SUMMARY
There are multiple links between residential location, built environment (BE), and neighborhood prevalence of
obesity and type 2 diabetes (T2D). Advances in GIS technologies have allowed a shift in geographic scale
from large counties and ZIP codes to more granular census tracts/blocks and individual tax parcels. Fine-
grained data permits more detailed characterization of BE with variables like bus ridership, sidewalk length,
and greenery coverage. Cross-sectional analysis of BE data does not allow causal inferences. Only
longitudinal cohort studies can address the critical question: does residential BE have a causal effect on body
weight changes and glycemic control in the long term? Identifying BE variables most strongly related to health
outcomes would help inform health policies, encourage health-supportive urban planning, and support
consumer residential location decisions. We propose a paradigm shift, modeled after the randomized social
experiment, the Moving to Opportunity Study. We will use data from Group Health (GH), a large integrated WA
State health care system, serving broad economic strata. By attaching a geographic context to anonymized
electronic medical records (EMR) of 320,000 adults (30,000 with T2D) and 90,000 children residing in King
County, WA, we will examine longitudinal relationships between individual-level BE and changes in weight and
glycemic control. More than 100,000 adults and 40,000 children in the cohort moved at least once between
2005 and 2016, allowing a unique opportunity to examine the impact of major changes in BE on long-term
weight and T2D status. Aim 1 will determine whether baseline BE variables can predict longitudinal changes in
body weight and HbA1c over up to 12 years follow-up, independent of baseline measures of socioeconomic
status (SES), demographic variables, and comorbid conditions. Aim 2 will determine whether immediate,
moving-induced changes in BE can predict body weight and HbA1c changes, independent of changes in SES
and other covariates. Aim 3 will determine whether more gradual, secular changes in BE can predict weight
and HbA1c changes, in people who do not move residence, independent of changes in SES and other
covariates. We will develop new spatial analysis methods to address the complexity of intertwining longitudinal
and time-to-event outcomes and spatially-dependent time-varying predictors. The project will afford novel “big
data” linkages between 40+ residential BE characteristics and detailed, longitudinal clinical data, allowing us to
examine body weight trajectories, glycemic control, and incidence of obesity and T2D over 12-years' follow-up.
This natural experiment examining sudden (for movers) and gradual (for non-movers) changes in the BE for a
very large cohort will provide unprecedented insights into the impact of place on health. Our findings will help
urban planners and policymakers target different BE features for intervention, based on local and regional
realities. Consumers and developers can use this information to make informed decisions about neighborhood
features likely to be most supportive of health.
