|
An important aspect of the pilot project was to pilot test
the use of existing datasets for Tracking, as well as to
development and use various methods for analysis. See
below for descriptions of some of the data and methods used
in the pilot project.
Birth outcomes data source
Asthma data sources
Estimating pollution
levels
Linking
health records to pollution exposure
Map making protocols
Publications
Birth outcomes data source
All birth data are drawn from Vital Records, which is the
birth certificate information compiled by the State of
California. To protect confidentiality, the only
identifying information we used from these files was the
mothers' addresses, which were used to make the "smooth
surface" maps (see Map
making protocols). Birth certificates in
California record basic information such as birthweight and
estimated gestational age, along with demographic data such
as race and ethnicity. For this project, we used data
from 2001. (back to
the top)
Asthma data sources
One of the goals of the Alameda County Pilot Project was to
explore the potential of using health event information from
administrative and billing records for asthma
surveillance. We combined databases from Kaiser
Permanente of Northern California and the fee-for-service
portion of Medi-Cal, the state's Medicaid program.
Kaiser Permanente is the region's largest healthcare provider and represents a relatively complete
cross-section of the county population, while Medi-Cal would
provide representation for very low-income families who
experience a disproportionate burden of asthma.
The rationale of combining the two datasets was that we
hoped to get as accurate a picture of the asthma-related
health events as possible. One should note, however,
that this is not the same thing as collecting information
from a random sample of Alameda County residents, which is
an ideal way to do this. The combined dataset is useful for making
comparisons about rates of asthma-related events between
different parts of the county, but it is difficult to use
the information to compare these rates to those in other
counties, the state, or the country (see
Publications). The dataset includes information on
about 176,000 children under 17 in the county, which is
about half of the total that live there. For this
project, we used data from 2001.
(back to the top)
Estimating pollution
levels
To estimate levels of traffic pollution for Alameda
County in 2001, we used a method called land use regression. In summary, this involved placing
sampling tubes in various locations in the county during the spring
and fall to measure levels of pollutants in these locations. Then we compared the amount of pollution at
each location with the characteristics of the location (such as distance from roadways, how many cars drove
on nearby roads, etc). We used this information to create a mathematical model that could predict amounts of
NO2 for other locations in the county based on these characteristics.
(back to the top)
Linking
health records to pollution exposures
Ideally, the way to understand the health effects of air
pollution is to chemically analyze the air that people are
breathing while monitoring them for changes in their health
status, but this is very difficult and resource intensive on a large scale.
For this study, we only knew the residence address recorded
by Medi-Cal or Kaiser Permanente for asthma and maternal
address for birth outcomes. Not only were we unable to
directly monitor people's health, we were only able to make
assumptions about their pollution exposure based on
knowledge of where they lived.
As part of the Alameda County pilot project, we experimented
with ways to show pollution levels in different parts of the
county based on traffic patterns and variations in land use
categories. While different methods appear to have met
with different levels of success, one should keep in mind
that individual exposure levels are assigned based on the
addresses of residence we have recorded for people.
People are of course exposed to different air at work,
school, or en route between home and other places.
Therefore, no matter how good our exposure estimates are,
assigning exposure based on home addresses should still be
considered a very crude approach to this problem.
(back to the top)
Map making protocols
Making maps of health outcomes can be a very effective way
of communicating information. In choosing
how to make these maps, we wanted a way to make
high-resolution images, so that people could look at
variations smaller than cities or ZIP codes and think about
what was going on in the actual communities they knew.
At the same time, we wanted to preserve confidentiality so
that there was no way to look at the maps and learn the
health information about any one person.
The process we used is called density estimation mapping, which was made
possible by first converting all the addresses in the
dataset into X and Y coordinates similar to longitude and
latitude numbers. Once this was done, a computer could
be used to draw circles on a map of Alameda County, in this
case circles a single mile across. The computer could
then calculate the rate of any event (such as preterm births
or emergency room visits) for the people living within that
circle. By moving this circle in half-mile increments
across the county, we could calculate rates for overlapping
areas throughout the county for any place that had
sufficient population density to calculate a rate.
This process resulted in "grids" of health event rates
containing all the information we needed regarding
variations throughout the county, but protected confidential
information for any single person involved.
Conventional software packages were then used to create
"smoothed surface" images showing the variations in rates
across the entire space covered by the grids we had created.
See Publications for more
details.
(back to the top)
Publications
Progress in Pediatric Asthma Surveillance I: The Application
of Health Care Use Data in Alameda County, California
Roberts EM, English PB,
Van den Eeden SK, Ray GT. Progress in pediatric asthma
surveillance I: the application of health care use data in
Alameda County, California. Prev Chronic Dis [serial
online] 2006 Jul [date cited]. Available from: URL:
http://www.cdc.gov/pcd/issues/2006/jul/05_0186.htm.
Progress in Pediatric Asthma Surveillance
II: Geospatial Patterns of Asthma in
Alameda County, California
Roberts EM, English PB,
Wong M, Wolff C, Valdez S, Van den Eeden SK, et al.
Progress in pediatric asthma surveillance II: geospatial
patterns of asthma in Alameda County, California. Prev
Chronic Dis [serial online] 2006 Jul [date cited]. Available
from: URL: http://www.cdc.gov/pcd/issues/2006/jul/05_0187.htm.
(back to
the top)
Eric Roberts, MD PhD
erobert1@dhs.ca.gov
Return to
the pilot project findings page
Pilot Project
Findings | Overview of Pilot Project Data and Methods |
Glossary of Scientific and Statistical Terms |
Additional Resources
|
