1
Traffic- Related Air Pollution:A Critical Review of the Literature on Emissions,
Exposure, and Health Effects
Clean Air Act Advisory CommitteeCrystal City, VA
October 6 – 7, 2009
Dan Greenbaum, PresidentHealth Effects Institute
Goals of the Review
Summarize and synthesize relevant information on air pollution from traffic and its health effects, linking:• Emissions and exposure to traffic air pollution• Exposure to traffic air pollution and health effects• Toxicological data and epidemiologic associations
A preprint of the report was released in May 2009The final Report, following extensive QA/QC, will be published in fall 2009
2
HEI Traffic Review PanelIra Tager—UC Berkeley, ChairKenneth Demerjian—SUNY, AlbanyMark Frampton—U Rochester Michael Jerrett—UC Berkeley Frank Kelly—King’s College Lester Kobzik—Harvard SPHNino Künzli—IMdM, BarcelonaBrian Leaderer—Yale SPHThomas Lumley—U of Washington SPH Frederick Lurmann—Sonoma Tech. Inc Sylvia Richardson—Imperial CollegeJon Samet—Johns HopkinsMichael Walsh—Consultant
Emissions from Motor VehiclesThe Current Context
Significant progress has been made in reduction of pollutant emissions from motor vehicles despite increases in number of vehicles and vehicle miles traveled
Increased urbanization and urban populations have:
• Increased dependence on motor vehicles and traffic congestion
• Changed land use patterns such that more people are near traffic sources of pollution
3
Report Organization
There are many studies (over 400) that have attempted to look at traffic exposure and effects
•However, they are not all of equal quality
4
1. How should we assess Exposure?
• Who is likely to be exposed?• What exposure assessment methods used in
epidemiologic studies?• Pollutant surrogates for traffic exposures (e.g., NO2, EC/BS,
CO, UFPM, benzene, etc.)• Distance from and/or length of roadways• Estimate of traffic density or intensity• Modeling of primary traffic-generated pollutant exposure
VOC (TraceAir) Distance Decay Around Highway 401, Toronto
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
-1000 -800 -600 -400 -200 0 200 400 600
Distance from Expressway (m)
Stan
dard
ized
VO
CLe
vels
Benzene (TraceAir)
THC (TraceAir)
NO2 (Ogawa)
Who is Likely to be Exposed?Highest levels within 300 – 500 meters of a major road
Toronto, Beckerman et al. (2008)
5
Near Roadway Exposure Can Include Large Populations Toronto Example: ~45%
(within 500 meters of an expressway; 100 meters of a major road)
Los Angeles Example: (~44%)
6
Portland, Oregon ~37%
What Markers or Surrogates?
• Pollutant surrogates for traffic exposures (e.g., NO2, EC/BS, CO, UFPM, benzene, etc.)
• Criteria for what is a good surrogate:1. Traffic as the major source2. Emissions vary with other motor vehicle constituents3. Can be measured at low concentrations by reasonably
inexpensive and accurate methods4. Not have independent health effects
7
NO2 as a surrogate
•There is substantial variability in average concentrations by locations.
•NO2 is a potential surrogate for vehicle emissions if it is measured on a fine spatial resolution.
PM2.5 as a Surrogate•Use of PM2.5 as a surrogate is of limited value because many sources contribute to urban PM2.5 and PM2.5concentrations are well mixed within a region
•Current central monitors do not provide sufficient spatial resolution for assessing the contribution of traffic to ambient PM2.5
8
What Markers or Surrogates?
• Pollutant surrogates for traffic exposures (e.g., NO2, EC/BS, CO, UFPM, benzene, etc.)
• Criteria for what is a good surrogate:1. Traffic as the major source2. Emissions vary with other motor vehicle constituents3. Can be measured at low concentrations by reasonably
inexpensive and accurate methods4. Not have independent health effects
• Can provide useful information but none meet all these criteria…
Can We Use Exposure Models?Models used• Proximity models• Geostatistical
interpolation models• Dispersion models• Land-use regression
models• Hybrid models
• Combine a model with time-activity data, or personal/microenvironmental monitoring
• Proximity models are least effective:• Can be confounded by
Socioeconomic Status, Noise, other factors
• Newer models of exposure are better• But should be validated
against some real-world data.
9
Criteria for Inclusion of Toxicology and Epidemiology Studies
• Quality of exposure assessment was key…• Studies had to include 1 or more of the following
exposure methods:• Distance from and/or length of roadways• Estimate of traffic density or intensity• Modeling of primary traffic-generated pollutant exposure• Studies of occupations characterized by exposure to traffic• Pollutant surrogates for traffic exposures (e.g., NO2, EC/BS,
CO, benzene, etc.) only if data provided to validate the pollutant as a reasonably specific surrogate for such exposure
2. What Can We Learn from Toxicology?(Example from a somewhat limited database):
Effects of Traffic Exposure on Asthmatics (Zhang HEI 2009)Lung function decline in asthmatics comparing Hyde Park and Oxford Street, London
(although symptoms did not increase…)
10
3. What can we learn from epidemiology?Criteria for Causal Inference
Four categories to test whether traffic causes effects, based on:- how well studies controlled for confounding- consistency of the findings with other studies- quality of the method to estimate exposure
• Sufficient evidence• Suggestive but not sufficient• Inadequate and insufficient evidence• Suggestive of no association
Epidemiology Health Outcomes Evaluated
• Mortality (all cause, cardiopulmonary)• Cardiovascular morbidity• Respiratory disease
• Asthma—childhood/adult• General respiratory symptoms• Lung function-childhood/adult/COPD• Health care utilization
• Non-asthmatic allergy• Birth Outcomes• Cancer
11
1.13 1.161.49
0.87 0.94 1.09 1.09
1.72 1.74
1.13
1.6
2.94
1.58
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
per 1
0.3
ug/m
3 in
NO
2 - 2
nd y
ear o
f life
per 1
0.6
ug/m
3 in
NO
2 - 4
th y
ear o
f life
per 1
0.6
ug/m
3 in
NO
2 - 4
th y
ear o
f life
(Alle
rgic
Chi
ldre
n)
per 8
.5 u
g/m
3 N
O2
- 1st
yea
r of l
ife
per 8
.5 u
g/m
3 N
O2
- 2nd
yea
r of l
ife
per I
QR
NO
2 - 1
st y
ear o
f life
per I
QR
NO
2 - 2
nd y
ear o
f life
Gau
derm
an 2
005:
Per
5.7
ppb
NO
2
Jans
sen
2003
: per
18
ug/m
3 N
O2
Hirs
ch 1
999:
per
10u
g/m
3 N
O2
Nor
dlin
g 20
08: p
er 4
4ug/
m3
NO
x
Van
Roo
sebr
oeck
200
8: p
er 1
7ug/
m3
Nic
olai
200
3: >
=57.
4 vs
. < 5
7.4
ug/m
3
Brauer2002
Brauer 2007 Gehring 2002 Morgenstern 2006 Other Authors
OD
DS
RA
TIO
(95%
CI)
Exacerbation of Asthma Symptoms
Increase in Wheeze Per Increment NO2
Synthesis of EvidenceExacerbations withasthma—Sufficientfor causal association
ReasonsLarge number of studies with adequate control for confounding and mostly precise effect estimates
Traffic Exposure and Doctor-Diagnosed Asthma Incidence in Children
NO2 Distance Density Spatial
Synthesis of EvidenceSufficient OR suggestive evidenceReasonsStudies that included both traffic-specific pollutantsand density measures most consistent
12
Long-Term Traffic Exposure and Cardiopulmonary Mortality
1.22
1.38
1.84
1.70
1.05 1.05 1.041.07
1.451.39
1.13
1.95
0.9
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
2.1
Jerr
ett 2
009:
Circ
ulat
ory
Fick
elst
ein
2005
:C
ardi
ovas
cula
r
Fick
elst
ein
2005
:St
roke
Geh
ring
2006
:C
ardi
opul
mon
ary
Hoe
k 20
02:
Car
diop
ulm
onar
y
Mah
esw
aran
2003
: Stro
ke
Bele
en 2
008c
:C
ardi
ovas
cula
r
Beel
en 2
008c
:C
ardi
ovas
cula
r
Beel
en 2
008c
:C
ardi
ovas
cula
r
Jerr
ett 2
009:
Circ
ulat
ory
(All)
Jerr
ett 2
009:
Circ
ulat
ory
(pro
xim
ity to
traffi
c)
Jerr
ett 2
005
Residential distance from major road or highway Per10ug/m3
BS
Per10ug/m3
NO2
Per 4 ppb NO2 Per10ug/m3PM2.5
Rel
ativ
e R
isks
(95%
CI)
Proximity BS NO2 PM
Synthesis of EvidenceSuggestive to infer causal association but not yet sufficient
ReasonsToo few studiesRelative imprecision of most estimates
Effects of Traffic Exposure onBirth Outcomes
• Synthesis of Evidence• Insufficient evidence
• Reasons• Only 4 studies met
criteria for inclusions
0.5
0.75
1
1.25
1.5
1.75
2
Qui
ntile
2 v
s. 1
Qui
ntile
3 v
s. 1
Qui
ntile
4 v
s. 1
Qui
ntile
5 v
s. 1
Qui
ntile
2 v
s. 1
Qui
ntile
3 v
s. 1
Qui
ntile
4 v
s. 1
Qui
ntile
5 v
s. 1
Hig
h S
ES
Mid
dle
SE
S
Low
SE
S
Hig
h S
ES
Mid
dle
SE
S
Low
SE
S
Qua
rtile
2 v
s. 1
Qua
rtile
3 v
s. 1
Qua
rtile
4 v
s. 1
NO
2 pe
r 10u
g/m
3
NO
per
10u
g/m
3
PM
2.5
per 1
ug/
m3
<50m
Hig
hway
<150
m H
ighw
ay
<50m
Maj
or ro
ad
NO
2 pe
r 10u
g/m
3
NO
per
10u
g/m
3
PM
2.5
per 1
ug/
m3
Pre-term birth &traffic density
(Wilhelm 2003)
Pre-term birth +LBW & traffic
density (Wilhelm2003)
Pre-term birthin summer &traffic density(Ponce 2005)
Pre-term birthin winter &
traffic density(Ponce 2005)
Birthweight<3000g &
PM2.5exposure
(Slama 2007)
Small for gestational age(Brauer 2008)
LBW at > 37wks (Brauer
2008)
RR
(95%
CI)
Pre-Term PT/LBS Summer Winter LBW SGA LBWDensity Density PT, Density PM2.5 NOx PM2.5 Proximity
13
Conclusions
Exposure
• Traffic-related pollutants impact ambient air quality on a broad spatial scale ranging from roadside, to urban, to regional background
• Based on synthesis of evidence, 300 to 500 meters from major road was identified as the near-source area most impacted by traffic; • variations exist depending on meteorology, background
pollution, and local factors
14
Issues for Exposure Assessment
• None of the pollutant surrogates considered met all criteria foran ideal surrogate• CO, benzene, and NOx [NO2] found in on-road vehicle emissions are
also major components of emissions from all sources• UF PM has not been used in epidemiologic studies so far. It is difficult to
model them because there are no emission inventories • Exposure models are important, but have various degrees of
utility to health studies• The proximity model is the most error-prone • Other models are better:
• Dispersion models (need adequate data)• Land use regression models• Several approaches together (hybrid)
Overall Conclusions
• The data are incomplete on emissions, their transformations, and exposure assessment
• There were, however, enough studies to find• Sufficient evidence for a causal association with
exacerbation of asthma• Suggestive evidence for a number of other health
effects (mortality, lung function, respiratory symptoms, and others)
15
Overall Conclusions II
• Limited evidence of effects but inadequate and insufficient to infer causal associations:• Adult onset asthma• Health care utilization• COPD• Non-asthmatic allergy• Birth outcomes• Cancers
Overall Conclusions III
• A caution: epidemiology studies are based on past estimates of exposure• they may not provide an accurate guide to estimating
health associations in the future• However, given the large number of people living
within 300- 500 meters of a major road, the Panel concluded that exposures to primary traffic generated pollutants are likely to be of public health concern and deserve attention.