Asthma Distribution patterns and their relationship with the urban landscape and social conditions in Newark NJ

1
Asthma Distribution patterns and their
relationship with the urban landscape and social
conditions in Newark NJ

• Authors
• Francisco Artigas, Leonard Beilory, Richard
  Holowczak, Kumar Patel
• Primary author affiliation
• CIMIC - Rutgers University, NJ
• artigas_at_cimic.rutgers.edu
• IHGC 2000Sunday March 19, 2000

2
Problem Statement

• Recent estimates suggest that roughly 50 of
  school children in the City of Newark suffer from
  some form of asthma. Similar urban areas across
  the country exhibit much lower rates.
• Hospital admissions
• 110 per 100,000 in Newark
• 46 per 100,000 in surrounding Suburban/rural

3
Research Objectives

• Build a robust spatial data-set about asthma
  cases in Newark (focus area).
• Find spatial correlation between asthma case
  locations and urban landscape features

4
Data Sources

• Admission records from UMDNJ University Hospital
  1997-1998 (n 542 and n 624)
• Landsat 5 thermal images (1997)
• High resolution aerial photographs (1995)
• Geo-coded street address vector coverage of
  Newark
• Census tracts from 1990

5
Overview of Data

• 1997 1998
• Male 280 305
• Female 261 329
• Black 485 561
• White 2 3
• Filipino 0 1
• Other 4 35
• Unknown 49 34

• Age 1997
• Min /Max 0 / 82
• Average 17
• Median 8
• Len.of Stay 1997 1998
• Min 1 1
• Max 24 87
• Average 3 3.1
• Median 2 2

6
Analytical Tools

• ARC/INFO and ARCVIEW
• MapObjects
• IDRISI Image processing software
• SPSS statistics software
• Wizsoft data mining software

7
Research Approach

• Clean and organize asthma case data from UMDNJ
  University Hospital
• Generate X and Y coordinates from address lists
  for 1997 and 1998
• Perform cluster analysis
• Intersect asthma cases with census data
• Spatial analysis of asthma cases with Landscape
  texture and features

8
Assumption

• Asthma cases are uniformly distributed across
• Streets
• Landscape texture
• Socio-economics indicators
• Asthma cases are uniformly distributed from
• Emission focal points

9
Cluster Analysis

• Method Use K-means cluster analysis (K5, K10
  and K15) on X, Y coordinates
• Characteristics of clusters
• Size (membership) of clusters
• Location of cluster centers
• Cluster migration from year to year
• Cluster homogeneity

10
1997 data K5
Clusters tend to align with Newark Ward
boundaries Black - cluster center H UMDNJ
Hospital
H
11
1997 data K15
Clusters tend to align with neighborhood
boundaries H UMDNJ Hospital
H
12
1998 data K5
Clusters tend to align with Newark Ward
boundaries Black - cluster center H UMDNJ
Hospital
H
13
1998 data K15
Clusters tend to align with neighborhood
boundaries H UMDNJ Hospital
H
14
Cluster Migration 97 to 98
Rt. 280
Blue 1997 Cluster centers Red 1998 Cluster
centers Central ward clusters tend to migrate
less H UMDNJ Hospital
H
15
Cluster Homogeneity

• Compare Race in population with Race of cases

16
Cluster Homogeneity
17
Cluster Homogeneity

• We expected the number of Asthma Cases to be
  proportional to the Racial makeup of the clusters
• However, our data suggests that asthma cases
  among Blacks are disproportionately higher
  compared to the racial makeup of the clusters

18
Spatial analysis

• Intersection of tract census data with asthma
  cases
• Observation of asthma cases and urban landscape
  texture
• Asthma cases at the street level
• Spatial relationship between diesel fume sources
  and asthma cases
• Spatial correlation between urban heat islands
  (UHI) and asthma cases.

19
Intersection of census tract information and
asthma cases
20
1997 Cases in terms of Public Assistance
Less than half on PAV
Half on PA
More than half on PA
Cluster Centers
21
1998 Cases in terms of Public Assistance
Less than half on PA
Half on PA
More than half on PA
Cluster Centers
22
Landscape Texture

• Expect to see more cases in high-density housing
  areas than in low-density housing areas

23
1997 cases in terms of population density
Low pop. density
Medium pop. density
High pop. density
Cluster Centers
24
1998 cases in terms of population density
Low pop. density
Medium pop. density
High pop. density
Cluster Centers
25
Urban Landscape Texture
High density housing
South Orange Ave.
Low density housing
26
Urban Landscape Texture
Low density housing
High density housing
27
Housing Density Effect

• Cluster centers which had the greatest
  recruitment of cases occurred in low density
  neighborhoods in central ward (many vacant lots)

28
Sick Streets

• All streets should exhibit a proportional number
  of cases

29
Sick Streets
S. 11th St.
1997 Yellow 1998 Green H UMDNJ Hospital S.
Orange Ave.
Fairmount Cemetery
H
30
Sick Streets
Manufacturing Facility
1997 Yellow 1998 Green S. Orange Ave.
31
Sick Streets

• An unusually high number of asthma cases
  congregate along specific streets
• We need to further investigate the impact of
  nearby manufacturing facilities and TRI sites

32
Spatial Relationship between Diesel Fumes and
Asthma

• Extracted addresses from digital yellow pages of
  trucking facilities in Newark where trucks are
  likely to congregate
• X and Y coordinates were extracted for each
  facility
• Trucking facility locations were mapped together
  with asthma case locations

33
Diesel fume sourcesAsthma case
34
Newark Urban Heat Islands

• Landsat 5 Thermal
• Ground level ozone is a photo chemical reaction
• Greater ozone levels are expected in hotter areas
  of the city

35
Newark Urban Heat Islands

• 1997 asthma cases correlated against urban heat
  islands

36
Newark Urban Heat Islands

• 1998 asthma cases correlated against urban heat
  islands

37
Newark Urban Heat Islands
38
Conclusions

• Asthma cases tend to congregate in the central
  ward
• Great majority of cases are
• African American
• Less than 10 years old
• Under public assistance
• From low density neighborhoods (Social
  dislocation effect, Wallace et al)

39
Conclusions (continued)

• Cluster centers tend to persist in the central
  ward and along heavy traffic corridors.
• Some streets in mixed industrial/residential
  neighborhoods have an unusually high number of
  asthma cases
• According to our data (limited number of years
  and only 1 hospital) we found no significant
  correlation between diesel fume sources or urban
  heat islands and asthma

40
Conclusions (continued)

• The evidence suggests that the external
  environmental conditions we studied are not
  strong indicators of asthma

41
Future Work

• Continue to build asthma database for different
  years and from different hospitals in Newark
• Incorporate daily and seasonal air quality
  measurements from monitoring stations to the data
  set
• Map TRI sites in Newark
• Employ more robust statistical tools
• Investigate temporal relationships (seasons vs.
  admissions)

42
End
43
High-D vs. Low-D housing
44
Socio-economics

• Expect asthma cases uniformly distributed across
  all socio-economic indicators
• Race
• Income Public assistance
• Home ownership/Rentals
• Population density
• People per census tract
• People per household

45
Image 11
46
Image 12
47
Image 13
48
Image 15
49
Image 16
50
Windrose