Abstract
In this study, we aimed to evaluate the relationship between individual total exposure to air pollution and airway changes in a group of 51 wheezing children.
Respiratory status was assessed four times (January 2006, June 2006, January 2007 and June 2007) during a 1-week period through a standardised questionnaire, spirometry, exhaled nitric oxide fraction and pH in exhaled breath condensate (EBC). Concentrations of particles with a 50% cut-off aerodynamic diameter of 10 µm (PM10), O3, NO2 and volatile organic compounds were estimated through direct measurements with an ad hoc device or air pollution modelling in the children's schools and at their homes in the same 4 weeks of the study. For each child, total exposure to the different air pollutants was estimated as a function of pollutant concentrations and daily activity patterns.
Increasing total exposure to PM10, NO2, benzene, toluene and ethylbenzene was significantly associated with a decrease of forced expiratory volume in 1 s (FEV1) and with an increase of change in FEV1. Increasing exposure to NO2 and benzene was also related to a significant decrease of FEV1/forced vital capacity. Increasing exposure to PM10, NO2, benzene and ethylbenzene was associated with acidity of EBC.
This study suggests an association in wheezing children between airway changes and total exposure to air pollutants, as estimated by taking into account the concentration in the various microenvironments attended by the children.
- Breath condensate analysis
- NO2
- particles with a 50% cut-off aerodynamic diameter of 10 µm
- spirometry
- total exposure
- volatile organic compounds
Various cross-sectional [1–3] and longitudinal [4–7] studies have shown the effects of both indoor and outdoor air pollutants on the respiratory health of children. Most of these studies have focused on the relationship with symptoms or diseases, and few have investigated the links with lung function tests [8–10] and inflammatory markers [8, 9, 11]. However, evidence of changes over time in parameters related to air pollutant variations in the literature is scarce. In addition, assessment of exposure to air pollutants has not been exhaustive in most studies.
Background pollutant concentrations measured at central monitoring stations have usually been used [12, 13], followed up by traffic density/distance to major roads [13, 14] or dispersion models [3]. Only more recently was exposure to indoor environments, such as schools [15, 16] or houses where children spend >80% of their time [17], also considered [18, 19]. Taken together, these studies have shown that there is a need for a more specific and precise assessment of exposure to maximise the potential information to be derived from epidemiological studies. Although continuous measurements of personal exposures for all study subjects for a complete study period might be considered the desired "gold standard" for exposure assessment, this is difficult to achieve due to feasibility constraints, as each participant has to carry special equipment 24 h per day for several days. To date, such studies [20, 21] have been scarce, focused on only some pollutants (e.g. particle matter [20]) and taken into account only part of the day.
An alternative strategy to achieve the assessment of total exposure to a pollutant could result from outdoor dispersion models [22], combined with direct objective measurements of air pollution in the different indoor environments attended by children. In addition, repeated measures for the same individual can be introduced to overcome the reduced sample size issue. This is a common limitation of such studies that look for a more detailed approach to calculate exposure and aim to evaluate the respiratory status at the same time.
In this prospective study, we aimed to evaluate the relationship between total exposure to various air pollutants and airway changes in a group of wheezing children. The main peculiarities of this study were to estimate: 1) individual total exposure for each child for a wide range of air pollutants, taking into account different environments and the time spent by the child in these environments; 2) airway changes through spirometry and inflammatory biomarkers; and 3) repeated measures through consistently timed data methodology.
MATERIAL AND METHODS
Population and protocol
The Saud'Ar prospective panel study took place in Viseu, a non-industrial city of Portugal with ∼50,000 inhabitants, located 60 km from the coast in the northern central area of Portugal. After approval by the local ethics committee, the International Study of Asthma and Allergies in Childhood questionnaire was distributed to 806 children in four primary schools in order to identify those who had suffered from wheezing in the previous 12 months. The parents of children reporting wheezing symptoms were proposed, after signing the informed consent, to permit the children's participation in the study. The study consisted of four visits in January 2006, June 2006, January 2007 and June 2007. All of the participants were evaluated in the same week of the month at the Dept of Pneumology of São Teotónio Hospital, the main hospital in Viseu. With the agreement of the ethics committee, parents were invited to suspend asthma control medication (montelukast or inhaled corticosteroids) 3 weeks prior to evaluation if, and only if, the child's asthmatic status allowed such suspension. Stopping treatment was discussed with the doctor treating the child for asthma; however, children could take other medications.
In agreement with the ethics committee, parents were instructed to re-introduce asthma medication as soon as necessitated by the child’s status, i.e. new symptoms, worsening of symptoms or attacks. Additionally, during the entire survey, a 24-h telephone line was available to the parents in order to answer to any questions on the child's health and an “open-door policy” was set up at the hospital outpatients' clinic for the study participants. Regular follow-up calls were also scheduled with the parents to question them on the health status of the child.
At the beginning of the study, only four children were receiving treatment (montelukast n=2, low-dose inhaled corticosteroids n=2). The remaining children used only rescue medication. No child needed immediate re-introduction of asthma medication during the survey.
At all visits, a short questionnaire concerning the presence of wheezing symptoms, use of rescue medication and emergency department visits in the previous 6 months was answered by parents. After returning the questionnaire, children performed the following clinical tests on the same day in the following order: exhaled breath condensate (EBC) collection for pH analysis, measurement of exhaled nitric oxide fraction (FeNO) and spirometry. Skin-prick tests to common inhalant allergens were performed during the first and last visit. More details about the methodology are presented in the online supplementary material.
Respiratory outcomes of interest were as follows. 1) Forced expiratory volume in 1 s (FEV1), FEV1/forced vital capacity (FVC) ratio, forced expiratory flow between 25–75% of FVC (FEF25–75%) and improvement of FEV1 15 min after administration of 200 μg of salbutamol (ΔFEV1). Lung function was expressed as the percentage of the predicted normal value. Spirometry was performed using a portable pneumotachograph (Vitalograph Compact; Vitalograph, Buckingham, UK). 2) FeNO in ppb and pH of EBC (after extraction of CO2 with argon) were the bronchial inflammatory outcomes. Measurement of FeNO was performed prior to spirometry using a portable analyser (Niox® Mino; Aerocrine, Solna, Sweden), in which the expiratory flow rate was maintained at 50 mL·s−1. An RTube (Respiratory Research Inc., Austin, TX, USA) was used for EBC collection. 3) Wheezing symptoms, use of rescue medication (bronchodilators such as salbutamol or terbutaline) and emergency department visits in the previous 6 months constituted the clinical outcomes.
Air quality measurements
Air quality assessment took place in the town of Viseu during the four previously mentioned periods. The measured outdoor parameters included ambient air concentrations of O3, NO, NO2, CO and BTEX (benzene, toluene, ethylbenzene and xylenes), and particles with a 50% cut-off aerodynamic diameter of 10 µm and 2.5 μm (PM10 and PM2.5, respectively), which were measured continuously in three mobile laboratories: in the city centre, the school courtyard of the urban location and a suburban school courtyard. Additionally, at the urban school, both ambient PM10 and PM2.5 were measured by gravimetric methods (24-h mean) while for the suburban school only PM10 was measured, as only one set of equipment for PM2.5 was available. O3 and NO2 concentrations were also measured using diffusive samplers located at 20 points throughout the city of Viseu, in an area of ∼40 km2.
Volatile organic compound (VOC; such as formaldehyde and BTEX), O3 and NO2 measurements were performed in the four schools (courtyard and classrooms) and children's houses (bedrooms) using diffusive samplers. All parents of children included in the medical visits were asked to participate in the air quality studies.
Indoor PM10 and PM2.5 concentrations were measured at the urban school by gravimetric methods (24-h mean). In the suburban school, only PM10 was measured for the same reason as mentioned previously. Further details concerning air quality measurements are presented in online supplement.
Air quality modelling
The mesoscale modelling system MM5/CHIMERE (for full details see the online supplement) was applied to characterise the spatial distribution of air pollutants concentrations. More information concerning air quality modelling is presented in the online supplement.
Exposure assessment estimation
To estimate individual exposure to air pollutants, two main tasks were performed: 1) the estimation of the daily activity pattern of each child, which allowed the microenvironments frequented by children and the time spent in each one to be identified; and 2) the air quality characterisation of those microenvironments. The daily activity pattern was established through questionnaires administered to parents and children and a 5-day study week (Monday to Friday) was considered. The air quality evaluation in the identified microenvironments (outdoors and indoors) was performed using a multi-strategy approach: measurements during field campaigns and air quality modelling simulations to characterise areas where it was not possible to obtain measurements.
The individual exposure of each child to the different air pollutants was then estimated using a microenvironment approach and calculated (for a 5-day study week) according to the following equation:
Expi = 
In this equation, Expi is the total exposure for the person i over the specified period of time, Cj is the pollutant concentration in microenvironment j, and ti,j is the time spent by the person i in microenvironment j. Results are presented as the mean (minimum and maximum) of the means of exposure over a week. More details are presented in the online supplement.
Statistical analysis
Exploratory analysis of the variables of interest was performed. The Chi-squared test was used to compare proportions and Friedman's test was used to compare pollutants, weather conditions, and spirometric and inflammation outcomes during the four visits. A generalised estimating equation (GEE) approach with an exchangeable working correlation was used to estimate the association between exposure to each air pollutant and spirometric, inflammatory and clinical outcomes. Crude regression coefficients were calculated first. Given the potential for factors to confound or modify the associations between air pollution and respiratory health, we decided a priori to adjust all the models for age, sex, parental smoking, parental education, atopic status, time of the visit, average temperature and relative humidity on the day of the medical visit. Other variables drawn from the literature, such as height, weight, older siblings, mould or dampness at home, fireplace at home and presence of pets at home, were also included as they were associated with at least one health outcome in the univariable analyses (p<0.15) and caused a change (>10%) on at least one pollutant's estimated effect when included in the models. Factors such as body mass index, use of natural gas for cooking/heating and presence of air conditioning were excluded. For coherency, all the models were adjusted for the same confounders. This decision was taken as the studied outcomes share many common aspects. After stepwise analyses, variables were included all at once in the models. Spirometric variables (not adjusted for age, sex, height and weight) were expressed as percentage of predicted value. In order to take the significant correlations between the pairs of air pollutants into account, two-pollutant models including the concentrations of the two considered pollutants as independent variables were implemented. Crude and adjusted regression coefficients and corresponding 95% confidence intervals were calculated for an increased exposure of 10 μg·m−3·week−1 of air pollutant.
Correlation matrix between the pollutants was computed at each visit using the Spearman's rank correlation.
The level of significance considered was α=0.05, although p-values >0.05 and <0.1 are still reported. Stata (StataCorp LP, TX, USA) for Windows was used to analyse the data.
RESULTS
Out of 806 distributed questionnaires, 645 (80%) were completed by the parents. Among the children whose parents replied, 77 (11.7%) reported wheezing in the previous 12 months, a frequency similar to the prevalence estimated in the paediatric Portuguese population. After contacting their parents, 54 children were allowed to participate. Herein, we present data from 51 children who reached the end of the study.
The description of the sample at the beginning of the study is presented in table 1. The majority of children were males (55%), with a mean±sd age at the end of the study of 8.8±1.1 yrs. 27 (53%) were atopic and all were Caucasian.
For each visit, baseline concentrations of VOCs for houses (indoors) and schools (indoors and outdoors) are presented in table 2. Indoor levels of NO2 and O3 were too low and were not considered for the exposure quantification. PM10 baseline concentrations were taken both outdoors and indoors in schools.
PM10 concentrations were high inside the schools. Both houses and schools constituted an important source of VOCs. VOC concentrations were higher in indoor environments (with the exception of benzene) than outdoors. There was a different pattern across the visits, with PM10 and NO2 reaching higher concentrations in January than in June, whereas for O3 the opposite was true.
Quantifications of children's total exposure to the different pollutants are presented in table 3. Children had similar daily activity patterns with the majority of the exposure time spent in indoor environments, namely their homes (65%) and schools (20%). Regarding the seasonal variability, time spent indoors in winter was slightly higher than in summer, without reaching statistical significance (p=0.283). Children spent >22 h in indoor environments each day, despite the time of the year. The only difference was that in summer, children spent a little more time in the courtyard of the school and outside their homes (on average <30 min more). Considering the child's exposure, air pollutants were correlated with each other at each visit (see online supplement). Statistically significant positive correlations existed among various air pollutants, namely for VOCs such as toluene-ethylbenzene, toluene-xylene, xylene-ethylbenzene and benzene-ethylbenzene. NO2 and ozone also presented a trend to be positively correlated. Conversely, PM10 and NO2, and PM10 and O3 were inversely correlated.
The results of spirometry, inflammation assessment and clinical outcomes across the studied seasons are presented in table 4. Most children did not present abnormal changes during the study period but, in each evaluation, ∼25% presented a ΔFEV1 >12%. There were differences in spirometric changes (FEV1: p=0.010; FEV1/FVC: p=0.092) between the four evaluations, but more significant differences were found for the inflammation parameters (p<0.001). This is in accordance with the known age-related changes in FeNO [23] and for the predicted spirometric values [24] that can spuriously introduce a trend of lung function deterioration without clinical significance.
According to GEE (table 5, figs 1 and 2), after adjustment, increasing individual exposure to PM10 in the studied week was associated with a trend of airway deterioration, reaching significance with a decrease of the FEV1 (regression coefficient -1.64, 95% CI -3.20– -0.10) and pH on EBC (-0.21, 95% CI -0.30– -0.12) and an increase of ΔFEV1 (1.19, 95% CI 0.12–2.26). Associations were also found for an increase of NO2 exposure: decrease of FEV1 (-6.31, 95% CI -11.87– -0.76), FEV1/FVC (-2.79, 95% CI -5.71–0.14), FEF25−75% (-10.20, 95% CI -18.80– -1.59) and pH on EBC (-0.69, 95% CI -1.04– -0.35), and an increase of ΔFEV1 (4.72, 95% CI 0.91–8.53). No associations were found for O3 after adjustment.
Per cent changes and corresponding 95% confidence intervals (whiskers) in forced expiratory volume in 1 s (FEV1) for 10-μg·m−3·week−1 increments of air pollutant, after adjustment. PM10: particles with a 50% cut-off aerodynamic diameter of 10 µm.
Exhaled breath condensate (EBC) pH changes and 95% confidence intervals (whiskers) for 10-μg·m−3·week−1 increments of air pollutant, after adjustment. PM10: particles with a 50% cut-off aerodynamic diameter of 10 µm.
Benzene, toluene and ethylbenzene were the VOC for which we found a significant association between increasing exposure and airway changes (table 5 and fig. 1), with all three being related to a deterioration of lung function. For benzene, we found a decrease of FEV1 (regression coefficients -4.33, 95% CI -7.13– -1.53), FEV1/FVC (-1.71, 95% CI -3.24– -0.18) and FEF25–75% (-5.89, 95% CI -10.16– -1.62) and an increase of ΔFEV1 (2.79, 95% CI 0.92–4.65). For toluene, we found a decrease of FEV1 (-1.10, 95% CI -1.97– -0.23) and an increase of ΔFEV1 (0.97, 95% CI 0.44–1.50). For ethylbenzene we observed a decrease of FEV1 (-1.79, 95% CI -3.32– -0.25) and FEF25–75% (-2.48, 95% CI -4.81– -0.16) and an increase of ΔFEV1 (1.30, 95% CI 0.27–2.35).
For VOCs, EBC pH exhibited a consistent trend of negative associations (fig. 2), reaching significance in the case of benzene (-0.24, 95% CI -0.42– -0.06) and ethylbenzene (-0.14, 95% CI -0.23– -0.04). VOCs also presented a trend of positive associations with FeNO, namely ethylbenzene (1.99, 95% CI -0.00–3.99).
Clinical outcomes were positively associated with toluene, for which an increase of rescue medication and emergency department visits were related with increasing exposure: 0.21 (95% CI 0.01–0.42) and 0.26 (95% CI 0.06–0.46), respectively (table 6). Positive associations were also found between need of rescue medication and benzene (0.76, 95% CI -0.11–1.62) and ethylbenzene (0.45, 95% CI 0.02–0.87). The negative associations between increasing exposure to PM10 (-0.70, 95% CI -1.14– -0.25) and NO2 (-2.08, 95% CI -3.59– -0.58) and a decrease of reported wheezing symptoms were unexpected.
The results from the application of two-pollutant models for FEV1 and pH on EBC are presented in the online supplement. Results that were similar, although not significant, to those of the one-pollution model were observed in the case of FEV1; benzene was the only pollutant that persisted with statistically significant negative associations but for pH, PM10 was the pollutant that prevailed.
DISCUSSION
In this study, we methodically assessed total exposure of children to different pollutants. To this extent, we took into account the different indoor and outdoor environments where children spend their day, which can provide different concentrations of pollutants. Our data illustrate that mean levels of major air pollutants in just one microenvironment could either underestimate or overestimate the real exposure of the child. Differences in air quality were found between houses and schools. Moreover, there were also differences between houses.
During the week of the study, increasing total exposure to PM10, NO2, benzene, toluene and ethylbenzene was associated with a decrease of FEV1 and an increase of ΔFEV1. Increased exposure to NO2 and benzene was also associated with a decrease of FEV1/FVC and FEF25–75% while ethylbenzene was also associated with a decrease of FEF25–75%. Increasing exposure to PM10, NO2, benzene and ethylbenzene was associated with acidity of EBC. Ethylbenzene was the only pollutant with a significant positive association with FeNO. Toluene was the only pollutant positively associated with symptoms in the previous months.
Exposure to PM10, as estimated using a function of both concentrations and daily activity patterns, was related to lung function decline, even in a non-industrial city. We did not find any significant association between PM10 and FeNO as suggested in other studies [9, 25], although we adjusted the analysis for important confounding factors [23]. However, it should be noted that in our study, airway acidity was related to different pollutants, namely PM10, NO2, benzene and ethylbenzene, suggesting a link between airways inflammation.
The spirometric findings for PM10 and NO2 are not in agreement with the negative associations found for wheezing in the previous months. This could to be in accordance with a previous study [26] where increasing exposure to PM10 was associated with reduction in peak expiratory flow, but not with wheezing or dyspnoea.
Benzene, toluene and ethylbenzene were the VOCs associated with lung function deterioration and airway inflammation. Toluene is a known cause of occupational asthma [27], where exposure to higher levels is common. Low levels of VOCs seem to increase the risk of childhood [19] and adult [28] asthma symptoms but associations with lung function have been more difficult to find [29]. Ethylbenzene has also been suggested as a significant risk factor for asthma [19], a fact that is in accordance with the airways changes found in our study.
To our knowledge, this is the first time that associations between individual's exposure to air pollution, using both indoor and outdoor air pollution monitoring and modelling techniques, and various respiratory outcomes were observed simultaneously. Our final results are in agreement with most previous epidemiological evidence on either outdoor or indoor measurements. Other panel studies with asthmatic children looked for associations between air pollutants and airways changes through a comprehensive assessment that included spirometry, FeNO and EBC analysis [9, 30]. However, in the majority of previous studies, only outdoor air pollutants were taken into account. In a panel study with 53 asthmatic children, Delfino et al. [21] assessed PM2.5 and NO2 through personal active air samplers worn in a backpack, reaching a negative association between FEV1 and increasing personal exposure to PM2.5 and NO2.
A major strength of our study is the fact that for each child, the exposure calculation took into account different indoor (home and school) and outdoor microenvironments through direct measurements or modelling methods, combining daytime activities. It has recently been indicated that future research will require increasing specificity of exposure assessment to identify the potential roles of individual exposure to air pollution components. This could elucidate potential mechanisms and facilitate studies of mixtures and gene–air pollution interactions [31]. It could be important, namely for PM10, NO2 and benzene, as they usually have urban transport as their major sources. However, in our study, traffic exhaust did not seem to be the most important source of PM10 and benzene for individual exposure. In addition, the correlation matrix among the pollutants showed that the patterns were strongly influenced by the daytime activity profile and by the different environments used for the exposure calculation, which is difficult to take into account.
Additionally, we chose a prospective study with repeated measures and with uniform medical evaluations and exposure assessments to different pollutants. The use of different objective outcomes (spirometric and inflammatory) is also a point that we should emphasise.
A limitation of our investigation could result from the fact that we considered only 4 weeks of a child's life which prevents the study of causality. We did not consider the weekends and we are aware that this approach may have affected the total exposure to important indoor residential air pollutants. Furthermore, the proper role of each pollutant could not be identified in our study. However, new data have shown that multi-pollution is an important phenomenon to take into account in the assessment of health effects of air quality [32]. In order to deal with this issue, we applied two-pollutant models. However, this approach does not allow us to fully take into account the effects of the other pollutants. Therefore, our data do not allow the separate effects of the various pollutants to be established but the observed effects can be indicative of a pollution mix. In addition, we decided to study only the effects of individual exposure to air pollutants over the airways in a susceptible group: wheezers. A future research area to clarify the links between air pollution and respiratory effects should include a control group.
Conclusion
Our study suggests a relationship between total exposure to air pollutants assessed in various environments and airways changes in wheezing children. It also suggests that attention should be dedicated to air quality in houses and schools in childhood as the majority of the children's time is spent in these environments, and namely to VOCs. These pollutants that are very common nowadays seem to have an impact on airways even at low concentrations [33].
Acknowledgments
The authors would like to thank the children and their parents. We thank also the school teachers and the authorities of Viseu, Portugal, who welcomed the study.
Footnotes
This article has supplementary material available from www.erj.ersjournals.com
Support Statement
The Saud'Ar study was supported by Fundação Calouste Gulbenkian. The authors thank the Portuguese Foundation for Science and Technology for the PhD grant of J. Valente (SFRH/BD/22687/2005).
Statement of Interest
None declared.
- Received February 9, 2011.
- Accepted June 5, 2011.
- ©ERS 2012
REFERENCES
