Geographic analysis of immunization patterns in Genesee County

1
Geographic analysis of immunization patterns in
Genesee County

• Kyle S. Enger, MPH
• Vaccine-Preventable Disease Epidemiologist
• Division of Immunization
• Michigan Dept. of Community Health (MDCH)

2
Data source MCIR(Michigan Care Improvement
Registry)

• Operational since 1998
• Birth-to-death immunization registry
• lt 20 y age restriction removed June 2006
• 95 of providers participate
• Part of CDCs IIS sentinel site project
• Michigans population is large
• 10 million people
• 130,000 births/year

3
Methods Geocoding

• MCIR receives responsible party addresses from
  birth certificate or clinicians
• Address is converted to geographic coordinates by
  the MI Dept. of Human Services (DHS) using
  Postalsoft
• 85 of addresses yield good coordinates
• Very easy to view using GIS (ArcView)

4
Methods Density mapping

• Converts point data to a continuous surface
• Similar to a drip castle each point gets a
  drip, and drips can stack on each other
• High areas are bright low areas are dark
• Brightnesskids/mile2
• ArcGIS Spatial Analyst

5
Methods Cluster analysis

• Determines where a factor is more common than
  population would suggest
• Generates circular clusters provides relative
  risk P-value for each cluster
• SaTScan (free software)

6
Study population

• Children with a Genesee County address
  geographic coordinates
• aged 19-35 months on 28 Feb. 2007
• on record in MCIR
• Density maps and cluster analysis
• For various childhood vaccines
• DTaP, MMR, polio, Hib, hep. B, varicella, PCV7
• Completely unimmunized
• Partially immunized

7
Population density
8
Density of kids with 0 DTaP - Genesee
9
Density of kids with 0 DTaP - Flint
10
Density of DTaP needed - Genesee
11
Density of DTaP needed - Flint
12
DTaP clustering
For illustration only. Cluster locations sizes
have been randomly changed.
13
PCV7 clustering
For illustration only. Cluster locations sizes
have been randomly changed.
14
Questions?
Acknowledgements
Mark Valacak, Sherry Wood, Fatema Mamou,
Genesee County Health Dept. Kevin Czubachowski,
Mich. Dept. Comm. Health

• Kyle Enger
• engerk_at_michigan.gov

15
SaTScan settings

• Max cluster size 50 of population
• Clusters of high RR only
• Bernoulli model (1/0)
• (immunized/unimmunized)
• No pairs of centers both in each others clusters

16
Density map settings

• Density type kernel
• Cell size 50m
• Kernel radius 1 km
• Output area units kids/mile2 or iz./mile2
• Color steps Factor of 1.67