American Journal of Epidemiology
ª The Author 2009. Published by the Johns Hopkins Bloomberg School of Public Health.
All rights reserved. For permissions, please e-mail: [email protected].
Vol. 169, No. 11
DOI: 10.1093/aje/kwp067
Advance Access publication April 24, 2009
Original Contribution
Association of Early-life Exposure to Household Gas Appliances and Indoor
Nitrogen Dioxide With Cognition and Attention Behavior in Preschoolers
Eva Morales, Jordi Julvez, Maties Torrent, Rafael de Cid, Mònica Guxens, Mariona Bustamante,
Nino Künzli, and Jordi Sunyer
Initially submitted September 25, 2008; accepted for publication February 27, 2009.
The authors investigated the association of early-life exposure to indoor air pollution with neuropsychological
development in preschoolers and assessed whether this association differs by glutathione-S-transferase gene
(GSTP1) polymorphisms. A prospective, population-based birth cohort was set up in Menorca, Spain, in 1997–
1999 (n ¼ 482). Children were assessed for cognitive functioning (McCarthy Scales of Children’s Abilities) and
attention-hyperactivity behaviors (Diagnostic and Statistical Manual of Mental Disorders, 4th Edition) at age
4 years. During the first 3 months of life, information about gas appliances at home and indoor nitrogen dioxide
concentration was collected at each participant’s home (n ¼ 398, 83%). Genotyping was conducted for the
GSTP1 coding variant Ile105Val. Use of gas appliances was inversely associated with cognitive outcomes
(b coefficient for general cognition ¼ 5.10, 95% confidence interval (CI): 9.92, 0.28; odds ratio for inattention symptoms ¼ 3.59, 95% CI: 1.14, 11.33), independent of social class and other confounders. Nitrogen
dioxide concentrations were associated with cognitive function (a decrease of 0.27 point per 1 ppb, 95% CI:
0.48, 0.07) and inattention symptoms (odds ratio ¼ 1.06, 95% CI: 1.01, 1.12). The deleterious effect of
indoor pollution from gas appliances on neuropsychological outcomes was stronger in children with the GSTP1
Val-105 allele. Early-life exposure to air pollution from indoor gas appliances may be negatively associated with
neuropsychological development through the first 4 years of life, particularly among genetically susceptible
children.
air pollution, indoor; attention deficit disorder with hyperactivity; child development; cognition; fossil fuels; glutathione transferase; nitrogen dioxide; polymorphism, genetic
Abbreviations: ADHD, attention deficit hyperactivity disorder; GSTP1, glutathione S-transferase P1.
Evidence of a link between air pollution and increased
risk of cerebrovascular and neurodegenerative diseases is
emerging (1, 2). Furthermore, indices of deleterious effects
of air pollution on children’s neurodevelopment have recently been reported (3, 4). Prenatal exposure to airborne
polycyclic aromatic hydrocarbons has been associated with
a lower mental development index at age 3 years (3).
A long-term concentration of black carbon particles from
mobile sources has been associated with decreases in cognitive
test scores among children aged 8–11 years (4). In addition,
some environmental factors have been related to development of attention deficit hyperactivity disorder (ADHD)
symptoms (5), but no studies are known to have examined
the influence of air pollution.
Although outdoor air pollution first brought the issue of
the health effects of air pollution to the public’s attention,
indoor air pollution likely has the greatest impact on children’s health and constitutes a public health problem in both
developed and developing countries (6, 7). Among the major sources of indoor air pollution are combustion byproducts from heating and cooking. Because the prevalence
of gas stoves in developed countries is high (50%–70%),
any evidence of a deleterious effect on health represents
a major public health issue (8). Domestic use of gas-fueled
Correspondence to Dr. Jordi Sunyer, Center for Research in Environmental Epidemiology, Parc de Recerca Biomèdica de Barcelona,
C / Dr. Aiguader 88, Barcelona E-08003, Catalonia, Spain (e-mail: [email protected]).
1327
Am J Epidemiol 2009;169:1327–1336
1328 Morales et al.
cooking and heating appliances can produce high indoor
levels of nitrogen dioxide, the most toxic of oxides of nitrogen and the most extensively studied indoor pollutant (9).
Nitrogen dioxide cell damage is mediated by oxidant injury,
inflammatory response, and lipid peroxidation (10). The glutathione S-transferase P1 gene, GSTP1, encodes major phase
II xenobiotic metabolizing enzymes involved in antioxidant
defenses to protect against oxidative stress (11). GSTP1 is
the most strongly expressed glutathione S-transferase isoenzyme in the human brain during early life (12). Given the
oxidative nature of nitrogen dioxide, we hypothesize that
detoxification genes may modify the neurologic effect of
nitrogen dioxide.
In this study, we aimed to investigate the effect of earlylife exposure to household gas appliances and indoor
nitrogen dioxide on cognitive functioning and inattentionhyperactivity symptoms in preschoolers. We also assessed
whether these effects differ by polymorphisms of GSTP1.
MATERIALS AND METHODS
Study population
A population-based birth cohort was recruited on the island of Menorca (Spain) (13). Recruited were all women
presenting for prenatal care in Menorca over 12 months
starting in mid-1997. Subsequently, 482 children (94% of
those eligible) were enrolled, and complete outcome data
were provided for 422 (87%) up to 4 years of age. Of these
482 children, 411 (85%) were genotyped. The present paper
is based on 398 children with complete information on neurodevelopmental assessment, gas appliances at home, and
indoor nitrogen dioxide concentrations.
After the study was explained to the parents, their written
informed consent was obtained. The ethics committee of the
Institut Municipal d’Investigació Mèdica approved this
study.
Data collection procedures
In the longitudinal study, detailed data on maternal and
paternal years of education, maternal and paternal social
class (using the United Kingdom Registrar General’s 1990
classification according to parental occupation by ISCO-88
(International Classification of Occupations) code), maternal marital status, maternal health and obstetric history, parity, dietary intake during pregnancy, child’s sex, maternal
alcohol consumption during pregnancy, parental tobacco
smoking habits, and type and duration of breastfeeding were
obtained through questionnaires administered in person.
Gestational age and anthropometric measures at birth were
collected from clinical records.
Outcomes of interest
Two certified psychologists performed the neuropsychological testing of the children at age 4 years, including assessment of cognitive functioning, motor abilities, and
ADHD symptoms, as described in detail previously (14–
16). This process was supervised by the project’s consulting
psychologist (including intra- and interpsychologist validity
at the beginning, midpoint, and end of the study), who rescored the tests. Staff members involved in the neuropsychological testing were blinded to a child’s exposure.
Neurocognitive functions and motor skills were measured
with the Spanish version of the McCarthy Scales of Children’s Abilities test, consisting of 6 different scales (general
cognitive, verbal, perceptual-performance, quantitative,
memory, and motor) (17). To further improve our understanding of the neurocognitive functions, we reorganized
the items from the McCarthy test into a new outcome, the
executive function subscale, according to present neuropsychological assessment standards. The executive function involves coordinating complex behaviors such as attention
and working memory. The new subscale showed good psychometric characteristics (14). Extensive information about
the outcome assessments, including scale and subscale descriptions, evaluation procedures, and validity of the instruments, has been published elsewhere (14).
ADHD was evaluated by using Diagnostic and Statistical
Manual of Mental Disorders, 4th Edition, criteria (18). This
internationally recognized questionnaire comprises 18 items
designed to evaluate attention-deficit (items 1–9), hyperactivity (items 10–15), and impulsivity (items 16–18) symptoms in children. The original Diagnostic and Statistical
Manual of Mental Disorders, 4th Edition, list was used,
which rates items as yes or no (15, 16). The ADHD Diagnostic and Statistical Manual of Mental Disorders, 4th
Edition questionnaire was administered to children’s teachers and mothers at the same time to assess interobserver
reliability. We found a moderate, but significant association
(Spearman r ¼ 0.27; P < 0.001) between mothers and
teachers, similar to that reported in other studies (19).
Exposure assessment
During a subject’s first 3 months of life, a trained field
technician visited the home to complete a questionnaire on
household characteristics (cooking appliances, heating and
cooling systems, number of hours of ventilation, and size of
the house), to collect dust samples, and to measure nitrogen
dioxide, as previously described (20). Briefly, average
2-week nitrogen dioxide concentrations were measured by
passive diffusion tubes installed in the living room wall at
a height of 2 m and away from any window or air conditioner (21). Nitrogen dioxide concentrations were measured
in a single laboratory by colorimetric reaction, as previously
described (20).
Genotyping methods
DNA was extracted from blood (87%) and saliva (13%)
samples. Two semiautomated assays were implemented to
facilitate detection of the coding variant Ile105Val in
GSTP1; that variant was analyzed by using pyrosequencing
technology (Biotage, Uppsala, Sweden) in a single assay.
All assays were performed by technicians blinded to neurodevelopmental outcomes and exposure information. GSTP1
genotypes were in Hardy-Weinberg equilibrium in the total
analyzed cohort (P > 0.05).
Am J Epidemiol 2009;169:1327–1336
Indoor Air Pollution and Neurodevelopment
Table 1. Comparison of the Distribution of Child and Maternal Characteristics, Exposure
Variables, and GSTP1 Genotype Between Preschoolers Included and Not Included in the
Menorca Cohort, Spain, 1997–1999
Included
(n 5 398)
Not Included
(n 5 84)
P Valuea
No.
%
No.
%
Female sex
198
49.7
36
42.9
McCarthy test score
398
23
106.9 (19.9)
97.5 (19.6)
0.036b
Perceptual-performance, mean (SD)
40.4 (7.7)
38.4 (5.9)
0.239b
Memory, mean (SD)
23.4 (7.5)
21.3 (6.8)
0.187b
Quantitative, mean (SD)
18.1 (4.7)
15.9 (4.2)
0.033b
Verbal, mean (SD)
48.4 (11.5)
43.5 (11.8)
0.051b
Motor, mean (SD)
34.5 (6.8)
34.2 (6.6)
0.892b
Executive function, mean (SD)
44.1 (11.0)
39.0 (9.6)
0.035b
Child variables
General cognitive, mean (SD)
ADHD information available
ADHD symptoms, yes
365
0.251
21
58
15.9
4
19.0
Inattention, yes
43
11.8
3
14.3
0.702
0.730
Hyperactivity, yes
25
6.8
2
9.5
0.640
0.012
Maternal variables
Maternal social class
Professional
384
82
50
13.0
9
11.0
205
53.4
34
41.5
Partially skilled and unskilled
49
12.8
22
26.8
Unemployed
80
20.8
17
20.7
Skilled manual and nonmanual
Education
384
82
Secondary or higher
164
42.7
30
36.6
Primary or less
220
57.3
52
63.4
0.307
Smoking during pregnancy, yes
78
19.6
24
28.6
0.067
Alcohol consumption during
pregnancy, yes
87
21.9
16
19.5
0.637
Yes
283
71.1
66
79.5
0.118
No (electric)
115
28.9
17
20.5
Yes
94
23.6
25
29.8
No
304
76.4
59
70.2
Exposure variables
Gas stove
Gas fire
Indoor measured NO2 concentration, ppb
398
58
Mean (SD)
15.8 (15.6)
19.4 (17.9)
Median
11.5
13.5
Minimum–maximum
GSTP1 genotype
0.235
0.41–98.9
360
0.108b
1.10–96.0
51
Ile/Ile
157
43.6
28
54.9
Ile/Val
171
47.5
20
39.2
Val/Val
32
8.9
3
5.9
0.301
Abbreviations: ADHD, attention deficit hyperactivity disorder; GSTP1, glutathione S-transferase
P1 gene; NO2, nitrogen dioxide; SD, standard deviation.
a
Unless otherwise indicated, P value derived from v2 test.
b
P value derived from Student’s t test.
Am J Epidemiol 2009;169:1327–1336
1329
1330 Morales et al.
Statistical analysis
Continuous cognitive outcomes were standardized to
a mean of 100 with a standard deviation of 16 to homogenize all the scales. The criterion for ADHD was the presence
of either 6 or more symptoms of inattention or 6 or more
symptoms of hyperactivity-impulsivity. Both components
(inattention and hyperactivity) were also studied separately.
To limit exposure misclassification, we decided to use all
available exposure predictors and estimate indoor nitrogen
dioxide concentration by using a regression model, regressing the 2-week average values of nitrogen dioxide against
type of stove (electric, gas), gas fire (no, yes), use of extractor fan when cooking (always, sometimes, never), number
of gas appliances at home, and season (winter, spring, summer, autumn). Predicted nitrogen dioxide derived from the
regression model was used as the exposure variable. To
evaluate the linearity of the relation between predicted nitrogen dioxide as continuous and McCarthy Scales of Children’s Abilities’s cognitive outcomes, we used adjusted
general additive models (through nonparametric depiction
of the predictor when the effects of the other variables had
been taken into account) (22).
Multivariate linear regression models were used to examine the association between gas appliances or indoor nitrogen dioxide concentration (both continuous and in quartiles)
and cognitive outcomes at age 4 years. We used logistic
regression to measure the associations of indoor air pollution with ADHD. All the variables significantly related to
the outcomes of interest (P < 0.20) were included in the
multivariate model, and they were retained only if they
modified the coefficient of predictor variables during pregnancy by more than 10%. Final multivariable models were
adjusted for maternal social class, maternal education,
school trimester at testing, evaluator (neuropsychologist),
maternal smoking during pregnancy, number of smokers
at home, maternal alcohol consumption during pregnancy,
and home location (urban vs. nonurban). Allowance for
other possible confounding variables such as methylmercury, prematurity, and low birth weight did not materially
alter the estimates. To evaluate whether the effect of the
evaluated exposures differed by polymorphisms of GSTP1,
we stratified the analysis according to GSTP1 genotypes.
The presence of gene-environment interactions between indoor nitrogen dioxide and GSTP1 polymorphisms was assessed by including interaction terms in the regression
model. All statistical analyses were conducted with Stata
8.0 statistical software (Stata Corporation, College Station,
Texas).
RESULTS
Compared with nonparticipants, participants had higher
general cognitive scores and were of a higher social class
(Table 1). There were nonstatistically significant differences
between participants and nonparticipants regarding ADHD
symptoms prevalence, gas appliances at home, and indoor
nitrogen dioxide concentrations. The prevalence of cooking
with gas was 71.1% and of heating with gas fires was 23.6%.
A total of 213 (53.5%) homes had 1 gas appliance, and 82
Table 2. Determinants of Indoor NO2 Concentration (ppb),
Menorca Cohort, Spain, 1997–1999
Mean
(SE)
Minimum Maximum P Value
Type of stove
Electric (n ¼ 115)
Gas (n ¼ 283)
7.2 (0.7)
0.41
57.64
19.3 (0.9)
0.46
98.9
<0.001
Gas fire
No (n ¼ 304)
13.1 (0.8)
0.41
98.9
Yes (n ¼ 94)
24.5 (1.9)
0.9
82.0
<0.001
Use of extractor fan
Always (n ¼ 256)
13.4 (0.8)
0.41
73.71
Sometimes (n ¼ 75)
18.4 (2.0)
0.6
85.0
Never (n ¼ 65)
22.6 (2.7)
0.5
98.9
0.41
22.7
<0.001
No. of gas appliances
0: No gas cooking or
gas fire (n ¼ 103)
6.10 (0.5)
1: Gas cooking or
gas fire (n ¼ 213)
16.7 (1.0)
0.46
98.9
2: Gas cooking and
gas fire (n ¼ 82)
25.7 (2.1)
0.90
82.0
0 (n ¼ 99)
14.4 (1.8)
0.50
98.9
1 (n ¼ 178)
15.4 (0.9)
0.57
71.4
2 (n ¼ 117)
17.4 (1.6)
0.41
86.3
<0.001
No. of parental smokers
at home
0.341
Season
Summer (n ¼ 69)
10.1 (0.9)
1.56
40.1
Spring (n ¼ 152)
13.2 (1.1)
0.41
86.3
Autumn (n ¼ 98)
17.5 (1.6)
0.6
82.0
Winter (n ¼ 79)
23.8 (2.3)
0.5
98.9
<0.001
Home location, %
Nonurban (n ¼ 130)
17.1 (1.4)
0.46
98.9
Urban (n ¼ 268)
15.2 (0.9)
0.41
86.3
0.249
Abbreviations: NO2, nitrogen dioxide; SE, standard error.
(20.6%) had 2 gas appliances. GSTP1 allelic frequencies did
not differ between participant and nonparticipant preschoolers. The GSTP1 genotype prevalences of the polymorphic Ile105Val were 43.6% for Ile/Ile, 47.5% for Ile/
Val, and 8.9% for Val/Val.
Concentrations of indoor nitrogen dioxide were, as expected, higher in homes in which gas was used for cooking,
when an extractor fan was not used, and in which individual
gas fires were used (Table 2). Indoor nitrogen dioxide concentrations significantly increased with increasing number
of gas appliances. Although indoor nitrogen dioxide concentration increased with number of parental smokers at
home, no statistically significant association was found.
Seasonal variations were observed, with winter measurement levels being higher; however, no significant differences were found by home location.
Predicted indoor nitrogen dioxide concentration increased significantly with low social class and low level of
Am J Epidemiol 2009;169:1327–1336
Indoor Air Pollution and Neurodevelopment
1331
Table 3. Distribution of Child and Maternal Variables in Different Quartiles of Indoor Predicted
NO2 Concentration, Menorca Cohort, Spain, 1997–1999
NO2 Concentration (ppb)
<10
(n 5 90)
10–14.9
(n 5 101)
15–21
(n 5 102)
>21
(n 5 103)
P Value
Child variables
Female sex, %
52.2
48.5
52.9
45.6
Gestational age <37 weeks, %
6.7
4.9
2.9
2.9
0.71
0.52
Birth weight, g, mean
3,205
3,240
3,160
3,200
0.72
Birth height, cm, mean
49.2
49.0
48.9
49.0
0.88
Breastfeeding, yes, %
91.1
81.2
81.4
80.6
0.16
0.24
Breastfeeding, weeks, %
<2
11.1
21.8
20.6
23.3
2–15.9
27.8
18.8
27.5
28.2
16–27.9
28.9
35.6
24.5
27.2
28
32.2
23.8
27.5
21.4
Nonmanual
44.9
36.4
33.7
29.3
Manual
55.1
63.6
66.3
70.7
54.6
45.4
36.0
35.0
Maternal variables
Maternal social class, %
0.15
Education, %
Secondary or higher
Primary or less
0.02
45.4
54.6
64.0
64.9
Smoking during pregnancy, yes, %
15.6
18.8
18.6
25.2
0.38
Alcohol consumption during
pregnancy, yes, %
16.7
22.8
20.6
27.2
0.35
Abbreviation: NO2, nitrogen dioxide.
education. Sex, gestational age, birth weight and height,
and breastfeeding of the child as well as maternal smoking
and maternal alcohol consumption during pregnancy were
not associated with indoor nitrogen dioxide concentration
(Table 3).
Exposure to gas appliances at home was inversely associated with general cognitive, memory, verbal, and executive function scores, with an increasing relation between
number of gas appliances and cognitive scores (Table 4).
Children exposed to gas appliances were at higher risk of
ADHD symptoms. The 2 components of ADHD were analyzed separately, and results showed that exposure to gas
appliances was associated with inattention but not with hyperactivity. Adjustment by parental social class, parental
educational level, breastfeeding, maternal alcohol consumption and tobacco smoking during pregnancy, and number of
smokers at home did not confound these associations.
We observed a linear relation between predicted nitrogen
dioxide concentration and the different cognitive outcomes
of interest (P gain > 0.30), indicating a linear relation with
neurodevelopment outcomes at age 4 years and indoor nitrogen dioxide concentration (Figure 1). A statistically significant negative dose-response effect was found between
indoor nitrogen dioxide levels and cognitive outcomes at
age 4 years, including general cognitive function, verbal
function, and executive function (Table 4). Moreover, chilAm J Epidemiol 2009;169:1327–1336
dren exposed to higher levels of indoor nitrogen dioxide
were at higher risk of developing ADHD symptoms (odds
ratio ¼ 1.04, 95% confidence interval: 1.00, 1.09;
P ¼ 0.04). The negative effect was restricted to the inattention
component (odds ratio ¼ 1.06, 95% confidence interval:
1.01, 1.12; P ¼ 0.024). Similar results were obtained for
children in the highest quartile of indoor nitrogen dioxide:
ADHD symptoms (odds ratio ¼ 2.52, 95% confidence interval: 0.96, 6.64; P ¼ 0.06) and inattention (odds ratio ¼
2.88, 95% confidence interval: 0.91, 9.11; P ¼ 0.07).
To evaluate residual confounding by socioeconomic factors and parental characteristics, a sensitivity analysis was
carried out. Exposure to gas appliances at home was also
associated with a decrease (in points) in children’s global
cognitive score at age 4 years for children of a high social
class (b ¼ 8.73, 95% confidence interval: 14.04, 3.41)
and from families with a high educational level (b ¼
10.92, 95% confidence interval: 25.11, 3.26) as well as
for children not exposed to active maternal smoking during
pregnancy (b ¼ 5.21, 95% confidence interval: 10.03,
0.38). The results were essentially the same for indoor
nitrogen dioxide exposure.
Adverse effects of exposure to gas appliances and indoor
nitrogen dioxide on cognitive functioning and ADHD symptoms were found in children with any GSTP1 Val-105 allele
but not in children with the Ile/Ile genotype (Table 5).
1332 Morales et al.
Table 4. Adjusted Associations of Cognitive Functioning and ADHD Symptoms at Age 4 Years With Gas
Appliances at Homea and Indoor Predicted NO2 Concentration (ppb), Menorca Cohort, Spain, 1997–1999
(N ¼ 398)b
Neuropsychological
Outcomes
General cognitive function
(b coefficient)
Gas Stove or Gas Fire
(n 5 213 children)
Gas Stove and Gas Fire
(n 5 82 children)
Estimate
95% CI
Estimate
1.43
5.29, 2.42
5.10
NO2 Concentration
(Increase per 1 ppb)
95% CI
Estimate
9.92, 0.28
0.27
95% CI
0.48, 0.07
Perceptual-performance
0.46
4.26, 3.33
1.51
6.25, 3.24
0.13
0.33, 0.08
Memory
0.20
4.15, 3.75
4.22
9.16, 0.71
0.18
0.39, 0.04
Quantitative
0.48
2.48, 1.53
0.91
2.74, 0.92
0.12
0.25, 0.01
Verbal
1.86
5.84, 2.11
6.16
11.12, 1.19
0.29
0.51, 0.08
Motor
1.40
5.13, 2.33
1.59
6.25, 3.07
0.13
0.33, 0.08
Executive function
2.40
6.32, 1.53
5.34
10.25, 0.43
0.30
0.52, 0.09
ADHD symptoms (odds ratio)
1.41
0.59, 3.34
2.72
1.01, 7.28
1.04
Inattention
1.42
0.49, 4.12
3.59
1.14, 11.33
1.06
1.00, 1.09
1.01, 1.12
Hyperactivity
1.57
0.49, 5.03
1.30
0.29, 5.83
1.01
0.95, 1.07
Abbreviations: ADHD, attention deficit hyperactivity disorder; CI, confidence interval; NO2, nitrogen dioxide.
Reference category: no gas stove or gas fire (n ¼ 103).
b
Each row represents a different model adjusted for maternal social class, maternal education, school trimester at
testing, evaluator (neuropsychologist), maternal smoking during pregnancy, number of smokers at home, maternal
alcohol consumption during pregnancy, and home location. ADHD symptoms are further adjusted for sex.
a
Exposure to gas appliances was associated with a decrease
of 9.72 (95% confidence interval: 17.13, 2.30) points in
general cognitive score and with an increased risk of inattention for children with any GSTP1 Val-105 allele. In addition, indoor nitrogen dioxide exposure was negatively
associated with general cognitive (P for interaction ¼
0.04), verbal (P for interaction ¼ 0.30), and executive
(P for interaction ¼ 0.13) function scores. An increased risk
of developing inattention symptoms was found for children
with any GSTP1 Val-105 allele (P for interaction ¼ 0.26).
DISCUSSION
We found that early-life exposure to household gas appliances is associated negatively with general cognitive functioning and with a higher risk of developing ADHD
symptoms at age 4 years. A negative dose-response effect
was found even after adjustment for social class, educational
level, and tobacco smoke and alcohol exposures. A higher
effect was found for some cognitive subareas, specifically
verbal and executive function, and for the inattention component of ADHD. Moreover, we found a decrease in cognitive
functioning and an increased risk of developing inattention
symptoms in children exposed to higher levels of indoor
nitrogen dioxide. Children with the GSTP1 Val-105 allele
were at higher risk of the adverse effects of gas appliances
and indoor nitrogen dioxide exposure on cognitive function
and ADHD symptoms. A statistically significant interaction
between indoor nitrogen dioxide concentration and GSTP1
polymorphisms was found for general cognitive functioning.
Several biologic mechanisms could explain the negative
association between exposure to gas appliances at home and
neuropsychological outcomes in preschoolers. Gas applian-
ces produce complex mixtures of volatile organic compounds, sulfur dioxide, water vapor, particulates, carbon
dioxide, and oxides of nitrogen (8). It has been shown that
urban particulate materials induce proinflammatory cytokines
in human bronchial epithelial cells (23), lung epithelial cells
(24), and macrophages (25). The interaction of macrophages
with epithelial cells amplifies cytokine production in those
cells, and these cytokines are also present in the blood of
subjects during an episode of acute atmospheric air pollution.
Furthermore, there is evidence that oxidative stress and induced inflammation translate systemically (26, 27). We postulate that an inflammatory systemic response associated with
indoor pollution derived from use of gas appliances might
adversely affect child neurodevelopment.
Inhalation of nitrogen dioxide can induce structural and
functional changes in lungs; alteration of the body’s defense
system; and abnormal biochemical, physiologic, and enzymatic reactions (28). Epidemiologic studies have shown a relation between indoor nitrogen dioxide exposure and
respiratory and allergic diseases (29, 30). The inflammatory
airway response to nitrogen dioxide was not assessed in our
study. However, it is well established that nitrogen dioxide–
mediated lung injury may be causally related to generation
of higher levels of proinflammatory mediators including
nitric oxide and other cytokines, such as interleukin-8,
interleukin-1b, and tumor necrosis factor-a. In addition,
early changes in bronchial cellular events may lead to a cascade of more widespread systemic inflammatory responses
of the lung, which may cause further damage to the surrounding cells and tissue and even may have systemic
effects in other organs including the heart and brain.
Information about neurologic effects of nitrogen dioxide
exposure is limited. In guinea pigs, nitrogen dioxide
Am J Epidemiol 2009;169:1327–1336
Indoor Air Pollution and Neurodevelopment
Prediction of General Cognitive
Function
A)
5
0
-5
-10
-15
0
5
10
15
20
25
30
35
Indoor NO2, ppb
Prediction of Verbal Skill
B)
5
0
-5
-10
-15
0
5
10
15
20
25
30
35
Indoor NO2, ppb
Prediction of Executive Function
C)
5
0
-5
-10
-15
0
5
10
15
20
25
30
35
Indoor NO2, ppb
Figure 1. Relation (and 95% confidence interval) between predicted indoor nitrogen dioxide (NO2) concentration and cognitive function scores adjusted for maternal social class, maternal education,
school trimester at testing, evaluator (neuropsychologist), maternal
smoking during pregnancy, number of smokers at home, maternal
consumption of alcohol during pregnancy, and home location. Adjusted general additive models for A) general cognitive, B) verbal,
and C) executive function, Menorca cohort, Spain, 1997–1999. The
þ symbols on the x-axes indicate NO2 observations.
Am J Epidemiol 2009;169:1327–1336
1333
intoxication (exposure to 5 ppm and 10 ppm for 2 hours
daily for 5 weeks) has been reported to induce significant
depletion of the total lipids and phospholipids content as
well as the cholesterol diminution content of the cerebral
hemisphere, cerebellum, and midbrain (31). Interestingly,
the rate of lipid peroxidation was found to be enhanced in
all brain areas investigated, and, with increasing dosage, the
effect was correspondingly pronounced (31). It is noteworthy that lipids are essential components of all cellular elements of the central nervous system. Biomembranes and
subcellular organelles are the major sites of the lipid peroxidation damage. The peroxidation changes triggered by free
radicals in the brain lipids may be of importance in the
development of brain cell damage.
A strong, adverse effect of indoor nitrogen dioxide exposure on cognitive function was observed regarding executive
function (i.e., cognitive tasks performed by a predominant
activation of the prefrontal cortex). This neurologic area,
including its neural circuitry, is innervated by the monoamines dopamine, noradrenaline, and serotonin. Alterations
in brain concentrations of catecholamines have been reported in relation to other pollutant gases. Experimental
studies have found that prenatal ozone exposure disrupts
the cerebellar catecholamine system, showing a decrease
in dopamine and noradrenaline mainly in the early postnatal
period (32). When ozone is inhaled, it produces reactive
oxygen species that can reach the central nervous system
through the bloodstream, producing oxidative stress. Oxidative stress caused by ozone exposure increases lipid peroxidation in different brain structures. Dopamine has an
oxidative metabolism that makes dopaminergic neurons
and their fibers especially sensitive to oxidative stress.
Tyrosine hydroxylase is the initial and rate-limiting enzyme
in the biosynthesis of the neurotransmitter dopamine. Tyrosine hydroxylase is inhibited and nitrated at tyrosine residues
in vitro by nitrogen dioxide and in vivo by drugs that damage
dopamine neurons (33). In addition, the dopaminergic system
plays a role in the pathogenesis of ADHD (34). Nitrogen
dioxide could probably impair dopaminergic neuron functioning by inducing oxidative stress and by inhibiting dopamine biosynthesis, thus resulting in a poor executive function
score and development of ADHD symptoms.
We found a stronger adverse effect of indoor nitrogen
dioxide concentration in children with any GSTP1 Val-105
allele. GSTP1 is involved in antioxidant defenses to protect
against oxidative stress (11), detoxifying lipid peroxidation
products and DNA oxidation products, and it represents the
most strongly expressed glutathione S-transferase isoenzymes in the human brain during early life (12). The GSTP1
Ile105Val substitution is located near the substrate binding
site of the enzyme, resulting in a less active enzyme (35).
Our results suggest that, for those children with the less
active GSTP1 Val-105 variant, brain cells are more susceptible to biochemical changes induced by early-life exposure
to nitrogen dioxide.
Our study has several strengths, including a populationbased birth cohort with scales individually administered to
parents and teachers. Additionally, we validated the McCarthy
Scales of Children’s Abilities new scale (executive function) and applied a strict protocol to reduce interobserver
1334 Morales et al.
Table 5. Adjusted Associations of Cognitive Functioning and ADHD Symptoms With Gas Appliancesa and Indoor
Predicted NO2 Concentration by GSTP1 Genotype, Menorca Cohort, Spain, 1997–1999b
Gas Stove or
Gas Fire
Estimate
95% CI
Gas Stove and
Gas Fire
Estimate
NO2 Concentration
(Increase Per 1 ppb)
95% CI
Estimate
95% CI
General cognitive function
(b coefficient)
Ile/Ile (n ¼ 152)
2.08
4.06, 8.23
2.06
3.12
8.50, 2.26
9.72
17.13, 2.30 0.55*
0.86, 0.25
Ile/Ile (n ¼ 152)
0.15
6.49, 6.78
2.79
10.62, 5.04
0.40, 0.30
Ile/Val or Val/Val (n ¼ 198)
2.08
7.56, 3.40
7.72
15.30, 0.14 0.46
Ile/Val or Val/Val (n ¼ 198)
5.19, 9.30
0.10
0.22, 0.42
Verbal skills (b coefficient)
0.05
0.78, 0.15
Executive function (b coefficient)
Ile/Ile (n ¼ 152)
1.84
4.52, 8.19
0.66
4.83
10.23, 0.57
9.61
Ile/Ile (n ¼ 138)
0.98
0.17, 5.68
0.84
0.12, 5.88
0.98
0.90, 1.07
Ile/Val or Val/Val (n ¼ 181)
2.01
0.67, 6.05
6.84
1.69, 27.68
1.09
1.02, 1.16
Ile/Val or Val/Val (n ¼ 198)
6.83, 8.15
0.01**
17.08, 2.14 0.56
0.34, 0.32
0.86, 0.25
ADHD symptoms (odds ratio)
Inattention symptoms (odds ratio)
Ile/Ile (n ¼ 138)
0.84
0.09, 7.66
0.90
0.09, 9.09
0.98***
0.88, 1.09
Ile/Val or Val/Val (n ¼ 181)
1.82
0.47, 7.05
10.29
1.93, 54.87
1.11
1.03, 1.20
Abbreviations: ADHD, attention deficit hyperactivity disorder; CI, confidence interval; GSTP1, glutathione S-transferase
P1 gene; NO2, nitrogen dioxide.
*P for interaction ¼ 0.04; **P for interaction ¼ 0.13; ***P for interaction ¼ 0.26.
a
Reference category: no gas stove or gas fire (n ¼ 103).
b
Each row represents a different model adjusted for maternal social class, maternal education, school trimester at
testing, evaluator (neuropsychologist), maternal smoking during pregnancy, number of smokers at home, maternal
alcohol consumption during pregnancy, and home location. ADHD symptoms are further adjusted for sex.
variability during testing, increasing robustness of the psychometric measures. Thus, we found similar effects in
related areas measured with 2 different methods (the executive function from the McCarthy test performed by
a neuropsychologist and the inattention symptoms from
the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, test reported by teachers). Moreover,
measurement of exposure was obtained from the model
of a long-term mean based on variables regressed against
2-week nitrogen dioxide measurements collected at each
participant’s home, and the neurodevelopmental areas affected were the same as those found to be related to gas
appliances when measured through questionnaire. Finally,
the gene-environment interaction we found reinforces the
plausibility of neurotoxic effects of nitrogen dioxide.
Children included in the study did not differ from nonparticipants regarding gas appliances and indoor nitrogen
dioxide exposure. Higher cognitive scores and social class
of participants in comparison with nonparticipants, if anything, would have diminished the strength of our findings.
Moreover, data on individual exposure to household gas
appliances, nitrogen dioxide measures, and the presence of
GSTP1 polymorphisms were obtained without knowledge
of neuropsychological outcomes. Thus, differential information bias is unlikely to have affected these associations.
However, our study has some limitations. The small numbers of subjects in each subgroup limit the conclusions
about interactions between indoor nitrogen dioxide concentrations and GSTP1 genotypes in a single study, although
control of missing observations by using the method of
multiple imputations did not change the results. We included a large number of covariates, but we were unable
to control for parental IQ, mental health, or problematic
family functioning, which might be linked to assessed exposures. Moreover, it is still possible that the associations
we found in this study with nitrogen dioxide could be
attributable to unmeasured indoor air contaminants, such as
particulates derived from gas appliances. This is supported
by the fact that, in Menorca, the principal source of gas
appliances is butane bottles, a fuel more toxic than natural
gas (36). In addition, information about the contribution of
outdoor nitrogen dioxide to indoor nitrogen dioxide levels
was lacking. However, Menorca is an island with little industry and traffic, and the association with nitrogen dioxide
was invariable after we adjusted for urban home location. In
fact, an increase in hours of ventilation in our study population was associated with a statistically significant decrease
in levels of measured indoor nitrogen dioxide (20), indicating that nitrogen dioxide in our population mostly originated
from indoor sources. Furthermore, adjustment for organochlorinated compounds and methylmercury, as well as maternal smoking or alcohol consumption during pregnancy,
did not confound the observed associations (data not
shown). Finally, exposure assessment was limited to early
Am J Epidemiol 2009;169:1327–1336
Indoor Air Pollution and Neurodevelopment
life only, and conclusions could not be extended to further
time periods of exposure, although we expected the larger
effect during early life, when children spend most of their
time at home and critical brain growth occurs. To our knowledge, this study is the first to analyze the effect of indoor air
pollution on cognitive functioning and ADHD symptoms in
preschoolers and the influence of glutathione S-transferase
polymorphisms. We cannot exclude the possibility that findings may have been confounded by other genetic variants in
linkage disequilibrium with the variant under study or by
population stratification. However, GSTP1 is the most
strongly expressed of the glutathione S-transferase isoenzymes in the human brain, its expression is observed as early
as 12 weeks of gestation, and it is highly expressed in the
blood-brain barrier, which supports the hypothesis that
GSTP1 may be the susceptibility gene rather than a linkage
disequilibrium marker. Moreover, the role of maternal genotypes in modifying the adverse effects of early-life exposure
to household gas appliances and indoor nitrogen dioxide, as
well as maternal-fetal gene interaction, should be assessed
in future studies.
In conclusion, current data provide preliminary evidence
that early-life exposure to indoor air pollution from gas
appliances may be related to impaired cognitive functioning
among preschoolers and may increase their risk of developing ADHD symptoms. Children with the GSTP1 Val-105
allele appear more susceptible to adverse neuropsychological effects of indoor air pollution from gas appliances, suggesting oxidative stress as a potential mechanism. In view of
widespread use of gas appliances, confirmation of the present
results in coming studies could have important public health
implications.
ACKNOWLEDGMENTS
Author affiliations: Center for Research in Environmental
Epidemiology (CREAL), Barcelona, Catalonia, Spain (Eva
Morales, Jordi Julvez, Mònica Guxens, Nino Künzli, Jordi
Sunyer); Municipal Institute of Medical Research (IMIMHospital del Mar), Barcelona, Catalonia, Spain (Eva
Morales, Jordi Julvez, Mònica Guxens, Nino Künzli, Jordi
Sunyer); Preventive Medicine and Public Health Educational Unit IMAS-UPF-ASPB, Barcelona, Catalonia, Spain
(Eva Morales); Area de Salud de Menorca, IB-SALUT,
Menorca, Spain (Maties Torrent); Genes and Disease Program,
Centre for Genomic Regulation (CRG), Barcelona, Catalonia, Spain (Rafael de Cid, Mariona Bustamante); Centro
Nacional de Genotipado (CeGen), Barcelona, Catalonia,
Spain (Rafael de Cid, Mariona Bustamante); CIBER Epidemiologı́a y Salud Pública (CIBERESP), Barcelona, Catalonia, Spain (Eva Morales, Mònica Guxens, Mariona
Bustamante, Nino Künzli, Jordi Sunyer); and Department
of Experimental Sciences, Universitat Pompeu Fabra
(UPF), Barcelona, Catalonia, Spain (Jordi Sunyer).
This work was supported by grants from the Spanish
Ministry of Health ((FIS-PI041436, FIS-PI041705, and FISPI051187), Instituto de Salud Carlos III (Red INMA G03/
176 and CB06/02/0041), and CIBER en Epidemiologı́a y
Am J Epidemiol 2009;169:1327–1336
1335
Salud Pública (CIBERESP)); the Generalitat de CatalunyaCIRIT (1999SGR 00241); and Genome Spain.
The authors acknowledge all teachers and parents of the
children from the island of Menorca for patiently responding to the questionnaires; the psychologists who coordinated
the fieldwork; and the nurses and administrative personnel
from the Primary Health Care Center of Maó, who provided
administrative, technical, and material support.
Conflict of interest: none declared.
REFERENCES
1. Lisabeth LD, Escobar JD, Dvonch JT, et al. Ambient air pollution and risk for ischemic stroke and transient ischemic
attack. Ann Neurol. 2008;64(1):53–59.
2. Migliore L, Coppedè F. Environmental-induced oxidative
stress in neurodegenerative disorders and aging. Mutat Res.
2009;674(1–2):73–84.
3. Perera FP, Rauh V, Whyatt RM, et al. Effect of prenatal exposure to airborne polycyclic aromatic hydrocarbons on neurodevelopment in the first 3 years of life among inner-city
children. Environ Health Perspect. 2006;114(8):1287–1292.
4. Suglia SF, Gryparis A, Wright RO, et al. Association of black
carbon with cognition among children in a prospective birth
cohort study. Am J Epidemiol. 2008;167(3):280–286.
5. Banerjee TD, Middleton F, Faraone SV. Environmental risk
factors for attention-deficit hyperactivity disorder. Acta
Paediatr. 2007;96(9):1269–1274.
6. Smith KR, Mehta S, Fuez M. Indoor air pollution from
household use of solid fuels. In: Ezzati M, Rodgers AD,
Lopez AD, et al, eds. Comparative Quantification of Health
Risks: Global and Regional Burden of Disease Due to Selected
Major Risk Factors. Vol 2. Geneva, Switzerland: World Health
Organization; 2003.
7. World Health Organization. The World Health Report 2002:
Reducing Risks, Promoting Healthy Life. Geneva, Switzerland: World Health Organization; 2002. (http://whqlibdoc.
who.int/hq/2002/WHO_WHR_02.1.pdf). (Accessed February
15, 2008).
8. Chauhan AJ. Gas cooking appliances and indoor pollution.
Clin Exp Allergy. 1999;29(8):1009–1013.
9. Pilotto LS, Douglas RM, Attewell RG, et al. Respiratory effects associated with indoor nitrogen dioxide exposure in
children. Int J Epidemiol. 1997;26(4):788–796.
10. Mustafa MG. Health effects and toxicology of ozone and nitrogen dioxide. In: Nrigau JO, Simmons MS, eds. Environmental Oxidants. New York, NY: John Wiley; 1994:351–404.
11. Hayes JD, Strange RC. Potential contribution of the glutathione S-transferase supergene family to resistance to oxidative
stress. Free Radic Res. 1995;22(3):193–207.
12. Raijmakers MT, Steegers EA, Peters WH. Glutathione
S-transferases and thiol concentrations in embryonic and early
fetal tissues. Hum Reprod. 2001;16(11):2445–2450.
13. Polk S, Sunyer J, Muñoz-Ortiz L, et al. A prospective study of
Fel d1 and Der p1 exposure in infancy and childhood wheezing. Am J Respir Crit Care Med. 2004;170(3):273–278.
14. Julvez J, Ribas-Fitó N, Torrent M, et al. Maternal smoking
habits and cognitive development of children at age 4 years in
a population-based birth cohort. Int J Epidemiol. 2007;
36(4):825–832.
15. Julvez J, Ribas-Fitó N, Forns M, et al. Attention behaviour and
hyperactivity at age 4 and duration of breast-feeding. Acta
Paediatr. 2007;96(6):842–847.
1336 Morales et al.
16. Ribas-Fitó N, Torrent M, Carrizo D, et al. Exposure to hexachlorobenzene during pregnancy and children’s social behavior at 4 years of age. Environ Health Perspect. 2007;115(3):
447–450.
17. McCarthy D. Manual for the McCarthy Scales of Children’s
Abilities. New York, NY: Psychological Corporation; 1972.
(Spanish adaptation: Madrid, Spain: TEA Ediciones SA;
1996).
18. American Psychiatric Association. Diagnostic and Statistical
Manual of Mental Disorders: DSM-IV-TR. 4th ed. Washington, DC: American Psychiatric Association; 1994.
19. Amador-Campos JA, Forns M, Guardia J, et al. DSM-IV.
Attention deficit hyperactivity disorder symptoms: agreement
between informants in prevalence and factor structure at different ages. J Psychopathol Behav Assess. 2006;28(1):
23–32.
20. Garcı́a Algar O, Pichini S, Basagaña X, et al. Concentrations
and determinants of NO2 in homes of Ashford, UK and Barcelona and Menorca, Spain. Indoor Air. 2004;14(4):298–304.
21. Yanagisawa Y, Nishimura H. A badge-type personal sampler
for measurement of personal exposure to NO2 and NO in
ambient air. Environ Int. 1982;8(1–6):235–242.
22. Hastie T, Tibshirani R. Generalized Additive Models. New
York, NY: Chapman & Hall; 1990.
23. Fujii T, Hayashi S, Hogg JC, et al. Particulate matter induces
cytokine expression in human bronchial epithelial cells. Am J
Respir Cell Mol Biol. 2001;25(3):265–271.
24. Jiménez LA, Drost EM, Gilmour PS, et al. PM(10)-exposed
macrophages stimulate a proinflammatory response in lung
epithelial cells via TNF-alpha. Am J Physiol Lung Cell Mol
Physiol. 2002;282(2):L237–L248.
25. van Eeden SF, Tan WC, Suwa T, et al. Cytokines involved in
the systemic inflammatory response induced by exposure to
particulate matter air pollutants (PM(10)). Am J Respir Crit
Care Med. 2001;164(5):826–830.
26. Tao F, Kobzik L. Lung macrophage-epithelial cell interactions
amplify particle-mediated cytokine release. Am J Respir Cell
Mol Biol. 2002;26(4):499–505.
27. Hirano S, Furuyama A, Koike E, et al. Oxidative-stress
potency of organic extracts of diesel exhaust and urban fine
particles in rat heart microvessel endothelial cells. Toxicology.
2003;187(2–3):161–170.
28. Dawson SV, Schenker MB. Health effects of inhalation of
ambient concentrations of nitrogen dioxide. Am Rev Respir
Dis. 1979;120(2):281–292.
29. Belanger K, Gent JF, Triche EW, et al. Association of indoor
nitrogen dioxide exposure with respiratory symptoms in children with asthma. Am J Respir Crit Care Med. 2006;173(3):
297–303.
30. Kattan M, Gergen PJ, Eggleston P, et al. Health effects of
indoor nitrogen dioxide and passive smoking on urban asthmatic children. J Allergy Clin Immunol. 2007;120(3):618–624.
31. Farahani H, Hasan M. Effect of NO2 on lipids and lipid peroxidation in the CNS of the guinea-pig. Pharmacol Toxicol.
1990;66(2):146–149.
32. Gonzalez-Pina R, Escalante-Membrillo C, Alfaro-Rodriguez A,
et al. Prenatal exposure to ozone disrupts cerebellar
monoamine contents in newborn rats. Neurochem Res.
2008;33(5):912–918.
33. Kuhn DM, Geddes TJ. Tetrahydrobiopterin prevents nitration
of tyrosine hydroxylase by peroxynitrite and nitrogen dioxide.
Mol Pharmacol. 2003;64(4):946–953.
34. Biederman J, Faraone SV. Attention-deficit hyperactivity disorder. Lancet. 2005;366(9481):237–248.
35. Strange RC, Fryer AA. The glutathione S-transferases: influence of polymorphism on cancer susceptibility. IARC Sci Publ.
1999;(148):231–249.
36. Zhang Y, Chen BS, Liu GQ, et al. Natural gas and indoor air
pollution: a comparison with coal gas and liquefied petroleum
gas. Biomed Environ Sci. 2003;16(3):227–236.
Am J Epidemiol 2009;169:1327–1336