged from 0.45 to 0.99. No between-method differences in time spent in different modes of travel were observed. The between-method correlations of estimated participant-aggregated minutes spent on all trips and vehicle-based trips were moderate to strong. However, those related to walking trips and total trips were weak (<.38) and not statistically significant. The average observation-by-observation correspondence between GPS- and diary-based estimates of travel-mode ranged from 87.4% to 97.3% agreement and Kappa statistics ranged from 0.31 to 0.52. Children were significantly more active in parks and less active in restaurants compared with at home, when estimating locations by diaries as well as GPS monitors.

Conclusions
It is feasible to simultaneously measure and process GPS and accelerometer data on Hispanic-American preschool children. Good between-method correspondence was found for location and moderate correspondence was observed for mode of travel. Location was associated with different activity levels even in this small sample. The simultaneous use of GPS and accelerometers will be used by this team in larger studies to investigate where Hispanic-American preschool children are active.

Support/Funding
This project was funded by a R21 grant from NIH-NICHD (5R21HD060925).