We followed a cohort of 1,150 children for 3 yr to investigate long-term effects of ambient ozone. Nine study sites were selected on the basis of air-quality data to represent a broad range of ozone exposure. In 1994, 1995, and 1996 lung function was recorded biannually, always before and after summertime. The effect of ozone was analyzed with regression analyses and study-site, a child's sex, atopy, passive smoking, baseline lung function, and increase in height were considered as confounding variables. A negative effect of summertime ozone on the pre- to post-summer-time change in FEV1 (ml/d) was present in 1994 ( β = − 0.019 ml/d/ppb; p < 0.01) and in 1995 ( β = − 0.017 ml/d/ ppb; p < 0.05), but not in 1996 ( β = 0.004 ml/d/ppb; p = 0.6); corresponding estimates for FVC were in 1994: β = − 0.022 ml/d/ppb, p < 0.005; 1995: β = − 0.018 ml/d/ppb, p < 0.05; and 1996: β = 0.006 ml/d/ppb, p = 0.46. When all three study years were considered simultaneously, i.e., the changes in lung function between each of two subsequent surveys being the dependent variable, summertime ozone was associated with a lesser increase in FEV1 ( β = − 0.029 ml/d/ppb; p < 0.001), FVC ( β = − 0.018 ml/d/ppb; p < 0.001), and MEF50 ( β = − 0.076 ml/s/d; p = 0.001). No consistent associations were observed for lung function and NO2, SO2 and PM10. Long-term ambient ozone exposure might negatively influence lung function growth.
Ozone constitutes one of the major outdoor pollutants in industrialized countries. For the United States it was estimated that some 70 million people live in areas where the former national ambient air-quality standard of 120 ppb is frequently exceeded (1). Given that the ozone standard has now been changed to 80 ppb, an even greater proportion of the population is considered to live in potentially harmful environmental conditions. Current evidence further suggests that even this standard might be not stringent enough to protect ozone-sensitive (2) subjects from reductions in lung function. Small reductions in volume parameters can be observed in healthy children exercising at ambient ozone as low as 80 ppb (3). Whereas it can be argued that these alterations are trivial and not associated with respiratory morbidity, an increase in hospital admissions for asthma has been observed at similar ozone exposure (1). In places like Mexico City with extremely high chronic ozone exposure epithelial upper airways damage can be observed where atrophy of the nasal mucosa is present in the vast majority of children (4). The formation of oxygen radicals represents the biochemical basis for these observations (5). Ozone reacts slowly with water to give reactive hydroxyl radicals, which are able to oxidize a wide range of biomolecules. They can diffuse freely into cells and lead to ATP depletion, sulfhydril oxidation, and DNA strand breaks. Glycolysis and mitochondrial respiration are impaired, which may lead to cell death. In animal studies there is evidence that oxidants may affect the development of lung architecture. In rats exposed to an urban profile of ozone with an average exposure of 0.18 ppm, mild restrictive changes in lung function were observed, which histologically corresponded to deposition of collagen in the airway epithelium (6). An intermittent and seasonlike exposure in monkeys over an 18-mo period resulted in restrictive lung disease caused by obliterative bronchiolitis (7). However, few studies in humans suggest that ozone exerts chronic effects on the respiratory system. Reduced levels of FVC and FEV1 were observed in Canadian children (8) living in an environment with high exposure to ozone and sulfates. In the NHANES study (9) reduced levels of FVC and FEV1 were observed at chronic ozone exposure above 40 ppb. However, the main disadvantage of some of these studies (8) is their cross-sectional design, which makes them prone to various forms of bias such as selection bias or season of assessment of lung function. Furthermore, in cross-sectional studies the duration of residence is likely to be highly variable in the study community, which is a major limitation for chronic exposure studies. Conceptionally, a cause-effect relationship cannot be studied with cross-sectional designs where the outcome as well as the exposure is assessed at the same time. We wanted to study whether chronic effects can be observed in an unselected cohort of children living in communities with different ozone exposure. We therefore prospectively followed a cohort of primary school children by repeated measurements of lung function. The present analysis covers the time period between January 1994 and December 1996.
Sites were selected on the basis of the 1991–1993 annual average ozone concentrations in two counties of Austria (Lower Austria and Styria) (Figure 1) to represent a broad range of ambient ozone exposure. Data were provided by the governmental air-pollution surveillance agency of the respective two counties. Sites with major industry in the vicinity were excluded to reduce confounding by other pollutants. From the nine sites selected, seven were rural communities with a population size of less than 10,000, and two were towns with a population size of approximately 20,000.
All study sites were equipped with fixed monitor stations, which were run by local authorities and provided 30-min means of ozone, NO2, and SO2 over the whole study period. Each site was equipped with one monitor station, which was located according to national standards in order to sample air, representative for the specific population. Nearby sources of emission were not allowed. Furthermore, buildings that would have prevented free air movement at the station, taking into account the local wind direction, were also forbidden. Sampling heads for ozone were located approximately 3.5 m above ground. Ozone was measured by the chemoluminiscence technique (ML8810; Monitor Laboratories, San Diego, CA) according to EPA guidelines. Exposure was calculated for each child by taking the mean of all 30-min means collected between his/her repeated lung function tests. For the summertime interval (June to September) participants were asked which time period (which weeks) they had spent in their community. “Individual” ozone exposure was then calculated as the mean for the respective period when the children stayed at the site. Across the nine communities, children stayed at their sites during the summertime between 67 and 89% of the weeks (average, 79%). For the fall to spring interval the mean exposure between the two surveys was calculated for each child.
NO2 was measured using nitric oxide (NO/NOx) analyzers (ML8840; Monitor Labs). Temperature was continuously monitored at all stations. Information on humidity was provided for four sites. Because particulate exposure was measured by local authorities as TSP only, we equipped for this study all but one site with Harvard impactors (Harvard Apparatus Co., Millis, MA) to measure PM10. PM10 was measured gravimetically on filters by using samplers equipped with 10-μm inlets. We used 14-d collection periods since monitor stations were visited by technicians at these time intervals. Filters for exposure assessment were provided by the Harvard School for Public Health (J.S.), and all filters were also weighed there. For this analysis PM10 data were available from September 1994 to November 1996.
In case of there being more than one school in a community, school authorities suggested a specific school that was then invited to participate. Parents of children enrolled in Grades 1 and 2 were then asked to participate. Older children were not asked in order to avoid a prepubertal growth spurt during the course of the study. After having gained informed consent for the study a questionnaire was distributed in the schools to be filled out by the parents at home. History of respiratory symptoms suggestive for asthma was obtained using standardized questions of the ISAAC questionnaire. Information on smoking habits, single room heating, and parental education was also gathered using a modified ATS questionnaire.
A skin prick test with seven common aeroallergens (cat, dog, birch, hazel, wheats, house dust mites; Allergologisk Laboratories A/S, Horsholm, Denmark), positive (histamin, 10 mg/ml) and negative (NaCl) controls were performed on the left forearm of all children. A child was considered atopic when a positive test, defined as a skin wheal of at least 2 mm in diameter and half the size of the histamin wheal, to any allergen was recorded.
Lung function tests were performed twice a year, between March and May and September and November when ozone concentrations would presumably be low. Because we were interested in long-term effects of ozone, this design was chosen to minimize a confounding effect of short-term exposure during the period of measurement. The time period between the spring and fall tests is referred to as “summertime” and the time period between fall and the next spring tests as “wintertime.”
Forced expiratory flow-volume curves (Jaeger; Würzburg, Germany) were performed in the schools. At first, each child had a training period as long as 5 min to get accustomed to the instrument. Then the child recorded a forced expiratory maneuver on the basis of ATS criteria. However, the 6-s zero-flow criterium at the end of expiration was not feasible in these young children. A trial of three to eight repeated tests was performed, and the test was accepted when the reproducibility of FVC was within 5%. Before data analyses all records with MEF25, MEF50, and MEF75 in the lower 5% percentile or the upper 95% percentile were reviewed for acceptability and reproducibility. The study was approved by the ethical committee of the University of Vienna, and written consent was given from the parents or guardian prior to all measurements.
Between 1994 and 1996 children were measured six times twice a year, always in spring and fall. When estimating a child's growth of lung function, the five respective differences between each of two subsequent surveys were used (dependent variable: change in FVC, FEV1, and MEF50). Generalized estimation equations (GEE) (10) were used to account for the correlation between observations from the same child. GEE result in asymptotically consistent estimators, granted the number of clusters (incorporated by the number of children) is large enough. We considered our study population large enough to rely on asymptoticity. All models were adjusted for time-independent covariates (FVC, FEV1, MEF50 at the first survey, atopic status, sex, and site) and time-dependent variables (passive smoking, season, change in height). For each change in lung function between two of six surveys the respective ozone exposure (summertime or wintertime) was the independent variable. We introduced a dummy variable for season (winter = 0, summer = 1) in order to allow for different intercepts and slopes for distinct levels of ozone exposure. In addition to these season-related specifications, a model allowing for estimation of separate period effects (year of study) was investigated (Table 5).
δFEV1 | δFVC | δMEF50 | ||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Model | β | SE | p Value | β | SE | p Value | β | SE | p Value | |||||||||
1994† | −0.0189 | 0.007 | 0.005 | −0.022 | 0.007 | 0.003 | −0.049 | 0.019 | 0.01 | |||||||||
1995 | −0.017 | 0.008 | 0.02 | −0.018 | 0.008 | 0.04 | −0.025 | 0.022 | 0.24 | |||||||||
1996 | 0.0041 | 0.008 | 0.6 | 0.006 | 0.008 | 0.46 | 0.003 | 0.024 | 0.9 |
Analysis was repeated implementing mixed models (SAS, PROC MIXED) and allowing site to be a random-effect variable, whereas all other variables were considered fixed. Site was included as a random-effect variable because sites represented a sample from possible monitor stations. However, covariance parameter estimates arising from the random specification of the site variable turned out to be nonsignificant. This suggested to include site as a fixed-effect variable. Furthermore a model considering additional clustering by site was analyzed. We herein present only results from GEE models. This was done because GEE models do not require assumptions on distributions compared with parametric models. However, all parametric models yielded results similar to the reported data. The statistical analysis was carried out with the SAS package, Version 6.11.
For the nine sites participation at start of follow-up ranged between 92.2% and 96.3%. For 1,150 children with a mean age of 7.8 ± 0.7 yr included at the start of the study, 1,060 recorded valid lung function tests at the sixth survey (loss to follow-up, 7.9%). A sample of 173 nonparticipants was studied using a much shorter questionnaire. Results showed that the prevalence of a doctor's diagnosis of asthma was somewhat more common among participants (4.7%) than among nonparticipants (2.9%). The main reason for nonparticipation was that children objected to undergoing repeated testing (20.8%). Characteristics of the study population are shown in Table 1. A positive skin prick test was observed in 13.4 to 25.8% of children at different sites. Passive smoke exposure ranged from 21.9 to 46.7%. Single room heating was reported in 25% of homes only at the utmost northern site, which was situated in a rural setting. Because of stringent test criteria for acceptability and presumably the children's young age, valid lung function tests were available for 86 to 95% of all children at the first survey. These site-specific percentages increased over the study to 96 and 100% at the last survey.
n | Boys(%) | Atopy*(%) | PSE†(%) | Single Room Heating (%) | High Parental Education (%) | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Amstetten | 141 | 56 | 20.9 | 41.3 | 12.2 | 22.8 | ||||||
St. Valentin | 120 | 47.5 | 21.7 | 32.8 | 12.7 | 30.8 | ||||||
Krems | 170 | 52.9 | 14.9 | 46.7 | 17.7 | 35.4 | ||||||
Heidenreichstein | 84 | 50 | 17.1 | 32.5 | 25 | 21 | ||||||
Gänserndorf | 102 | 51 | 16.9 | 39.8 | 9.4 | 39.4 | ||||||
Mistelbach | 154 | 54.5 | 13.5 | 31.2 | 14.5 | 25 | ||||||
Wiesmath | 129 | 53.5 | 13.4 | 21.9 | 7.2 | 21.6 | ||||||
Bruck | 75 | 46.7 | 25.8 | 43.2 | 19.1 | 26.8 | ||||||
Pöllau | 175 | 50.9 | 16.8 | 31 | 15 | 16.2 | ||||||
Total | 1,150 | 51.9 | 17.4 | 35.4 | 14.3 | 26.2 |
For descriptive purposes exposure to pollutants was calculated as annual means (Table 2). With regard to long-term exposure the site with the highest ozone exposure (Pöllau, 1994– 1996 mean: 39.1 ppb) demonstrated levels approximately twice as high as at the site with the lowest exposure (Amstetten: 18.5 ppb). This was true whether exposure was calculated as the annual mean in 1994 or as the mean from 1994– 1996. Average summertime ozone was 34.8 ppb (SD, 8.7 ppb) and wintertime ozone was 23.1 ppb (SD, 7.7 ppb). PM10 exposure showed little variation between the study sites. Correlation coefficients were calculated for the 14-d mean exposures since PM10 was measured at these time-intervals. Ozone was negatively correlated with NO2 and humidity and positively correlated with temperature (Table 3). All other correlations were weak or insignificant.
PM10 *(μg/m3 ) | SO2 †(ppb) | NO2(ppb) | Ozone(ppb) | Ozone(94–96 ppb) | Days with Ozone (maximum > 60 ppb) | Altitude(m) | Temperature(°C ) | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Amstetten | 21.7 | 3.75 | 15.1 | 18.0 | 18.5 | 44 | 270 | 10.5 | ||||||||
St. Valentin | 17.6 | 3.00 | 9.7 | 21.3 | 19.7 | 78 | 242 | 11.5 | ||||||||
Krems | 22.9 | 3.75 | 14.2 | 23.0 | 20.4 | 57 | 190 | 11.7 | ||||||||
Heidenreichstein | 17.6 | 4.13 | 6.2 | 30.2 | 29.8 | 50 | 560 | 7.9 | ||||||||
Gänserndorf | 22.9 | 5.63 | 10.3 | 27.2 | 25.5 | 81 | 161 | 11.1 | ||||||||
Mistelbach | 21.4 | 5.25 | 8.5 | 26.7 | 27.0 | 66 | 250 | 10.5 | ||||||||
Wiesmath | 13.6 | 6.00 | 5.6 | 40.7 | 37.9 | 99 | 738 | 8.9 | ||||||||
Bruck | 20.5 | 4.88 | 7.3 | 33.1 | 30.8 | 96 | 210 | 11.2 | ||||||||
Pöllau | -- | 2.25 | 2.7 | 37.7 | 39.1 | 61 | 1,180 | 6.8 |
PM10 | Ozone | SO2 | NO2 | Humidity | Temperature | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
PM10 | −0.07 | 0.46 | 0.37 | 0.22 | −0.15 | |||||||
Ozone | −0.07 | −0.27 | −0.62 | −0.72 | 0.52 | |||||||
SO2 | 0.46 | −0.27 | 0.35 | 0.44 | −0.59 | |||||||
NO2 | 0.37 | −0.62 | 0.35 | 0.3 | −0.43 |
Maximum 30-min ozone in the last 24 h before the lung function test was 58.7, 58.4, and 51.4 ppb at the spring surveys and 37.9, 39.7, and 33.7 ppb at the fall surveys. Mean summertime ozone exposure decreased during the study: 1994, 37.3 ppb; 1995, 35.4 ppb; 1996, 32.4 ppb.
Average height was 128.2 (SD ± 6.3) cm at the first survey and 140.6 (SD ± 7.1) cm at the last survey. Increase in height was on average 1.9, 3.4, 2.2, 3.3, and 2.5 cm between each of two surveys, suggesting that children had not entered their prepubertal growth spurt. In this time period mean FVC increased from 1.85 (SD ± 0.3) L to 2.36 (SD ± 0.4) L. The average increase for FVC per day was 0.65, 0.55, 0.63, 0.67, and 0.74 ml between one specific survey and the following one.
To study short-term effects of ozone survey-specific analyses was performed (Table 4). For FVC, FEV1, and MEF50 a significant negative association with previous-day maximum ozone concentration was observed for the fall survey in 1994, but not for other surveys.
FEV1 | FVC | MEF50 | ||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Model | β | SE | p Value | β | SE | p Value | β | SE | p Value | |||||||||||
1994 | Spring | 0.1 | 0.4 | 0.8 | 0.62 | 0.56 | 0.3 | 0.24 | 1.43 | 0.9 | ||||||||||
Fall | −1.5 | 0.56 | 0.006 | −1.6 | 0.52 | 0.003 | −4.6 | 1.8 | 0.01 | |||||||||||
1995 | Spring | −0.48 | 0.79 | 0.5 | −0.33 | 0.79 | 0.7 | −2.19 | 2.25 | 0.3 | ||||||||||
Fall | 0.17 | 0.71 | 0.8 | 0.04 | 0.78 | 0.9 | −1.15 | 2.09 | 0.6 | |||||||||||
1996 | Spring | −1.3 | 0.58 | 0.03 | −0.83 | 0.53 | 0.1 | −4.05 | 1.72 | 0.02 | ||||||||||
Fall | −0.7 | 0.85 | 0.4 | −1.1 | 0.9 | 0.2 | 1.28 | 2.32 | 0.6 |
For the year-specific analyses the change in a specific lung function parameter between the spring survey and the corresponding survey in the fall of the same year was the outcome. Exposure to summertime ozone was negatively associated with change in both lung function parameters, the effect being significant for 1994 and 1995. For the 1996 surveys, no significant association with ozone was observed (Table 5). Change in MEF50 was associated with ozone in 1994 only.
When combining data from all three study years, exposure to summertime ozone, and to a lesser extent to wintertime ozone, was negatively associated with change in all three parameters (Table 6). When the analyses was restricted to children who had spent the respective whole summer period at their site the parameter estimate for summertime exposure decreased (i.e., increase in effect) to β = −0.033 (SE, 0.007; p < 0.001) for FVC and β = −0.034 (SE, 0.009; p < 0.001) for FEV1. In a FVC model where atopy was defined as allergic sensitization to pollen only, atopy was of borderline significance (β = −0.08; p = 0.07), suggesting slower lung growth for sensitized children. An interaction term between sensitization and ozone was insignificant (β = 0.0027; SE, 0.003; p = 0.37). In most models season of measurement affected change in FVC. A smaller increase of FVC per day was suggested to take place during the winter period. For a hypothetical child living in a community where the summer ozone concentration for all three years was 10 ppb higher than for a child from another community with identical other background variables the model would predict a 48.7 ml FVC loss. This FVC loss was derived, assuming 90-d summer periods for each of three years: FVC loss (ml) = β*10*90*3. This FVC loss corresponds to an approximately 2% smaller FVC than the average FVC of 2.36 L.
δFEV1 | δFVC | δMEF50 | ||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Model | β§ | SE | p Value | β | SE | p Value | β | SE | p Value | |||||||||||
Ozone | Season† | 0.561 | 0.104 | < 0.001 | 0.587 | 0.109 | < 0.001 | 0.667 | 0.299 | 0.026 | ||||||||||
Summertime O3 ‡ | −0.029 | 0.005 | < 0.001 | −0.018 | 0.005 | < 0.001 | −0.076 | 0.014 | < 0.001 | |||||||||||
Wintertime O3 ‡ | −0.024 | 0.006 | < 0.001 | −0.01 | 0.006 | 0.08 | −0.084 | 0.016 | < 0.001 | |||||||||||
SO2 | Season | 0.577 | 0.071 | < 0.001 | 0.467 | 0.078 | < 0.001 | −1.24 | 0.495 | < 0.001 | ||||||||||
Summertime SO2 | −0.018 | 0.004 | < 0.001 | −0.009 | 0.004 | 0.02 | −0.059 | 0.010 | < 0.001 | |||||||||||
Wintertime SO2 | 0.003 | 0.001 | < 0.001 | 0.002 | 0.001 | 0.03 | 0.003 | 0.003 | 0.26 | |||||||||||
NO2 | Season | −0.011 | 0.066 | 0.87 | 0.005 | 0.07 | 0.95 | 0.058 | 0.192 | 0.76 | ||||||||||
Summertime NO2 | −0.004 | 0.009 | 0.65 | 0.02 | 0.01 | 0.04 | −0.058 | 0.03 | 0.051 | |||||||||||
Wintertime NO2 | −0.015 | 0.007 | 0.03 | −0.007 | 0.007 | 0.32 | −0.027 | 0.022 | 0.21 | |||||||||||
PM10 § | Season | 0.442 | 0.19 | 0.02 | 0.226 | 0.193 | 0.25 | 1.912 | 0.564 | < 0.001 | ||||||||||
Summertime PM10 | 0.003 | 0.012 | 0.77 | 0.023 | 0.012 | 0.06 | −0.074 | 0.034 | 0.029 | |||||||||||
Wintertime PM10 | 0.014 | 0.004 | < 0.001 | 0.018 | 0.004 | < 0.001 | 0.019 | 0.013 | 0.15 |
To investigate whether other metrics of measure would produce similar results further analyses were performed. Ozone exposure ⩾ 60 ppb was observed on average in 235 h during each of the three summer periods. When hours with ozone ⩾ 60 ppb was entered as explanatory variable the result indicated a 0.6-ml loss in FVC (SE, 0.001; p < 0.001) per hour of elevated exposure. No effect was observed for the winter period (three years mean of ozone ⩾ 60 ppb: 86 h, β = −0.39; SE, 0.09; p = 0.25).
The mean of daily exposure peaks (mean of daily maximum) was 50 ppb over the three summer periods and 34 ppb over the three winter periods. No effect on lung function was observed for this measure in the longitudinal analysis (for summer: β = −0.0058; SE, 0.004; p = 0.14; for winter: β = −0.0004; SE, 0.004; p = 0.9). This result is most likely due to measurement error because only one exposure value per day is used, whereas the cumulative ozone analysis uses 48 30-min values per day. To investigate a nonlinear relationship between ozone and FVC, a variable was entered into the model indicating exposure above or below the median ozone exposure (28.6 ppb). Parameter estimates were similar for both exposure levels (for ozone < 28.6 ppb: β = −0.014; SE, 0.007; p = 0.03; for ozone ⩾ 28.6 ppb: β = −0.015; SE, 0.006; p = 0.006), suggesting a linear relationship. When investigating the shape of the exposure-response curve a second-order polynominal, a third-order polynominal, and a model with log-transformed ozone data were fitted. However, the data were best described by the linear model.
For SO2, effects were not consistent. A negative parameter estimate was observed for SO2 exposure during the summer but a positive estimate for SO2 exposure during the winter period. For NO2 a negative effect was seen for winter NO2 on FEV1 only. For FVC the estimate for summer NO2 was positive, presumably because of the high negative correlation with ozone. In all models except for the NO2 analysis, season effected change in lung function, suggesting smaller lung growth in the winter season. Only in the NO2 model were the parameter estimates for season insignificant, demonstrating that NO2 confounded the effect of season.
For PM10, data were available from 1994 to 1996. Increase in ventilatory function was positively related to wintertime exposure. To investigate a confounding effect of temperature, PM10 and ozone models were rerun, including temperature. The correlation coefficients between temperature and PM10, and temperature and ozone, were −0.28 and 0.41, respectively. Results show that in three of four models summertime temperature was positively associated with lung function, whereas the PM10 effect was no longer present. However, results on ozone remained virtually unchanged (for FVC(summertime): β = −0.029; SE, 0.007; p < 0.001; FVC(wintertime): β = −0.04; SE, 0.009; p < 0.001; FEV1(summertime): β = −0.03; SE, 0.006; p < 0.001; FEV1(wintertime): β = −0.03; SE, 0.008; p < 0.001).
A further model was run, including an interaction term between passive smoking and ozone, which was insignificant for the former variable and did not change the ozone effect to a meaningful extent. Another model included a variable for an acute respiratory tract illness (defined as rales or wheeze on auscultation at the time of lung function measurement). Although the parameter estimate for the variable itself was significant (and negative) for δFEV1, it did not confound the effects of ozone. Stratified analyses for subgroups, i.e., those with a diagnosis of asthma or children with respiratory symptoms suggestive for asthma, showed no differerential effects of ozone. There was no effect of parental education on the change in lung function parameters either, nor did parental education confound the relationship between lung function and ozone. In order to exclude regression to the mean because of differences in baseline lung function associated with ozone exposure, analyses was rerun without initial lung function. Results show that parameter estimates were virtually unchanged for FEV1 (β(summertime) = −0.029; SE, 0.005; p < 0.0001) and FVC (β(summertime) = −0.018; SE, 0.005; p < 0.0007). The longitudinal analysis (Table 6) when using parametric models (SAS PROC MIXED) obtained the same results. An additional model also taking into account clustering by site gave the same results.
There is convincing evidence that ozone at ambient concentrations causes irritation of the upper airways, which at higher concentrations may lead to persisting epithelial damage (4). Animal models have convincingly shown that after chronic exposure at ambient concentrations, monkeys develop chronic respiratory broncholitis with replacement of proximal alveolar type I and type II cells by airway cells (called “bronchiolarization”) (7). These alterations may lead to remodeling of the lung infrastructure, which in some species has been shown to be irreversible (1). In humans, functional impairment of the respiratory system caused by ozone is characterized primarily by a decrease in lung volume parameters (11), caused by an inhibition of maximal inspiratory capacity and presumably inflammatory mechanisms elicited by ozone (5). Adverse effects are based on the inhaled dose, which depends next to concentration and inhalation time on ventilation rate. There are few studies that have provided data on small airways dysfunction such as effects on FEF25–75 (13), which might be relevant for chronic effects.
Hence, a high physical activity in children is a risk factor for adverse effects of ozone. Apart from exposure patterns that differ between children and adults, a higher susceptibility to air pollutants might apply to children. At ambient ozone exposure exercising children demonstrate larger decrements of lung function than do adults (1). Most epidemologic studies, both in children and in adults, have focused on short-term effects, and there is a lack of data on the issue of chronic effects. To our knowledge this is the first longitudinal study looking at effects of ozone on lung function growth. There was considerable growth of the children who were studied. Average height increased by approx. 12 cm, and an almost 26% increase in FVC was observed over the study period. However, we studied children before they entered their prepubertal growth spurt, thereby decreasing the variability of lung function growth at different years of the study. Rigorous ATS criteria were employed at lung function testing, and the same two field workers (F.H., A.V.) performed all the tests. Furthermore, there was very little loss to follow-up.
Epidemiologic studies are the only approach to investigate a possible link between chronic exposure to ozone and the occurrence of human health effects. Principal problems of these studies relate to the specification of individual exposure, to the coincident effects of other oxidant species (Nox) or other particulate pollutants, and to seasonal effects related possibly to meteorologic factors that affect pulmonary function tests. In the NHANES study (9), which included both adults and children, Schwarz reported a nonlinear relationship between chronic ozone exposure below 0.12 ppm and lung function (FVC, FEV1, PEFR). However, in their regression models other pollutants (TSP,NOx) showed similar relationships to lung function. A sample of 3,945 children 7 to 11 yr of age living in 10 towns with either high or low ozone exposure (90th percentile for 1-h maximum exposure, 80 versus 47 ppb) was studied by Stern and colleagues (8). They observed small decrements of FEV1 and FVC in the range of 2%, which corresponds to our results. Their findings were limited by a high correlation between ozone and sulfate exposure. In the classic studies by Detels and colleagues (12), the problem of disentangling multiple pollutants again remains the main issue when interpreting their results (i.e., loss of lung function with increasing ambient oxidant exposure). Effective lifetime exposure to ozone was estimated by a recent paper by Tager and colleagues (13) in first-year college students. Weighted averages of lifetime exposure were calculated by data from outdoor monitor stations and activity patterns in different ozone environments. Alterations in lung function were observed that suggested chronic effects on distal airways. Apart from lung function abnormalities other end points suggestive of chronic effects studied are COPD symptoms (14), histologic changes (15), or immunologic outcomes (16). All these studies show an association with ozone but were not able to unequivocally disentangle effects of individual pollutants. However, taken together the current evidence is suggestive of possible health effects from chronic ozone exposure (17).
In our study children lived at different levels of ozone exposures, with the highest and lowest exposure being different by a factor of two. Children were repeatedly measured at nine different sites, which were selected according to ozone exposure and other pollutants at very low levels. One problem in the analyses of our observation lies in the fact that children from the same community had very similar exposure data, thus artificially increasing the correlation between pollutants. The correlation between ozone and PM10 was only −0.07 when calculated from the ambient pollutant records, but it was −0.46 when calculated from the data set for the multivariate analyses. This factitious correlation made it impossible to adjust the ozone effect for other pollutants in the statistical models. However, in comparison with the models for the other pollutants, the ozone model showed robust results. Limiting the analysis to children who spent the complete summer at their community, we observed a much stronger effect of ozone. This finding somehow validated our assumption of a biologic ozone effect. One cocontaminant with summertime ozone exposure is fine sulfate particles, which were not measured in our study. Controlled exposure studies on the combination of ozone and acid aerosols are suggestive of an enhancement of the ozone effect by sulfate particles (1). In a children's summer camp study (18) we performed in the vicinity of the present study region, sulfates were measured concurrently with ozone. In this study an acute effect of acid particles and ozone on lung function was observed, with ozone exposure remaining at low levels. However, there was only a weak correlation between sulfates and ozone (r = 0.34). Hence, it is extremely unlikely that sulfate correlated with ozone over 3 yr at nine sites.
For the other pollutants measured in our study no consistent results were observed. SO2 was negatively associated with lung function in the summer but positively in the winter, which makes a causal relationship unlikely. One explanation might relate to the observation that a positive correlation between SO2 and ozone was seen in the summer period, but a negative correlation was seen in the winter period. For PM10 a positive effect was seen for the winter exposure, which was completely confounded by temperature.
Several controlled exposure studies on ozone have shown that atopic subjects might be more susceptible for the effects of ozone than nonatopic subjects (19). In our analyses no interaction between atopy and ozone was observed. However, our study was not designed to show such an interaction. Therefore, we have not measured allergen exposure.
In summary, we have observed during a 3-yr study small but consistent decrements in lung function in a cohort of school children, which were associated with ambient ozone exposure. This is the first study that suggests chronic effects of ozone on lung function growth in children. Thus, ozone would constitute a risk factor for premature respiratory morbidity during later life.
The writers are grateful to Dr. Bert Brunekreef for a thorough evaluation of the manuscript.
Supported by a grant from the government of Lower Austria (NU3) and by the Austrian Science Foundation (P-09507med).
Supported by the FWF 9507 MED Ministry of Science.
1. | Bascom R.State of the art: health effects of outdoor air pollution. Part 1. Am. J. Respir. Crit. Care Med.1531996356 |
2. | McDonnell W. F., Horstman D. H., Abdul-Salaam S., House D. E.Reproducibility of individual responses to ozone exposure. Am. Rev. Respir. Dis.13119853640 |
3. | Spektor D. M., Lippmann M., Lioy P. J., Thurston G. D., Citak K., James D. J., Bock N., Speizer F. E., Hayes C.Effects of ambient ozone on respiratory function in active, normal children. Am. Rev. Respir. Dis.1371988313320 |
4. | Calderon-Garciduenas L., Osorno-Velasquez A., Bravo-Alvarez H., Delgado-Chavez R., Barrios-Marquez R.Histopathologic changes of the nasal mucosa in southwest metropolitan Mexico City inhabitants. Am. J. Pathol.1401992225232 |
5. | Frischer Th., Pullwit A., Kühr J., Haschke N., Studnicka M., Lubec G.Aromatic hydroxylation in nasal lavage fluid following ambient ozone exposure. Free Radic. Biol. Med.221997201207 |
6. | Chang L., Huang Y., Stockstill N. L., Graham E., Grose E., Menache M., Miller F. J., Costa D. L., Crapo J. D.Epithelial injury and interstitial fibrosis in the proximal alveolar regions of rats chronically exposed to a simulated pattern of urban ambient ozone. Toxicol. Appl. Pharmacol.1151992241252 |
7. | Tyler W. S., Tyler N. K., Last J. A., Gillespie M. J., Barstow T.Comparison of daily and seasonal exposures of young monkeys to ozone. Toxicology501989131144 |
8. | Stern B. R., Raizenne M. E., Burnett R. T., Jones L., Kearney J., Franklin C. A.Air pollution and respiratory health: exposure to sulfate and ozone in 10 Canadian rural communities. Environ. Res.661994125142 |
9. | Schwarz J.Lung function and chronic exposure to air-pollution: a cross sectional analysis of NHANESII. Environ. Res.501989309321 |
10. | Carr G. J., Chi E. M.Analysis for repeated measures data: a generalized estimating equations approch. Statist. Med.11199210331040 |
11. | Hazucha M. J., Bates D. V., Bromberg P. A.Mechanism of action of ozone on the human lung. J. Appl. Physiol.67198915351541 |
12. | Detels R., Taskin D. P., Sayre J. W., Rokaw S. N., Massey F. J., Coulson A. H., Wegman D. H.The UCLA study population of CORD: X. A cohort study of changes in respiratory function associated with chronic exposure to Sox, Nox and hydrocarbons. Am. J. Public Health811991350359 |
13. | Tager I., Künzli N., Lurmann F., Ngo L., Segal M., Balmes J.Methods development for epidemiologic investigations of the health effects of prolonged ozone exposure: II. An approach to retrospective estimation of lifetime ozone exposure using a questionnaire and ambient monitoring stations. Investigator's report. Health Effects Inst. Res. Rep.8119982778 |
14. | Abbey D. E., Petersen F., Mills P. K., Beeson W. L.Long term ambient concentrations of total suspended particulates, ozone and sulfur dioxide and respiratory symptoms in a non-smoking population. Arch. Environ. Health4819933346 |
15. | Sherwin R. P.Air pollution: the pathobiologic issues. J. Toxicol. Clin. Toxicol.291991385400 |
16. | Zwick H., Popp W., Wager C., Reiser K., Schmoger J., Bock A., Herkner K., Radunsky K.Effects of ozone on the respiratory health, allergic sensitization, and cellular immune system in children. Am. Rev. Respir. Dis.144199110751079 |
17. | Environmental Protection Agency. 1996. Air Quality Citeria for Ozone and Related Photochemical Oxidants, Vol III. EPA/600/P-93/004cF. |
18. | Studnicka M., Meinert R., Studnicka-Benke A., Hajek K., Spengler J., Neumann M., Frischer Th.Acidic particles and lung function in children: a summer camp study in the Austrian alps. Am. J. Respir. Crit. Care Med.1511995423430 |
19. | Joerres R., Nowak D., Magnusson H.The effect of ozone exposure on allergen responsiveness in subjects with mild asthma or rhinitis. Am. J. Respir. Crit. Care Med.15319965665 |