INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Over the past decade there has been a worldwide increase in morbidity and mortality from asthma, despite a better understanding of the underlying pathophysiology of this disease and availability of effective medications for treating it (1). Increases in hospitalization rates, emergency department (ED) visits, and deaths from asthma have been greatest in inner-city areas (2, 3). In New York City (NYC), the mortality rate from asthma has been estimated to be two to three times the national average (3). In addition, although NYC has only 3% of the population of the United States, it has 6% of all asthma hospitalizations in the United States (3). Within NYC, small area analyses have shown that excessive deaths and hospitalizations from asthma cluster in specific inner-city neighborhoods (2, 3).

Many theories have been postulated to explain why asthma morbidity and mortality have increased precipitously in some neighborhoods. Various components of the urban environment have been suggested as potential etiologic agents. In particular, ambient air pollution has been proposed as a probable contributor to global trends in increased incidence and prevalence of asthma. Constituents of ambient air pollution include ozone (O₃), sulfur dioxide (SO₂), nitrogen dioxide (NO₂), carbon monoxide (CO), and particulate matter of 10 µm diameter (PM₁₀). Population-based studies have demonstrated relationships between high levels of ambient air pollution and respiratory symptoms (4, 5). Epidemiologic studies have shown associations between increased levels of outdoor air pollutants and respiratory morbidity and mortality in patients with underlying lung diseases, particularly asthma (4).

A host of epidemiologic studies have demonstrated an association between high atmospheric O₃ levels and both ED visits and hospital admissions for asthma (5, 10). Controlled human exposure studies have shown that O₃ exposure can produce decrements in FEV₁ and FVC, as well as increases in bronchial reactivity in both normal and asthmatic subjects (4, 13). The question remains of whether there are susceptible subpopulations of individuals with asthma who have increased vulnerability to the deleterious effects of O₃.

Because cigarette smoking is prevalent in populations of low socioeconomic status (16, 17), and because the urban poor are at high risk for asthma-related morbidity, the possible interaction between cigarette smoking and environmental factors is of particular interest. Epidemiologic studies have not examined the relationship between personal tobacco use and ambient O₃-associated morbidity from asthma. We hypothesized that tobacco use might influence the susceptibility of individuals with asthma to the adverse effects of increased levels of ambient O₃. In order to determine whether asthmatic individuals who smoke cigarettes are at increased risk for asthma-related morbidity during periods of high ambient O₃, we examined the relationship between cumulative lifetime cigarette use and O₃-associated ED visits for asthma in the inner city.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The study population consisted of a cohort of subjects obtained from the adult Bellevue Hospital Primary Care Asthma Clinic (BHPCAC), a large public clinic in NYC. All patients enrolled in the clinic between July 1992 and December 1995 and who completed the BHPCAC database questionnaire were included in the analysis (n = 1,216). This previously validated questionnaire provided baseline information on demographic characteristics, asthma symptom frequency, functional status, and smoking history. Clinical diagnoses were made by physicians' assessment at the time of enrollment of a patient in the clinic. The diagnosis of asthma was based on the guidelines of the National Asthma Eduation and Prevention Program (NAEPP) of the National Institutes of Health (NIH) (18).

Detailed information about the duration and amount of cigarette use was available for 1,115 patients. The cumulative amount of tobacco use was calculated as pack-yr of cigarettes smoked. Never smokers (NS; n = 552) were defined as having zero pack-yr of cigarette use. Among patients who had smoked at some time in their lives, the median number of pack-yr of cigarette use was 13. Heavy smokers (HS; n = 285) were defined as those with  13 pack-yr of smoking; light smokers (LS; n = 278) were defined as those with < 13 pack-yr of tobacco use.

Outcome Variables

Data on all ED visits made by the study subjects to any of the 11 NYC public hospitals was obtained from the NYC Health and Hospitals Corporation database for the period between January 1, 1989 and December 31, 1993. Discharge diagnoses were coded by the ED physicians and were identified by International Classification of Diseases, Ninth Revision (ICD-9) codes. ED visits for asthma were identified by discharge diagnosis codes 493 to 493.9. ED visits for other respiratory conditions were tracked separately, using the following ICD-9 discharge diagnosis codes: 496 for chronic obstructive pulmonary disease (COPD), 490 for bronchitis, 492 for emphysema, and 494 for bronchiectasis. ED visits for nonrespiratory causes were also identified by discharge diagnosis codes.

Ambient Air Pollution and Weather Measurements

Daily atmospheric air pollution levels were tracked with the U.S. Environmental Protection Agency's Aerometric Information Retrieval System (AIRS). Hourly observations of O₃, CO, NO₂, and SO₂ were obtained from AIRS, and 24-h daily averages were computed for each day of the study period. Daily data were not collected for particulate matter in NYC at this time. Available pollution readings were obtained from sites throughout NYC. We selected sites with continuous records during the study period, and with site characterizations indicating no strong impact from industrial emission sources (i.e., "land use" = "residential" or "commercial," but not "industrial"). Hourly temperature and relative humidity readings measured at LaGuardia Airport in NYC were retrieved from the National Climactic Data Center's Surface Hourly Observation database (Earth Info., Inc.). Daily average temperature and relative humidity values for each day of the study period were computed.

Statistical Methods

Time-series regressions were done of ED visits on O₃ and covariates including weather, season/long-wave fluctuations, and day of week. To accommodate small daily counts, Poisson regression models (log-linear generalized linear models) were used. To control for seasonal patterns and trends, locally estimated smoothing splines (loess) of the time unit (day), with a span that captured periodicity of 3 mo or longer, were included. Poisson regression models that included loess were established with the generalized additive models method (19) using the statistical package Splus (StatSci; MathSoft Inc., Seattle, WA, 1993). Indicator variables were used to control for day-of-week patterns. To allow for nonlinear effects of temperature, we included loess of temperature (span = 0.5 of the data range) with lags of up to 4 d to determine optimum temperature effects. This span was chosen on the basis of preliminary exploration and past experience with hospital admission data analyses. Effects of relative humidity were likewise examined.

Controlling for confounding variables of season, temperature, and day of week, we studied the relationship between daily ED visits for asthma and 0- to 3-d lags of O₃. As with temperature, we explored possible nonlinear effects of O₃ by including loess of the lagged O₃ values. These analyses were done both with daily average and with 1-h daily maximum O₃ levels. Assuming a linear O₃ predictor, we report the relative risk (RR) of ED visits for asthma and the relevant 95% confidence intervals (CIs). For the Poisson model, the RR was exp(O₃ coefficient × O₃ increment). We used an O₃ increment of 50 ppb for daily average and of 100 ppb for daily 1-h maximum O₃ levels. These increments approximate the magnitude of (maximum  mean) the respective measured daily levels during summer months in NYC. To compare the RRs estimated for the four gaseous pollutants, we also used the interquartile range of each pollutant as the increment with which to calculate the RRs. ED visits for nonrespiratory reasons were aggregated separately, and the relationship between these ED visits and ambient O₃ levels was examined in a similar manner.

In order to test the relationship between personal tobacco use and O₃-associated ED visits for asthma, we subgrouped subjects by smoking status. As previously described, subjects were categorized as NS, LS, or HS. Time-series regressions of ED visits for asthma on daily average O₃ levels, as well as on daily 1-h maximum O₃ levels, were done for each subpopulation. Dispersion parameters in the residuals for all the subcategories were less than 1.02. Autocorrelation was negligible after smoothing of time and day-of-week variables in the regressions. Time-series regressions of ED visits for asthma on 0- to 3-d lags of daily average SO₂, CO, and NO₂ levels were subsequently done for the HS in order to to determine whether other ambient pollutants had a similar effect.

To determine whether observed associations were a reflection of increased ED use by subjects with certain characteristics, we studied the relationship between frequency of ED visits for asthma and individual patient characteristics. Frequency of ED visits for asthma was modeled through Poisson regression analysis, with the individual characteristics of age, gender, race/ethnicity, and number of pack-yr of cigarette use as covariates.

To confirm the results of the time-series analyses, logistic regression analyses were performed, with ED visits on high-O₃ days coded as 1 and visits on other days coded as 0 as a dependent variable. The covariates included age, gender, race/ethnicity, and 5 pack-yr categories (0, 1 to 5, 6 to 13, 14 to 30, and 31+ pack-yr) as indicator variables. High-O₃ days were defined as the highest quartile of the average 1- and 2-d lags for the daily 1-h maximum and daily average O₃ levels. The references for noncontinuous variables were: female for gender, white and other for race/ethnicity, and 0 pack-yr for the pack-year categories. To account for the possibility of multiple ED visits by the same patients, we also repeated the selected models method using generalized estimation equations (GEEs) (20). All of the logistic regression analyses were done with the SAS (SAS Institute, Cary, NC) statistical package. The publicly available SAS GEE macro (version 2.02) was also used.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Demographic and Clinical Characteristics of the Study Population

The demographic characteristics of the study population are given in Table 1. The median age of the study cohort was 39 yr (range: 18 to 84 yr). The cohort consisted predominantly of women (60%), and the majority of patients were from ethnic/ racial minorities: (58% Hispanic, 25% African-American). Many of the patients qualified for Medicaid (48%), and a large percentage were uninsured (39%), indicating that most were among the poor or working poor, respectively. All but 19 of the patients were residents of NYC, and most were from neighborhoods at highest risk for asthma morbidity (3). Most of the subjects in the cohort had severe asthma as based on symptom frequency (18); 57% had daily wheezing and 57% had nightly sleep disturbances from asthma. Similarly, many subjects suffered from substantial functional impairment because of respiratory symptoms; 34% were unable to walk more than two city blocks, and 65% were unable to climb more than two flights of stairs.

Forty-nine percent of the cohort (n = 552) had never smoked cigarettes, and were designated as NS. Of the 563 subjects who reported cigarette use, 49% (n = 278) were characterized as LS and 51% (n = 285) as HS. The median number of pack-yr of cigarette use by the LS was 5; the HS had a median of 30 pack-yr.

Subject smoking status was used to define three subpopulations. The number of ED visits for asthma made by each of these subpopulations was similar (n = 1,038, 1,005, and 977 for the NS, LS, and HS, respectively). The demographic characteristics of each subpopulation are described in Table 1. The LS were younger than the HS and the NS. More of the NS were female than of the LS or HS. The vast majority of subjects in all three smoking categories were from racial/ethnic minorities of African-American or Hispanic descent (82%, 85%, and 83% of the NS, LS, and HS, respectively). The percentage of patients who received Medicaid or who were uninsured was not significantly different among HS, LS, and NS (85% 92%, and 87%, respectively). Most of the subjects in each of the three smoking subpopulations had severe asthma as defined by the NIH/NAEPP guidelines (18). For example, a similar percentage of subjects in each of the smoker subpopulations had daily wheezing (n = 57%, 55%, and 60% of NS, LS, and HS, respectively) or nightly sleep disturbances from asthma (n = 56%, 51%, and 61% of NS, LS, HS, respectively).

ED Data

Between 1989 and 1993, the 1,216 patients in the study cohort made a total of 6,335 visits to the EDs of the 11 public hospitals in NYC. Most of the visits (n = 3,880; 61%) were to the Bellevue Hospital ED. The remainder of the visits were distributed among the other 10 HHC facilities located throughout NYC. Most ED visits made by the study population had ICD-9 discharge diagnosis codes that fell under the broad category of respiratory diseases (n = 3,497; 59%). Of these visits for respiratory diseases, the vast majority had discharge diagnosis codes 493 to 493.9, indicating a diagnosis of asthma (n = 3,024; 87%). ED visits for nonrespiratory conditions (n = 2,392) had a wide range of discharge diagnosis codes.

Air Pollution Data

Maximum hourly O₃ levels infrequently exceeded 80 ppb, with the 90th percentile at < 80 ppb for the period between 1989 and 1994. During the study period, the maximum 1-h O₃ value was 174 ppb and the mean was 37.2 ppb. The maximum 24-h average O₃ value was 74 ppb. The distributions of 24-h average levels of CO, NO₂, and SO₂ during the study period are shown in Table 2. Daily average temperature and humidity readings are also described in Table 2.

Time-Series Analyses

Time-series regression analysis was done on ED visits for asthma in response to 0- to 3-d-lagged 24-h average O₃ levels. This analysis revealed a trend toward an increase in the RR for total ED visits for asthma in response to increases in O₃ levels. The trend was most significant for 2-d-lagged 24-h average O₃ levels (RR = 1.22 per 50 ppb; 95% CI: 0.98 to 1.53) (Figure 1). This lag pattern remained the same regardless of the lags and averaging of temperature in this model. Temperature was negatively associated with ED visits for asthma. Same-day and 1-d lags were most significant in combinations of O₃ and temperature; however, other lags of temperature were also negatively associated with ED visits for asthma. Without temperature in the model, the RR for 2-d-lagged O₃ was 1.15 (95% CI: 1.00 to 1.31). With same-day temperature in the model, the RR per 50 ppb increase in 2-d-lagged O₃ was 1.22 (95% CI: 0.98 to 1.53). This model, with loess of 2-d-lagged O₃, is shown in Figure 1. Including the average of all the temperature lags (0 to 3 d) did not alter the RR for the 2-d-lagged O₃ (RR = 1.22; 95% CI: 0.96 to 1.55). We therefore fixed the temperature lag at zero in the subsequent analyses. Relative humidity was not a significant predictor of ED visits for asthma. The temperature-humidity interaction was not significant, and reduced the loess-smoothed temperature significance, but did not affect the O₃ coefficients. The fitted seasonal cycles appear to have captured spring and fall peaks in ED visits for asthma. As expected, analysis of day-of-week indicator variables revealed an increased RR for ER visits for asthma on Mondays. There was a slight positive association between non-respiratory-related ED visits and ambient O₃, which did not achieve significance (RR = 1.11 per 50 ppb; 95% CI: 0.89 to 1.38).

View larger version (13K): [in this window] [in a new window]   Figure 1.   Time series regression analysis of the relationship between O3 and ED visits for asthma for the total study population. Dotted lines represent the pointwise standard error bands for analysis.

Time-series regression analyses were done on ED visits for asthma made by NS, LS, and HS. HS displayed a monotonic increase in the RR of ED visits for asthma in response to increases in 2-d-lagged daily average O₃ levels (RR per 50 ppb = 1.72; 95% CI; 1.13 to 2.62) (Figure 2). In contrast, no association between ED visits for asthma and ambient O₃ levels was found for either LS or NS (RR = 0.88 per 50 ppb; 95% CI: 0.6 to 1.28; and RR = 0.88 per 50 ppb; 95% CI: 0.57 to 1.12, respectively). Similar results were obtained when time-series analyses were done on ED visits for asthma with 1-h daily maximum O₃ levels. In this analysis, the RR was 1.59 per 100 ppb O₃ (95% CI: 1.08 to 2.36) for HS, compared with an RR of 0.89 per 100 ppb (95% CI: 0.62 to 1.16) for LS and an RR of 0.8 per 100 ppb (95% CI: 0.58 to 1.14) for NS. There was no significant association between ED visits for asthma and 0- to 3-d lags of SO₂, NO₂, or CO when using 24-h average daily levels of these pollutants (Table 3). After controlling for seasonal cycles and day-of-week effects, the dispersion parameters were negligible (< 1.02).

View larger version (13K): [in this window] [in a new window]   Figure 2.   Time series regression analysis of the relationship between O3 and ED visits for asthma for each subpopulation of asthmatic subjects. Never smokers were defined as smoking for 0 pack-yr, light smokers as smoking for < 13 pack-yr, and heavy smokers as smoking for  13 pack-yr. Dotted lines represent the pointwise standard error bands for analysis.

The question arose of whether the observed trend in O₃- associated ED visits for asthma among HS could be explained by an overall increased frequency of such visits by HS compared with NS and LS. To evaluate this possibility, we analyzed ED visits for asthma as a function of individual subject characteristics (Table 4). Being male and African-American are associated with a higher frequency of ED visits for asthma. There was no definable trend toward increased frequency of ED visits for asthma in association with increased personal tobacco use. The strongest association between ED use and cigarette smoking was noted in the 6 to 13-pack-yr smokers who constituted most of the LS, thus supporting the findings in Table 1. The absence of a linear association between cigarette use and ED use suggests that cigarette smoking is not the only determinant of ED use in this population.

Logistic regression analysis was done to further assess the relationship between ED visits for asthma and specific subject characteristics. This analysis showed that patients with the greatest amount of personal tobacco use were at greater risk for ED visits for asthma within 1 to 2 d of a high-O₃ day (Table 5). Neither age nor race/ethnicity were independent predictors of ED visits for asthma following high-O₃ days. Male gender was a negative predictor of ED visits for asthma following high-O₃ days. The analysis was repeated with the GEE approach, to take into account multiple visits by the same patient; however, the coefficients and significance were unchanged.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ozone is a major component of urban air pollution. Healthy individuals without underlying respiratory disease may experience adverse respiratory responses to high levels of O₃, and a substantial body of evidence indicates that asthmatic individuals are particularly at risk for such responses. Previous epidemiologic studies have shown an association between ambient O₃ and morbidity from asthma in both adults and children (4, 6, 21). However, the question remains of whether there are susceptible subpopulations of asthmatic individuals who are particularly at risk during periods of high O₃ levels. Previous large-scale studies have been unable to fully address such susceptibility because they have lacked sufficient detailed demographic and clinical information about study subjects.

Our study cohort was obtained from the BHPCAC, a well-characterized urban asthma clinic population, and we therefore had the opportunity to use our detailed clinic database to determine individual patient characteristics. The subjects were evaluated by a physician at the time of their enrollment in the BHPCAC, and a clinical diagnosis of asthma was made on the basis of NAEPP guidelines (18). Although there can be overlap between the diagnoses of asthma and COPD, for the purposes of this study, only emergency room visits coded with the ICD-9 discharge diagnosis code for asthma were evaluated. We evaluated the relationship between ED visits for asthma and ambient O₃ levels in this cohort of adult inner-city asthmatic subjects at high risk for asthma-related morbidity. We observed an overall trend toward an increased RR for ED visits for asthma in response to increases in 2-d-lagged O₃ levels that was most evident above a 40 ppb daily average level of O₃.

To determine whether a particularly susceptible subpopulation of asthmatic individuals could be identified, we investigated whether cigarette smoking was an effect modifier in O₃-associated asthma morbidity. In pack-yr-specific analyses, we found that HS, but neither NS nor LS, had an increased RR for ED visits for asthma in association with increases in 2-d-lagged O₃ levels. Further analysis of additional ambient pollutants (SO₂, NO₂, and CO) failed to show a similar association, suggesting that the finding was specific for O₃. These findings were supported by logistic regression analyses that determined heavy cigarette use to be an effect modifier of ED visits for asthma on days following high O₃ levels. These data suggest that excessive tobacco use may increase the risk for ED visits for asthma on days following high O₃ concentrations.

The question arises of whether our observations about O₃-associated ED visits for asthma could be explained by confounding factors such as demographic characteristics. To resolve this issue, we performed logistic regression analysis to assess whether individual subject characteristics were predictors of ED visits for asthma during periods of high O₃ levels. Neither age nor race/ethnicity were independent predictors of ED visits on days following high O₃ concentrations.

One might question whether differences in the clinical severity of asthma in our three study subpopulations might account for our observations. In fact, NS, LS, and HS were similar in terms of the severity of their asthma. Most of the subjects in each of the three subgroups had moderate-to- severe persistent asthma, as defined by symptom frequency. Therefore, the observations about differences in O₃-associated ED visits for asthma among the three groups are unlikely to be due simply to more severe asthma in the HS than in the LS and NS.

Although our analyses included 0- to 3-d-lags in O₃, time-series analysis detected an association only between ED visits for asthma and 2-d-lagged O₃. Similar lagged effects have been reported by other investigators (25). Several factors may account for the delayed effects. Although patients with asthma may experience increased symptoms during periods of high O₃, there is often a delay between the onset of symptoms and a patient's arrival in the ED. Patients may initially attempt to manage worsening asthma at home before going to the ED. Indeed, Steib and coworkers conducted detailed patient interviews and reported the median interval between symptom onset and an ED visit to be 2.2 d (26). Hence, ED use may be a late and insensitive indicator of asthma-related morbidity in response to any trigger. Another explanation for the 2-d lag between symptom onset and an ED visit may lie in the time course for the development of airway inflammation in response to O₃ exposure. This well-described inflammatory response is characterized by an accumulation of neutrophil-associated inflammatory mediators such as interleukin-6 (IL-6) and IL-8 (4, 27, 28). The response has been shown to progressively intensify over time, reaching significant levels 18 h after exposure (29). Thus, the delays between exposure and effect are biologically plausible.

The mechanism by which excessive personal tobacco use may influence O₃-associated ED visits for asthma is unclear. Exposure to significant doses of cigarette smoke may prime the airways of asthmatic individuals for inflammatory reactions to other airborne triggers. Alveolar macrophages of smokers are known to release greater amounts of reactive oxygen intermediates than those of nonsmokers (30). In addition, smoking is associated with airway inflammation characterized by neutrophil influx and increased levels of inflammatory mediators, including IL-6 and IL-8 (31). This may translate into a more intense inflammatory response to ambient O₃.

Controlled human exposure studies comparing the O₃ responsiveness of healthy smokers to that of nonsmokers have yielded conflicting results. Kerr and colleagues found no significant decrease in the lung function of smokers subjected to short-term O₃ exposure (32). Frampton and colleagues found that healthy smokers exhibited fewer symptoms and smaller decreases in lung function than did nonsmokers after short-term exposure to 0.22 ppm O₃ with exercise (33). In contrast, Hazucha and associates reported that exposure to high O₃ levels had significantly greater effects in smokers than in nonsmokers (34). However, these studies involved small numbers of healthy subjects, and did not include asthmatic individuals. The O₃ responses of healthy subjects, whether smokers or nonsmokers, may not adequately predict the responses of asthmatic subjects. In fact, chamber studies of asthmatic subjects exposed to O₃ have yielded perplexing results. Kreit reported that in response to O₃ exposure, normal and asthmatic subjects had similar increases in bronchial responsiveness, lung volume, and respiratory symptoms, but that asthmatic subjects had a greater degree of airway obstruction (14). Koenig and coworkers found little difference in physiologic responses of normal and asthmatic subjects to O₃ (13). Short-term controlled exposure studies may not adequately reflect the pathophysiologic response of the airways of asthmatic individuals to prolonged exposure to high levels of ambient O₃. Furthermore, chamber studies do not measure the effect of prolonged O₃ exposure on the response of asthmatic individuals to exposure to common environmental triggers of asthma.

A preponderance of epidemiologic evidence indicates that patients with asthma are at risk for significant morbidity during periods of increased ambient O₃ (9, 11, 12, 22, 23). Although our patient cohort was much smaller than the populations in most of the studies cited here, we were able to demonstrate an overall trend toward an increased RR for O₃-associated ED visits for asthma with increases in ambient O₃ levels. In fact, we were able to demonstrate this association despite the substantially lower ambient O₃ levels in NYC than in cities in which other published studies were performed, such as Mexico City (12). Moreover, when we subgrouped our population by pack-yr, the subpopulations became even smaller and our power to detect an association was even less; hence, adult NS and LS did not show an increased RR for O₃-associated ED visits for asthma. However, we cannot conclude that these subpopulations were unaffected by ambient O₃ concentrations, since small sample size may have precluded our ability to detect an association. Moreover, in our study, ED use was the only outcome variable used to assess effects of O₃. An ED visit is an indicator of extreme respiratory morbidity. Respiratory symptoms were not assessed in this study, and less severe adverse responses to O₃ may therefore have occurred but remained undetected unless they resulted in an ED visit.

Additionally, our study population consisted entirely of adult asthmatic individuals. The results therefore cannot be extrapolated to other populations such as children. Asthmatic children have been clearly shown to be susceptible to the adverse effects of ambient O₃ even in the absence of exposure to environmental tobacco smoke (22, 23). Nevertheless, the finding of a strong association between ED visits for asthma and ambient O₃ in a small cohort of HS indicates that tobacco use may be a potent effect modifier for O₃-associated asthma morbidity in our adult urban population.

In summary, our data suggest that heavy tobacco use may act as an effect modifier for O₃-associated ED visits for asthma. Adult asthmatic individuals with extensive personal tobacco use may represent one subpopulation particularly susceptible to the adverse effects of high levels of ambient O₃. Prospective investigations, with this and other populations, are needed to further delineate the combined effects of personal tobacco use and exposure to ambient air pollution on asthma-related morbidity. Other, as yet unidentified factors may also sensitize populations at risk, particularly children, for exacerbations of asthma.