Neighborhood Greenness, Residential Density and Two-Year Changes in Children

1
Neighborhood Greenness, Residential Density
and Two-Year Changes in Childrens BMI

• Janice F. Bell, PhD MPH MN
• University of Washington
• Department of Health Services
• Child Health Institute

2
Background

• Childhood obesity at least doubled since 1980
• In 2003 2004
• 17 obese (gt95th percentile of national growth
  standards)
• 34 overweight (gt 85th lt95th percentile)
• Increased risk for risk for health problems,
  later obesity, high costs

3
Built Environment and Healthy Weight

• Adult studies
• Supermarkets, land use mix, residential density,
  intersections
• Child studies
• Proximity to parks, time spent outdoors

4
The Role of Urban Vegetation

• Improves aesthetics, pollution, regulates health
  exchange
• Could influence childrens BMI by promoting
  physical activity
• Cross-sectional evidence of associations between
  greenness and BMI in areas with high residential
  density

5
Purpose

• To test whether
• Neighborhood greenness and residential density
  have independent effects on 2-year changes in
  childrens BMI
• Residential density mediates or moderates
  greenness-BMI relations

6
Data Source and Sample

• Geo-coded electronic medical records
• Ages 3 14 (n3,625)
• Same-day clinical measures of height and weight
  two years apart
• Index year 1996 2000
• Same address over 24 months

7
Variables

• Dependent
• 2-year change in BMI z-Score (t2-t1)
• Covariates
• Age, race/ethnicity, sex, baseline z-BMI, health
  insurance status, census tract median family
  income (log), index year

8
Independent Variables

• Residential density residential housing units/
  acres residential land use
• Greenness (2-km buffers)
• Radial
• Road network-based

9
Normalized Difference Vegetation Index (NDVI)

• Pixel values of remote satellite images converted
  to NDVI
• Ranges from -1 (water) to 1 (dense vegetation)

10
Methods

• GeoDa (spatial dependence diagnostics)
• OLS regression of z-BMI change
• SE adjusted for clustering
• A priori criteria for mediation
• Chow tests

11
Sample Characteristics (n3,625)
12
Linear Regression Models of 2-year change in
z-BMI
13
Other Findings

• Residential density did not mediate or moderate
  NDVI-BMI relations
• Results robust by buffer shape
• Possible effect modification by age potential
  role for residential density among children 6 -10
  years

14
Conclusions

• Higher greenness associated with declines in
  childrens z-BMI over two years
• Residential density appears to play a less
  important role in childrens overweight than
  among adults
• Similar findings with network circular buffers
  (walkability mechanism may not apply to
  children)

15
Strengths Limitations

• Large, at-risk sample with clinical measures of
  BMI over time
• Limited generalizability
• Possible confounding by SES
• Selection bias
• No standards for urban form specification

16
Potential Pathways and Mechanisms

• Places that promote outside play and physical
  activity
• Territorial personalization
• Improved mental health, self-esteem, reduced
  stress

17
Implications

• Green space may be more important to child BMI
  than residential density
• Modifiable aspect of the built environment
• Viable target for childhood obesity prevention

18
Thank You

• Agency for Healthcare Research and Quality (grant
  T32-HS013853-01)
• Gilbert C. Liu, Pediatrics, Indiana University
• Jeffery S. Wilson, Geography, Indiana University