Indoor Air Pollution From Gas Cooking and Infant Neurodevelopment

ORIGINAL ARTICLE Indoor Air Pollution From Gas Cooking and Infant Neurodevelopment Martine Vrijheid,a,b,c David Martinez,a,b,c Inma Aguilera,a,b,c Mariona Bustamante,c,d Ferran Ballester,c,e,f Marisa Estarlich,c,e,f Ana Fernandez-Somoano,c,g Mònica Guxens,a,b,c Nerea Lertxundi,h,i M. Dolores Martinez,j Adonina Tardon,c,g and Jordi Sunyera,b,c,k; on behalf of the INMA Project Background: Gas cooking is a main source of indoor air pollutants, including nitrogen dioxide and particles. Because concerns are emerging for neurodevelopmental effects of air pollutants, we examined the relationship between indoor gas cooking during pregnancy and infant neurodevelopment. Methods: Pregnant mothers were recruited between 2004 and 2008 to a prospective birth cohort study (INfancia y Medio Ambiente) in Spain during the first trimester of pregnancy. Third-trimester questionnaires collected information about the use of gas appliances at home. At age 11 to 22 months, children were assessed for mental development using the Bayley Scales of Infant Development. Linear regression models examined the association of gas cooking and standardized mental development scores (n ⫽ 1887 mother– child pairs). Submitted 1 April 2011; accepted 19 July 2011. From the aCenter for Research in Environmental Epidemiology (CREAL), Barcelona, Spain; bHospital del Mar Research Institute (IMIM), Barcelona, Spain; cCIBER Epidemiología y Salud Pública (CIBERESP), Spain; dGenes and Disease Program, Centre for Genomic Regulation (CRG), Barcelona, Spain; eEnviroment and Health Area, Center for Public Health Research (CSISP), Valencia, Spain; fDepartment of Nursing, University of Valencia, Valencia, Spain; gPreventive Medicine and Public Health, University of Oviedo, Asturias, Spain; hFaculty of Psychology, University of the Basque Country (EHU-UPV), San Sebastián, Spain; i Health Research Institute (BIODONOSTIA), San Sebastián, Spain; jDirección de Calidad Ambiental del Departamento de Medio Ambiente del Gobierno Vasco, Donostia-San Sebastian, Spain; and kFaculty of Health and Life Sciences, Pompeu Fabra University, Barcelona, Spain. This study was funded by grants from Spanish Ministry of Health and Instituto de Salud Carlos III (Red INMA G03/176, CB06/02/0041, FIS-PS09/00090, FIS-FEDER 03/1615, FIS-PI04/1436, 04/1509, 04/ 1112, 04/1931, 04/2018, 05/1079, 05/1052, 06/1213, 06/0867, 07/0314, 08/1151, 09/02311, 09/02647, 09/02311), Generalitat de CatalunyaCIRIT 1999SGR 00241, Conselleria de Sanitat Generalitat Valenciana, Universidad de Oviedo, Obra social Cajastur, Department of Health of the Basque Government (2005111093 and 2009111069), the Provincial Government of Gipuzkoa (DFG06/004 and DFG08/001), and Fundación Roger Torné. The authors have no conflicts of interest to declare. Supplemental digital content is available through direct URL citations in the HTML and PDF versions of this article (www.epidem.com). Editors’ note: A commentary on this article appears on page ●●●. Correspondence: Martine Vrijheid, Centre for Research in Environmental Epidemiology, Doctor Aiguader, 88, 08003 Barcelona, Spain. E-mail: mvrijheid@creal.cat. Copyright © 2011 by Lippincott Williams & Wilkins ISSN: 1044-3983/12/2301-0001 DOI: 10.1097/EDE.0b013e31823a4023 Epidemiology • Volume 23, Number 1, January 2012 Results: Gas cookers were present in 44% of homes. Gas cooking was related to a small decrease in the mental development score compared with use of other cookers (⫺2.5 points 关95% confidence interval ⫽ ⫺4.0 to ⫺0.9兴) independent of social class, maternal education, and other measured potential confounders. This decrease was strongest in children tested after the age of 14 months (⫺3.1 points 关⫺5.1 to ⫺1.1兴) and when gas cooking was combined with less frequent use of an extractor fan. The negative association with gas cooking was relatively consistent across strata defined by social class, education, and other covariates. Conclusions: This study suggests a small adverse effect of indoor air pollution from gas cookers on the mental development of young children. (Epidemiology 2012;23: 000 – 000) T he presence of gas cookers— or gas stoves—inside the home is common in developed countries (50% to 70%) and has long been recognized as a main source of indoor air pollution.1 Gas cooking produces a complex mixture of volatile organic compounds, sulfur dioxide, particulates, carbon monoxide, carbon dioxide, nitric oxide, and nitrogen dioxide.2 Of these, nitrogen dioxide (NO2) is the most extensively studied indoor air pollutant.3 Gas cooking is a main predictor of indoor NO2 concentrations in homes in developed countries, together with cigarette smoking and outdoor traffic-related NO2.4 – 8 Homes with gas appliances can have NO2 concentrations twice as high as other homes; concentrations may exceed the WHO guideline for average annual outdoor NO2 of 40 ␮g/m3.1,3 Gas combustion also produces particulate matter smaller than 10 ␮m (PM10)9 and ultrafine particles in the nm range.10 Women and young children are especially exposed because they spend a larger part of their day at home and in the kitchen; they may experience more air pollution from indoor than from outdoor sources. Air pollution may impair neurodevelopment.11 Air pollution has been implicated as a chronic source of neuroinflammation and oxidative stress that can produce neuropathology and central nervous system diseases.12 Because maturation of the brain cortex is intensive in the first few www.epidem.com | 1 Epidemiology • Volume 23, Number 1, January 2012 Vrijheid et al years of life, this period of neurodevelopment may be particularly vulnerable to environmental pollutants.13 Recent studies have observed adverse cognitive and behavioral effects of perinatal outdoor air pollution,14 –20 and these studies have raised concerns about similar effects from indoor air pollution. A small birth-cohort study in Menorca, Spain21 was the first to report a negative association of exposure to gas appliances and indoor NO2 with general cognition and inattention symptoms in 4-year-old children. These effects were modified by the Ile105Val polymorphism in the detoxification gene glutathione-S-transferase P1 (GSTP1). Other recent research has highlighted the potential role of antioxidant and anti-inflammatory nutrient intake as modifiers of harmful effects of outdoor air pollution.22–25 Thus, detoxification and antioxidant factors also require examination in indoor air pollution studies. We examined the relationship between exposure to gas cookers during pregnancy and mental development of children aged 1 to 2 years in a large Spanish birth cohort study and the role of potential modifying factors. METHODS The population-based birth cohort study INMA (INfancia y Medio Ambiente-Environment and Childhood) recruited pregnant women from 4 Spanish regions (Asturias; Gipuzkoa, Basque-Country; Sabadell, Catalonia; and Valencia) between 2004 and 2008.26 Women were enrolled during the first trimester of pregnancy at the primary health care center or hospital (depending on the region) if they fulfilled the following inclusion criteria: age at least 16 years, intention to give birth in the reference hospital, no problems in communication, singleton pregnancy, and no assisted conception.26 The study was approved by the Hospital Ethics Committees in the participating regions; participants signed informedconsent forms. Information was collected by questionnaires in the first and third trimesters of pregnancy, and when the child was approximately 6 months and 14 months old. Questions included maternal and paternal education, social class (coding maternal and paternal occupation according to the International Standard Classification of Occupations-88 system27), country of birth, maternal health and obstetric history, parity, medication use, alcohol consumption, and active and passive cigarette smoking. Breastfeeding practices were reported at the last 2 time points. Maternal diet during the pregnancy was assessed through a validated food frequency questionnaire in trimesters one and three.28 Participation rates at each follow-up are reported elsewhere in detail26; 54% of eligible women agreed to participate in the study, and of those, 90% were still being followed up when the child was 14 months of age. Women from higher educational levels were somewhat more likely to participate and to continue participation.26 In one region (Sabadell), we analyzed data on the child’s GSTP1 Ile105Val polymorphism, which had been 2 | www.epidem.com found in the previous study in Menorca, to modify the effect of gas cooking.21 DNA extraction from cord blood was performed at the Spanish National Genotyping Centre using the Chemagic Magnetic Separator technology (Chemagen). DNA samples were quantified using dsDNA fluorescent detection (PicoGreen, Molecular Probes, Eugene, Oregon) and normalized to 75 to 100 ng/␮L. GSTP1 Ile105Val polymorphism (rs1695, A/G) was genotyped with KASPar technology (KBioscience, Hoddesdon, UK). GSTP1 Ile105Val genotypes were in Hardy-Weinberg equilibrium. Gas Cooking A questionnaire on environmental exposures was administered in person to the mothers during the third trimester of pregnancy. This questionnaire included questions on age and size of the home, number of persons resident, type of cooker used inside the house (natural gas, butane gas, propane gas, electric, other), use of an extractor fan (always, sometimes, never), type of heating, number of hours per day when windows were opened for ventilation (by season), passive smoking in the house, and other indoor and outdoor environmental exposures. We categorized information on type of cooker as “gas cooker” (any type of gas cooker) and “no gas cooker” (electric cooker and other). Validation studies with cohort participants in 2 regions (Sabadell and Valencia) characterized gas cooking as a major determinant of indoor NO2 concentrations: in subsamples of 108 pregnant mothers29 and 352 1-year-old children,8 average indoor NO2 concentrations in homes with gas cookers were nearly twice those in homes with electric cookers, and indoor exposure correlated well with personal exposure.29 We created further subcategories of gas-cooking exposure by type of gas (natural gas, butane/propane gas), and we categorized houses with and without gas cookers by use of an extractor fan (always, sometimes/never) and by ventilation of the house (⬍2 hours, ⱖ2 hours per day). Literature and results from our cohort suggest that more NO2 is emitted by butane and propane gas than by natural gas.8 Also, more frequent use of extractor fans is expected to reduce indoor levels of air pollutants from cooking. The net effect of ventilation is more complex as ventilation may lower exposure from indoor sources but increase exposure from outdoor sources; these effects are likely to depend on season.8 Very few subjects (⬍5%) used gas fires or heaters other than central heating, so we did not analyze exposure from heaters. Neurodevelopment testing Children’s neurodevelopment was assessed around 14 months of age (range, 11 to 22 months) using the Bayley Scales of Infant Development.30 The Bayley mental development scale (163 items) assesses age-appropriate cognitive development in areas such as performance ability, memory, and first verbal learning. The Bayley psychomotor scale consists of 81 items assessing fine and gross motor develop© 2011 Lippincott Williams & Wilkins Epidemiology • Volume 23, Number 1, January 2012 ment. Because earlier studies found effects of air pollution mainly on cognitive development,14 –18,20,21 we focused on the Bayley mental score as the main outcome, and analyzed psychomotor scores only in a sensitivity analysis. All testing was done at the primary care center, in the presence of the mother, by specially trained psychologists who were blind to any exposure information. The assessments followed a strict fieldwork protocol, including interobserver reliability tests. Psychologists flagged children whose Bayley tests were of poor quality because of basic pathologies (Down syndrome, autism) or less-than-optimal cooperation of the child (due to tiredness, bad mood, illness, etc), and these were excluded from main analyses. The Bayley mental and psychomotor indices (MDI, PDI) are based on a U.S. reference sample, and their appropriateness outside the United States has been questioned.31 In our study population, age of the child still had an influence on the MDI and PDI. We therefore used our own population to standardize the raw Bayley scores, based on the child’s age in days at test administration. Standardized scores were then computed from the standardized residuals, under the assumption of a normal distribution with a mean of 100 and a standard deviation (SD) of 15. The correlation between this standardized score and the MDI was 0.92. Children born preterm (⬍37 weeks’ gestation) have a particular developmental pattern and were therefore excluded from the main analysis. Also, the few children tested after 23 months of age were excluded. Statistical Analysis Multivariate linear regression models were used to examine the association of gas cooking with the mental development score. Regression models always included child’s sex and region as covariates. We then introduced maternal education and maternal social class; these 2 variables did not show evidence for collinearity in the model. Other, covariates were evaluated as potential confounders if they were reported to be associated with neurodevelopment in the literature or showed a relationship with gas cooking in our data. Covariates evaluated were birth weight, gestational age, season of birth, breastfeeding, nursery attendance, paternal social class, paternal education, paternal country of birth, maternal country of birth, maternal age, parity, maternal smoking at any time during pregnancy, maternal cotinine concentration (in trimester 3 spot urine samples), presence of a smoker in the house, alcohol consumption during pregnancy, fish consumption during pregnancy (total of all fish types), consumption of fruit and vegetables during pregnancy, age of the house, type of house, number of people in the house, urban or rural area, cord blood mercury concentration, and average outdoor NO2 concentration during the pregnancy (as estimated from a land use regression model).32 Covariates were retained only if they modified the coefficient of the gas cooking variable in the basic model by ⬎5%. The final multivariate model included sex, region, maternal edu© 2011 Lippincott Williams & Wilkins Gas Cooking and Neurodevelopment cation, maternal social class, gestational age, maternal country of origin, and type of house. Because information was missing for some covariates (up to 22%, Table 1) and to address the potential bias and loss of precision that could result from incomplete case analyses, we used multiple imputations to replace missing values in covariates. We created 20 imputations, generating 20 complete datasets that we analyzed following the standard combination rules for multiple imputations (eTable 1, http://links.lww.com/EDE/A523).33 Neurodevelopment tests of the children were spread over a relatively wide age range, from 11 to 22 months. Because it is difficult to compare results in children of very different ages, we stratified analyses by test age using the median Bayley test age as cutoff point (⬍14 months, ⱖ14 months or more). We evaluated the robustness of results over various strata of potential confounding variables. We also stratified by consumption of antioxidant nutrients (breastfeeding, fish consumption, and fruit and vegetable intake), which are possible modifiers of air pollution effects as reported in the literature. In the Sabadell region, we had data on the child’s GSTP1 Ile105Val polymorphism; we stratified analyses following a genetic dominant model (Ile/Ile vs. Ile/Val and Val/Val), while restricting to children with parents of Spanish origin and white ethnic background to limit the potential for population-stratification bias. The presence of interactions between gas cooking and stratification variables was assessed by likelihood ratio tests of the interaction terms in the regression model. All statistical analyses were conducted with Stata 10.0 statistical software (Stata Corporation, College Station, TX). RESULTS A total of 2644 mother– child pairs were enrolled in the first trimester; of whom, 2260 (85%) participated in the Bayley scale testing at 14 months. Of these, 2113 were term births (ⱖ37 weeks) with a Bayley test before 23 months of age and thus eligible for this study. Of these, 1887 pairs (89%) had good quality Bayley test results, and data on gas cooking (189 were excluded for poor test quality, 16 for having basic pathologies, and 21 for missing gas cooking data); they formed the main sample for analysis. The included mother– child pairs were generally of higher social class and higher educational achievement than the original cohort of enrolled mothers (eTable 2, http://links.lww.com/EDE/A523). Forty-four percent of mothers (n ⫽ 825) reported having a gas cooker at home in the third trimester of pregnancy. Of these, most were connected to natural gas (n ⫽ 583); the rest used propane or butane gas cookers. Of those who did not use gas cookers, 98% had electric cookers. Twelve percent of mothers changed cooker status between the third trimester interview and the time of neurodevelopment testing of the child. Mothers with gas cookers were of lower social class and educational achievement, and were more likely to have www.epidem.com | 3 Epidemiology • Volume 23, Number 1, January 2012 Vrijheid et al TABLE 1. Summary Measuresa of Covariates by Presence of Gas Cooker in the House, and Mean Bayley Mental Development Score by Covariate Categories (n ⫽ 1887) Gas Cooking Covariates Child’s sex Girl Boy Birth weight (g); mean (SD) Gestational age (weeks); mean (SD) Breastfeeding (months) ⬍6 ⱖ6 Missing Nursery attendance Yes No Missing Age at Bayley test administration (months) ⬍14 ⱖ14 Region Asturias Gipuzkoa Sabadell Valencia Maternal social class Professionals and managers (I, II) Other nonmanual (III) Skilled, semiskilled, and unskilled manual (IV, V) Maternal education Primary or secondary High school University Missing Maternal country of birth Spain Other Missing Parity (no. previous pregnancies) 0 ⱖ1 Missing Maternal smoking during pregnancy No Yes Missing Maternal cotinine in urine (ng/mL) ⬍4 4–99.9 ⱖ100 Missing Yes (n ⴝ 825) % No (n ⴝ 1062) % Mental Score Mean (SD) 48 52 3293 (433) 39.9 (1.2) 52 48 3302 (414) 39.8 (1.2) 102.0 (14.8) 97.9 (14.8) — — 53 46 1 54 43 3 99.3 (14.9) 100.8 (15.1) 29 68 3 36 58 7 100.2 (15.1) 100.0 (14.9) 51 49 31 69 100.2 (15.2) 99.8 (14.8) 8 9 37 47 25 37 18 21 102.0 (15.3) 98.0 (14.9) 99.7 (14.5) 100.6 (15.1) 20 26 54 24 29 47 101.6 (14.4) 100.6 (15.7) 98.9 (14.8) 28 43 29 ⬍1 18 41 42 ⬍1 99.2 (15.0) 100.0 (14.8) 100.3 (15.2) 87 13 ⬍1 96 4 ⬍1 100.2 (15.1) 97.2 (13.9) 54 46 ⬍1 60 40 0 100.5 (14.9) 99.2 (15.1) 65 35 ⬍1 73 27 ⬍1 99.9 (15.0) 100.0 (15.0) 34 37 22 8 40 38 15 7 101.0 (14.7) 99.1 (15.1) 100.5 (14.7) (Continued) 4 | www.epidem.com © 2011 Lippincott Williams & Wilkins Epidemiology • Volume 23, Number 1, January 2012 Gas Cooking and Neurodevelopment TABLE 1. (Continued) Gas Cooking Covariates Yes (n ⴝ 825) % Smoking in the house No 62 Yes 38 Missing ⬍1 Urban/rural area Urban 75 Semirural/Rural 25 Missing Age of the house (years) ⬍15 44 ⱖ15 53 Missing 4 Type of house Flat 84 Other 16 Missing ⬍1 No. people per room ⬍1 82 ⱖ1 18 Missing ⬍1 Maternal fish consumption (g/day)b ⬍69.6 59 ⱖ69.6 40 Missing 1 Maternal consumption of fruit and vegetables (g/day)b ⬍540.4 55 ⱖ540.4 45 Cord blood mercury (␮g/L); mean (SD) 11.2 (9.7) Missing (%) 23 Outdoor NO2 concentration (␮g/m3)b ⬍29.1 37 ⱖ29.1 60 Missing 3 a No (n ⴝ 1062) % Mental Score Mean (SD) 73 27 ⬍1 100.3 (15.2) 99.3 (14.6) 89 11 ⬍1 100.1 (14.9) 99.4 (15.3) 55 43 2 100.2 (15.0) 99.8 (15.0) 78 21 ⬍1 100.0 (14.9) 99.6 (15.5) 88 11 1 100.1 (15.0) 99.1 (15.2) 52 46 2 99.8 (15.0) 100.0 (14.9) 56 45 11.2 (8.3) 22 99.7 (15.1) 100.3 (14.9) — 67 30 3 99.4 (14.9) 100.4 (14.9) Values presented as percentages unless otherwise indicated for continuous variables. Cutoffs defined by mean values. b been born outside Spain (Table 1). They also reported higher levels of smoking, had higher cotinine levels, and reported more often that cigarette smoking occurred inside the house. Houses with gas cookers were more likely to be older (at least 15 years) and less likely to have central heating and air conditioning than houses without gas cookers. Houses with gas cookers also had more inhabitants and a higher density of persons per room. Gas cookers were more prevalent in the eastern Spanish regions of Sabadell and Valencia (around 63% of houses had gas cookers) than in the northern regions of Asturias and Gipuzkoa (18% had gas cookers). Gas cookers were used more often in areas classified as rural. Mothers living in © 2011 Lippincott Williams & Wilkins houses with gas cookers reported lower consumption of fish, but intake of fruit and vegetables and duration of breast-feeding were similar to mothers not using gas cookers. Outdoor NO2 levels were higher near houses with gas cookers; mercury levels in cord blood were similar. Mental development scores differed by child’s sex, gestational age, breastfeeding, social class, maternal country of birth, and region, but not by any of the other covariates (Table 1). Sex, gestational age, social class, and region had a substantial effect on mental scores in the final multivariable adjusted model (eTable 3, http://links.lww.com/EDE/A523). The presence of a gas cooker in the home was associated with decreased mental development scores in both the model www.epidem.com | 5 Epidemiology • Volume 23, Number 1, January 2012 Vrijheid et al TABLE 2. Association of Exposure to Gas Cooking With Bayley Mental Development Score Gas Cooking No. Score Mean (SD) Nod Yes 1062 825 100.7 (15.4) 99.0 (14.3) Nod Yes 329 424 100.9 (15.8) 99.7 (14.8) Nod Yes 733 401 100.6 (15.2) 98.3 (13.9) Unadjusteda Score (95% CI) Adjusted for Socioeconomic Statusb Score (95% CI) Overall 0 0 ⫺2.3 (⫺3.9 to ⫺0.8) ⫺2.3 (⫺3.8 to ⫺0.8) Child age at test <14 months 0 0 ⫺1.2 (⫺3.6 to 1.1) ⫺1.4 (⫺3.7 to 1.0) Child age at test >14 months 0 0 ⫺3.3 (⫺5.3 to ⫺1.3) ⫺3.1 (⫺5.1 to ⫺1.2) Fully Adjusted Modelc Score (95% CI) 0 ⫺2.5 (⫺4.0 to ⫺0.9) 0 ⫺1.6 (⫺4.0 to 0.7) 0 ⫺3.1 (⫺5.1 to ⫺1.1) a Model including sex and region. Model additionally adjusted for maternal social class, and maternal education. c Model additionally adjusted for maternal country of birth, type of house, and gestational age. d Reference category. b adjusted only for sex and region (⫺2.3 points 关95% confidence interval (CI)⫽ ⫺3.9 to ⫺0.8兴), and in the model adjusted for sex, region, maternal social class, and education (⫺2.3 points 关⫺3.8 to ⫺0.8兴) (Table 2). Final adjustment for maternal country of birth, gestational age, and type of house resulted in a decrease of ⫺2.5 points (95% CI ⫽ ⫺4.0 to ⫺0.9). The decrease in the mental score was greater in children tested after 14 months of age (⫺3.1 points 关⫺5.1 to ⫺1.1兴) than in children tested at younger ages (⫺1.6 points 关⫺4.0 to 0.7兴; test for interaction P ⫽ 0.07). Results did not change when mothers who changed gas cooker status between the third trimester questionnaire and the 14-month follow-up were excluded, when preterm births were included, or in the completecase analysis (eTable 4, http://links.lww.com/EDE/A523). Gas cooking was not related to the Bayley psychomotor score (eTable 4, http://links.lww.com/EDE/A523). Use of butane/propane gas cookers was related to a greater reduction in the mental score at test age ⱖ14 months (⫺6.2 points 关⫺10.3 to ⫺2.2兴) but not overall (Table 3). Infrequent use of an extractor fan had a small negative effect on the mental score, in houses both with and without gas cookers. Reductions in the mental score were greatest in houses with gas cookers and infrequent use of the extract fan (both test ages, and overall: ⫺3.7 points 关⫺5.9 to ⫺1.4兴). There were no clear differences by reported ventilation of the house. When analyses were stratified by covariates (Table 4), we observed that the negative effect of gas cooking was fairly homogeneous in most subcategories and that there was little evidence for interactions between these variables and the gas cooking effect (test for interaction P ⬎ 0.1). The strongest reductions in the mental development score related to gas cooking were found in the highest and lowest social class strata, in mothers with medium educational level, in mothers from non-Spanish origin, in older houses, in urban areas and areas with higher outdoor NO2 concentrations, and in non6 | www.epidem.com smoking households. The gas-cooking effect also appeared somewhat stronger in mother– child pairs with lower intake of protective nutrients as indicated by shorter breast-feeding, lower fish consumption, and lower consumption of fruits and vegetables. Findings were largely similar when children were stratified by test age ⬍14 months and ⱖ14 months (eTable 5, http://links.lww.com/EDE/A523). In the subgroup from the Sabadell region, the gas cooking association was somewhat stronger in children with the Ile/Val and Val/Val genotypes (⫺6.5 points 关⫺11.2 to ⫺1.8兴) compared with those who had the Ile/Ile genotype (⫺3.2 points 关⫺7.8 to 1.5兴). There was little evidence of a gene– environment interaction overall (P ⫽ 0.32), although the interaction effect was stronger (P ⫽ 0.06) in the group of children with test age ⱖ14 months (eTable 5). DISCUSSION The presence of a gas cooker at home during pregnancy was associated with slower mental development of young children—particularly those tested after the age of 14 months. The associations were relatively consistent across social and educational classes. Given the consistency of exposure over time within families, it is not possible to determine how timing of exposure (prenatal versus postnatal) might affect the association. These findings support those of a recent study in Menorca, Spain,21 that found adverse cognitive functioning and higher risk of inattention symptoms at age 4 years in relation to early-life exposure to gas appliances and indoor NO2. The Menorca study found that measured indoor 48-hour NO2 concentrations were closely correlated with the use of gas appliances; the study observed similar associations with neurobehavioral outcomes with both exposure measures. The replication of an association of gas cooking with neurodevelopment in 2 cohorts of different age groups, using different © 2011 Lippincott Williams & Wilkins Epidemiology • Volume 23, Number 1, January 2012 Gas Cooking and Neurodevelopment TABLE 3. Association of Type of Gas Used, Use of Extractor Fan, and Ventilation of the House With Bayley Mental Development Scorea Child Age at Test <14 Months Overall Type of gas used No gas cookerb Natural gas Butane/propane gas Use of extractor fanc No gas cooker Always use extract fanb Sometimes/never use extract fan Gas cooker Always use extract fan Sometimes/never use extract fan Ventilation of housec No gas cooker Ventilate the house ⬎2 hour per Ventilate the house ⱕ2 hour per Gas cooker Ventilate the house ⬎2 hour per Ventilate the house ⱕ2 hour per a Child Age at Test >14 Months No. Score (95% CI) No. Score (95% CI) No. Score (95% CI) 1062 583 242 0 ⫺2.8 ⫺1.6 (⫺4.5 to ⫺1.1) (⫺3.9 to 0.6) 329 237 187 0 ⫺2.7 ⫺0.4 (⫺5.4 to 0.1) (⫺3.3 to 2.6) 733 346 55 0 ⫺2.5 ⫺6.2 (⫺4.6 to ⫺0.4) (⫺10.3 to ⫺2.2) 817 227 0 ⫺1.0 (⫺3.2 to 1.2) 263 62 0 ⫺0.5 (⫺4.7 to 3.8) 554 165 0 ⫺1.4 (⫺3.9 to 1.2) 510 258 ⫺2.2 ⫺3.7 (⫺4.0 to ⫺0.4) (⫺5.9 to ⫺1.4) 262 136 ⫺1.6 ⫺2.8 (⫺4.4 to 1.3) (⫺6.2 to 0.6) 248 122 ⫺2.5 ⫺4.7 (⫺4.8 to ⫺0.1) (⫺7.6 to ⫺1.7) dayb day 600 437 0 ⫺1.0 (⫺2.8 to 0.9) 203 118 0 ⫺1.9 (⫺5.4 to 1.7) 397 319 0 ⫺0.9 (⫺3.0 to 1.2) day day 563 258 ⫺2.7 ⫺3.0 (⫺4.9 to ⫺0.5) (⫺4.9 to ⫺1.2) 321 100 ⫺1.3 ⫺2.7 (⫺5.0 to 2.5) (⫺5.4 to 0.1) 242 158 ⫺3.6 ⫺3.6 (⫺6.3 to ⫺0.8) (⫺6.0 to ⫺1.1) Model includes sex, region, maternal social class, maternal education, maternal country of birth, type of house, and gestational age. Reference category. Excluding persons with missing extractor fan or ventilation data. b c age-appropriate neurodevelopment assessments and different exposure assessment methods, strengthens the evidence for a causal association, although similar confounding structures cannot be excluded as explanation. A limitation of this study is that assessment of exposure to gas cookers was based on only one question in the interviews. We assume that women are able to report the type of cooker without much error, and in this prospective study, there is little concern for recall bias. We did not, however, measure levels of NO2, fine and ultrafine particles, or other gas-cooking-related pollutants, and we had no information about factors that can influence exposure, such as the time spent cooking or the location of the kitchen in relation to other living spaces. We did find somewhat stronger associations with less frequent use of extractor fans and butane/ propane cookers, which are also predictors of higher indoor NO2 concentration.8,29 Further, there is little doubt that gas cooking is a main predictor of indoor NO2 levels, together with outdoor NO2 and smoking. Studies, including our own validation studies, have reported concentrations of NO2 1.5to 2-fold higher in living rooms of houses with gas cookers.4,5,7,8,29 However, it is possible that other unmeasured indoor air pollutants are responsible for our findings. Ultrafine particles, volatile organic compounds, and carbon monoxide are also associated with gas cookers,1,5,10,34,35 and particulate matter and polycyclic aromatic hydrocarbons have © 2011 Lippincott Williams & Wilkins been related to the cooking of food, irrespective of the type of cooker.6,9,34,36 Better characterization of gas-cooking-related exposures, preferably through the measurement of associated pollutants, will be important in future studies. We assessed the potential for confounding for a wide range of data on socioeconomic factors, diet and life-style habits, and exposure to environmental pollutants. Still, we cannot exclude the possibility that confounding by unmeasured risk factors (eg, maternal IQ, child’s maternal attachment, and home environment indicators) produced the association between gas cooking and neurodevelopment. Gas cooking was related to social class, education, maternal country of birth, and other variables such as maternal cigarette smoking. Variables related to housing conditions, such as the age of the house, were also strongly associated with gas cooking, but not with neurodevelopment. Adjustment for these variables produced little change in the regression coefficients. Moreover, the negative effect of gas cooking did not clearly differ among social variables, making it less likely that social-class confounding was responsible for the findings. Stratified analyses did tend to show somewhat stronger associations in older houses, in houses with a higher concentration of outdoor NO2, and in urban areas. Such findings may indicate a somewhat higher vulnerability of children in the presence of other pollution exposures, or when living in adverse housing conditions. www.epidem.com | 7 Epidemiology • Volume 23, Number 1, January 2012 Vrijheid et al TABLE 4. Association of Exposure to Gas Cooking With Mental Development Score, Stratified by Covariates No. Overall 1887 Maternal social class I and II 420 III 519 IV and V 948 Maternal education Primary or secondary 419 High school 785 University 679 Maternal country of origin Spain 1739 Other 145 Maternal smoking No 1309 Yes 573 Smoking in house No 1280 Yes 599 Age of the house (years) ⬍15 945 ⱖ15 887 Urban/rural area Urban 1527 Semirural/rural 359 Outdoor NO2 concentration (␮g/m3) Low ⬍29.1 1017 High ⱖ29.1 816 Breast-feeding (months) ⬍6 1014 ⱖ6 830 Maternal fish consumption (g/day) ⬍69.6 1034 ⱖ69.6 821 Consumption of fruits and vegetables (g/day) ⬍540.4 1042 ⱖ540.4 845 GSTP1 genotypeb Ile/Ile 150 Ile/Val and Val/Val 172 Score (95% CI)a ⫺2.5 (⫺4.0 to ⫺0.9) ⫺3.5 (⫺6.6 to ⫺0.3) ⫺0.6 (⫺3.6 to 2.4) ⫺3.4 (⫺5.5 to ⫺1.3) ⫺1.6 (⫺4.6 to 1.5) ⫺3.4 (⫺5.7 to ⫺1.1) ⫺1.7 (⫺4.4 to 1.0) ⫺2.1 (⫺3.7 to ⫺0.6) ⫺7.3 (⫺12.9 to ⫺1.7) ⫺2.4 (⫺4.2 to ⫺0.5) ⫺2.5 (⫺5.1 to 0.2) ⫺2.5 (⫺4.4 to ⫺0.6) ⫺1.9 (⫺4.3 to 0.6) ⫺1.3 (⫺3.4 to 0.9) ⫺3.8 (⫺6.0 to ⫺1.6) ⫺2.8 (⫺4.4 to ⫺1.2) ⫺0.9 (⫺4.9 to 3.2) ⫺1.9 (⫺4.1 to 0.3) ⫺2.9 (⫺5.1 to ⫺0.8) ⫺3.5 (⫺5.6 to ⫺1.5) ⫺1.4 (⫺3.7 to 0.9) ⫺3.4 (⫺5.4 to ⫺1.4) ⫺0.9 (⫺3.2 to 1.4) ⫺2.8 (⫺4.8 to ⫺0.8) ⫺2.1 (⫺4.4 to 0.3) ⫺3.2 (⫺7.8 to 1.5) ⫺6.5 (⫺11.2 to ⫺1.8) a Model includes sex, region, maternal social class, maternal education, maternal country of birth, type of house, and gestational age. b Sabadell only. Our study shows a stronger effect of gas-cooking exposure on the mental development of children tested after the age of 14 months, compared with those tested at younger age. It is well recognized that these tests can be more variable and less sensitive in younger children.37 Indeed, studies of the neurodevelopmental effects of other pollutants have also observed effects on the Bayley mental development index at older, but not younger, test ages.14,38 This may be due to difficulties in testing very young children or due to the fact 8 | www.epidem.com that effects of pollutants may become apparent only with maturation of the brain over time. Continued follow-up will be needed to evaluate possible effects of indoor air pollution at older ages. There is increasing experimental evidence that air pollution can cause chronic neuroinflammation and damage the central nervous system.12 The main mechanisms proposed for effects of air pollution on the central nervous system are through systemic effects or through peripheral damage. Inflammatory response of the lung to air pollution exposure may translate to a wider systemic effect, with oxidative stress damage in other organs, including the brain.11,12,39,40 Ultrafine particles and compounds absorbed by them (eg, polycyclic aromatic hydrocarbons and metals) may traverse the blood– brain barrier and cause neuroinflammation, lipid peroxidation, and other damage directly in the brain.12 The exact mechanisms by which damage is caused are under discussion. Little is known about direct neurotoxic effects of NO2, but speculations include a possible mechanism through the peroxidation of brain lipids or through interference with dopamine biosynthesis.21 GSTP1 is included in the super family of glutathione S-transferase enzymes involved in the phase-II detoxification of xenobiotic substances, and that protect against the toxic effects of reactive oxygen and related oxidative stress.41 GSTP1 is mainly expressed in adult lung, placenta, breast, and urinary bladder, and it is the most strongly expressed GST isoenzyme in almost all embryonic and fetal tissues, including the brain,41 making it, of particular, a priori interest. The GSTP1 Ile105Val substitution is located near the substrate-binding site, resulting in a less active enzyme.42 Other studies have reported higher risks of childhood asthma and allergic disease associated with traffic-related air pollution in subjects with this particular polymorphism.43,44 In the Menorca study,21 greater neurodevelopmental effects of NO2- and gas-appliances exposure were found in the Ile/Val and Val/Val genotypes; we find a similar result in a small subcohort of our new study, although we did not have statistical power to draw strong conclusions. Taken together, the 2 studies suggest that an oxidative-stress mechanism modified by detoxification genes may be involved in the neurodevelopmental toxicity of gas-cooking-related exposures. The effect of gas cooking on mental development tended to be somewhat stronger in mothers who consumed less fish and fewer fruits and vegetables and who breast-fed for a shorter duration, lending some support to the oxidativestress mechanism. Fruits and vegetables are a rich source of antioxidant micronutrients,23,45 and fish and breast milk have high contents of long-chain polyunsaturated fatty acids and other nutrients that may reduce inflammation and oxidative stress.46 It should be noted, however, that these dietary factors were related to social class (data not shown), and that, © 2011 Lippincott Williams & Wilkins Epidemiology • Volume 23, Number 1, January 2012 although we adjusted for socioeconomic status indicators, it is hard to disentangle protective effects of antioxidant intakes from those of other social-class-related factors. In conclusion, this study suggests a small adverse effect of indoor air pollution from gas cookers on mental development. We cannot exclude unmeasured confounding as an explanation. The potential public health implications of these findings, if true, are substantial because of the frequent use of gas cookers and because neurodevelopmental disorders and disabilities impose a social and economic burden. Future studies should, wherever possible, use designs that more accurately assess exposure to indoor air pollutants. Gas Cooking and Neurodevelopment 17. 18. 19. 20. 21. ACKNOWLEDGMENTS We thank all the cohort participants for their collaboration. A full roster of the INMA Project Investigators can be found at http://www.proyectoinma.org/presentacion-inma/ listado-investigadores/en_listadoinvestigadores.html. REFERENCES 1. Chauhan AJ. Gas cooking appliances and indoor pollution. Clin Exp Allergy. 1999;29:1009 –1013. 2. Ng TP, Seet CS, Tan WC, Foo SC. Nitrogen dioxide exposure from domestic gas cooking and airway response in asthmatic women. Thorax. 2001;56:596 – 601. 3. WHO. WHO guidelines for indoor air quality: selected pollutants. 2010. Available at: http://www.euro.who.int/__data/assets/pdf_file/ 0009/128169/e94535.pdf. 4. Garcia-Algar O, Zapater M, Figueroa C, et al. Sources and concentrations of indoor nitrogen dioxide in Barcelona, Spain. J Air Waste Manag Assoc. 2003;53:1312–1317. 5. Raw GJ, Coward SK, Brown VM, Crump DR. Exposure to air pollutants in English homes. J Expo Anal Environ Epidemiol. 2004;14(suppl 1): S85–S94. 6. Baxter LK, Clougherty JE, Laden F, Levy JI. Predictors of concentrations of nitrogen dioxide, fine particulate matter, and particle constituents inside of lower socioeconomic status urban homes. J Expo Sci Environ Epidemiol. 2007;17:433– 444. 7. Gillespie-Bennett J, Pierse N, Wickens K, et al. Sources of nitrogen dioxide (NO2) in New Zealand homes: findings from a community randomized controlled trial of heater substitutions. Indoor Air. 2008;18: 521–528. 8. Esplugues A, Ballester F, Estarlich M, et al. Indoor and outdoor concentrations and determinants of NO2 in a cohort of 1-year-old children in Valencia, Spain. Indoor Air. 2010;20:213–223. 9. Dick CA, Dennekamp M, Howarth S, et al. Stimulation of IL-8 release from epithelial cells by gas cooker PM(10): a pilot study. Occup Environ Med. 2001;58:208 –210. 10. Dennekamp M, Howarth S, Dick CA, Cherrie JW, Donaldson K, Seaton A. Ultrafine particles and nitrogen oxides generated by gas and electric cooking. Occup Environ Med. 2001;58:511–516. 11. Sunyer J. The neurological effects of air pollution in children. Eur Respir J. 2008;32:535–537. 12. Block ML, Calderon-Garciduenas L. Air pollution: mechanisms of neuroinflammation and CNS disease. Trends Neurosci. 2009;32:506 –516. 13. Grandjean P, Landrigan PJ. Developmental neurotoxicity of industrial chemicals. Lancet. 2006;368:2167–2178. 14. 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;114:1287–1292. 15. Perera F, Li TY, Zhou ZJ, et al. Benefits of reducing prenatal exposure to coal-burning pollutants to children’s neurodevelopment in China. Environ Health Perspect. 2008;116:1396 –1400. 16. Suglia SF, Gryparis A, Wright RO, Schwartz J, Wright RJ. Association © 2011 Lippincott Williams & Wilkins 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. of black carbon with cognition among children in a prospective birth cohort study. Am J Epidemiol. 2008;167:280 –286. Perera FP, Li Z, Whyatt R, et al. Prenatal airborne polycyclic aromatic hydrocarbon exposure and child IQ at age 5 years. Pediatrics. 2009; 124:e195– e202. Edwards SC, Jedrychowski W, Butscher M, et al. Prenatal exposure to airborne polycyclic aromatic hydrocarbons and children’s intelligence at 5 years of age in a prospective cohort study in Poland. Environ Health Perspect. 2010;118:1326 –1331. Freire C, Ramos R, Puertas R, et al. Association of traffic-related air pollution with cognitive development in children. J Epidemiol Community Health. 2010;64:223–228. Guxens M, Aguilera I, Ballester F, et al. Prenatal exposure to residential air pollution and infant mental development: modulation by antioxidants and detoxification factors. Environ Health Perspect. 2011 Aug 25. [Epub ahead of print]. Morales E, Julvez J, Torrent M, et al. Association of early-life exposure to household gas appliances and indoor nitrogen dioxide with cognition and attention behavior in preschoolers. Am J Epidemiol. 2009;169:1327– 1336. Jedrychowski W, Perera F, Mrozek-Budzyn D, et al. Higher fish consumption in pregnancy may confer protection against the harmful effect of prenatal exposure to fine particulate matter. Ann Nutr Metab. 2010; 56:119 –126. Kannan S, Misra DP, Dvonch JT, Krishnakumar A. Exposures to airborne particulate matter and adverse perinatal outcomes: a biologically plausible mechanistic framework for exploring potential effect modification by nutrition. Environ Health Perspect. 2006;114:1636 – 1642. Romieu I, Castro-Giner F, Kunzli N, Sunyer J. Air pollution, oxidative stress and dietary supplementation: a review. Eur Respir J. 2008;31: 179 –197. Villarreal-Calderon R, Torres-Jardon R, Palacios-Moreno J, et al. Urban air pollution targets the dorsal vagal complex and dark chocolate offers neuroprotection. Int J Toxicol. 2010;29:604 – 615. Guxens M, Ballester F, Espada M, et al. Cohort Profile: The INMA– INfancia y Medio Ambiente–(Environment and Childhood) Project. Int J Epidemiol. In press. ILO (International Labor Organisation). International Standard Classification of Occupations 1988;1990. Available at: http://www.ilo.org/public/ english/bureau/stat/isco/isco88/index.htm. Vioque J, Weinbrenner T, Asensio L, Castello A, Young IS, Fletcher A. Plasma concentrations of carotenoids and vitamin C are better correlated with dietary intake in normal weight than overweight and obese elderly subjects. Br J Nutr. 2007;97:977–986. Valero N, Aguilera I, Llop S, et al. Concentrations and determinants of outdoor, indoor and personal nitrogen dioxide in pregnant women from two Spanish birth cohorts. Environ Int. 2009;35:1196 –1201. Bayley N. Escalas Bayley de Desarrollo Infantil. Madrid: TEA Ediciones; 1977. Wu YT, Tsou KI, Hsu CH, Fang LJ, Yao G, Jeng SF. Brief report: Taiwanese infants’ mental and motor development— 6 –24 months. J Pediatr Psychol. 2008;33:102–108. Estarlich M, Ballester F, Aguilera I, et al. Residential exposure to outdoor air pollution during pregnancy and anthropometric measures at birth in a multicenter cohort in Spain. Environ Health Perspect. 2011; 119:1333–1338. Spratt M, Carpenter J, Sterne JA, et al. Strategies for multiple imputation in longitudinal studies. Am J Epidemiol. 2010;172:478 – 487. Afshari A, Matson U, Ekberg LE. Characterization of indoor sources of fine and ultrafine particles: a study conducted in a full-scale chamber. Indoor Air. 2005;15:141–150. Willers SM, Brunekreef B, Oldenwening M, Smit HA, Kerkhof M, Vries H. Gas cooking, kitchen ventilation, and exposure to combustion products. Indoor Air. 2006;16:65–73. Bhangar S, Mullen NA, Hering SV, Kreisberg NM, Nazaroff WW. Ultrafine particle concentrations and exposures in seven residences in northern California. Indoor Air. 2011;21:132–144. Dietrich KN, Eskenazi B, Schantz S, et al. Principles and practices of neurodevelopmental assessment in children: lessons learned from the www.epidem.com | 9 Epidemiology • Volume 23, Number 1, January 2012 Vrijheid et al 38. 39. 40. 41. 42. Centers for Children’s Environmental Health and Disease Prevention Research. Environ Health Perspect. 2005;113:1437–1446. Eskenazi B, Marks AR, Bradman A, et al. In utero exposure to dichlorodiphenyltrichloroethane (DDT) and dichlorodiphenyldichloroethylene (DDE) and neurodevelopment among young Mexican American children. Pediatrics. 2006;118:233–241. Calderon-Garciduenas L, Franco-Lira M, Torres-Jardon R, et al. Pediatric respiratory and systemic effects of chronic air pollution exposure: nose, lung, heart, and brain pathology. Toxicol Pathol. 2007;35:154 –162. MohanKumar SM, Campbell A, Block M, Veronesi B. Particulate matter, oxidative stress and neurotoxicity. Neurotoxicology. 2008;29: 479 – 488. Strange RC, Spiteri MA, Ramachandran S, Fryer AA. GlutathioneS-transferase family of enzymes. Mutat Res. 2001;482:21–26. Moyer AM, Salavaggione OE, Wu TY, et al. Glutathione s-transferase 10 | www.epidem.com 43. 44. 45. 46. p1: gene sequence variation and functional genomic studies. Cancer Res. 2008;68:4791– 4801. Carlsten C, Dybuncio A, Becker A, Chan-Yeung M, Brauer M. GSTP1 polymorphism modifies risk for incident asthma associated with nitrogen dioxide in a high-risk birth cohort. Occup Environ Med. 2011;68:308. Melen E, Nyberg F, Lindgren CM, et al. Interactions between glutathione S-transferase P1, tumor necrosis factor, and traffic-related air pollution for development of childhood allergic disease. Environ Health Perspect. 2008;116:1077–1084. Balsano C, Alisi A. Antioxidant effects of natural bioactive compounds. Curr Pharm Des. 2009;15:3063–3073. Rodriguez-Palmero M, Koletzko B, Kunz C, Jensen R. Nutritional and biochemical properties of human milk: II. Lipids, micronutrients, and bioactive factors. Clin Perinatol. 1999;26:335–359. © 2011 Lippincott Williams & Wilkins