Presentation on theme: "KEEP THIS TEXT BOX this slide includes some ESRI fonts. when you save this presentation, use File > Save As > Tools (upper right) > Save Options > Embed."— Presentation transcript:
KEEP THIS TEXT BOX this slide includes some ESRI fonts. when you save this presentation, use File > Save As > Tools (upper right) > Save Options > Embed TrueType Fonts (all characters) this will allow vector maps created with common ESRI symbols to show on computers that do not have ESRI software loaded a a a a a a a a a a a a a a a New Measures of the Food Environment in Seattle-King County NUTR 500, Phil Hurvitz, PhC Urban Form Lab College of Built Environments University of Washington KEEP THIS TEXT BOX this slide includes some ESRI fonts. when you save this presentation, use File > Save As > Tools (upper right) > Save Options > Embed TrueType Fonts (all characters) this will allow vector maps created with common ESRI symbols to show on computers that do not have ESRI software loaded a a a a a a a a a a a a a a a 1 of 44
Acknowledgements UWCOR/NIH-NIDDK Urban Form Lab – Eric Scharnhorst (food source classification) Group Health Cooperative – David Arterburn (health outcomes data) 2 of 44
Overview What does food environment mean? – Food environment at different scales Time Space Built Environment/Food Environment Food Environment in a Spatial Framework So what? Conclusions 3 of 44
What does food environment mean? 4 of 44
Food environments across temporal scales What is a healthful or harmful food? What is a healthful or harmful diet? 5 of 44
Environmental effects across spatial scales 6 of 44 McGarigal and Marks 1995
Food environments across spatial scales Cell Organ System Organism/Individual Community/Neighborhood City Region Country/Continent/Globe 7 of 44
Food environments across spatial scales Cell Organ System Organism/Individual Community/Neighborhood City Region Country/Continent/Globe 8 of 44 basic biological bench science built environment/ urban form EPI
Built Environment/Food Environment Food environment = places to procure & consume food – Stores – Restaurants – Emergency food system What does access mean? 9 of 44
Built Environment/Food Environment For food access, what conditions are necessary and what conditions sufficient'? Spatial proximity? – Necessary if transportation is limited Ease/convenience of getting to/from? – Necessary if mobility is limited Affordability – Necessary if income is limited 10 of 44
Food Environment in a Spatial Framework How is food environment measured? How can food environments be summarized over area or population of interest? How are food environments related to other spatially explicit factors? Tools: – Microsoft Access: data storage, database processing – ArcGIS 9.3: spatial analysis & mapping – R (with RODBC library): statistics & graphics 11 of 44
Measuring the Food Environment Food processing & selling establishments are regulated by public health agencies Main responsibility = protect against foodborne illnesses Requires up-to-date address & identification information 12 of 44
PHSKC Food License Data 13 of 44
PHSKC Food License Data 14 of 44 ???
Food source classification No standards exist We developed an ad hoc classification system 15 of 44
Food source classification An L4 class was assigned to each of the >10,000 food sources Hierarchical nesting of classes to top level 16 of 44
Address Geocoding 17 of 44
Address Geocoding Algorithm 18 of 44 ~20%
Address Geocoding 19 of 44
Extending the Map Having a map of the location of different food sources, while intrinsically interesting, is not an end goal in itself Food sources need to be conceptualized as part of an overall environment How are food locations related to other spatial factors? 20 of 44
Different ways to summarize Count within a spatial tolerance Proximity to closest Mean distance within a spatial tolerance (combines the 2 above approaches) Area-based summaries Kernel density estimators 22 of 44
Count within a spatial tolerance 23 of 44
Proximity to closest 24 of 44
Mean distance within a spatial tolerance 25 of 44 mean Euclidean distance to all: 717 m
Mean distance within a spatial tolerance 26 of 44 mean network distance to all: 911m
Area-based summaries: FFRD by tract 27 of 44
Area-based summaries: FFRD by ZIP code area 28 of 44
Area-based summaries: MAUP 29 of 44
Area-based summaries: MAUP 30 of 44 does this location have lower exposure
Area-based summaries: MAUP 31 of 44 does this location have lower exposure than this?
Area-based summaries: MAUP 32 of 44 why this pattern?
Kernel density estimators 33 of 44 cross-sectional view summation of XY Gaussians 3D & planimetric view
Kernel density estimators 34 of 44
So what? Relating measures to other things that matter – Mortality-based deprivation index – Median household income – Percent of residents living below poverty – Race/ethnicity – Health outcomes Obesity Diabetes 35 of 44
Multivariate census-based deprivation index Singh, G.K., Area deprivation and widening inequalities in US mortality, Am J Public Health, (7): p US Census Variables Higher SES Lower SES 36 of 44
Mortality rates higher among most deprived Singh, of 44
Relationship(s) between FFRD, SES, health? Based on existing data, do any relationships that appear? Caveats: – Area-based summaries (census tracts) – GHC data may not be generalizable to tracts 38 of 44
MALE there seems to be a relationship between deprivation & some health outcomes 39 of 44
40 of 44 MALE do you see a relationship between FFRD & other area- based variables?
41 of 44 FEMALE there seems to be a relationship between deprivation & some health outcomes
Median household income 42 of 44 FEMALE there seems to be a relationship between FFRD & some health outcomes – but in the direction we would expect?
Conclusion Health outcomes follow SES gradients Area-based fast food restaurant densities do not appear to be related to either SES gradients or health outcomes Area-based spatial measurement & summary methods are fundamentally problematic Other built environment factors frequently ignored (e.g., road density, land use mix, employment density, transit hubs) Individual eating behavior, SES, and built environment data will be necessary to investigate causal relationships 43 of 44
Phil Hurvitz gis.washington.edu/phurvitz 44 of 44 Higher SES Lower SES does this location have lower exposure than this?