A REVIEW OF PUBLIC HEALTH AIR SURVelU-AUCE EVALUATION PROJECT

V BOOTH€2, F DIMMICK1, V HHL6V3, C PRULU4, M B6KK6DRL5, D HOllfiND', T TfilSOT3, fi SMITH", M UU€RN€R5, £ BfilDRIDGS1,

, (1	F OF H EALTH. AUGUSTA

(5):WISC0NSIN'DEPARTMENT OF PUBLIC'HEALTH, MADISON. (6) APEX EPIDEMIOLOGY RESEARCH,.BALTIMORE

D MINTZ1, T FITZ-SIMONS1, T BRTGSON6, T UUflTKINS1

Background:

CDC, EPA, and the health departments of
New York, Maine, and Wisconsin, have
been collaborating on an evaluation of
different air characterization methods for
use in Environmental Public Health
Tracking (EPHT).

Approach:

The three public health departments
collected health tracking data associated
with cardiovascular and respiratory health
events. In doing so, the states developed
consistent case definitions and addressed
spatial qualities of the health data. EPA
provided air quality data based on four
different approaches of estimating
exposure. With these air quality data sets,
the three states applied a "case-crossover"
analysis technique to evaluate the associa-
tion between the health and air quality data.

Methods:

We examined the sensitivity of the associations between the health
outcome data and fine particles or ozone levels for four different air
datasets provided by EPA.

Ambient PM monitors- nearest monitor was
assigned to each case. In NYS, 95% of monitors take
measurements only every third day (only these monitors
were used).

CMAQ (Community Multiscale Air
Quality Model)- meteorology and emissions
inventories (point, area, biogenic, and mobile sources)
were used to estimate air quality levels and transport.

Interpolated- third day monitor data were first
interpolated in time using splines, and then in space
using ordinary kriging.

Combination monitor/CMAQ- Hierarchical
Bayesian method that provided more weight to monitor-
ing data in areas where monitoring exists, and greater
weight to the CMAQ model in areas far from monitors.

Results:

The statistical technique produced less error than the CMAQ predictions when compared to a set of independent air
quality monitor data. Figure 1 compares the fine particle CMAQ predictions and the ambient fine particle data in
metropolitan New York City. Figure 2 compared the statistical technique estimates and the ambient fine particle
data in metropolitan New York City. The qualitative contrast shows the effect of the statistical technique.

AQ

Monitoring
Data

Discussion and Conclusions:

These results are one piece of a larger evaluation of air characterization methods by the Public Health Air Surveillance
Evaluation (PHASE) team to select a method that could be routinely used in a sustainable EPHT Network. Note that
all of these methods provide only surrogate measures of true personal exposure.

The PHASE collaboration demonstrated the ability to link air quality and surveillance health data. This will enable
State public health departments to assess the impact of air pollution in rural and urban populations and to take appropriate
action at the community level. The techniques have been evaluated and implementation activities are underway.

Next steps include: a menu-driven software tool to link health and air quality data and readily available air quality data
available. This will enable health department professionals to link and analyze air quality and health data routinely.

Disclaimer:

The views of these authors may not necessarily reflect official New York , Maine Wisconcin, CDC, or EPA agency policy.

-------