Background Few studies examine relationships between built environment (BE) and type 2 diabetes mellitus (T2DM) using spatial models, investigate BE domains apart from food environment or physical activity resources or conduct sensitivity analysis of methodological choices made in measuring BE. We examine geographic heterogeneity of T2DM, describe how heterogeneity in T2DM relates to BE and estimate associations of T2DM with BE.
Methods Individual-level electronic health records (n=41 203) from the Duke Medicine Enterprise Data Warehouse (2007–2011) were linked to BE based on census block. Data on housing damage, property disorder, territoriality, vacancy and public nuisances were used to estimate BE based on four different construction methods (CMs). We used race-stratified aspatial and spatial Bayesian models to assess geographic heterogeneity in T2DM and associations of T2DM with BE.
Results Among whites, a 1 SD increase in poor quality BE was associated with a 1.03 (95% credible interval 1.01 to 1.06) and 1.06 (95 % credible interval 1.02 to 1.11) increased risk of T2DM for poor quality BE CM1 and CM2, respectively. Among blacks/African Americans, associations between T2DM and BE overlapped with the null for all CMs. The addition of BE to white models reduced residual geographic heterogeneity in T2DM by 4%–15%, depending on CM. In black/African–American models, BE did not affect residual heterogeneity.
Conclusion Associations of T2DM with BE were sensitive to CM and geographic heterogeneity in T2DM differed by race/ethnicity. Findings underscore the need to consider multiple methods of estimating BE and consider differences in relationships by race/ethnicity.
- spatial analysis
- public health
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Contributors Each co-author made substantive contributions to this manuscript. MAB designed and conducted the analysis and drafted the manuscript. RA assisted with the analysis and drafting of the manuscript. MLM provided feedback on all aspects of the project, including edits to multiple manuscript drafts.
Funding This work was supported by the Bristol-Myers Squibb Foundation “Together on Diabetes” Program and by Cooperative Agreement Number 1C1CMS331018-01-00 from the Department of Health and Human Services, Centers for Medicare & Medicaid Services.
Competing interests None declared.
Ethics approval Duke University Institutional Review Board and Rice University Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.
Patient consent for publication Not required.