Improving Environmental Measures in Obesity Research using Innovative Technology

Obesity and physical inactivity are significant public health issues, particularly in low-income and minority populations. Recent research has highlighted the built environment as a potential causal factor for obesity and physical inactivity. A number of methods can be used to measure built environment characteristics including self-report measures, archival data, direct observation and, more recently, web-based audits (e.g., Google Street View); however, each of these methods has notable limitations. In the proposed study, we seek to test the use of an innovative technology for measuring the built environment, namely GigaPan. GigaPan is a robot system that automates obtaining numerous photos of an area using a basic camera housed within its apparatus. The resulting photos are then stitched together to form a single high resolution photo that is highly navigable. Using GigaPan, we will document and characterize features of the built environment on 683 street segments and in 287 target areas across 19 park/playgrounds. These measures will be compared to measures obtained via direct observation (collected as part of a separate grant) and web-based audits (i.e., Google Street View for street segments and Bing Maps for parks/playgrounds). We hypothesize that the GigaPan technology will embody most of the benefits of direct observation while significantly reducing the time and cost burden associated with training and monitoring field staff. In addition, we hypothesize that measures obtained via GigaPan will be more valid than measures obtained from Google Street View and better suited for studies of built environment change as GigaPan data are time sensitive. The study will capitalize on individual-level data from a population-based cohort of adults (N=1052) living in low-income minority communities that are experiencing significant built environment changes. Individual data include objectively measured height and weight, physical activity levels from both accelerometry and self-report, correlates of physical activity, perceptions of the environment, and demographic data. The specific aims of the grant are as follows: 1) To assess the reliability of coding environmental constructs from GigaPan photos across multiple raters; 2) To determine the validity of measures obtained via GigaPan by comparing them to the same measures obtained through direct observation (gold-standard) and web-based audits (less time- and cost-intensive); 3) To determine the predictive validity of each measure; Specifically, whether the association between built environment characteristics and accelerometer-based physical activity levels, self-reported walking levels, and objectively-measured obesity status depends on the built environment measurement method (i.e., direct observation, GigaPan, and web-based audits); and 4) To document the costs of obtaining built environment data via direct observation, GigaPan and web-based audits. We believe that this study will demonstrate GigaPan’s promise as a relevant tool for future multi-regional and longitudinal studies examining the effect of built environment changes on health behavior.