Environmental Public Health Tracking in California (???????????)

1
Environmental Public Health Tracking in
California (???????????)

• Eric M. Roberts, MD PhD (??????)
• Paul English, PhD MPH, Principal Investigator
• Geoff Lomax, DrPH, Research Director
• Michelle Wong, MPH, Health Educator
• Craig Wolff, MS Eng, IT/GIS Manager

2
Background

• Chronic disease accounts for significant
  morbidity and mortality among Californians (along
  with injuries, responsible for 75 of deaths)
• Many chronic diseases are increasing in
  prevalence
• Asthma
• Auto-immune diseases
• Autism and learning disabilities (maybe)

3
Background

• Diseases with known or suspected environmental
  links

• Asthma
• Neurodevelomental disorders
• Autoimmue diseases
• Cancer

• Alzheimers, Parkinsons
• Endocrine disruption
• Endometriosis
• Heart disease

• Besides pain and suffering, treatment of
  environmental diseases costs at least US10
  billion annually in California alone

4
Background

• Problem
• We have limited data to track trends in health
  conditions which have suspected links to the
  environment
• We have limited ability to monitor human exposure
  to toxic chemicals, and we know little about what
  the public is exposed to and at what level

5
Environmental Public Health Tracking

• EPHT is the systematic, ongoing collection,
  collation, and analysis of data related to
• Environmentally related disease
• Environmental exposures
• EPHT also includes the timely dissemination of
  information to those who need to know so that
  action can be taken

6
Functions of an EPHT System

• Track environmental hazards, exposures, and
  diseases to help monitor hotspots where
  exposure to environmental hazards is excessive
  and requires reduction
• Track trends over time to help evaluate the
  success of environmental protection and public
  health measures
• Link environmental hazard information and disease
  information to help generate hypotheses about
  possible connections and
• Provide the foundation researchers need in order
  to do scientific studies designed to identify the
  causes of disease.

7
Planning for an EPHT system
Level of agency
Federal EPHT Program
CDC funded program in EPHT
Federal (??) State (???) County, city, public
(?,??,??)
Legislature mandated that DHS develop program in
EPHT
EHIB, DHS
Planning Consortium
Interested parties convened by EHIB
8
The Process of EPHT

• coordinate between agencies
• develop IT infrastructure
• format and process data

Disparate sources of data

• tabulation
• statistical analysis
• map making

Useable datasets

• stakeholder input
• develop and field test materials
• create mechanisms for access and
  dissemination

Results
Information for action
9
EPHT Process Discussion

• Assembling data from a variety of agencies
  through automated processes
• Collaboration with private-sector data sources
• Data visualization
• As an analytic tool
• For communication with community stakeholders

10
1. Data Assembly

• The keepers of information are spread out over
  many agencies, many of which may not be
  accustomed to working with each other

• Center for Vital Statistics (Birth certificate
  records)
• Department of Pesticide Regulation
• Childhood Lead Poisoning Prevention Branch

• Air Resources Board
• Department of Disability Services
• Public and private sector providers of health
  services

11
Inter-agency Data Communication Infratructure

• California Air Resources Board (ARB) collects
  emissions data from thousands of industrial
  facilities in the state
• Using geography, weather patterns, and dispersion
  modeling, ARB is attempting to model ground-level
  concentrations (GLCs) of air toxics

Emissions (inventoried)
Concentrations (modeled GLCs)
GLC grid layer
12
Geographic Information Systems (GIS) data queries
between agencies

• Input from DHS to ARB is circular buffer or
  polygon
• Output from ARB to DHS is proportional summation
  of metrics from all overlapping grids
• Resolution varies grid size (d) may be as small
  as 250 m

d
13
Remember the limitations!

• All modeling is based on inventories of air
  toxics released into the environment
• Large-scale industries must report releases to
  the government, but what if they are inaccurate?
• Small-scale industries are not required to report
  releases

No amount of technical modeling will allow us to
overcome inaccurate reporting
14
Remember the limitations!

• The assumptions going into any modeling procedure
  are very numerous
• The precision for any GLC estimate in any area is
  likely to be low
• Exposure data to use are low, medium, and
  high rather than actual concentrations of air
  toxics

Emissions (inventoried)
Concentrations (modeled GLCs)
GLC grid layer
15
2. Partnering with private-sector data sources

• Kaiser-Permanente is a very large private source
  of health services in Alameda County, California
• In 2001, they provided services to 577,687
  people, or 40 of the County population
• Sample of patients is broadly representative of
  County population

16
Income group representation of Kaiser vs. General
Population
17
Asthma-related health utilization

• 577,687 Kaiser Permanente patients in Alameda
  County in 2001
• 587 hospitalizations
• 2,694 emergency room visits
• 51,087 outpatient visits
• 218,205 prescriptions for asthma medications

18
EPHT Process Discussion

• Data visualization
• As an analytic tool
• For communication with community stakeholders

19
Geocoding health outcome data

• Geocoding assigns to every address record
• The postal code of residence
• The tract and block group used by the US Census
• x and y coordinates

y
x
20
Preterm singleton births by postal code Alameda
County, 2001
21
Problems with Postal Code maps

• Small communities with very high or low rates do
  not show up within postal codes
• Crossing the street from one postal code to
  another should not appear to take you from one
  level of risk to another
• Some large codes have very few people living in
  them
• Solution smoothed maps based on geocoded
  address data

22
Smoothed maps

• Ignore the postal code (or any other) boundaries
• Calculate small area rates at regular intervals.

23
Preterm singleton births Alameda County, 2001
24
Statistical significance on smoothed surface
maps

• Difficult to indicate which regions of map have
  statistically significant variations in preterm
  birth
• Overlapping and adjoining circular areas violates
  assumption of independence of rates (spatial
  autocorrelation)
• Must use Monte Carlo simulation approach to
  assess real distribution

25
Statistical significance on smoothed surface
maps

• Monte Carlo simulation
• Assuming uniform distribution of preterm births,
  generate 1,000 hypothetical preterm birth rate
  maps
• At each point can compare the measured rate to
  the distribution of hypothetical rates

26
Statistical significance on smoothed surface
maps

• Monte Carlo simulation
• For plt0.05, expect about 5 of measured rates to
  appear significant
• Therefore, this is a test to locate significant
  rates that you assume exist somewhere in the
  County
• For health conditions with well established
  disparities in the United States (preterm birth,
  asthma) this assumption (of the existence of
  these significant rates) is valid

27
Elevated preterm singleton birth rates Alameda
County, 2001
28
??!
Funding Centers for Disease Control and
Prevention, Environmental Public Health Tracking
Program

• Community Health Education
• Michelle Wong, MPH
• Mimi Johnson, MPH
• Eddie Oh, MPH
• University of California Center for Excellence
• Jonathan Balmes, MD
• Ira Tager, PhD
• Amy Kyle, PhD

• Principle Investigator
• Paul English, PhD MPH
• Research Director
• Geoff Lomax, DrPH
• IT/GIS Manager
• Craig Wolff, MS Eng
• Administration
• Mailie Newman