Acute ozone (O
3
) exposure is associated with reversible decrements in pulmonary function (1,2), and O
3
inhalation can precipitate symptoms of lower respiratory irritation such as cough, shortness of breath, and pain with inspiration (3,4). Controlled chamber exposure studies have shown that O
3
is associated with acute declines in lung function and increased respiratory symptoms at exposure levels ranging from 80 to 400 ppb (2,5). Far less is known about the response of adults with more diverse demographics and fitness levels who are exposed to low-level O
3
during vigorous exercise in the ambient environment. Furthermore, there are inconsistencies between results of controlled exposure and observational studies regarding the exposure-response relationship and, in particular, whether there are levels of O
3
exposure below which acute health effects are not detectable.
We evaluated the acute effects of ambient O
3
and concomitant fine particulate matter (PM
2.5
) and aerosol acidity exposures on pulmonary function in a diverse population of healthy adults exposed to varying O
3
levels while hiking on Mt. Washington in the White Mountain National Forest of New Hampshire.
Photochemical reactions of nitrogen oxides and hydrocarbons from upwind industrial and urban areas of the central and northeastern United States produce episodically high O
3
levels in the Mt. Washington area. For example, hourly mean O
3
concentrations on the mountain between 1987 and 1993 ranged from 0 to 148 ppb (6). The White Mountain National Forest is a popular site for outdoor recreation, with over 7 million visitor days each year, including 60,000 hikers on Mt. Washington and additional hikers on neighboring peaks who are exposed to elevated O
3
while exercising.
Volunteers were solicited from adults (18-65 years of age) beginning a day hike on Mt. Washington from the trail entrance at Pinkham Notch, New Hampshire, on 78 days during the summers of 1991 and 1992. A study researcher, spirometry equipment, and a sign soliciting volunteers were placed in full view of all hikers who used this trail entrance. All passing hikers who agreed to participate were evaluated. The study researcher and hikers were unaware of the ambient O
3
or other pollutant levels. After explaining the study, the researcher obtained written informed consent from each subject before his/her evaluation. This study was approved by the Human Research Committee of Brigham and Women's Hospital. Only current nonsmoking subjects were included in these analyses.
Spirometry measurements
. Pulmonary function was measured outdoors at 620 m above sea level next to the entrance of the main trails to the summit of Mt. Washington. Each participant performed a minimum of three and a maximum of eight forced expiratory maneuvers before the day's hike and again after returning to the base. Prehike tests generally were performed between 0800 and 1030 hours and posthike tests were performed between 1500 and 1930 hours. Most (90%) posthike testing was performed within 25 min of completion of the hike. Subjects were tested, while seated and wearing nose clips, with a 12-liter Morgan Spiroflow spirometer (P.K. Morgan, Andover, MA) attached to a personal computer with customized software. The spirometer was calibrated twice daily with a 3-liter syringe, before the prehike and posthike assessments.
For each hiker, mean values for forced expiratory volume in 1 sec (FEV
1
) and forced vital capacity (FVC) were the means of the two or three best acceptable (7) and reproducible (± 5%) values. Values for forced expiratory flow between 25 and 75% of FVC (FEF
25-75%
) and peak expiratory flow rates (PEFR) were taken from the acceptable and reproducible maneuvers that had the maximum sum of FEV
1
and FVC. All spirometry values were corrected to body temperature and pressure saturated with water (BTPS). Percentage change in pulmonary function was calculated as 100 times the posthike minus prehike pulmonary function divided by the prehike value.
Questionnaire measurements
. Each subject was given a diary to complete during the hike and instructed in taking his/her own pulse. The diary included a record of each hiker's location, the time of day when the location was reached, and the hiker's pulse and respiratory symptoms at four points on the hike--beginning, half-way, high point, and finish.
After completion of his/her climb, each participant completed a modified American Thoracic Society-Division of Lung Disease (ATS-DLD) questionnaire (8) with questions regarding demographic information, past respiratory illnesses and symptoms, tobacco use, and weekly aerobic exercise. Information on medication use was not obtained.
Environmental exposure measurements
. Continuous O
3
measurements were made both at the Mt. Washington Observatory on the summit (1,910 m above sea level) and on the east side of the mountain base near the beginning of the Mt. Washington Auto Road (480 m above sea level). Summit monitoring used a chemiluminescent O
3
analyzer (Model 8410-E Ozone Analyzer; Monitor Labs, San Diego, CA) in 1991 and an ultraviolet photometric O
3
analyzer (Model 1008-RS; Dasibi Environmental Corp., Glendale, CA) in 1992. The base site used an ultraviolet photometric O
3
analyzer (Model 49-100; ThermoEnvironmental Corp., Franklin, MA). At the summit site, external O
3
monitoring performance was audited by the EPA and the state of New Hampshire Air Resources Division. At the base site, internal calibrations were conducted twice monthly. Ambient O
3
concentrations were expressed as the hourly mean of continuous O
3
monitoring. Hiker-specific mean O
3
exposures were defined as the average of both the summit and the base hourly ambient O
3
concentrations over the hours during which the individual hiked.
Fine particulate matter or PM
2.5
(aerodynamic diameter
2.5 µm in micrograms per cubic meter) and strong aerosol acidity (expressed here as sulfuric acid equivalent mass concentration in micrograms per cubic meter) were measured near the base of the mountain. Fine particulate matter was sampled on a Teflon filter with a 10-l/min Harvard Impactor (Air Diagnostics and Engineering, Inc., Harrison, ME) (9). Strong aerosol acidity was sampled with a modified Harvard Impactor with an ammonia denuder to minimize acid neutralization (10). Fine particulate matter and strong aerosol acidity samplers operated for approximately 10 hr (0800 to 1800 hrs) on most days during which spirometry was performed and results were reported as a single integrated daytime measure. Fine particulate matter and strong aerosol acidity samplers were audited for internal performance approximately halfway through each summer; measurements were corrected to standard temperature and pressure and reviewed for compliance with quality assurance and quality control specifications based on a modification of previously established standards (Harvard 24 City Acid Aerosol Study: Data Processing Procedures; July 1992). On one day of sampling, PM
2.5
and strong aerosol acidity samples failed to meet these specifications and were excluded from these analyses.
Ambient temperature was measured at the base of the mountain during each hiker's prehike and posthike spirometry. In addition, daily ambient mean temperatures were obtained from National Oceanographic and Atmospheric Administration weather sites on the summit (the Mt. Washington Observatory) and at the Appalachian Mountain Club facility at the base of the mountain.
Hiking (or exposure) time was defined as the difference between the times of the prehike and posthike spirometry tests for each participant.
Statistical methods
. Three pollutant exposure variables were considered in these analyses: the average of both the base and summit hourly O
3
concentrations for the hours of each hiker's hike (hiker's mean O
3
per hour of hiking); daily PM
2.5
concentrations; and daily strong aerosol acidity concentrations. Each hiker's mean O
3
exposure was calculated if hourly base and summit O
3
data were available for at least 50% of the hike. Five outcome variables were studied: the posthike percentage change in each of FEV
1
, FVC, FEV
1
/FVC, FEF
25-75%
, and PEFR. The relationship of O
3
to lung function was assessed using five different methods: quintile analysis, nonparametric smoothing, linear regression, piecewise linear regression, and logistic regression. Potential confounders of the relationship between O
3
exposure and acute changes in pulmonary function that were considered as linear terms in the models included ambient temperature (mean daily temperatures at the summit and base of the mountain as well as mean temperatures at the time of prehike and posthike spirometry) and daily PM
2.5
and strong aerosol acidity concentrations. Other covariates considered as linear terms were age, height, hours hiked, a proxy measure of fitness (average hours of aerobic exercise per week), and proxy measures of hike work [estimated backpack weight, self-reported maximum pulse over the hike, and percentage of age-predicted maximum pulse achieved during the hike (age-predicted maximum pulse was defined as 220 minus age)]; covariates considered as categorical variables were sex, smoking status (former versus never), a history of physician-diagnosed asthma, a history of chronic bronchitis, a history of any wheeze, a history of severe wheeze symptoms (at least two episodes of wheeze associated with shortness-of-breath over the previous year), year of hike, and proxy measures of hike work (carrying a backpack and reaching the summit of Mt. Washington). Covariates that were consistently significant in multivariate models (coefficients with two-sided t-tests with p<0.15) or that were considered to be of a priori importance (ambient temperature, sex, smoking status, history of physician-diagnosed asthma or severe wheeze, and certain proxy measures of hike work) were retained in the final models.
To assess effect modification by categories of respiratory disease history, smoking status, sex, hours hiked, and proxy measures of hike work, a Wald chi-square test for interaction between O
3
and each covariate was used, with hours hiked divided at the median. In addition, age was divided into 10-year age groups to compare the O
3
effect across four age categories (18-27, 28-37, 38-47,
48).
Hikers' mean O
3
exposures were divided into quintiles and plotted against the percentage change in each pulmonary function measure. The adjusted mean (± standard error) percent change in each pulmonary function measure for each quintile of mean O
3
exposure was estimated by the least-squares means option of a general linear model program (PROC GLM; SAS Institute Inc., Cary, NC). The shape of the exposure-response function was also assessed by nonparametric smoothing with the locally weighted regression loess function (11) in S-plus (Statistical Sciences Division of Math Soft Inc., Seattle, WA). The significance of nonlinearities was tested with nonparametric F tests of generalized additive models (12).
|
Figure 1
. Posthike percentage changes in (A) forced expiratory volume in 1 sec (FEV
1
) and (B) forced vital capacity (FVC) versus mean O
3
exposure after adjustment for age, hours hiked, sex, former versus never smoker, history of physician-diagnosed asthma or severe wheeze symptoms, carrying a backpack, reaching the summit, and mean ambient temperature during prehike and posthike spirometry. Error bars indicate 95% confidence intervals for mean percentage change in FEV
1
and FVC for each quintile of O
3
. Quintile test for trend: p = 0.08 for FEV
1
; p = 0.007 for FVC.
|
Linear regressions were modeled by ordinary least squares estimation and by methods that accounted for the potential intraclass correlation among hikers hiking on the same day (13). In addition, two linear regression models of the percent change in each of FEV
1
, FVC, FEV
1
/FVC, FEF
25-75%
, and PEFR as a function of mean O
3
exposure were assessed: one assuming a single slope relationship and a second assuming a piecewise relationship as a conservative approximation of nonlinearities suggested by the quintile and nonparametric analyses. For the piecewise models, an inflection point of 40 ppb mean O
3
was chosen on the basis of visual inspection of the results of the quintile and nonparametric analyses (Fig. 1). The piecewise model took the form
y =
+ ß
1
(x)(
) + ß
2
(x -40)(1-
), (1)
where y = percentage change in each pulmonary function measure, x = each hiker's mean O
3
exposure,
= 1 if the mean O
3
exposure was <40 ppb, and * = 0 if the mean O
3
exposure was
40 ppb. An F statistic was calculated to test the difference between the two coefficients for O
3
in the piecewise models.
Logistic regression modeling was used to assess the relationship between pollution exposures and the odds of having a greater than 10% posthike decline in each measure of pulmonary function.
Study population
. The study recruitment protocol (see Methods) did not assess the total number of eligible hikers; as a consequence, overall participation rates are not available. Of the 766 hikers who volunteered to participate over the 1991-1992 study period, 595 (78%) provided acceptable and reproducible spirometry data both before and after the hike. Of the balance (n = 171), 120 (70%) hikers did not return for posthike spirometry, and the remainder did not meet acceptability (n = 45) or reproducibility (n = 6) criteria for spirometry (see Methods). Current smokers (n = 19), hikers with incomplete ambient O
3
data (n = 35), and hikers missing smoking status, respiratory disease histories, or other necessary covariates (n = 11) were excluded, leaving 530 hikers tested on 74 days for analysis. Of these hikers, 507 (96%) had measures of PM
2.5
and strong aerosol acidity concentrations available. For the remaining 23 hikers (4%), 15 (3%) hiked on days when PM
2.5
and strong aerosol acidity measurements were not taken and 8 (2%) hiked on a day when the available PM
2.5
and strong aerosol acidity measurements did not comply with quality assurance and quality control specifications. The majority of hikers in the analysis population were healthy, white (97%), male (71%) never smokers (76%) (Table 1). Ages ranged from 18 to 64 years, with a mean of 35 years.
Hikers excluded because of unacceptable or unreproducible spirometry, current smoking, or missing O
3
or other covariates were slightly younger. More hikers were excluded in 1992. Otherwise, hikers excluded from these analyses did not differ substantially from the analysis population (Table 1).
Air pollutant exposures
. Peak O
3
levels at the mountain's summit were generally seen in early morning, with minimal hourly variation during midday (Fig. 2), a pattern seen at other high-altitude sites (6,14). The inverse of this diurnal pattern is typical at lower altitudes and in regions with significant local sources of O
3
pollution (15). As a result of daytime vertical mixing, O
3
levels at the mountain's summit and base were very similar during the usual hiking time of approximately 0900-1700 hours (Fig. 2). Hourly base and summit O
3
concentrations were available, on average, for 98% of hiking hours and ranged from 6 to 94 ppb. The average of the hourly O
3
concentrations during each hike ranged from 21 to 74 ppb with a mean ± standard deviation (SD) of 40 ± 12 ppb.
Figure 2
. Maximum and mean hourly ozone levels (ppb) at the summit and base of Mt. Washington for 74 days on which study hikers were evaluated in summer 1991 and 1992. EDT, Eastern Daylight Savings Time.
The median daily PM
2.5
concentration was 10 µg/m
3
with a maximum of 60 µg/m
3
; the median strong aerosol acidity concentration was 0.3 µg/m
3
with a maximum of 20 µg/m
3
. For most (89%) of the study days, strong aerosol acidity concentrations did not exceed 5 µg/m
3
(102 nmol/m
3
).
Hikers' mean O
3
exposures were correlated with daytime PM
2.5
(Spearman r = 0.77) and strong aerosol acidity levels (Spearman r = 0.62). The Spearman partial correlation (adjusted for 24-hr mean O
3
) between daytime PM
2.5
and strong aerosol acidity concentrations was 0.42.
Temperature
. Summer temperatures on Mt. Washington are generally mild at the base and cool on the summit. The average temperature at the base during prehike and posthike spirometry was 19 ± 3°C (range 9-29°C). On test days, the average temperature was 17 ± 3°C (range 8-25°C) at the mountain's base and 8 ± 3°C (range -2-16°C) at the summit. Although peak hourly O
3
levels do not correspond to the hottest periods of the day at this site, average daily temperatures and mean 24-hr O
3
levels were positively correlated at both the summit (r = 0.69) and the base (r = 0.59).
Exercise
. The average hiking time between the prehike and posthike spirometry was 8 ± 1.5 hr (range 2-12 hr). Most of the 530 hikers (75%) completed the 1,300-m climb to the summit of Mt. Washington, and 94% carried a backpack during the day's hike. For the 517 hikers with diary data, the mean self-reported maximum pulse rate attained during this exercise was 122 ± 26 beats/min or 66 ± 14% of age-predicted maximum pulse. Hikers varied widely in their usual activity levels, with self-reported aerobic exercise (a proxy measure of fitness) varying from 0 to over 50 hr/week, with a median of 6 hr/wk.
Pulmonary function and ozone
. After adjustment for multiple covariates, linear models demonstrated a 2.6% decline in FEV
1
[95% confidence interval (CI), 0.4-4.7] and a 2.2% decline in FVC (CI, 0.8-3.5) for each 50-ppb increment in hikers' mean O
3
exposure (Table 2). Use of the maximum rather than the mean (see Methods) acceptable and reproducible values for each pulmonary function measure did not change the results. These models did not include adjustment for the intraclass correlation among hikers hiking on the same day because these values were quite small (<5%) and nonsignificant. Adjustment for PM
2.5
and strong aerosol acidity concentrations did not change the observed inverse relationship between changes in FEV
1
and hikers' mean O
3
exposures, but this association was no longer significant (Table 2). The inverse association between ambient O
3
and changes in FVC was diminished and nonsignificant after adjustment for the other pollutants (Table 2).
While the linear model was an acceptable description of the dose-response relationship in this study, both the quintile analysis and a nonparametric smoothing function suggested nonlinearity in the relationship between posthike percentage changes in FEV
1
and FVC and hikers' mean O
3
exposures, with the steepest changes occurring at mean O
3
exposures above approximately 40 ppb (Fig. 1). These deviations from linearity were not significant at the p = 0.05 level in generalized additive models (p = 0.07 for FEV
1
; p = 0.08 for FVC). In adjusted piecewise models, there was a significant difference between the estimated effect of O
3
above and below 40 ppb (p = 0.005 for FEV
1
; p = 0.001 for FVC). Above 40 ppb, there was a 4.4% decline in FEV
1
(CI, 0.5-8.2) and a 3.2% decline in FVC (CI, 0.7-5.6) for each 50-ppb increment of mean O
3
exposure (Table 2). These inverse relationships above 40 ppb mean O
3
, although no longer significant, did not change substantially after adjustment for ambient PM
2.5
and aerosol acidity concentrations (Table 2).
Hikers with a self-reported history of physician-diagnosed asthma or severe wheeze symptoms over the previous year had significantly greater O
3
-associated decrements in FEV
1
(7.5% decline for each 50-ppb increment in O
3
) than other hikers (1.8% decline for each 50-ppb increment in O
3
) (Table 3). Adjustment for ambient PM
2.5
and aerosol acidity concentrations did not change the magnitude or significance of this difference. Male hikers had greater O
3
-related FEV
1
and FVC declines than female hikers; never smokers had greater O
3
-related FVC declines than former smokers; and longer hikes were associated with greater O
3
-related FEV
1
and FVC declines than shorter hikes, although none of these differences was significant (Table 3). There was no association between O
3
responsiveness and age (Table 3).
There were insufficient numbers of hikers with greater than 10% declines in FEV
1
(n = 11), FVC (n = 8), and FEV
1
/FVC (n = 4) to evaluate this group separately. However, 109 and 71 hikers had greater than 10% posthike declines in FEF
25-75%
and PEFR, respectively. After adjustment for multiple covariates, O
3
was significantly associated with greater than 10% declines in FEF
25-75%
[odds ratio (OR) = 3.67; CI, 1.25-10.78; p = 0.02 for each 50-ppb increase in mean O
3
].
Pulmonary function and fine particulate matter and aerosol acidity
. There were consistent associations between ambient PM
2.5
exposures and changes in FVC (0.4% declines; CI, 0.2-0.6 across the interquartile range for PM
2.5
concentration of 9 µg/m
3
) and PEFR (0.8% declines; CI, 0.01-1.6 across the interquartile range for PM
2.5
) that persisted but were nonsignificant after adjustment for ambient O
3
(Table 4). Changes in PEFR had the largest inverse association with strong aerosol acidity exposures (0.4% decline; CI, 0.1-0.7 declines across the interquartile range for aerosol acidity of 1.3 µg/m
3
or 26.5 nmol/m
3
). This association persisted (but was not statistically significant) after adjustment for mean O
3
(Table 4). Although there were associations between posthike changes in FEV
1
and both PM
2.5
(0.4% decline across the interquartile range of 9 µg/m
3
) and aerosol acidity (0.1% decline across the interquartile range of 1.3 µg/m
3
) exposures, these associations were diminished and no longer significant after adjustment for O
3
(Table 4). Changes in the other measures of pulmonary function (FEV
1
/FVC and FEF
25-75%
) had no consistent or significant associations with ambient PM
2.5
or strong aerosol acidity. These effect estimates were of similar magnitude to the estimated O
3
effect. For example, in multivariate models, each 15-ppb increase in mean O
3
(the interquartile range of mean O
3
exposures) was associated with a 0.8% decline in FEV
1
and a 0.6% decline in FVC (Table 2).
We assessed the effects of simultaneous exposures to ambient O
3
, PM
2.5
, and strong aerosol acidity on pulmonary function in an adult population with broad ranging ages and fitness levels; this has not been done in most other field studies of adult O
3
exposure (16-18), despite the fact that exposure to pollutant mixtures is characteristic of the ambient environment, which in turn affects diverse populations. The same magnitude of O
3
-associated declines in FEV
1
was seen after simultaneous adjustment for PM
2.5
and acid aerosols (Table 2). There were consistent, albeit diminished, O
3
-associated declines in FVC after adjustment for other ambient pollutant exposures (Table 2).
Based on the average ambient O
3
during exercise, the estimated effects of O
3
exposure were larger than have been described in most previous field studies (16-18) and experimental chamber studies (1,3). Among the Mt. Washington hikers with mean FEV
1
of 4.08 liters and mean FVC of 5.14 liters, our findings approximate declines of 106 ml and 113 ml, respectively, in each of these measures for each 50-ppb increment in mean ambient O
3
level. When restricted to mean O
3
exposures >=40 ppb, our effect estimates were even larger (Table 2). In a study of 24 male and female joggers by Selwyn et al. (16), results were consistent with a smaller effect estimate of 20-ml decline in FEV
1
for each 50-ppb increment in O
3
. Our results are closer to those of Brunekreef et al. (17) in their study of 23 male amateur cyclists and those of Spektor et al. (18) in a study of 30 exercising adults; their effect estimates were consistent, respectively, with 78- and 70-ml declines in FEV
1
and 111- and 105-ml declines in FVC for each 50-ppb increment in mean O
3
concentration during exercise. Differences in dose-response modeling are unlikely to explain differences in effect estimates among field studies. Previous field studies (16-18) also used a linear dose response as the best description of the ozone-pulmonary function relationship and were conducted at ambient O
3
concentrations similar to those observed on Mt. Washington.
Differences in exposure duration may explain the greater mean O
3
effect estimates seen in the current study. In three previous field studies (16-18), most subjects' exercise periods were less than 2 hr, compared with an average exercise period of 8 hr on Mt. Washington. On Mt. Washington, the number of hours hiked was an independent predictor of declines in measures of pulmonary function for FEV
1
and FVC [after adjustment for O
3
, each hour hiked was associated with a 0.3% decline (p = 0.05) in FEV
1
; data not shown]. In addition, longer hikes were associated with a greater mean O
3
effect (Table 3).
Although other factors contributing to the effects of O
3
exposure--minute ventilation, O
3
concentration pattern, and participants' ages, for example--may have increased or decreased the effect of mean O
3
exposure in this setting compared with other studies, estimates of each hiker's minute ventilation were not available. Proxy measures of hike work--reaching the mountain summit, carrying a backpack, maximum self-reported pulse, or percentage of age-predicted maximum pulse--were not significant independent determinants of decrements in pulmonary function, and the interaction between each of these measures and mean O
3
were not significant (data not shown). In addition, there was minimal hourly O
3
variation during hiking periods (Fig. 1), a circumstance that likely attenuates rather than enhances mean O
3
effects (19). Lastly, the hikers were, on average, older than participants in other field (17,18) and chamber studies of O
3
exposure (1-3), a characteristic associated with decreased O
3
responsiveness in other studies (20) but not here (Table 3).
Although we did not collect information on nonparticipating eligible hikers, it is unlikely that selection bias influenced the results because hikers' willingness to participate is very unlikely to have been related either to air pollutant levels or their responsiveness to pollutants. It is possible that unrecognized confounding influenced our results. However, the observed associations were adjusted for exposure to important potential confounders of the O
3
-pulmonary function relationship: concomitant pollutants and temperature. Humidity information was not available for these analyses, but, given the diurnal O
3
pattern (Fig. 1) and moderate-to-cool temperatures characteristic of the site, humidity is unlikely to have been an important confounder. In addition, positive confounding by exercise or exercise-induced asthma is unlikely to have occurred for several reasons. First, hikers with a history of asthma or severe wheeze symptoms over the previous year were no more likely than healthy hikers to hike on more polluted days (e.g., on days with O
3
concentrations above the median, 6% of hikers had a history of asthma or wheeze compared to 9% of hikers who hiked when O
3
concentrations were below the median; p = 0.19, data not shown). Second, the same work or exercise was performed by most participants; adjustment for the hours hiked, reaching the mountain summit, carrying a backpack, maximum self-reported pulse, or percentage of age-predicted maximum pulse did not change the estimated pollution effects (data not shown).
In general, O
3
-associated changes in pulmonary function are greater in natural than in controlled exposure settings (18). Although synergism or interaction among a variety of uncontrolled environmental factors have been hypothesized to play a role in this finding, the explanation for this discrepancy is unknown. As a consequence, there is ongoing controversy regarding the appropriate exposure-response relationship for O
3
-associated pulmonary function changes. Our findings provide additional support for increased effect estimates under conditions of naturally occurring O
3
exposure, particularly under prolonged exposure conditions such as hiking, and even after adjustment for concurrent environmental exposures (PM
2.5
, acid aerosols, and temperature).
Exposures to both particulate (21,22) and strong aerosol acidity (23,24) pollutants have been independently associated with acute decrements in pulmonary function. As was the case for O
3
, the observed associations between measures of pulmonary function and exposures to ambient PM
2.5
and aerosol acidity were larger than observations from other field studies (22,23). Furthermore, there were consistent effects of ambient PM
2.5
on both FVC and PEFR and of ambient aerosol acidity on PEFR after adjustment for O
3
(Table 4). Dockery and Pope (22) estimated a 0.15% decrease in FEV
1
(they did not report FVC results) and a 0.08% decline in PEFR for each 10 µg/m
3
increase in inhalable particulate matter (aerodynamic diameter 
10 µm) in an analysis combining results from several studies of school children. Assuming that 6 µg/m
3
of PM
2.5
corresponds to 10 µg/m
3
of inhalable particulate matter (22), our results are consistent with a greater effect estimate--a 0.25% decrease in FEV
1
and a 0.52% decline in PEFR after adjustment for multiple covariates (Table 4). At the same PM
2.5
concentrations and after additional adjustment for O
3
, the observed declines in FEV
1
and PEFR persist, but are nonsignificant and diminish to 0.09% and 0.44%, respectively (Table 4).
For each 6 µg/m
3
(125 nmol/m
3
) increment in acid aerosol exposure, our results include a 1.8% decline in PEFR (or 1.5% decline after adjustment for O
3
) (Table 4), a larger effect estimate than the 0.4-0.8% decrements described elsewhere, with a similar exposure in children (23).
Despite our relatively large effect estimates, the mean changes in pulmonary function in this setting were small and unlikely to result in clinical symptoms in most individuals. However, there is a wide range of individual susceptibility to pollutants (25,26). Susceptible individuals are likely to have substantial declines in pulmonary function that are not apparent in an assessment of population means. As a result, a relatively minor change in the overall mean effect of exposure (as reported here) may still result in substantial changes for certain sensitive subgroups. For example, several studies support increased O
3
susceptibility among asthmatics demonstrated by greater lung function response (25,27) or enhanced pulmonary inflammatory response (28). Hikers with a history of physician-diagnosed asthma or severe wheeze were a sensitive subgroup in these analyses, with a fourfold greater responsiveness to O
3
than other hikers (Table 3). This finding of increased susceptibility to O
3
was not affected by adjustment for concomitant pollutant exposures or exercise. In addition, because asthmatics taking medication may be less sensitive to O
3
, we are likely to have underestimated the maximum increased sensitivity of this subgroup because we lacked information regarding effect modification by medication. Lastly, our results demonstrated that a substantially increased fraction of the exercising population had significant declines in lung function (>10% declines in FEF
25-75%
, for example) on days with higher O
3
.
We observed significant effects of mean ambient O
3
, PM
2.5
, and strong aerosol acidity on pulmonary function in a wilderness area designated for air quality protection (6). Large numbers of visitors engage in prolonged outdoor exercise in this area and are thereby at risk for acute health effects related to ambient pollution exposures. Even without considering differences in individual susceptibility, the observed effect estimates are notable for having occurred among hikers exposed to hourly O
3
concentrations averaging 40 ppb, a relatively low level characteristic of much of the continental United States and well below the currently operant National Ambient Air Quality Standard of 0.12 ppm (120-124 ppb) for hourly O
3
. Furthermore, this is the first epidemiologic study to report the effects of long-term ambient O
3
exposures, which are applicable to recent revisions to the O
3
standard that are based on an 8-hr average 0.08-ppm (80-84 ppb) exposure limit (29). Physicians, public health officials, and the general public should be aware of the potential acute health impact of relatively low-level air pollutants not only among residents of urban and industrial regions but also among individuals engaged in outdoor recreation in certain wilderness areas.
