American Journal of Respiratory and Critical Care Medicine

The associations between lung function measures (spirometry and peak expiratory flow lability) and estimated 20-yr ambient concentrations of respirable particles, suspended sulfates, sulfur dioxide, ozone, and indoor particles were studied in a sample of 1,391 nonsmokers followed since 1977. Differences in air pollutants across the population were associated with decrements of lung function. An increase of 54 d/yr when particles < 10 μ m in diameter (PM10) exceeded 100 μ g/m3 was associated with a 7.2% decrement in FEV1, as percent of predicted, in males whose parents had asthma, bronchitis, emphysema, or hay fever and with increased peak expiratory flow lability of 0.8% for all females and 0.6% for all males. An increase in mean SO4 concentration of 1.6 μ g/m3 was associated with a 1.5% decrement in FEV1, as percent of predicted, in all males. An increase of 23 ppb of ozone as an 8-h average was associated with a 6.3% decrement in FEV1, as percent of predicted, in males whose parents had asthma, bronchitis, emphysema, or hay fever.

Few cohort studies of the effects of long-term ambient exposure to air pollutants on lung function have been conducted in adults. A cohort study of lung function in Los Angeles communities showed more rapid deterioration in lung function in more polluted cities, but the study suffered extensive loss to follow-up and was unable to differentiate between effects of different pollutants because it only included three communities (1). Other cohort studies, such as the study in six eastern or midwestern U.S. cities known as the Six Cities Study, have not yet reported follow-up adult lung function data associated with long-term concentrations of air pollutants. The Six Cities Study reported elevated respiratory symptom rates in children to be associated with air pollutants, but it found no associations between ambient air pollutants and lung function measures (2).

All of these cohort studies lacked continuous measures of ambient air pollutants according to exposure history before enrollment and throughout the follow-up period. Most lacked an evaluation of the interaction of multiple pollutants. The ongoing cohort study that enrolled nonsmoking Seventh-day Adventists throughout California in 1977 (the AHSMOG Study) has found associations between elevated incidence of respiratory disease and long-term concentrations of ambient air pollutants, but until recently that study has lacked lung function measures (3).

The study described here has retrospective exposure histories back to 1967 and updated exposures throughout the period of follow-up. The purpose of this paper is to relate lung function measures collected on this cohort in 1993 to long-term ambient concentrations of respirable particles less than 10 μm in diameter (PM10) as well as to conduct multipollutant analyses for other ambient air pollutants: suspended sulfates (SO4), sulfur dioxide (SO2), and ozone (O3).

Enrollment and Follow-up of Cohort

Other papers detail the methods of enrollment and follow-up used in our study (4). Briefly, a cohort of 6,338 non-Hispanic white nonsmokers who were 25 yr or older in 1976 and had lived 10 yr or longer within 5 miles of their present residence were enrolled in 1977 for a study on the health effects of long-term ambient concentrations of air pollutants. In April of 1977, 1987, and 1992, study subjects completed respiratory symptom questionnaires that contained questions from the standardized American Thoracic Society (ATS) questionnaire. The questionnaires also ascertained residence and work location histories as well as lifestyle and housing factors pertinent to relative ambient and indoor air pollutant exposure. Respiratory symptoms were ascertained in the low air pollution season of the year to reduce acute effects and always at the same season to cancel out seasonal influences.

Lung Function Measures

In 1993, individuals less than 80 yr of age as of January 1, 1993, who had completed all three questionnaires (1977, 1987, and 1992), who had not smoked since 1977, and who lived and worked within 20 miles of a California air-quality monitoring station were invited to participate in lung function testing. Of the 1,914 eligible, 1,510 (79%) actually performed spirometry, and data from 1,391 remained after exclusion criteria were considered (incomplete or unacceptable spirometry tests, reported cigarette smoking, current respiratory infection, severe kyphosis or scoliosis, reported congestive heart failure, pneumoconiosis, lung cancer, or body mass index > 45 kg/m2). Under the supervision of a Registered Respiratory Therapist, each volunteer performed spirometry on a dry rolling seal spirometer (Minjhart; S&M Instruments, Doylestown, PA). Slow vital capacity (SVC) maneuvers were performed until two acceptable measures within 5% of each other were obtained. Up to eight FVC maneuvers were performed until three acceptable ones were recorded with the two largest FEV1 measures within 5% (5). The largest FEV1 and the largest VC from either the SVC or FVC maneuvers were selected. The FEV1/VC was calculated as the quotient of these largest values. The percent of predicted FEV1 (PPFEV1) was calculated as the observed FEV1 × 100%, divided by the predicted FEV1. Predicted values were generated internally for each sex by regressing the FEV1 values from healthy subsets of males (n = 199) and females (n = 366) of this population on age and height.

Participants were then trained in the proper use of the Mini-Wright peak flow meter and were asked to perform peak expiratory flow (PEF) maneuvers in triplicate at home four times daily for 1 wk. They were instructed to record the highest PEF from each set of three maneuvers and record it on a diary sheet; 1,223 of the 1,391 had at least two acceptable observations per day on Days 3 through 7; their data were used to determine PEF lability (6). The first 2 d of home measures were excluded to allow for a training effect. Daily lability was defined as the highest PEF minus the lowest PEF, divided by the day's average PEF, and converted to percent. Peak expiratory flow lability was defined as the average of the two highest daily labilities from Days 3 through 7 with at least two valid observations in a day. At least 3 d of valid PEF data were required for a participant's data to be included in the analysis.

Estimating Ambient Air Pollutants

Ambient air pollutant concentrations for 1967 through 1993 were estimated by interpolating monthly statistics from the statewide network of air monitoring stations in California to zip code centroids according to residence and work location histories (4, 7). Pollutants included PM10, 8-h average O3, and mean concentrations of SO4 and SO2. Before 1987, indirect estimates of PM10 were formed from site/seasonal regressions on total suspended particulate (TSP), because PM10 was not monitored statewide before then. Average annual concentrations for 1973 through the month before lung function testing in 1993 were formed for each pollutant except SO4, which has been monitored since 1977.

Several alternative metrics for PM10 were formed, including mean concentration and days per year when PM10 exceeded a number of different cutoff levels—40, 50, 60, 80, and 100 μg/m3, abbreviated PM10(40), etc. These latter indices were termed “exceedance frequency indices.” Excess concentrations—number of days multiplied by concentration—were also calculated for each cutoff. These indices all had correlations greater than 0.85 with exceedance frequency indices for corresponding cutoffs.

A priori hypotheses, and most statistical analyses concerning associations with ambient air pollutants, were restricted to three primary lung function measures: (1) percent of predicted FEV1 (PPFEV1) using internally generated reference equations, (2) FEV1/VC, and (3) PEF lability. Some secondary analyses were also conducted for raw FEV1, bronchodilator response, percent of predicted FVC, and percent of predicted FEF25–75%.

Development and Checking of Statistical Models

Sex-specific multiple linear regressions were used to study the association of lung function measures with PM10(100) annual averages for 1973–1993, adjusting for covariates. The details for development and checking of the multiple linear regression models are given in the Appendix. All of the covariates listed in Tables 1 and 2 were evaluated for possible inclusion in each model. Any covariate was retained in a model if its exclusion resulted in a substantial change in the PM10 regression coefficient or if the covariate substantially improved the precision of the model (8). A substantial change was defined as one that caused a change of 1% or more in the predicted value of the percent-of-predicted lung parameter as estimated from the multiple linear regression. Covariates that could modify exposure to PM10 were evaluated for possible inclusion as interaction terms with PM10. Subjects who, at the time of lung function testing, were taking medications known to affect lung function but for conditions unrelated to air pollution were excluded from the modeling process; subjects who had short-term respiratory conditions were likewise excluded. Once the regression models were developed, they were rerun to include these individuals, with indicator variables used to represent medication use and short-term respiratory conditions.


Females (n = 872)VariableMales (n = 519)
(mean)(min)* (max)* (% 0)(mean)(min)* (max)* (% 0)
49.417.973.80.0O3, 8-h mean (ppb)49.317.670.60.0
26.911.240.70.0O3, 24-h mean (ppb)27.110.340.40.0
31.30.0145.60.0PM10, average d/yr exceeding 100 μg/m3 35.30.3112.40.0
52.721.380.60.0PM10, mean (μg/m3)
5.00.810.10.0SO2, mean (ppb)4.80.810.00.0
7.42.710.10.0SO4, mean (μg/m3), average d/yr exceeding 200 μg/m3
95.939.7155.40.0TSP, mean (μg/m3)98.240.6155.40.0
Continuous candidate covariates (1993) 66.343.080.00.0 of past smoking lived with smoker to 1993 worked with smoker to 1993 dust at work to 1993 of education to 1977 of childhood colds,§ of fume exposure since 1987
27.214.946.50.0Body mass index, kg/m2∥ 27.214.844.50.0 exercise index, of floor carpeted per bedroom, 1993 home heated with wood through 1992 home heated with gas through 1992 vigorous exercise outdoors, 1977** outdoors, 1977**

*Averaged over monthly values of study subjects having less than 20% missing data, 1973–1993, except for SO4 (1977–1993).

Means based on nonzero values.

Primary candidate covariate.

§Frequency of childhood colds, measured on a 5-point scale (1 = much less, 2 = less, 3 = same, 4 = more, 5 = much more than other children of the same age).

Secondary candidate covariate.

An algorithm classified individuals according to frequency of vigorous occupational and leisure exercises (1 = none/light, 2 = low, 3 = moderate, 4 = high).

**A priori interaction candidate covariate.


% with Characteristic
Asthma, bronchitis, or pneumonia before age 16* 9.27.5
Parental history of AOD or hay fever* 40.025.0
Hobbies with dust or fumes 2.55.4
Pets in home, 1993 45.240.8
Dirt road near home, 1993 19.415.5
Parents with hay fever or AOD 40.025.0
Air conditioning at home in 1977 54.353.0
ACE inhibitor use, 1993§ 6.77.5
β-blocker use, 1993§ 9.16.7
Calcium-channel blocker use, 1993§ 8.09.6
Digitalis/diuretics use, 1993§ 16.911.2
Diabetes medication use, 1993§ 3.66.0
Cold/respiratory illness on test date, 1993§ 11.610.4

*Primary candidate covariate.

Secondary candidate covariate.

A priori interaction candidate covariate.

§Subjects with these characteristics excluded from modeling steps, then included in final model using indicator variables.

The distributions of the estimated 1973–1993 average concentrations for PM10(100) and PM10 mean concentration are given in Figures 1 and 2. Tables 1 and 2 give sex-specific descriptive statistics for ambient air pollution concentrations and participant characteristics. There was a higher percentage of female never-smokers (90.8%) than of males (78.6%). Fewer females (45.1%) than males (52.8%) had never lived with a smoker, but a higher percentage of females (51.9%) than males (45.6%) had never worked with a smoker. Females had minimal exposures to dusts at work (89.6% reporting no exposure) compared with 67.2% of males. Parental history of airway obstructive disease (AOD) was reported by 40% of females but only 25% of males.

Tables 3 and 4 show the sex-specific associations of lung function parameters with the two indices of PM10(100) and PM10 mean concentration. Results were similar for the two indices. Table 3 expresses significant interactions of PM10 and other covariates with increased PM10 for subgroups of the interacting covariate, whereas Table 4 gives the regression coefficients and standard errors for the linear and interaction terms.


95% CISex/Subgroup*
0.9 −0.8, 2.5All females (n = 826)
0.3* −2.2, 2.8Males whose parents did not have asthma, bronchitis, emphysema, or hay fever (n = 357)
−7.2* −11.5, −2.7§ Males whose parents had asthma, bronchitis, emphysema, or hay fever (n = 119)
FEV1/VC, %
−0.2 −0.9, 0.5All females (n = 724)
−1.5* −2.7, −0.4 Male never-smokers (n = 381)
−0.2* −1.4, 1.1Male past smokers evaluated at 7 pack-yr (n = 107)
PEF Lability, %
0.8 0.2, 1.6 All females (n = 708)
0.6 −0.1, 1.3All males (n = 441)

Data were estimated from sex-specific multiple linear regression models.

*For subgroups other than sex, the effects for subgroups are estimated from the entire sex-specific regression using estimated PM10 regression coefficient × 54.2 d × interactive variable evaluated at stated value (n = number of subjects in subgroup).

The interactive variable for PPFEV1 was whether or not parents had asthma, bronchitis, or hay fever; for FEV1/VC the interactive variable was pack-years of past smoking.

For linear terms, estimated regression coefficient × 54.2 d.

§p < 0.05.

p < 0.01.


βSEn* βSEn*
PPFEV1 PM10 (μg/m3)0.0130.034826−0.0020.526476
PAOD × PM10 ns −0.279§ 0.106
FEV1/VC, %PM10 −0.0140.014815−0.058§ 0.024488
PKYR × PM10 ns 0.006 0.002
PEF Lability0.0240.0147080.0200.015441

Definition of abbreviations: PAOD = (0, 1) indicator variable for a parental history of asthma, bronchitis, emphysema, or hay fever; PKYR = pack-years of past smoking.

*n = number of subjects with less than 20% missing data

Covariance (β1, β2) for PM10 and PKYR × PM10 = −0.00001634.

These interaction terms were not significant; therefore, they were not used for female models.

§p < 0.05.

Covariance (β1, β2) for PM10 and PAOD × PM10 = −0.002746.

p < 0.01.

**p < 0.001.

An interquartile range increase in PM10(100) of 54.2 d/yr was associated with a 7.1% lower PPFEV1 (p < 0.01) for males whose parents had asthma, bronchitis, emphysema, or hay fever, but not for other males or females (Table 3). This interquartile range difference of PM10(100) was also associated with a 1.5% lower FEV1/VC in never-smoking males (p < 0.01), but PM10 was not significantly associated with lower FEV1/VC in females or males who were former smokers. The interquartile range difference in PM10(100) was associated with a higher PEF lability of 0.8 percent in all females (p < 0.01), 0.6% in all males (p = 0.09), and 0.9% in never-smoking males (p = 0.02). In secondary analyses, no associations with PM10 were seen in either sex for bronchodilator response, percent of predicted VC, or FEF25–75%.

Sex-specific regressions, substituting exceedance frequency indices for each of the other cutoffs of PM10—40, 50, 60, and 80 μg/m3—as well as mean concentration, for PM10(100) showed associations of all indices for PPFEV1 in males whose parents had respiratory disease, for cutoffs of 50 μg/m3 and higher with FEV1/VC in male never-smokers, and for cutoffs of 80 μg/m3 and higher with PEF lability for females. Associations were seen for all cutoffs with the excess concentration indices.

Other significant covariates in the models for PM10 included respiratory problems before age 16, use of cardiac medications, age, height, body mass index (kg/m2), and former smoking (Tables 5 and 6).


Point Estimate95% CIOther Covariates
−5.5−8.8, −2.2 Asthma, bronchitis, or pneumonia before age 16 (n = 76)
−4.0−6.3, −1.7 On β-blockers, diuretics, or digitalis (n = 180)
For regression: n = 826, R 2 = 0.028
−2.0−2.5, −1.5 10-yr difference in age
−0.5−0.6, −0.3 2.5-cm increment in height
1.10.7, 1.5 5-unit increment in body mass index (kg/m2)
1.40.3, 2.5§ Increment of 1 person per bedroom
0.6−0.2, 1.44-yr increment in education
−0.5−1.1, 0.17 pack-yr of past smoking (n = 75)
−0.2−0.5, 0.110 yr lived with smoker through 1993 (n = 443)
For regression: n = 815, R 2 = 0.146
% PEF Lability
0.60.2, 1.1 10-yr increment in age
1.50.2, 2.9§ Asthma, bronchitis, or pneumonia before age 16 (n = 68)
−0.3−0.5, −0.1 2.5-cm increment in height
0.3−0.5, 0.75-unit increment in body mass index (kg/m2)
0.2−0.4, 0.77 pack-yr of past smoking (n = 62)
0.30.0, 0.610 yr lived with smoker through 1993 (n = 385)
0.2−0.1, 0.51-unit increase in vigorous exercise#
0.7−0.3, 1.7Use of ACE inhibitors, digitalis, or diuretics (n = 147)
For regression: n = 708, R 2 = 0.072

*Estimated female lung function change associated with selected increments of covariates in PM10-based lung function regression models. (Increments selected to be 80th percentile.)

p < 0.001.

Throughout table, n = number of subjects having nonzero covariate values.

§p < 0.05.

p < 0.01.# An algorithm classified individuals according to frequency of vigorous occupational and leisure exercises in 1976 (1 = none/light, 2 = low, 3 = moderate, 4 = high).


Point Estimate95% CIOther Covariates
−1.0−2.0, −0.1 7 pack-yr of past smoking (n = 103, past smokers)
−6.4−10.3, −2.5 Calcium-channel blockers or diabetic medication used (n = 87)
0.5−0.8, 1.820% increment in carpeted floor
1.6−2.8, 6.1Parent(s) had hay fever or AOD (n = 119)
−2.7−7.1, 1.8Cold or respiratory illness on test date (n = 49)
For regression: n = 476, R 2 = 0.063
−2.4−3.2, −1.6§ 10-yr increment of age
−1.1−1.7, −0.5 7 pack-yr of past smoking (n = 107, past smokers)
−0.3−0.6, −0.1 2.5-cm increment in height
−0.9−1.8, 0.14-yr increment in education
−0.5−1.2, 0.110 yr worked with smoker through 1993 (n = 271)
−1.8−4.2, 0.7Asthma, bronchitis, or pneumonia before age 16 (n = 39)
1.4−1.2, 4.1Use of β-blockers (n = 33)
For regression: n = 488, R 2 = 0.115
PEF Lability, %
0.80.3, 1.3 10-yr increment in age
0.40.1, 0.8 10 yr worked with a smoker through 1993 (n = 244)
For regression: n = 441, R 2 = 0.037

*Estimated male lung function change associated with indicated increments of covariates in PM10-based lung function regression models. (Increments selected to be 80th percentile.)

p < 0.05.

p < 0.01.

§p < 0.001.

Single and Multiple Pollutant Analyses for Other Pollutants

Table 7 shows correlations between 20-yr averages of air pollution concentrations as estimated for study participants (1973–1993). These did not vary by sex. In single-pollutant analyses of other pollutants, PPFEV1 was associated with both SO4 and O3, but not SO2, for males. In males whose parents had asthma, bronchitis, emphysema, or hay fever, an increase by the interquartile range, 23 ppb, of O3 (8-h average, 1973– 1993) was associated with a decrement in PPFEV1 of 6.3% (95% CI, −10.8% to −1.8%; p = 0.02). An increase by the interquartile range of SO4 (1.6 μg/m3) was associated with a decrement of 1.5% (95% CI, −2.9% to −0.1%; p = 0.04) in all males.


PM10(mc)TSP(200)O3(8 h)O3(24 h)SO2 SO4 *
PM10, mc0.830.850.720.160.61
O3, 8-h0.870.080.57
O3, 24-h−0.150.38
SO2 0.85

Definition of abbreviations: mc = mean concentration; TSP(200) = average d/yr of TSP > 200 μg/m3, 1973–1987. The number of subjects with less than 20% missing data for both pollutants varied from 890 for SO2 and SO4 to 1,316 for PM10 and TSP.

*Average, 1977–1993.

Two pollutant models were run in which each of the other pollutants in turn were added to the PM10(100) final models. Individuals with more than 20% missing data for both pollutants were excluded. The regression coefficients for PM10 remained stable or increased in magnitude with the same sign when other pollutants—O3, SO2, or SO4—were added one at a time to the multiple linear regression models. The estimated standard errors of regression coefficients also remained stable. This was not true for the other pollutants, except SO4, which retained a strong association with PPFEV1 in a two-pollutant model with PM10 in all males. Thus, our multipollutant analyses indicated that the observed effects of PM10 are not due to confounding by O3, SO2, or SO4, and that SO4 has effects in all males that are not solely due to other components of PM10.

Suspended sulfate appears to be the best indicator of the overall mix of pollutants, as it had the highest multiple correlation coefficient when each pollutant was regressed in turn on the other three. For males, these coefficients were 0.50, 0.61, 0.71, and 0.76 for PM10, O3, SO2, and SO4, respectively; for females, they were 0.60, 0.67, 0.72, and 0.76.

Environmental Tobacco Smoke Exposure

Concomitant environmental tobacco smoke (ETS) exposure was defined as contributing 1 h/d exposure or more to ETS at work or at home as reported in 1987, 1992, or 1993. Stratified analyses indicated that the PM10 lung function associations for males were stronger for those with concomitant ETS exposure. This was not true for females.

FEV1 versus PPFEV1

Sex-specific multiple linear regressions of raw FEV1 (unadjusted for age and height) with PM10(100), using the same covariates as for PPFEV1 but adding age and height, gave very close and consistent results to those reported in Table 3. The decrement of FEV1 associated with an interquartile range difference in PM10(100) of 54.2 d/yr was equivalent to 7.3 yr of aging for males whose parents had respiratory disease.

Peak Expiratory Flow Lability—Missing Values

Our definition of PEF lability allowed up to two of the four daily values to be missing. Subjects who were missing a first morning PEF value, which often tends to be close to the day's minimum, might have lower lability. As a check on this, we formed an alternative definition of PEF lability requiring that a first morning PEF be present on all days used in the lability calculation. This reduced the number of subjects available for analyses by one female. The sex-specific correlations of the two lability indices were greater than 0.99, and the observed relationships between PM10 and PEF lability were essentially unchanged; PM10 regression coefficients changed by less than 2% of their former value.

Lack of Baseline Lung Function

“Lack of baseline lung function measures” means that we are unable to compare baseline lung function measures of those lost to follow-up to the rest of the eligible cohort, in order to determine if there is a possible bias in lung function/air pollution associations due to a “healthy worker effect” (9). This effect is a bias toward a weaker association, because those who survive and are not lost to follow-up may have less of a lung function/air pollution association. Although lung function data were lacking in previous years, we did have symptom data as an indicator of lung health. Sex- and age-specific prevalences in 1992 of overall AOD, asthma, and chronic bronchitis for those selected for lung function testing were very similar to those of the entire group of survivors who completed the 1992 symptom questionnaires. To evaluate the possibility of a healthy worker effect further, we compared the sex-specific prevalences of AOD in 1977 among the 1,391 selected for lung function testing with the entire cohort of 6,339 in 1977 who would have been less than 80 yr old on January 1, 1993, and thus eligible for lung function testing. The prevalence of AOD in 1977 in females selected for lung function testing was 10.4% (95% CI, 8.4–12.5%), compared to 15.2% for eligible females in the entire cohort. For males selected for lung function testing, the prevalence of AOD in 1977 was 14.5% (95% CI, 11.5– 17.6%), compared to 16.8% among eligible males in the entire cohort. Thus, the healthy worker effect may be stronger among females in this cohort.

To address this issue further, and also to determine if PM10 effects were only seen in those with already compromised lung health in 1977, the multiple linear regressions relating PPFEV1 and FEV1/VC to PM10(100) for males were repeated on those males who had AOD in 1977 versus those who did not. There were stronger associations between PM10 and PPFEV1, as well as PM10 and FEV1/VC, for males without AOD in 1977 than for those with AOD in 1977. No associations of PM10 with PPFEV1 or FEV1/VC were noted for either females with or without AOD in 1977. There was little change in the effect size of PM10/PEF lability associations for females or males when stratified on AOD in 1977.

Short-term Exposures

We measured lung function during the low PM10/O3 season in California, January through April. To further rule out the possibility of short-term O3 exposures affecting the observed associations between PM10(100) and the primary lung function measures, we ran the lung function regression models for PPFEV1, percent FEV1/VC, and PEF lability, excluding subjects who lived or worked within 30 miles of a monitoring station where a maximal hourly average over 120 ppb was recorded on the day of or the day before lung function testing or whose lung function testing site was within 30 miles of a station that recorded more than 120 ppb on the day of lung function testing. This resulted in excluding 39 females and 19 males for PPFEV1, 37 females and 19 males for percent FEV1/ VC, and 31 females and 17 males for PEF lability. The regression coefficients for PM10(100) changed by less than 6%, and confidence intervals were close to what they had been before.

To ascertain whether the observed associations between long-term ambient concentrations of PM10 and lung function were due instead to short-term effects of PM10, we excluded subjects from the lung function multiple regression models for whom estimated ambient concentrations of PM10 exceeded 100 μg/m3 for 1 d or more for the month of or the month before lung function testing. This resulted in excluding 38 females and 25 males for PPFEV1, 37 females and 27 males for FEV1/VC, and 29 females and 22 males for PEF lability. For both females and males, the effects of the long-term averages of PM10(100) were just as strong or stronger.

Because PM10 is only monitored every sixth day in California, we cannot rule out the possibility that PM10 exceeded 100 μg/m3 on a day that was not monitored. However, PM10 and O3 concentrations in California are highly correlated (Table 4), so excluding subjects who may have had high short-term O3 exposures as done in the previous sensitivity analysis would also tend to exclude subjects whose PM10 concentrations may have been high, at least on the day of or day before testing. Indicator variables to represent short-term effects of PM10 or O3 were also added to each of the models. These were not found to be significantly related to reduced lung function, nor did their addition substantially change the significant long-term regression coefficients.

Population Density

Population density has been found to be associated with elevated rates of respiratory disease (10). Our study design largely controlled for population density, since over 90% of subjects were selected from urban areas. Our baseline questionnaire data ascertained population density according to a three category measure; however, because of the urban selection criteria there were only sufficient frequencies to form a two-category variable.

As a check on potential confounding of PM10(100) and population density, we used a two-category, high/low, population density (0, 1) indicator variable. High population density was defined as more than 10 homes in a two-block (1/4 mile) radius, with 87.3% of females and 89.5% of males meeting this criteria. Population density had no association with PM10(100) nor with any of the lung function parameters for females or males, except with PEF lability for females, whereas those with low population density had less PEF lability. Stratified by population density, analyses of PM10(100) with PEF lability showed the effect size of the PM10(100) coefficient decreased by only 5% in the low-population-density group and by 26% in the high-population-density group. Thus, population density does not appear to be confounded with PM10 in these analyses.

Indirect Estimates of PM10 and Time Trends

Our analyses used indirect estimates before 1987 to form the 1973–1993 averages for PM10. We repeated the multiple linear regressions relating lung function measures to PM10, substituting directly estimated PM10(100) 1987–1993 for PM10(100) 1973–1993, and found that the regression coefficients changed by less than 8% and that the estimated standard errors of regression coefficients were also very similar, with the exception of FEV1/VC for females, which still showed no association with PM10. Similar substitutions, using average days per year from 1973–1987 when TSP exceeded 200 μg/m3 (the metric of TSP most closely associated with PM10[100]), gave consistent results with those reported in Table 2, except that the association with PEF lability for males was stronger (p < 0.04).

We found that PM10 concentrations in our cohort gradually decreased over time. The average days per year of ambient concentrations of PM10 in excess of 100 μg/m3 for 1973–1977, 1977–1987, and 1987–1993 were 39.5, 34.1, and 32.6, respectively. The sensitivity analyses restricting PM10 to 1987 to 1993 indicates that the conclusions reached using PM10 averaged over the entire time period, 1973–1993, still apply.

Summary and Comparison with Previous Reports on This Cohort

We have found elevated ambient PM10 concentrations over a 20-yr period to be associated with decreased PPFEV1 in nonsmoking California males with a parental history of respiratory disease, with decreased FEV1/VC in never-smoking males, and with increased PEF lability in females and never-smoking males. These results do not appear to be due to short-term ambient concentrations and do not appear to be solely due to other measured air pollutants—O3, SO2, or SO4. Ozone was associated with decreased PPFEV1 only in males with a parental history of respiratory disease. Suspended sulfate was associated with decreased PPFEV1 in all males.

These results are consistent with previous findings from this cohort (3, 4, 7) that recorded associations between PM10 or TSP and either 1977 prevalence or 1977–1987 incidence of AOD (asthma, chronic bronchitis, or emphysema). Suspended sulfate was found to be associated with development of asthma in this cohort (11). Ozone was previously reported to be associated with development of asthma in males but not in females (12). Sex differences in asthma/O3 associations as well as in lung function/PM10 associations may be the result of sex differences in exposures: in 1977, males in our cohort spent more time outdoors, based on a year-round average, than females: 16.1 h/ wk versus 9.2 h/wk (p < 0.0005). Also, males had greater exposures to major indoor sources of PM10, including ETS, occupational dusts, and hobbies (Table 1). The higher exposures to PM10 and O3 experienced by males in our cohort may trigger a response in those males with an inherited predisposition to AOD.

Comparisons with Other Studies

A number of other studies (1, 10, 13-15) have shown decrements in lung function of adults to be associated with long-term elevated levels of particulates. Peak expiratory flow lability has previously been found to be a sensitive indicator of acute effects of SO2 in subjects with COPD (16). Our multi- pollutant analyses of long-term effects of PM10 on lung function, which remained stable when O3, SO2, or SO4 were added to the models, are consistent with findings from short-term studies that found PM10 effects both with and without high concentrations of acid aerosols, SO2, and O3 (17). The associations we have observed between lower lung function in males and long-term concentrations of TSP and PM10 are in general agreement with associations reported for both sexes with cross-sectional measures of lung function from the National Health and Nutrition Surveys (NHANES) I and II of more than 40 U.S. cities and the SAPALDIA studies of adults in Switzerland, except that we did not observe consistent associations between PM10 or TSP and PPFEV1 across all subgroups of sex and past smoking (13, 15, 18). Sex inconsistencies in our cohort may be due to the marked sex differences in outdoor exposure to PM10 and O3 and occupational air pollutants as discussed above. The healthy-worker effect could also be influencing our prospective data and may partially account for sex differences. We saw evidence for a stronger healthy-worker effect in women in our cohort.

We found that parental history of respiratory disease was associated with increased susceptibility to the effects of air pollution in males; childhood history of respiratory problems was associated with lower lung function in females. Previous research found that both of these were risk factors for the development of lower lung function (19). There is also evidence of a strong genetic component associated with the risk of developing asthma (20). Other analyses of our data have shown asthma to be highly correlated with reduced PPFEV1. It is possible that our finding of reduced PPFEV1 in males with a parental history of respiratory disease is evidence of a gene– environment interaction, but further research is needed on this idea. Current medication usage showed strong associations with some lung function measures. These effects are likely to be primarily the effects of the cardiovascular and associated diseases for which they are taken (21). Past smoking and years worked with a smoker were associated with lung function measures in men, but not in women. Years lived with a smoker showed an association for women. These sex differences are consistent with sex differences in exposure (Table 1). Our finding of increased bronchial lability with chronically high PM10 exposures confirms the few longitudinal studies and those showing acute changes in bronchial lability associated with high particulate exposures (22).

The associations we have observed between decrements in lung function and long-term ambient concentrations of PM10 may be enhanced by concomitant exposures to O3 and SO4. Exposure to O3 has been found to enhance the effects of subsequent exposure to allergens in asthmatics (23) and subsequent exposure to asbestos fibers in rats (24).

Short-term Exposures

Short-term (less than 1 d) exposures to O3 concentrations as low as 120 ppb have been shown to be associated with alterations in lung function in heavily exercising adults (25). Decrements in lung function associated with short-term O3 concentrations have been found to persist to the day following exposure in children at summer camp (26).

Decrements in lung function measures have also been associated with short-term episodes of particulate pollution. In one study (27), PM10 levels higher than 150 μg/m3 were associated with mean declines in PEF equal to 1–6%. Decrements in lung function associated with a short-term episode of moderately elevated TSP and SO2 persisted for 16 (but not for 25) d (28). In order to avoid possible short-term effects, we conducted the lung function testing during the low O3 and PM10 season in California—January through early April. We additionally ruled out the possibility of short-term effects when we observed no change in our results after excluding subjects who might have experienced greater than 120 ppb O3 within a day of lung function testing or greater than 100 μg/m3 of PM10 within a month of testing.


General limitations of this study related to methods of estimating ambient air pollutants and epidemiologic methods have been discussed in previous papers (4, 7). Only the limitations specific to assessing lung function/air pollution associations are discussed here.

Lack of baseline lung function. Lung function was not measured at time of enrollment in 1977; however, it has been demonstrated that the age- and sex-specific coefficients of variation associated with changes in lung function measures derived from longitudinal studies are large compared with those from cross-sectional data (29). Therefore, there may be little loss of power for detecting the long-term effects of ambient air pollutants on lung function in this study.

Measurement error. We have unknown amounts of measurement error in estimated long-term ambient concentrations of pollutants as well as other covariates. This could bias our estimates of PM10 regression coefficients (30). A simulation study of O3/FEV1 relationships in school children in Southern California indicated that using estimated ambient O3 concentrations biased estimated regression coefficients between O3 exposure and FEV1 toward the null (31).

Indoor versus outdoor air pollution. An important limitation of assessing associations between long-term ambient concentrations of air pollutants and lung function is that our estimates address outdoor ambient concentrations, not personal exposure. Most studies that have correlated short-term personal exposures of PM10 over periods of 1 to 24 h have been limited to local areas and have found low correlations with varying outdoor ambient concentrations. A recent study of PM10 personal exposures in Riverside, California (near our highest PM10 concentrations), found a correlation of outdoor PM10 with personal exposure of 0.46 when longitudinal variation over time was correlated for each home (32). Dockery and Spengler (33) found that 48% of variation in short-term personal exposure was explained by outdoor concentrations. They also found that a five-variable model for predicting personal exposure was somewhat, but not significantly, better than either indoor or outdoor concentrations alone (33). It is possible that longer-term ambient concentrations (over many years) in areas differing widely in ambient concentrations may correlate more highly with relative long-term personal exposures, but to our knowledge no one has studied this.

We have dealt with the lack of indoor exposure measurements in two ways: (1) obtaining self-reports of personal exposure to indoor PM sources such as past smoking, passive smoking, occupation, hobbies, etc., as covariate candidates for the health effect models; and (2) by rerunning our final models using adjusted outdoor ambient concentrations obtained by applying an indoor/outdoor adjustment factor to PM10 mean concentration, according to the time spent indoors as reported by season for each study participant in 1977, 1987, 1992, and 1993. An indoor adjustment factor of 0.7 was used (34). This indoor adjustment factor is consistent with more recent studies of personal exposures to PM10 conducted in Southern California (32, 35). The results of using adjusted ambient concentration were consistent with those reported in Table 2.

Indirect estimates of PM10   . Before 1987, only indirect estimates of PM10 using site-specific and seasonal regression prediction equations based on TSP were available. However, only a small loss of precision occurred from using these indirect estimates over long time periods (7). Results consistent with those reported were obtained when we restricted PM10 to direct measures available for 1987–1992 or substituted TSP from 1973–1987.

The investigators wish to acknowledge the helpful assistance of Beverley Stocker for word processing and preparation of graphs and tables as well as Mark Olfert, Barbara Peters, and Valerie Whyte, respiratory therapists, and James A. Peters, M.D., Dr.PH., physician on-site, for performing the pulmonary function tests. They also thank Dane Westerdahl, John Moore, and Andy Alexis of the California Air Resources Board for providing the air quality data, and the participants for their continued willingness to participate in this study.

1. Detels R., Tashkin D. P., Sayre J. W., Rokaw S. N., Massey F. J., Coulson A. H., Wegman D. H.The UCLA population studies of CORD: X. A cohort study of changes in respiratory function associated with chronic exposure to SOx, NOx, and hydrocarbons. Am. J. Public Health811991350359
2. Dockery D. W., Speizer F. E., Stram D. O., Ware J. H., Spengler J. D., Ferris B. G.Effects of inhalable particles on respiratory health of children. Am. Rev. Respir. Dis.1931989587594
3. Abbey D. E., Lebowitz M. D., Mills P. K., Petersen F. F., Beeson W. L., Burchette R. J.Long-term ambient concentrations of particulates and oxidants and development of chronic disease in a cohort of non-smoking California residents. Inhalation Toxicology719951934
4. Abbey D. E., Petersen F. 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
5. American Thoracic SocietyStandardization of spirometry–1987 update. Am. Rev. Respir. Dis.136198712851298
6. Enright, P. L., J. A. Peters, R. J. Burchette, M. D. Lebowitz, W. F. McDonnell, and D. E. Abbey. 1998. Peak flow lability: association with asthma and spirometry in an older cohort. Chest (In press)
7. Abbey D. E., Hwang B. L., Burchette R. J.Estimated long-term ambient concentrations of PM10 and development of respiratory symptoms in a nonsmoking population. Arch. Environ. Health501995139150
8. Maldonado G., Greenland S.Simulation study of confounder-selection strategies. Am. J. Epidemiol.1381993923936
9. Eisen A. E., Wegman D. H., Louis T. A., Smith T. J., Peters J. M.Healthy worker effect in a longitudinal study of one-second forced expiratory volume (FEV1) and chronic exposure to granite dust. Int. J. Epidemiol.24199511541162
10. Holland W. W., Reid D. D.The urban factor in chronic bronchitis. Lancet11965445448
11. Abbey, D. E., F. F. Petersen, P. K. Mills, and L. Kittle. 1993. Chronic respiratory disease associated with long-term ambient concentrations of sulfates and other air pollutants. J. Expo. Anal. Environ. Epidemiol. 3(S1):99–115.
12. Greer J. R., Abbey D. E., Burchette R. J.Asthma related to occupational and ambient air pollutants in nonsmokers. J. Occup. Environ. Med.351993909915
13. Chestnut L. G., Schwartz J., Savitz D. A., Burchfiel C. M.Pulmonary function and ambient particulate matter: epidemiological evidence from NHANES I. Arch. Environ. Health461991135144
14. Xu X. P., Dockery D. W., Wang L. H.Effects of air pollution on adult pulmonary function. Arch. Environ. Health461991198206
15. Ackermann L. U., Leuenberger P., Schwartz J., Schindler C., Monn C., Bolognini G., Bongard J. P., Brandli O., Domenighetti G., Elsasser S., Grize L., Karrer W., Keller R., Keller-Wossidlo H., Kunzli N., Martin B. W., Medici T. C., Perruchoid A. P., Schoni M. H., Tschopp J. M., Villiger B., Wuthrich B., Zellweger J. P., Zemp E.Lung function and long term exposure to air pollutants in Switzerland. Study on air pollution and lung diseases in adults (SAPALDIA) team. Am. J. Respir. Crit. Care Med.1551997122129
16. Higgins B. G., Francis H. C., Yates C. J., Warburton C. S., Fletcher A. M., Reid J. A., Pickering C. A. C., Woodcock A. A.Effects of air pollution on symptoms and peak expiratory flow measurements in subjects with obstructive airways disease. Thorax501995149155
17. Pope C. A., Bates D. V., Raizenne M. E.Health effects of particulate air pollution: time for reassessment? Environ. Health Perspect.1031995472480
18. Schwartz J.Lung function and chronic exposures to air pollution: a cross-sectional analysis of NHANES II. Environ. Res501989309321
19. Lebowitz M. D.The trends in AOD morbidity in the Tucson epidemiological study. Am. Rev. Respir. Dis.1401989S35S41
20. Harris J. R., Magnus P., Samuelsen S. O., Tambs K.No evidence for effects of family environment on asthma: a retrospective study of Norwegian twins. Am. J. Respir. Crit. Care Med.15619974349
21. Enright P. L., Kronmal A., Higgins M., Schenker M., Haponik E. F.Spirometry reference values for women and men 65 to 85 years of age. Am. Rev. Respir. Dis.1471993125133
22. Lebowitz M. D.Epidemiological studies of the respiratory effects of air pollution. Eur. Respir. J.9199610291054
23. Molfino N. A., Wright S. C., Katz I.Effects of low concentrations of ozone on inhaled allergen responses in asthmatic subjects. Lancet3381991199203
24. Pinkerton K. E., Brody A. R., Miller F. J., Crapo J. D.Exposure to low levels of ozone results in enhanced pulmonary retention of inhaled asbestos fibers. Am. Rev. Respir. Dis.140198910751081
25. Folinsbee L. J., McDonnell W. F., Horstman D. H.Pulmonary function and symptoms responses after 6.6 hour exposure to 0.12 ppm ozone with moderate exercise. J. Air Pollut. Control Assoc.3819882835
26. Spektor D. M., Thurston G. D., Mao J., He D., Hayes C., Lippmann M.Effects of single- and multi-day ozone exposures on respiratory function in active normal children. Environ. Res.551991107122
27. Pope C. A., Dockery D. W.Acute health effects of PM10 pollution on symptomatic and asymptomatic children. Am. Rev. Respir. Dis.145199211231128
28. Dassen W., Brunekreef B., Hoek G., Hofschreuder P., Staatsen B., deGroot H., Schouten E., Biersteker K.Decline in children's pulmonary function during an air pollution episode. J. Air Pollut. Control Assoc.36198612231227
29. Ware J. H., Dockery D. W., Louis T. A., Xu X., Ferris B. G., Speizer F. E.Longitudinal and cross-sectional estimates of pulmonary function decline in never-smoking adults. Am. J. Epidemiol.1321990685700
30. Liu K.Measurement error and its impact on partial correlation and multiple linear regression analyses. Am. J. Epidemiol.1271988864874
31. Navidi W., Lurmann F.Measurement error in air pollution exposure assessment. J. Expo. Anal. Environ. Epidemiol.51995111124
32. Wallace L.Indoor particles: a review. J. Air Waste Manag. Assoc.46199698126
33. Dockery D. W., Spengler J. D.Personal exposures to respirable particulates and sulfates. J. Air Pollut. Control Assoc.311981153159
34. Winer, A. M., F. W. Lurmann, L. A. Coyner, S. D. Colome, and M. P. Poe. June 1989. Characterization of air pollution exposures in the California South Coast Air Basin: application of a new regional human exposure (REHEX) model. In Final Report, Contract No. TSA 106-01-88. South Coast Air Quality Management District, Statewide Air Pollution Research Center, University of California, Riverside, CA. 33–47.
35. Colome, S. D., N. Y. Kado, P. Jaques, and M. Kleinman. 1992. Indoor-outdoor air pollution relations: particulate matter less than 10 mm in aerodynamic diameter (PM10) in homes of asthmatics. Atmospheric Environment 26A:2173–2178.
Correspondence and requests for reprints should be addressed to David E. Abbey, Loma Linda University, CHR, Evans Hall 204, Loma Linda, CA 92350. E-mail:

This project was supported by Environmental Protection Agency cooperative agreement CR 819691-02-0 and the California Air Resources Board contract A933-160.

Although the research described in this article has been funded by the U.S. Environmental Protection Agency, it has not been subjected to Agency review and does not necessarily reflect the view of the Agency. The statements and conclusions in this article are those of the contractor and not necessarily those of the California State Air Resources Board.


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record