Physical inactivity, an unhealthy diet, and obesity are related to breast and colon cancer directly and through insulin and inflammation pathways. Greater access to healthy environments in residential neighborhoods is associated with higher physical activity (PA), a healthier diet, lower BMI, and in one study lower insulin resistance. Despite some significant findings, the effect sizes in built environment research have been small likely because of a mismatch between the environment assessed (home neighborhoods only) and location where the behaviors occur. Temporal variation in behaviors and locations within and across days has also been ignored. We propose to advance methods of cancer risk exposure assessment by measuring both neighborhood access and total environment exposure to healthy environments by dynamically integrating Global Positioning System (GPS) data with Geographical Information System (GIS) data. We hypothesize that dynamic GPS based measures of environmental exposure will be more strongly related to behavior and insulin and inflammation biomarkers than static addressed based GIS measures of access. We will study a large sample of adults (N=700), 40-75 years old, who have lived at their residence at least one year. We will recruit participants from census blocks specially selected to vary by income, walkability, and food environments to ensure environmental variability not achieved in a random sample. We will ensure balanced recruitment by census block type (walkability &food environments) across ethnicity, gender, age, and season. Half the sample will be Hispanic to explore potential interaction effects by ethnicity. Participants will complete surveys about their PA, sedentary behavior, environmental perceptions, self-selection, cancer risk, and demographics. PA and sedentary behavior will also be assessed by accelerometry and Machine Learning techniques will be employed to objectively identify specific behaviors likely related to the built environment e.g. walking, biking, riding in a car, screen time etc. Participants will complete the ASA 24 to assess diet, total calories and fat calories. A subsample (N=50) will wear a SenseCam to assess social context, validate GIS built environment measures, and validate the Machine Learned categories. Dynamic GIS measures of exposure will be created from 7 day person worn GPS data matched to GIS indicators of supportive PA and food environments (e.g. parks, walkable streets, fresh produce markets etc.) weighted by time, speed, transportation mode and features of the environment e.g. parcel size. Static residential GIS buffers of access to neighborhood resources will be created within a 1km street network buffer around a participant's home. Using multilevel statistical models adjusting for clustering, we will investigate whether GPS based Dynamic GIS measures of exposure to healthy food and PA supportive environments are more strongly associated with breast and colon cancer risk factors - including behaviors (PA, sedentary behavior, &diet), BMI, and biomarkers of insulin resistance and inflammation (e.g. CRP, IL-6, HOMA-IR) than Static GIS measures of access to neighborhood resources.
Where people spend their time during the day may be related to their risk of getting cancer. Previous studies have only looked at places near to home. This project will assess behaviors in different locations across the day and relate exposure to different environments to biological outcomes such as insulin resistance.
|Meseck, Kristin; Jankowska, Marta M; Schipperijn, Jasper et al. (2016) Is missing geographic positioning system data in accelerometry studies a problem, and is imputation the solution? Geospat Health 11:403|