Traffic-related air pollution and cognitive function in a cohort of older men.

Abstract

Traffic-related particles induce oxidative stress and may exert adverse effects on central nervous system function, which could manifest as cognitive impairment.
We assessed the association between black carbon (BC), a marker of traffic-related air pollution, and cognition in older men.
A total of 680 men (mean ± SD, 71 ± 7 years of age) from the U.S. Department of Veterans Affairs Normative Aging Study completed a battery of seven cognitive tests at least once between 1996 and 2007. We assessed long-term exposure to traffic-related air pollution using a validated spatiotemporal land-use regression model for BC.
The association between BC and cognition was nonlinear, and we log-transformed BC estimates for all analyses [ln(BC)]. In a multivariable-adjusted model, for each doubling in BC on the natural scale, the odds of having a Mini-Mental State Examination (MMSE) score ≤ 25 was 1.3 times higher [95% confidence interval (CI), 1.1 to 1.6]. In a multivariable-adjusted model for global cognitive function, which combined scores from the remaining six tests, a doubling of BC was associated with a 0.054 SD lower test score (95% CI, -0.103 to -0.006), an effect size similar to that observed with a difference in age of 1.9 years in our data. We found no evidence of heterogeneity by cognitive test. In sensitivity analyses adjusting for past lead exposure, the association with MMSE scores was similar (odds ratio = 1.3; 95% CI, 1.1 to 1.7), but the association with global cognition was somewhat attenuated (-0.038 per doubling in BC; 95% CI, -0.089 to 0.012).
Ambient traffic-related air pollution was associated with decreased cognitive function in older men.

Full text

Page 1

682
volume 119 | number 5 | May 2011 • Environmental Health Perspectives
Research
Declining or comparatively low cognitive
function in older adults is associated with loss
of independence in activities of daily living
(Greiner et al. 1996; McGuire et al. 2006),
nursing home admission (Gaugler et al. 2007;
Joray et al. 2004), hospitalization (Chodosh
et al. 2004), and mortality (Bassuk et al. 2000;
Nguyen et al. 2003). The burden associated
with deficits in cognitive function is expected
to grow as the global population ages (Ferri
et al. 2005), but few interventions to prevent,
delay, or halt the progression of cognitive
decline are currently available. The potential
relationship between exposure to common
environmental toxicants and cognition in
older adults has received relatively little con-
sideration compared with nonenvironmental
risk factors, despite evidence that many toxi-
cants may be neurotoxic and the modifiable
nature of many environmental exposures.
Air pollution may exert adverse effects on
central nervous system (CNS) function. In
autopsy studies of dogs and people, Calderón-
Garcidueñas et al. (2002, 2003, 2004, 2008b)
found evidence of increased levels of inflam-
matory mediators, β-amyloid deposition, and
markers of oxidative damage to DNA, as well
as evidence of blood–brain barrier disruption,
in the brains of those from a city with high lev-
els of air pollution compared with those from a
city with low levels. Children and dogs residing
in a highly polluted city were also more likely
to exhibit white matter hyperintensities in the
prefrontal cortex than were those residing in a
city with lower levels (Calderón-Garcidueñas
et al. 2008a). Although this ecological evidence
is insufficient to conclude that air pollution is
the causal agent, other lines of evidence sup-
port this claim. In animal studies, exposure to
ozone has been linked to neuroinflammation,
lipid peroxidation, and memory deficits (Block
and Calderón-Garcidueñas 2009; Dorado-
Martinez et al. 2001), effects that may be medi-
ated by the induction of systemic inflammation.
Controlled animal experiments demon strate
that ultrafine particulate matter (PM), a com-
ponent of traffic-related air pollution, can
translocate to the brain via the olfactory nerve
or systemic circulation (Elder et al. 2006;
Oberdörster et al. 2004; Peters et al. 2006),
where it may promote CNS dysfunction. In
animal studies, exposure to fine or ultrafine
PM has been associated with CNS inflamma-
tion and lipid peroxidation (Campbell et al.
2005; Kleinman et al. 2008; Zanchi et al.
2008), neuronal degeneration (Veronesi et al.
2005), and behavioral changes (Zanchi et al.
2008). Alternately, the cardiovascular effects of
air pollution (Brook 2007; Pope and Dockery
2006) may lead to CNS dysfunction through
promotion of vascular brain pathology.
One potential manifestation of CNS dys-
function in older adults is cognitive impair-
ment. We hypothesized that exposure to
black carbon (BC), a marker of traffic-related
air pollution, would be associated with cog-
nitive function in a cohort of older men,
because traffic-related air pollution is a major
source of exposure to ultrafine PM, which
may induce CNS dysfunction.
Materials and Methods
Study sample. We drew the study sample
from participants of the U.S. Department
of Veterans Affairs (VA) Normative Aging
Study (NAS), an ongoing longitudinal cohort
study of men established in 1963 (Bell et al.
1972). Participants in the NAS are invited
to undergo an in-person examination every
3 years. In addition to a physical examina-
tion and laboratory tests, participants provide
information on medical history, lifestyle, and
demographic factors at each study visit and,
starting in 1993, have been asked to complete
a battery of cognitive tests. At each visit, par-
ticipants provided written informed consent
as approved by the VA Boston Healthcare
System Institutional Review Board (IRB).
Address correspondence to M.C. Power, Harvard
School of Public Health, EOME, Landmark Center,
3rd Floor East, P.O. Box 15697, Boston, MA 02215
USA. Telephone: (617) 721-9984. Fax: (617) 384-
8994. E-mail: mcpower@hsph.harvard.edu
Supplemental Material is available online (doi:10.
1289/ehp.1002767 via http://dx.doi.org/).
We thank S. Melly for his help creating Figure 1
and J. Weuve for her contribution to the develop-
ment of this work.
M.C.P. is supported by a grant from the National
Institute of Environmental Health Sciences (NIEHS;
T32 ES007069). This study was supported by a grant
from the U.S. Environmental Protection Agency
(RD832416 and RD83479801) and by grants from the
NIEHS (ES00002, ES015172, ES014663, ES005257,
ES012044, and P01 ES009825). The Cognition and
Health in Aging Men Project is supported by the
Clinical Science Research and Development Service,
U.S. Department of Veterans Affairs, the National
Institutes of Health (grants R01-AG14345, R01-
AG18436, 5-P42-ES05947, and R01-ES05257), the
U.S. Department of Agriculture, Agricultural Research
Service (contract 53-K06-510), and Veterans Affairs
Merit Review and Research Career Scientist awards
to A.S. The Veterans Affairs Normative Aging Study
is supported by the Cooperative Studies Program/
Epidemiology Research and Information Center, U.S.
Department of Veterans Affairs.
The views expressed in this article are those of the
authors and do not necessarily represent the views of
the U.S. Department of Veterans Affairs.
The authors declare they have no actual or potential
competing financial interests.
Received 22 July 2010; accepted 20 December
2010.
Traffic-Related Air Pollution and Cognitive Function in a Cohort of Older Men
Melinda C. Power,1,2 Marc G. Weisskopf,1,2 Stacey E. Alexeeff,1 Brent A. Coull,3 Avron Spiro III,4,5
and Joel Schwartz1,2
1Department of Environmental Health, 2Department of Epidemiology, and 3Department of Biostatistics, Harvard School of Public Health,
Boston, Massachusetts, USA; 4Veterans Affairs Boston Healthcare System, Boston, Massachusetts, USA; 5Department of Epidemiology,
Boston University School of Public Health, Boston, Massachusetts, USA
Background: Traffic-related particles induce oxidative stress and may exert adverse effects on
central nervous system function, which could manifest as cognitive impairment.
oBjective: We assessed the association between black carbon (BC), a marker of traffic-related air
pollution, and cognition in older men.
Methods: A total of 680 men (mean ± SD, 71 ± 7 years of age) from the U.S. Department of
Veterans Affairs Normative Aging Study completed a battery of seven cognitive tests at least once
between 1996 and 2007. We assessed long-term exposure to traffic-related air pollution using a vali-
dated spatiotemporal land-use regression model for BC.
results: The association between BC and cognition was nonlinear, and we log-transformed BC
estimates for all analyses [ln(BC)]. In a multivariable-adjusted model, for each doubling in BC on
the natural scale, the odds of having a Mini-Mental State Examination (MMSE) score ≤ 25 was
1.3 times higher [95% confidence interval (CI), 1.1 to 1.6]. In a multivariable-adjusted model for
global cognitive function, which combined scores from the remaining six tests, a doubling of BC
was associated with a 0.054 SD lower test score (95% CI, –0.103 to –0.006), an effect size similar
to that observed with a difference in age of 1.9 years in our data. We found no evidence of heteroge-
neity by cognitive test. In sensitivity analyses adjusting for past lead exposure, the association with
MMSE scores was similar (odds ratio = 1.3; 95% CI, 1.1 to 1.7), but the association with global
cognition was somewhat attenuated (–0.038 per doubling in BC; 95% CI, –0.089 to 0.012).
conclusions: Ambient traffic-related air pollution was associated with decreased cognitive func-
tion in older men.
key words: aging, black carbon, cognitive dysfunction, epidemiology, particulate matter. Environ
Health Perspect 119:682–687 (2011). doi:10.1289/ehp.1002767 [Online 20 December 2010]

Page 2

Traffic-related air pollution and cognition
Environmental Health Perspectives • volume 119 | number 5 | May 2011
683
This study has been approved by the IRBs of
the participating institutions.
We limited the study sample to partici-
pants of the NAS who completed cognitive
testing between 1996 and 2007 and who lived
within the range of our BC exposure model.
These constraints ensured that both cogni-
tive test data and an estimate of BC exposure
before cognitive assessment were available for
each study participant. We also excluded par-
ticipants if they experienced a stroke before
study start. In addition, we restricted the pres-
ent study population to those self-reporting
white race/ethnicity.
Exposure assessment. Estimates of BC
exposure at the residence of each partici-
pant were used as a surrogate for individual
exposure to traffic-related air pollution. We
estimated BC exposures using a validated
spatio temporal land-use regression model
that, starting in 1995, provides daily estimates
of BC concentrations throughout the greater
Boston, Massachusetts, area, which roughly
includes the area within Interstate 495; details
of this model have been published previously
(Gryparis et al. 2007). Briefly, daily aver-
age BC estimates from 83 monitoring sites
were used to develop a BC prediction model.
Predictors in the final model include infor-
mation on meteorological conditions (e.g.,
wind speed), land use (e.g., traffic density),
daily BC concentrations at a central monitor,
and other descriptors (e.g., day of the week).
Based on the training set, the model R2 was
0.83, whereas the average correlation between
predicted values and observed BC levels in
four out-of-sample validation samples was
0.59. For the present study, addresses with
predicted daily BC concentrations outside the
range of the exposure measurements from the
training set were excluded. To create a metric
of long-term exposure, we took the average
of the 365 daily estimate at the participant’s
residential address prior to the date of the first
cognitive assessment completed on or after 1
January 1996, the first date for which compu-
tation of a prior 1-year average was possible.
Cognitive testing. Since the onset of cog-
nitive testing, the cognitive battery has been
revised, reflecting both the addition and sub-
traction of individual cognitive tests. Seven
cognitive tests have been administered with
relative consistency across the span of cogni-
tive testing and were considered in this study:
the Mini-Mental State Examination (MMSE),
the digit span backward test, a verbal fluency
task, constructional praxis, immediate recall
of a 10-word list, delayed recall of a 10-word
list, and a pattern comparison task. The
MMSE assesses multiple cognitive domains
and is widely used as a dementia screen-
ing and research instrument (Tombaugh
and McIntyre 1992); the six additional
tests assess a variety of domains, including
attention, memory, executive function, lan-
guage, and visuomotor ability. All cognitive
tests were drawn from established cognitive
test batteries, including the Consortium to
Establish a Registry for Alzheimer Disease
(CERAD) battery (Morris et al. 1989),
the Wechsler Adult Intelligence Scales for
Adults, Revised (WAIS-R) (Wechsler 1981),
the Neurobehavioral Evaluation System 2
(Letz 1991), and the Developmental Test
of Visual-Motor Integration (Beery and
Buktenica 1967).
The present study includes cognitive
data from study visits from 1996 to 2007.
Therefore, the baseline visit for each study
participant is defined as the first cognitive
assessment completed on or after 1 January
1996. This restriction ensures that we have an
estimate of BC exposure obtained before the
cognitive test data used in our analyses. Most
study participants have completed multiple
waves of cognitive testing, and all available
cognitive data obtained between 1996 and
2007 were used in analyses.
Statistical analysis. Because we observed
a ceiling effect for MMSE scores, with
many men achieving the maximum score,
we created a dichotomous variable for infe-
rior MMSE performance for use in analyses.
MMSE scores are commonly used to screen
for dementia (Tombaugh and McIntyre
1992), but few men in our cohort exhibited
scores that would typically trigger further
evaluation (< 24). We therefore chose to label
scores ≤ 25 as “low” MMSE scores (18% of
our observations). Cognitive test scores for
the remaining six tests were z-transformed
based on the mean and standard deviation of
the individual test scores at baseline and were
considered as continuous variables in analyses,
with positive z-scores indicating better perfor-
mance for all tests.
We evaluated the association between BC
and the odds of having a low MMSE score
using logistic regression with generalized esti-
mating equations and empirical variance esti-
mates to account for repeat measures across
individuals. For the remaining six tests, which
yield seven unique scores (the pattern com-
parison task produces two unique scores), we
treated each cognitive test score as a repeat
measure of global cognitive function in a ran-
dom effects linear mixed model. This model
included a random intercept for individual to
account for repeat measures within individu-
als and a random intercept for cognitive test
score to account for use of seven different
cognitive test score types as repeat measures of
global cognitive function. This analysis pro-
vides one hypothesis test for the association
between BC and global cognitive function.
We evaluated the presence of heterogeneity
of BC effect across cognitive test score type
by adding and testing the significance of a
random slope for BC effect by cognitive test
score type. If the slope were to be significant,
this would indicate that the BC effect is dis-
similar across cognitive test score types and
the global analyses would be inappropriate.
Final models were multivariable adjusted for
potential confounders or predictors of cog-
nitive function that are not plausible inter-
mediates for a BC–cognition relationship.
We obtained data on these covariates from
information collected at each NAS study visit.
Final multivariable-adjusted models were
adjusted for age at cognitive assessment as well
as information on several variables at baseline:
education (< 12, 12–16, > 16 years), alcohol
intake (< 2 drinks/day, ≥ 2 drinks/day), physi-
cal activity [< 12, 12 to < 30, ≥ 30 metabolic
equivalent hours (MET-hr) per week], diabe-
tes (yes/no), dark fish consumption (less than
once a week, equal to or greater than once
a week), computer experience (yes/no), first
language (English/not English), percentage of
the participant’s census tract that is nonwhite,
percentage of the adult residents in the par-
ticipant’s census tract with at least a college
degree, an indicator for whether the cognitive
data were from the participant’s first cogni-
tive assessment (yes/no), and an indicator for
whether the participant was a part-time resi-
dent of the greater Boston area (yes/no). We
excluded data from study visits for which we
lacked covariate information from all analyses
(1.4% of study visits).
We log-transformed BC estimates for use
in all analyses after examination of restricted
cubic splines indicated a log-linear relation-
ship. We additionally explored interactions
between BC and body mass index (BMI),
hypertension, smoking status, and diabetes
using multiplicative interaction terms and
conducted sensitivity analyses further adjust-
ing our multivariable models for smoking
and BMI. We used R (version 2.10.1; R
Development Core Team 2010) to examine
splines and SAS (version 9.2; SAS Institute
Inc., Cary, NC) for all other analyses.
Sensitivity analyses: potential confounding
by lead. Exposure to lead may confound the
relationship between BC and cognitive func-
tion. Traffic-related exposures were a primary
source of lead exposure in the era of leaded
gasoline, and it is reasonable to expect that
men with high current traffic-related air pol-
lution exposures had high past traffic-related
lead exposures. In addition, cumulative past
lead exposure may have an adverse impact on
cognitive function in older adults (Shih et al.
2006), even at low levels (Weisskopf et al.
2007; Wright et al. 2003). Thus, we conducted
sensitivity analyses to determine the influence
of potential confounding by exposure to lead.
Tibia bone lead concentrations, a bio-
marker for cumulative past lead expo-
sures (Hu et al. 1998), were available for

Page 3

Power et al.
684
volume 119 | number 5 | May 2011 • Environmental Health Perspectives
363 participants (53.4%). Tibia bone lead
concentrations were imputed for 299 partici-
pants (44.0%) using a published prediction
model developed in the NAS (reduced model)
(Park et al. 2009). Missing data precluded
imputation of bone lead concentrations for
18 participants (2.6%), and we excluded these
participants from the sensitivity analyses. To
test the sensitivity of our results to potential
confounding by exposure to lead, we repeated
the analyses described above additionally
adjusting for bone lead concentration and
imputation status (yes/no).
Table 1. Baseline characteristics of the cohort
(n = 680).
Results
A total of 680 men met the inclusion criteria
for this study. Within this group, the number
of men completing at least one wave of cogni-
tive testing varied across the cognitive tests:
MMSE (n = 671), digit span backward test
(n = 663), verbal fluency task (n = 670), con-
structional praxis (n = 626), immediate recall
of a 10-word list (n = 669), delayed recall of a
10-word list (n = 667), and pattern compari-
son task (n = 645).
Table 1 summarizes the baseline charac-
teristics of the cohort. The mean ± SD age of
the study sample at baseline was 71 ± 7 years
(range, 51–97 years), and most men had at
least some college or graduate-level educa-
tion (70.1%). On average, study participants
completed 2.14 cognitive assessments (range,
1–5). On the natural scale, 1-year average
BC exposure estimates ranged from 0.03 to
1.77 µg/m3 (mean ± SD, 0.58 ± 0.28 µg/m3)
and exhibited a skewed distribution. Figure 1
illustrates the substantial spatial variability of
BC over the study area in the year 1995.
Because the association between cogni-
tive test scores and BC exposure estimates
appeared log-linear, we log-transformed BC
for use in all analyses and report associations
for a doubling in BC concentration on the
natural scale, or approximately a 0.69 unit
change in ln(BC). BC exposure was signifi-
cantly associated with the risk of having a
low MMSE score [multivariable-adjusted
odds ratio (OR) = 1.3 for a doubling in BC
concentration; 95% confidence interval
(CI), 1.1–1.6; Table 2]. In the analysis of
Characteristic
Age (years)
50–59
60–69
70–79
80–89
90–99
Education (years)
< 12
12–16
> 16
First language
English
Not English/bilingual
Computer experience
Yes
No
Physical activity (MET‑hr/week)
< 12
12 to < 30
≥ 30
Alcohol (drinks/day)
≥ 2
< 2
Diabetes
Yes
No
Consumed dark fish (times/week)
≥ 1
< 1
Nonwhite (% of census tract)
< 5%
5 to < 10%
≥ 10%
≥ 25 years of age with at least a college degree (% of
census tract)
< 30% 214 (31.5)
30 to < 50% 281 (41.3)
≥ 50%185 (27.2)
Smoking
Never 192 (28.2)
Former453 (66.6)
Current35 (5.1)
BMI (kg/m2)
< 25 152 (22.4)
≥ 25 528 (77.6)
Hypertension
Yes 454 (66.8)
No 226 (33.2)
Tibia bone lead
Measured 363 (53.4)
Imputed299 (44.0)
Missing 18 (2.6)
n (%)
BC concntration
(mean ± SD)
48 (7.1)
284 (41.8)
273 (40.1)
72 (10.6)
3 (0.4)
0.550 ± 0.288
0.574 ± 0.277
0.583 ± 0.267
0.605 ± 0.333
0.466 ± 0.285
203 (29.9)
338 (49.7)
139 (20.4)
0.629 ± 0.304
0.572 ± 0.276
0.521 ± 0.239
585 (86.0)
95 (14.0)
0.578 ± 0.282
0.581 ± 0.266
277 (40.7)
403 (59.3)
0.542 ± 0.255
0.604 ± 0.294
385 (56.6)
178 (26.2)
117 (17.2)
0.574 ± 0.286
0.602 ± 0.293
0.558 ± 0.237
166 (24.4)
514 (75.6)
0.573 ± 0.242
0.581 ± 0.291
106 (15.6)
574 (84.4)
0.567 ± 0.254
0.581 ± 0.285
100 (14.7)
580 (85.3)
0.566 ± 0.283
0.581 ± 0.280
268 (39.4)
188 (27.6)
224 (32.9)
0.521 ± 0.298
0.541 ± 0.249
0.679 ± 0.255
0.626 ± 0.301
0.540 ± 0.277
0.583 ± 0.250
0.557 ± 0.255
0.584 ± 0.294
0.634 ± 0.208
0.587 ± 0.262
0.576 ± 0.285
0.587 ± 0.288
0.562 ± 0.264
0.599 ± 0.286
0.559 ± 0.271
0.507 ± 0.289
Figure 1. Predicted 1-year average BC exposures (μg/m3) for the greater Boston area in 1995.
0.02–0.07
0.08–0.19
0.20–0.28
0.29–0.35
0.36–0.41
0.42–0.48
0.49–0.55
0.56–0.61
0.62–0.82
0.83–1.55
Limited access highway
Town boundary

Page 4

Traffic-related air pollution and cognition
Environmental Health Perspectives • volume 119 | number 5 | May 2011
685
global cognitive function, which combined
the remaining seven cognitive test scores, a
doubling in BC was significantly associated
with a decrement of 0.054 SD in cogni-
tive test score in our multivariable-adjusted
model (95% CI, –0.103 to –0.006; Table
2), an effect similar to a 1.9-year differ-
ence in age in our data. We found no evi-
dence of heterogeneity of results across tests
[p-value = 0.25]. See Supplemental Material,
Tables 1 and 2 (doi:10.1289/ehp.1002767)]
for the full multivariable-adjusted models.
In our sensitivity analyses, we additionally
adjusted for smoking and BMI, and our
results were materially unchanged.
We found no evidence of interactions
between BC and diabetes, smoking, BMI or
hypertension on MMSE scores (p-values for
all interactions > 0.1). Our exploration of
potential interactions between BC and these
factors on global cognitive function, in mod-
els that combined the remaining seven test
scores, suggested that the adverse effects of
BC were concentrated in overweight and
obese individuals (p-value for interaction =
0.10) and in ever smokers (p-value for inter-
action = 0.07), but we found no evidence for
effect modification by hypertension or diabe-
tes (p-values for interactions > 0.1).
Among participants with measured tibia
bone lead concentrations, the mean ± SD bone
lead concentration was 19.2 ± 11.9 µg lead/g
bone mineral; among those with imputed
bone lead concentrations, it was 16.1 ± 7.0 µg
lead/g bone mineral. In the sensitivity analysis
that adjusted for estimates of cumulative lead
exposure, we observed that the multivariable-
adjusted association between BC and low
MMSE scores was almost identical (OR = 1.3
per doubling in BC; 95% CI, 1.1–1.7). The
association between BC and global cognitive
function was somewhat attenuated after we
adjusted for lead in our multivariable-adjusted
model (–0.038 per doubling in BC; 95% CI,
–0.089 to 0.012).
Discussion
The results of this study suggest an adverse
effect of traffic-related air pollution on global
cognitive function in older men. Confounding
by lead cannot fully account for these findings.
When we explored the potential for effect
modification, our results suggest that the effect
of traffic-related air pollution on cognition
may be greater in smokers or overweight and
obese individuals. Because these conditions are
proinflammatory, this is biologically plausible
but requires further confirmation.
To our knowledge, this is only the sec-
ond study to consider the association between
air pollution and cognitive function in older
adults. Ranft et al. (2009) considered the
association between two markers of air pollu-
tion, a measure of PM with an aerodynamic
diameter ≤ 10 µm (PM10) and distance to a
major road, and cognitive function in 399
women 68–79 years of age who had lived
at the same address for 20 years. Residence
within 50 m of a major road, but not PM10,
was significantly associated with a decrement
in the total score of the CERAD-Plus test
battery, the Stroop test, and a test of olfac-
tory function in multivariable-adjusted mod-
els. In addition, residence within 50 m of
a major road was adversely associated with
performance on all subtests of the CERAD
battery, although few individual associations
achieved statistical significance. Our findings
are consistent with these, because BC, a traffic
particle, is strongly associated with distance to
major road and weakly associated with PM10.
A handful of studies have evaluated the
association between air pollution and cogni-
tive function in younger populations, and, col-
lectively, they also suggest an adverse effect. In
a study of adults 20–59 years of age, Chen and
Schwartz (2009) reported that estimates of
exposure to ozone, but not PM10, were asso-
ciated with inferior performance on two of
three neurobehavioral tests in fully adjusted
models. Calderón-Garcidueñas et al. (2008a)
found the performance age on several sub-
tests of the Wechsler Intellegence Scale for
Children, Revised (WISC-R) to be signifi-
cantly behind chronological age in children
residing in cities with high levels of air pollu-
tion but not in those residing in cities with low
levels. In a study that compared neurobehav-
ioral test scores across students from two pri-
mary schools in the same city in China, Wang
et al. (2009) found that students who attended
the school with high levels of traffic-related
air pollution were more likely to exhibit poor
performance than were those who attended
the school with low levels. Franco Suglia et al.
(2008) found that higher residence-based
estimates of average lifetime exposure to BC
were associated with lower cognitive test scores
in children 8–11 years of age. Similarly, in
4- to 5-year-old children from the INMA
(Environment and Childhood) study, Freire
et al. (2010) observed that higher residence-
based estimates of nitric dioxide were associ-
ated with lower scores on all subscales of a
standardized version of the McCarthy Scales of
Children’s Abilities, although only one of these
associations achieved statistical significance.
Several biological mechanisms have been
proposed to explain how traffic-related air
pollution may have an adverse effect on
CNS function. First, ultrafine particles, such
as those found in diesel exhaust, are small
enough to pass through the air–blood barrier
of the lung, enter the systemic circulation,
and translocate to other body tissues, includ-
ing the brain (Oberdörster et al. 2004; Peters
et al. 2006), and may also translocate directly
to the brain via the olfactory nerve (Calderón-
Garcidueñas et al. 2003; Elder et al. 2006).
The presence of such particles in the brain is
associated with neuroinflammation and oxida-
tive stress (Calderón-Garcidueñas et al. 2003;
Elder et al. 2006). Second, traffic-related PM
may exert indirect effects on CNS function
through effects on cardiovascular health.
Exposure to PM is associated with a variety of
cardiovascular end points (Brook 2007; Pope
and Dockery 2006), including hypertension
(Auchincloss et al. 2008; Brook 2007) and
atherosclerosis (Künzli et al. 2005, 2010), as
well as markers of increased cardiovascular
risk, including progression of atherosclerotic
plaques (Floyd et al. 2009; Künzli et al. 2010;
Sun et al. 2005) and homocysteine (Baccarelli
et al. 2007; Park et al. 2008). Several of these
factors are also associated with worse cognitive
function (Komulainen et al. 2007; Qiu et al.
2005; Sander et al. 2010; Vidal et al. 2008),
which likely reflects the link between cogni-
tive impairment and vascular brain pathology.
The present study has several strengths. The
BC exposure model enabled us to report indi-
vidual estimates of long-term exposure to traf-
fic-related air pollution. In our study, we were
also able to control for socio economic status
(SES) using both personal and residence-based
characteristics, which should limit residual con-
founding by this factor. Furthermore, we were
able to address the potential for confounding
by past exposure to lead, another environmen-
tal pollutant that may affect cognitive function
in older adults, using a combination of meas-
ured and imputed biomarkers for cumulative
past lead exposure. Although correlated with
traffic-related air pollution, exposure to lead
cannot completely account for the observed
associations. Finally, our results are based on a
substantial number of observations, given that,
on average, each of our 680 participants com-
pleted 2.14 cognitive testing sessions.
Table 2. Adjusted associations for a doubling in BC concentration on the natural scale and cognitive test
score(s).
Effect per doubling in BC concentration (95% CI)
Analysis
Low MMSE score OR
Global analysis estimate
Model 1: age adjusted
1.4 (1.1 to 1.6)
–0.073 (–0.122 to –0.023)
Model 2: age and
education adjusted
1.3 (1.0 to 1.5)
–0.052 (–0.100 to –0.004)
Model 3: multivariable
adjusteda
1.3 (1.1 to 1.6)
–0.054 (–0.103 to –0.006)
aAdjusted for age, education, first language, computer experience, physical activity, alcohol consumption, diabetes,
dark fish consumption, percentage of residential census tract that is nonwhite, percentage of residential census tract
adults with a college degree, indicator for first cognitive assessment, and indicator for part-time resident.

Page 5

Power et al.
686
volume 119 | number 5 | May 2011 • Environmental Health Perspectives
Several potential sources of bias must be
considered when interpreting our results. The
use of exposure estimates based on residential
address may misclassify personal exposure lev-
els. However, the lack of significant occupa-
tional or commuting exposure in this largely
retired cohort should mean that residence-
based exposure estimates are an excellent
measure of personal exposures over the past
few years. Because 87% of our participants
have lived at the same address for at least
5 years before their baseline cognitive assess-
ment (mean ± SD duration, 18 ± 12 years)
and because people are likely to choose simi-
lar neighborhoods when relocating, misclas-
sification due to participant relocation is also
expected to be minimal. In addition, occa-
sional exposures encountered elsewhere are
unlikely to be correlated with cognitive func-
tion and so would result in nondifferential
misclassification of exposure, which would
be expected to attenuate our effect estimates
(Gryparis et al. 2009). Similarly, nondiffer-
ential misclassification of cognitive function
is expected when assessing cognitive func-
tion and would make it harder to detect a
true effect, although our use of global analy-
ses helps to reduce the impact of noise by
using multiple measures to capture under-
lying global cognitive function. We observed
statistically significant associations between
BC and cognitive function, when measured
as the risk of a “low” MMSE score and as
overall performance on our battery of cogni-
tive tests, despite the presence of misclassi-
fication, suggesting that the true association
may be even stronger than what we observed.
Although residual confounding cannot be
excluded, we do not believe confounding
to be a major source of bias in the present
study given the availability of information
on potential confounders and the use of sev-
eral different, related variables to account for
SES. Moreover, we do not believe confound-
ing by regional pollutants, which may have
an independent effect on cognition, to be a
major source of bias. Correlation between
BC and regional pollutants may be induced
through similar spatial or temporal patterns.
However, traffic-related air pollution exhib-
its significant spatial variation not shared by
more regional pollutants [e.g., ozone or par-
ticulate matter with an aerodynamic diameter
< 2.5 µm (PM2.5)], and temporal correlation
is quite weak between our 1-year residence-
based BC estimates (which have varying start
dates depending on the timing of the baseline
cognitive assessment) and the 1-year average
of ozone concentrations at four central moni-
tors in the Boston area (r = –0.03) or 1-year
average PM2.5 concentrations at a central site
monitor in Boston (r = 0.16) calculated over
the same time periods. Incomplete follow-up
may induce bias in the present study but is
unlikely to account for the observed associa-
tions. In our study, and generally in studies
of older adults, participants who are lost to
follow-up exhibit lower cognitive test scores
than do those who complete follow-up (Euser
et al. 2008). Because air pollution has been
related to cardiovascular morbidity and mor-
tality (Brook 2007; Pope and Dockery 2006)
and because we observed lower BC estimates
in participants with three or more follow-up
visits, we expect that a greater loss of those
with both lower cognitive test scores and
higher BC estimates to lead to an underesti-
mate of the association. Thus, the true asso-
ciation may be stronger than that observed.
One additional limitation lies in our
inability to attribute our findings to a particu-
lar traffic-related exposure. We have estimated
exposure to BC particles for each participant.
However, traffic-related air pollution is a com-
plex mixture of gases and particles, and BC is
correlated with other components of the mix-
ture. Although existing evidence suggests that
PM, specifically ultrafine PM, is a likely agent
for adverse effects to the CNS (Calderón-
Garcidueñas et al. 2003; Elder et al. 2006;
Oberdörster et al. 2004; Peters et al. 2006),
other components may be important. In
addition, measures of traffic-related air pol-
lution are often correlated with other envi-
ronmental exposures. Although we were able
to address the potential for lead to account
for the observed associations, other exposures
must be considered, including noise. Elevated
noise levels have been linked to metrics of
cognitive function in children in a handful
of studies (Clark et al. 2006; Stansfeld et al.
2005), and chronic occupational exposure
to noise is associated with neurophysiologi-
cal measures related to attention in adults
(Gomes et al. 1999). Given the high correla-
tion between measures of traffic-related air
pollution and noise (Allen et al. 2009; Davies
et al. 2009), it is possible that the associa-
tions observed for one exposure are attribut-
able to the other. Further research to evaluate
the rela tive contribution of particular traffic-
related pollutants or other exposures to the
observed association is warranted.
Conclusions
Our findings suggest that traffic-related air
pollution may have an adverse effect on cog-
nition in older men. This is the first study to
find an association between traffic-related air
pollution and cognition in older men, and
only the second to consider the relationship
in older adults. Given the ubiquitous nature
of the exposure, if traffic-related air pollution
is causally related to cognitive impairment in
older adults, implementation of interventions
to reduce exposure, including establishment
of more stringent emissions standards, would
be expected to have substantial benefits.
RefeRences
Allen RW, Davies H, Cohen MA, Mallach G, Kaufman JD,
Adar SD. 2009. The spatial relationship between traffic-
generated air pollution and noise in 2 U.S. cities. Environ
Res 109(3):334–342.
Auchincloss AH, Diez Roux AV, Dvonch JT, Brown PL, Barr RG,
Daviglus ML, et al. 2008. Associations between recent
exposure to ambient fine particulate matter and blood
pressure in the multi-ethnic study of atherosclerosis
(MESA). Environ Health Perspect 116:486–491.
Baccarelli A, Zanobetti A, Martinelli I, Grillo P, Hou L, Lanzani G,
et al. 2007. Air pollution, smoking, and plasma homo-
cysteine. Environ Health Perspect 115:176–181.
Bassuk SS, Wypij D, Berkman LF. 2000. Cognitive impairment
and mortality in the community-dwelling elderly. Am J
Epidemiol 151(7):676–688.
Beery KE, Buktenica NA. 1967. Developmental Test of Visual
Motor Integration. Chicago:Follett.
Bell B, Rose CL, Damon A. 1972. The Normative Aging Study:
an interdisciplinary and longitudinal study of health and
aging. Aging Hum Dev 3(1):5–17.
Block ML, Calderón-Garcidueñas L. 2009. Air pollution: mecha-
nisms of neuroinflammation and CNS disease. Trends
Neurosci 32(9):506–516.
Brook RD. 2007. Is air pollution a cause of cardiovascular
disease? Updated review and controversies. Rev Environ
Health 22(2):115–137.
Calderón-Garcidueñas L, Azzarelli B, Acuna H, Garcia R,
Gambling TM, Osnaya N, et al. 2002. Air pollution and brain
damage. Toxicol Pathol 30(3):373–389.
Calderón-Garcidueñas L, Maronpot RR, Torres-Jardon R,
Henriquez-Roldán C, Schoonhoven R, Acuna-Ayala H, et al.
2003. DNA damage in nasal and brain tissues of canines
exposed to air pollutants is associated with evidence
of chronic brain inflammation and neuro degeneration.
Toxicol Pathol 31(5):524–538.
Calderón-Garcidueñas L, Mora-Tiscareño A, Ontiveros E,
Gómez-Garza G, Barragán-Mejia G, Broadway J, et al.
2008a. Air pollution, cognitive deficits and brain abnor-
malities: a pilot study with children and dogs. Brain Cogn
68(2):117–127.
Calderón-Garcidueñas L, Reed W, Maronpot RR, Henríquez-
Roldán C, Delgado-Chavez R, Calderón-Garcidueñas A,
et al. 2004. Brain inflammation and Alzheimer’s-like pathol-
ogy in individuals exposed to severe air pollution. Toxicol
Pathol 32(6):650–658.
Calderón-Garcidueñas L, Solt AC, Henriquez-Roldán C, Torres-
Jardón R, Nuse B, Herritt L, et al. 2008b. Long-term air
pollution exposure is associated with neuroinflamma-
tion, an altered innate immune response, disruption of the
blood-brain barrier, ultrafine particulate deposition, and
accumulation of amyloid beta-42 and alpha-synuclein in
children and young adults. Toxicol Pathol 36(2):289–310.
Campbell A, Oldham M, Becaria A, Bondy SC, Meacher D,
Sioutas C, et al. 2005. Particulate matter in polluted air
may increase biomarkers of inflammation in mouse brain.
Neurotoxicology 26(1):133–140.
Chen J, Schwartz J. 2009. Neurobehavioral effects of ambi-
ent air pollution on cognitive performance in U.S. adults.
Neurotoxicology 30(2):231–239.
Chodosh J, Seeman TE, Keeler E, Sewall A, Hirsch SH,
Guralnik JM, et al. 2004. Cognitive decline in high-
functioning older persons is associated with an increased
risk of hospitalization. J Am Geriatr Soc 52(9):1456–1462.
Clark C, Martin R, van Kempen E, Alfred T, Head J, Davies HW,
et al. 2006. Exposure-effect relations between aircraft and
road traffic noise exposure at school and reading compre-
hension: the Ranch Project. Am J Epidemiol 163(1):27–37.
Davies HW, Vlaanderen JJ, Henderson SB, Brauer M. 2009.
Correlation between co-exposures to noise and air pollu-
tion from traffic sources. Occup Environ Med 66(5):347–350.
Dorado-Martinez C, Paredes-Carbajal C, Mascher D, Borgonio-
Pérez G, Rivas-Arancibia S. 2001. Effects of different ozone
doses on memory, motor activity and lipid peroxidation
levels, in rats. Int J Neurosci 108(3–4):149–161.
Elder A, Gelein R, Silva V, Feikert T, Opanashuk L, Carter J,
et al. 2006. Translocation of inhaled ultrafine manganese
oxide particles to the central nervous system. Environ
Health Perspect 114:1172–1178.
Euser SM, Schram MT, Hofman A, Westendorp RG,
Breteler MM. 2008. Measuring cognitive function with
age: the influence of selection by health and survival.
Epidemiology 19(3):440–447.