Although chronic airway disease (CAD) is known to be

Respiratory Symptoms and Pulmonary Function in an Elderly Nonsmoking Population* David J. Berglund, MD, MPH; David E. Abbey, PhD; Michael D. Lebowitz, PhD, FCCP; Synnøve F. Knutsen, PhD, MD; and William F. McDonnell, PhD, MD Objective: To examine risk factors for chronic airway disease (CAD) in elderly nonsmokers, as determined by pulmonary function tests (PFTs), and to correlate reported respiratory symptoms with PFT measures. Design: An observational survey. Setting: Several communities in California. Measurements: Exposures and respiratory history were assessed by standardized questionnaire. PFTs were performed and prediction equations derived. Results: Significant risk factors for obstruction on PFTs in multiple logistic regression included reported environmental tobacco smoke (ETS) exposure (relative risk [RR] 1.44), parental CAD or hay fever (RR 1.47), history of childhood respiratory illness (RR 2.15), increasing age, and male sex. The number of years of past smoking was of borderline significance (RR 1.29 for 10 years of smoking; p 0.06). The prevalence of obstruction on PFTs was 24.9% in those with definite symptomatic CAD, compared with 7.5% in those with no symptoms of CAD. The prevalence of obstruction was 36.0% among those with asthma and 70.6% among those with emphysema. Also, symptomatic CAD correlated with reduction in lung function by analysis of covariance. The mean percent predicted FEV 1 adjusted for covariates was 90.6% in persons with definite symptoms of CAD, compared with 97.8% in those without it (p < 0.001). Conclusions: Age, sex, parental history, childhood respiratory illness, and reported ETS exposures were significant risk factors for obstruction on PFTs. Self-reported respiratory symptoms also correlated significantly with PFTs. (CHEST 1999; 115:49 59) Key words: chronic airway disease; nonsmoking; passive smoking; pulmonary function; respiratory symptoms Abbreviations: AHSMOG Adventist Health Study of Smog; ANCOVA analysis of covariance; ATS American Thoracic Society; CAD chronic airway disease; CI confidence interval; ETS environmental tobacco smoke; FEF 25 75 forced expiratory flow from 25 to 75% of the vital capacity; PEF peak expiratory flow; PFT pulmonary function test; PM 10 particles less than 10 m in diameter; PM 10 (100) number of days per year that PM 10 exceeded 100 g/m 3 ; PPFEF 25 75 percent predicted forced expiratory flow from 25 to 75% of the vital capacity; PPFEV 1 percent predicted FEV 1 ; PPVCmax percent predicted maximum vital capacity; RR relative risk; VC vital capacity; VCmax maximum vital capacity * From the Center for Health Research (Drs. Berglund, Abbey, and Knutson), School of Public Health, Loma Linda University, CA; the Arizona Prevention Center (Dr. Lebowitz), University of Arizona, College of Medicine, Tucson, AZ; and the United States Environmental Protection Agency (Dr. McDonnell), Research Triangle Park, NC. Dr. Berglund is now with the National Center for Health Statistics (NCHS) in Hyattsville, MD. However, this study has not undergone review by NCHS. Supported by EPA Cooperative Agreement no. CR819619 02 and the California Air Resources Board Contract no. A933 160. Although the research described in this article has been funded by the United States 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 authors and not necessarily those of the California State Air Resources Board. Manuscript received March 10, 1998; revision accepted August 11, 1998. Correspondence to: David E. Abbey, PhD, Loma Linda University, School of Public Health, CHR, Evans Hall 204, Loma Linda, CA 92350; e-mail: dabbey@sph.llu.edu Although chronic airway disease (CAD) is known to be much more prevalent among smokers, 1 4 it is also common among nonsmokers. Chronic bronchitis and emphysema have a combined prevalence of 4 to 6% For editorial comment see page 4 among lifelong nonsmokers in the United States. 5 Asthma has a prevalence of 4 to 5% in the United States general population, 6 although it may vary by region. 7 While smoking is the major recognized risk factor for chronic bronchitis and emphysema, multiple risk factors have some importance for asthma, including air pollution, tobacco smoke exposure, and specific allergens. 7 10 CAD also increases with age in adulthood. 6 With the increase in the elderly population and the decrease in smoking in the United States, other risk factors for CAD may become of CHEST / 115 / 1/ JANUARY, 1999 49

more importance, particularly in the nonsmoking elderly. Such risk factors may include air pollution, environmental tobacco smoke (ETS) exposure, and genetic predisposition. 7,9 14 CAD may be assessed with historical information, clinical evaluation including auscultation of lung sounds, and pulmonary function tests (PFTs). Lability in the peak expiratory flow (PEF) measured by the subject at home may be used to indicate bronchial responsiveness, as an alternative to bronchial challenge test. 15,16 For large population-based studies, standardized questionnaires that assess history of symptoms and previous diagnoses are often most practical. 17 19 PFTs, PEF lability, and respiratory symptoms self-reported on the standardized American Thoracic Society (ATS) questions were obtained in a 1993 cohort study of the cumulative effects of decades of exposure to air pollution in California residents. 15,20 22 This study provides the opportunity to ascertain risk factors for CAD as measured by PFTs in a large community sample of middle-aged and elderly nonsmokers, as well as to correlate CAD as ascertained from the standardized ATS questions with CAD determined from PFTs. Study Population Materials and Methods Subjects were selected from 3,091 surviving men and women in an observational cohort study of health effects of ambient air pollutants in nonsmoking middle-aged and elderly Seventh-day Adventist California residents, begun in 1977. 20 The study is known as the Adventist Health Study of Smog (AHSMOG). Original selection criteria required that subjects be non-hispanic white individuals, over 25 years of age in 1977, and that they had lived within 5 miles of their 1977 residence for the previous 10 years. Two thirds of the participants resided in the South Coast Air Basin, which includes the heavily populated areas of Los Angeles, Orange, Riverside, and San Bernardino counties. The remaining one third resided in the San Diego or San Francisco Air Basins, or were from a 10% geographically ordered random sample of eligible participants from the rest of California. Selection criteria for this substudy required that subjects have complete air pollution data; live within 20 miles of an air quality monitoring station; have completed mailed questionnaires in 1977, 1987, and 1992; and be less than 80 years old on January 1, 1993. Of 1,870 candidates who met these criteria, 62 refused to participate, 70 were too sick, 11 could not be contacted, 61 were unable to come to a testing site, and 156 were clinic visit no-shows. The remaining 1,510 subjects completed PFTs at the clinic visit in 1993. Data from 1,391 subjects was used, after these additional exclusions: report of current tobacco smoking; missing questionnaire data; incomplete or unacceptable spirometry performance; and disorders other than obstructive lung disease which could affect spirometry, including respiratory infection on the day of testing, congestive heart failure, severe kyphosis or scoliosis, lung cancer, tuberculosis, pneumoconiosis, collapsed lung, or morbid obesity (body mass index of 45 kg/m 2 ). 15 Characteristics of this population are shown in Table 1. PFT Methods The PFT methods used in this study have been described elsewhere. 15,23 Briefly, spirometry was performed during low air pollution months (January to April), according to the ATS recommendations, with a dry rolling-seal spirometer automated by a personal computer to display flow-volume curves. Three to eight FVC measurements were performed under the direction of a trained respiratory therapist to obtain at least two acceptable maneuvers. Results were reviewed by a pulmonologist. 23 Slow vital capacity (VC) maneuvers were performed until two measures within 5% of each other were obtained. The maximum VC (VCmax) was the largest of the acceptable FVC and slow VC measures. The maximum FEV 1 was obtained, along with the forced expiratory flow from 25 to 75% of the VC (FEF 25 75 ) from the maneuver with the largest sum of FVC and FEV 1. Height, weight, and arm span were measured for use in Table 1 Prevalence of Obstructive Disease Identified by Pulmonary Function Tests and by Respiratory Symptom Complexes Total (N 1,391) Obstruction on PFTs (n 147 [10.6%]) CAD by Symptoms (n 209 [15.0%]) Sex Female 872 73 (8.4) 124 (14.2) Male 519 74 (14.3) 85 (16.4) Age group, yrs 40 to 50 104 1 (1.0) 20 (19.2) 51 to 60 282 11 (3.9) 35 (12.4) 61 to 70 518 49 (9.5) 86 (16.6) 71 to 80 487 86 (17.7) 68 (14.0) Education, yrs 12 or less 311 37 (11.9) 52 (16.7) 14 598 51 (8.5) 89 (14.9) 16 or more 480 59 (12.3) 68 (14.2) data missing 2 Parental CAD No 880 89 (10.1) 101 (11.5) Yes 462 57 (12.3) 99 (21.4) data missing 49 Smoking No 1,166 111 (9.5) 165 (14.2) Yes 225 36 (16.0) 44 (19.6) Any ETS history* No 616 55 (8.9) 65 (10.6) Yes 775 92 (11.9) 144 (18.6) Childhood respiratory illness No 1,272 125 (9.8) 165 (13.0) Yes 119 22 (18.5) 44 (37.0) Dust exposure (dichotomous) No 1,130 117 (10.4) 155 (13.7) Yes 261 30 (11.5) 54 (20.7) Fumes exposure (dichotomous) No 1,241 130 (10.5) 172 (13.9) Yes 150 17 (11.3) 37 (24.7) *Any history of ETS exposure lasting at least 1 h/d on average for at least 1 year. 50 Clinical Investigations

prediction equations for PFTs. The prediction equations were formulated for a healthy reference subgroup of 565 asymptomatic never smokers. 24 Prediction equations were used to obtain percent predicted values for PFTs, including the percent predicted FEV 1 (PPFEV 1 ), percent predicted VCmax (PPVCmax) and the percent predicted forced expiratory flow from 25 to 75% of the VC (PPFEF 25 75 ). 24 For analyses in this study, obstructive disease based on PFTs was defined a priori as either the ratio FEV 1 /VCmax 65% or the PPFEV 1 75%. Reversible airway obstruction was assessed using bronchodilator response and lability in PEF. The bronchodilator response was obtained as the percent improvement in FEV 1 following administration of albuterol. 23 This was completed in 1,365 subjects (26 subjects declined or were unable). Subjects were trained to perform PEF measures in triplicate at home four times daily for 1 week and record the results in a diary. Acceptable peak flow diaries were obtained from 1,223 of 1,391 subjects (requiring at least two acceptable measures per day on at least 3 of the last 5 days of the week). The daily lability was defined as: 100% (maximum PEF minimum PEF)/mean PEF The overall maximum lability in PEF was taken as the average of the two highest daily lability values for the week, after data for the first two days were dropped because of a learning effect. 3,15,25 Respiratory Symptom Algorithms Standardized questionnaires were completed in 1977, 1987, 1992, and 1993. These included questions contained in the ATS questionnaire, 20,26 along with history of exposures and other information. Computerized algorithms based on history and respiratory symptoms were used to classify subjects into categories of respiratory symptom complexes. 20 For a definite diagnosis of asthma, both reported symptoms of wheezing and history of physician-diagnosed asthma were required. Symptoms related to chronic bronchitis include chronic cough and chronic sputum production, both of which were considered separately. Definite diagnosis of chronic bronchitis of either cough type or sputum type required presence of symptoms on most days for at least 3 months, and also of at least 2 years duration. For a definite diagnosis of emphysema, both reported symptoms of shortness of breath with exertion and history of physician-diagnosed emphysema were required. Comparison is made between these symptom complexes and questions about physician diagnosis of asthma, chronic bronchitis, and emphysema. The term physician diagnosis will be used to mean questionnaire self-report of physician diagnosis. Analysis of chronic bronchitis was performed only after asthma and emphysema were excluded, as overlap in these diagnoses could otherwise influence the results. Similarly, analysis of emphysema was performed after exclusion of asthma. Analysis of asthma was primarily performed without emphysema or chronic bronchitis excluded, although a sensitivity analysis was done with these exclusions. Assessment of Potential Risk Factors and Covariates Exposure to particulate air pollution was assessed with methods described in detail elsewhere. 13,27,28 Directly measured values of particulate matter less than 10 m in diameter (PM 10 ) were used after they became available on a statewide basis in 1987. For earlier exposures back to 1973, estimates of PM 10 were made based on site- and season-specific regression from total suspended particles. 13 Individual estimates of exposure were derived by interpolating values from the nearest three monitoring stations to the ZIP code centroid for both home and work. 13,27,28 The analyses used both the mean PM 10 concentration and the number of days per year that PM 10 exceeded 100 g/m 3 [PM 10 (100)]. 13 Other potential risk factors and exposures were assessed by questionnaire, and these are listed in Tables 1 and 2. For reported ETS exposure, a daily average of at least 1 h was required. In addition to any ETS history (dichotomous variable), ETS was also classified more specifically as years living with a smoker as a child, years living with a smoker as an adult, and years working with a smoker. 12 An additional dichotomous ETS exposure variable was created to assess any exposure as an adult (combining home and work exposure), and was used in a sensitivity analysis. Reported occupational exposures included the total years of exposure to dust and the number of years of reported exposure to fumes from 1987 to 1993. A history of childhood respiratory illness (dichotomous) included asthma, bronchitis, and pneumonia. Table 2 Risk Factors for Obstruction Identified by Pulmonary Function Tests With Relative Risks and Confidence Intervals (N 1,332, With 146 Cases) Difference for RR Calculation* Regression Coefficients SE RR 95% CI Age 10 yrs 0.0967 0.0136 2.44 1.92, 3.07 Education 4 yrs 0.0367 0.0364 1.14 0.88, 1.45 Sex 1 male 0.5994 0.1968 1.70 1.21, 2.35 0 female Past smoking 10 yrs 0.0288 0.0147 1.29 1.00, 1.66 Ever had ETS exposure 1 yes 0.4112 0.1993 1.44 1.02, 2.01 0 no PM 10 (100) 42 d/yr 0.0001 0.0001 1.09 0.92, 1.30 Childhood respiratory illness 1 yes 0.9021 0.2736 2.15 1.38, 3.20 0 no Parental CAD 1 yes 0.4428 0.1932 1.47 1.06, 2.01 0 no Constant 10.1338 1.1524 *Difference used for relative risk calculation; for continuous variables, RR is for an increase equal to stated difference. ETS exposure required report of exposure lasting at least 1 h/d for at least 1 year. CHEST / 115 / 1/ JANUARY, 1999 51

Statistical Methods Multivariate logistic regression was used to assess the relationship between prevalence of CAD and potential risk factors and covariates. These variables were first examined with univariate testing before logistic regression models were run. Variables forced into the logistic model were age, sex, years of education, PM 10 (100) (or, alternatively, mean PM 10 ), and years of past smoking. Other variables tested in models included dichotomous history of any ETS (with later models also testing years of exposure at work, at home as a child, and at home as an adult), parental CAD or hay fever, childhood respiratory illness, years of reported dust exposure, and years of reported fumes exposure (1987 to 1993). A back-step method retained variables in the model if they were at least borderline in significance (p 0.1). Since CAD is common, odds ratios were converted to relative risks (RRs); a method described by Abbey et al 27 and data on prevalence of CAD in unexposed subjects in this cohort were used. Sex-specific models were also run, as were models with smokers excluded. The 2 test was used to compare the categorization of subjects who had obstruction determined by PFTs with the categorization of subjects who had CAD determined by self-reported respiratory symptoms. Analysis of covariance (ANCOVA) was used to compare PFT results across categories of definite, possible, or no respiratory disease as determined by algorithms based on selfreported symptoms, with adjustment made for age, sex, and prior smoking. Adjusted mean values of PFTs for different categories of respiratory disease are reported along with significance of differences between categories of CAD from ANCOVAs. Results Population Prevalence of Obstruction The crude population prevalence of obstruction by PFTs was 10.6%, while the crude population prevalence of CAD was 15.0%. Higher prevalence of obstruction was seen for males, increasing age, history of smoking, parental CAD, history of ETS exposure, and childhood respiratory illness. Table 1 shows characteristics of the population and prevalence rates of obstruction on PFTs and of CAD as determined by respiratory symptom algorithms, as defined in the Methods section. Risk Factors for Obstructive Disease Multivariate logistic regression was used to evaluate risk factors for obstructive disease determined by PFTs and to obtain RRs with confidence intervals (CIs) (Table 2). The risk of obstruction detected by PFTs was significantly positively related to the following factors: increasing age, male sex, parental CAD or hay fever, childhood respiratory illness, and any ETS history. Past smoking showed a borderline significance (RR 1.29 for 10 years of smoking; p 0.06), even though it had been 16 years or more since any subject had smoked. More detailed analysis of ETS exposure showed significance only for years living with a smoker as an adult (the other variables, years working with a smoker and years living with a smoker as a child, were not significant). Education level failed to show a significant relationship to risk of PFT-detected obstruction. Neither PM 10 (100) nor mean PM 10 were significant in models that combined men and women, but in sex-specific models in men only, these variables did show borderline significance. Years of occupational exposure to dust and fumes were not significant. In sex-specific analyses and with exclusion of smokers, the dichotomous ETS exposure variable decreased to borderline significance. However, years living with a smoker as an adult remained significant in both men and women. In men, years of exposure to fumes at work was significant. In women, parental CAD lost significance. Subclassification of CAD There is some overlap in the different types of CAD. Table 3 shows the numbers of subjects having each type of CAD by itself and in combination with other types of CAD. A total of 209 individuals had some type of CAD by symptom algorithms, while 222 individuals had some type of physician-diagnosed CAD. In this study, exclusion of chronic Table 3 Classification of Definite CAD into Specific Respiratory Diseases* Disease Classification CAD by Symptom Algorithms Physician-diagnosed CAD Asthma only 60 82 Chronic bronchitis only 108 82 Emphysema only 8 14 Asthma with chronic bronchitis 24 33 Asthma and emphysema, with or without chronic bronchitis 5 9 Emphysema and chronic bronchitis 4 2 Total with CAD 209 222 *N 1,391; CAD cases 209. For asthma and emphysema, this includes physician diagnosis and symptoms, while for chronic bronchitis it involves symptoms only. Chronic bronchitis by symptom algorithm was any of the following: chronic cough only (n 33), chronic sputum only (n 42), or both chronic cough and chronic sputum (n 33). 52 Clinical Investigations

bronchitis or emphysema for analysis of asthma had no effect on results, so combined results are presented. However, chronic bronchitis is presented with asthma and emphysema excluded. Physician-diagnosed CAD and CAD based on symptom algorithms differed for many individuals, though there is a large amount of overlap involving 125 subjects. Many of those with diagnosed asthma were currently asymptomatic, and thus were counted as having possible asthma using the symptom algorithms. Physician-diagnosed chronic bronchitis overlapped with symptoms of chronic sputum or chronic cough for only 36 subjects. Correlation of Symptomatic CAD and PFTdetected Obstruction by 2 Test Overall, CAD detected by symptom algorithms showed a strong association with presence of obstruction on PFTs, assessed by the 2 test (Table 4). Nearly one quarter of those with definite CAD by symptom algorithms had obstruction detected by PFTs, which is more than three times the frequency of obstruction in those without CAD. Of the specific types of CAD, emphysema by symptom algorithm has the most pronounced effect on PFTs. For those with definite emphysema, 70.6% have obstruction documented by PFTs (Table 4), which is more than seven times the figure for those without emphysema (9.6%). When past smokers are excluded, the number of subjects with emphysema drops from 17 to 12, but the percentage with notable obstruction by PFTs remains similar, at 66.7% (data not shown). Asthma identified by symptom algorithm also shows a highly significant relationship to obstruction on PFTs, with a prevalence of obstruction of 36.0% (Table 4). In addition, when examined for the entire group with the 2 test, chronic sputum production is significantly related to obstruction on PFTs (Table 4). However, significance is lost when asthma and emphysema are excluded (data not shown). Chronic cough shows a nonsignificant trend. Results relating self-report of physician diagnoses to PFTs are not shown in the tables. Among subjects with any physician diagnosis of CAD, the percentage who had PFT-detected obstruction was 23.0%, compared with 8.2% in those who did not report a physician diagnosis of CAD (p 0.001). For physician-diagnosed asthma, the percentage of subjects with obstruction by PFTs was 29.0%, compared with 8.8% in those without a physician diagnosis of asthma (p 0.001). For physician-diagnosed emphysema, the percentage of subjects with obstruction by PFTs is 64.0%, compared with 9.6% in those without a physician diagnosis of emphysema (p 0.001). For physician-diagnosed chronic bronchitis, the percentage of subjects with obstruction by PFTs is 16.5%, compared with 10.0% in those without a physician diagnosis of chronic bronchitis (p 0.026). However, most of this difference de- Table 4 Tests of Association and Percentages of Study Participants Having Significant Obstruction on Pulmonary Function Tests According to Different Respiratory Symptom Complexes Categories (N 1,391) Respiratory Symptom Algorithm for Symptom Algorithm Category Tests of Association None Possible Definite 2 p Value CAD % Obstructed on 7.5% 9.7% 24.9% 54.5 0.001 PFT* Total 872 310 209 Asthma % Obstructed on 8.4% 16.9% 36.0% 69.6 0.001 PFT* Total 1,237 65 89 Emphysema % Obstructed on 9.6% 57.1% 70.6% 82.3 0.001 PFT* Total 1,367 7 17 Chronic cough % Obstructed on 9.6% 12.9% 14.4% 4.0 0.135 PFT* Total 1,030 271 90 Chronic sputum % Obstructed on 9.2% 12.3% 22.0% 16.8 0.001 PFT* Total 1,080 211 100 *Significant obstruction on PFT was determined as a PPFEV 1 of 75% or an FEV 1 /VCmax of 65%. CHEST / 115 / 1/ JANUARY, 1999 53

rives from chronic bronchitis subjects who also had asthma or emphysema. When the subjects with concomitant asthma or emphysema were excluded, the percentage of subjects with obstruction on PFTs was similar in those with or without physiciandiagnosed chronic bronchitis. Next, subjects were classified according to whether or not airflow obstruction was detected in PFTs, and the percentage having definite, possible, or no symptoms of CAD was examined. Among those with airflow obstruction, 35.4% had definite CAD, 20.4% had possible CAD, and 44.2% had no CAD according to the respiratory symptom algorithm. Among those with no airflow obstruction, 12.6% had definite CAD, 22.5% had possible CAD, and 64.9% had no CAD based on the symptom algorithm. Of those with obstructive disease found by PFTs, 34.7% had reported a physician diagnosis of CAD. Among those with no obstruction, 13.7% reported a physician diagnosis of CAD. ANCOVAs: PFT Results for Different Categories of Respiratory Disease The relationships between PFTs and CAD symptom complexes were also assessed by ANCOVA, using lung function as a continuous variable. Adjusted mean values of PFT results were obtained from ANCOVA for different categories of respiratory disease based on symptom complexes, with adjustments made for age, sex, and past smoking. A significant test indicates that some differences in the adjusted mean PFT values exist among the disease severity categories. Overall, CAD defined by the symptom algorithms was significantly associated with all the obstructive pulmonary function measures (including PPFEV 1, FEV 1 /VCmax, and PPFEF 25 75 ), and with the measures of reversible obstruction (postbronchodilator response and maximum lability in the PEF), but not with the PPVCmax (Table 5). The association between asthma indicated by the symptom algorithm and all the PFTs examined was highly significant (Table 6), as was physiciandiagnosed asthma (Table 7). Those with definite asthma according to the algorithm had lower adjusted mean values of obstructive measures and higher PEF lability and bronchodilator response than those with a reported physician diagnosis alone. Emphysema identified by the symptom algorithm was significantly associated with the obstructive measures by ANCOVA, with the following adjusted mean values for subjects who had definite emphysema with asthma excluded: PPFEV 1 of 81.0% (p 0.001), FEV 1 /VCmax of 65.0% (p 0.001), and PPFEF 25 75 of 68.4% (p 0.003). Comparison values for subjects without emphysema are PPFEV 1, 97.5%; FEV 1 /VCmax, 75.2%; and PPFEF 25 75, 98.6%. Emphysema by symptom algorithm was not significantly associated with the PPVCmax, nor with changes in the measures of reversible obstruction. For evaluation of chronic bronchitis, including chronic cough and chronic sputum, ANCOVA was performed with asthma and emphysema excluded. This eliminated possible association due to overlap with these diagnoses. Obstructive measures all tended to be lower in those with definite chronic sputum (data not shown). However, when asthma and emphysema were excluded, the ANCOVA was significant only for the FEV 1 /VCmax, with an adjusted mean of 73.9%, compared with 75.2% in those without chronic sputum (p 0.026). Chronic cough failed to show any association with PFTs by ANCOVA with asthma and emphysema excluded Table 5 Adjusted Mean Pulmonary Function Values* with Significance from Analysis of Covariance for Categories of None, Possible, and Definite Overall CAD by Symptom Algorithm (N 1,391) CAD Category Significance Pulmonary Function Test None Possible Definite F Value p Value No. of subjects 872 310 209 PPFEV 1 97.8% 97.1% 90.6% 21.7 0.001 PPVCmax 97.9% 97.1% 95.5% 2.8 0.062 FEV 1 /VCmax 75.3% 75.6% 71.7% 28.0 0.001 PPFEF 25 75 99.0% 100.5% 84.2% 17.1 0.001 Postbronchodilator response 2.14 2.02 3.97 14.3 0.001 Lability 8.49 8.70 10.51 12.2 0.001 *Mean values adjusted for age, sex, and smoking history. Due to missing data on postbronchodilator response, total N is 1,365, with individual values as follows: for no CAD, 856; for possible CAD, 304; and for definite CAD, 205. Due to missing data on lability, total N is 1,223, with individual values as follows: for no CAD, 768; for possible CAD, 269; and for definite CAD, 186. 54 Clinical Investigations

Table 6 Adjusted Mean Pulmonary Function Values* with Significance from Analysis of Covariance for Categories of None, Possible, and Definite Asthma by Symptom Algorithm (N 1,391) Asthma Category Significance Pulmonary Function Test None Possible Definite F Value p Value No. of subjects 1,237 65 89 PPFEV 1 97.7% 91.3% 85.0% 38.5 0.001 PPVCmax 97.9% 93.0% 92.7% 9.8 0.001 FEV 1 /VCmax 75.3% 74.3% 69.2% 36.2 0.001 PPFEF 25 75 99.2% 90.8% 72.7% 25.2 0.001 Postbronchodilator response 2.07 2.88 6.34 37.0 0.001 Lability 8.53 9.36 12.92 28.3 0.001 *Mean values adjusted for age, sex, and smoking history. Due to missing data on postbronchodilator response, total N is 1,365, with individual values as follows: for no asthma, 1,213; for possible asthma, 64; and for definite asthma, 88. Due to missing data on lability, total N is 1,223, with individual values as follows: for no asthma, 1,089; for possible asthma, 57; and for definite asthma, 77. (results not shown). For reported physician-diagnosed chronic bronchitis, significant associations by ANCOVA were seen only for PPFEV 1 (adjusted mean 93.7%, compared with 97.7% for no chronic bronchitis; p 0.008) and PPVCmax (adjusted mean 93.9%, compared with 98.0% for no chronic bronchitis; p 0.007). Discussion Risk Factors for Lung Obstruction Parental CAD or hay fever was a significant risk factor for obstruction detected by PFTs, as was childhood respiratory illness. Such risk factors as smoking, male sex (which is often associated with smoking and other exposures), and aging are well known. The borderline significance of past smoking in our cohort is not unexpected, given the low prevalence and the fact that no subject had smoked in the past 16 years. History of reported ETS exposure averaging at least 1 h daily was associated with obstruction found on PFTs. More detailed analysis of ETS exposure showed that the number of years living with a smoker was significantly related to obstruction found on PFTs. Exposures from work or childhood were not significant. The number of subjects in the AH- SMOG Study with current ETS exposure decreased markedly over the study period, and by 1993 few subjects still had current exposure. Greater effects might be seen for more current exposure. Previous analyses of the AHSMOG cohort have shown associations of 1977 prevalence of CAD by symptom algorithms with 10 years or more living with a smoker (RR, 1.07; p 0.01), and with 10 years or more working with a smoker (RR, 1.10; p 0.001). 29 From 1977 to 1987, these RRs each increased to 1.13 for development of CAD. 30 When exposure to ETS in childhood, adult home, and work were combined, a RR of 2.03 (95% CI, 1.45 to 2.77) was found for Table 7 Adjusted Mean Pulmonary Function Values* with Significance from Analysis of Covariance for Categories of None and Definite Physician-Diagnosed Asthma (N 1,391) Asthma Category Significance Pulmonary Function Test None Definite F Value p Value No. of subjects 1,267 124 PPFEV 1 97.4% 87.6% 54.8 0.001 PPVCmax 97.7% 93.6% 10.8 0.001 FEV 1 /VCmax 75.2% 70.8% 53.0 0.001 PPFEF 25 75 98.9% 78.4% 39.4 0.001 Postbronchodilator response 2.11 5.12 50.4 0.001 Lability 8.59 11.47 32.8 0.001 *Mean values adjusted for age, sex, and smoking history. Due to missing data on postbronchodilator response, total N is 1,365, with individual values as follows: for no physician-diagnosed asthma, 1,242; and for definite physician-diagnosed asthma, 123. Due to missing data on lability, total N is 1,223, with individual values as follows: for no physician-diagnosed asthma, 1,114; and for definite physician-diagnosed asthma, 109. CHEST / 115 / 1/ JANUARY, 1999 55

development of new CAD between 1977 and 1987. 12 Asthma with symptoms showed an RR of 1.45 (95% CI, 1.21 to 1.75) for those who had worked with a smoker for 10 years or more. 9 There has been a great deal of interest in ETS exposure and possible associated risk of diseases, including CAD, lung cancer, and heart disease. 31 34 Some of the earliest studies to look at lung function and ETS showed little or no effect for the most part, 35 37 while some of those that showed an effect 38,39 had questionable methodology. 37 It has been suggested that self-reported duration of exposure to ETS has low reliability, and that this may explain the inability of earlier studies to show a significant dose-response relationship between ETS and lung cancer. 40 On the other hand, dichotomous questions on whether or not one was ever exposed to ETS have shown much better reliability. 41 Additionally, studies examining the relationships between ETS and symptoms of CAD have shown significant results. 9,12,42 More recent studies in adults have been reporting borderline significant relationships between ETS exposure and lung function, with some significant trends of decreased PFTs related to ETS exposure. 43 47 The findings on ETS reported here are consistent with others in the literature. 48,49 Public health efforts in the United States have attempted to increase awareness of potential effects of ETS, and may be leading to decreases in ETS exposure. This might be expected to contribute to decreased severity of asthma. 50 However, the prevalence of asthma has increased since the 1970s, for reasons that are not clear. 8,10 Also, depending on how similar the effect of ETS is to the effect of active smoking, cessation of ETS exposure could result in some reduction in age-related decline in lung function for exposed persons. 51 Efforts to further document such improvements may be beneficial, particularly if ETS exposure continues to decrease. Other risk factors studied in this cohort in previous papers included long-term ambient concentrations of air pollutants, examined here using both mean PM 10 concentrations and the frequency of PM 10 concentrations in excess of 100 g/m 3. 13,21 One of these measures was forced into models to adjust for air pollution. Particulate pollution (PM 10 ) may be related especially to chronic cough and chronic sputum. 22 Borderline significance of mean PM 10 and PM 10 (100) was seen in men in sex-specific models. These analyses may lack power, as the outcome variable was not continuous. Detailed analyses relating specific air pollutants and lung function in the AHSMOG Study have been conducted recently. 52 In these analyses, sex differences were noted. The greatest decrement in lung function associated with PM 10 (100) was found in male subjects whose parents had CAD or hay fever. 52 CAD Symptoms and Obstructive Disease Detected by PFTs Overall, definite CAD, asthma, and emphysema identified using algorithms based on self-reported respiratory symptoms all showed a marked association with PFT findings of obstruction in ANCOVAs. They all also showed a highly significant relationship with obstruction found in PFTs, assessed by the 2 test. These results confirm the use of these algorithms as chronic respiratory disease indicators for studying associations with ambient air pollutants and ETS exposures. 9,12,13,20,21,29,30 Previous papers have used the term airway obstructive disease. However, while asthma and emphysema do involve airway obstruction, the current analyses did not find chronic bronchitis to be independently related to airway obstruction. The association found for self-reported physician diagnosis of CAD with obstruction detected by PFTs was similar to the association for definite CAD by algorithm with obstruction detected by PFTs, although the association was not quite as marked. This is largely because the symptom algorithms for asthma and emphysema are more stringent, requiring both physician diagnosis and current symptoms. Both a finding of asthma based on symptom algorithm and a self-reported physician diagnosis of asthma showed a very significant association with all PFTs, with a fairly marked decrease in obstructive measures. Measures of reversible obstruction included PEF lability and postbronchodilator response, each of which was markedly elevated for both the algorithm and physician diagnosis of asthma, as expected. Mild chronic bronchitis may have a reversible component. 53 However, when asthma and emphysema were excluded, no significant association was seen between chronic bronchitis and reversible measures of obstruction. Although the small decrement of FEV 1 /VCmax seen with symptoms of chronic sputum in this study would not appear to be clinically significant, the questions on chronic cough and chronic sputum did contribute to detection of those with obstruction found on PFTs. Other epidemiologic studies have suggested that the use of a direct question about physician-confirmed chronic bronchitis is better correlated with PFTs than is use of the symptoms of chronic cough and chronic phlegm. 53,54 In this study, the physician diagnosis of chronic bronchitis was related to decrements in both PPFEV 1 and PPVCmax, though not with FEV 1 / 56 Clinical Investigations

VCmax. It did show more clinical significance than the symptoms in correlation with PFTs, although the decrements found were not large. Despite the small numbers of people with emphysema by either physician diagnosis or symptom algorithm, reductions in the adjusted mean values for obstructive measures of lung function were readily apparent, and more marked than for either asthma or chronic bronchitis. Both the symptom algorithms and physician diagnoses labeled some subjects as having obstructive diseases who had little obstruction found on PFTs. With the reversible nature of asthma and variable nature of chronic bronchitis, this is not unexpected. On the other hand, a significant proportion of those with obstructive disease by PFTs were not detected by either the symptom algorithms or report of physician diagnoses. Others have also reported that CAD may often be undiagnosed, particularly in the elderly. 2,55,56 Study Limitations This study has several limitations. The AHSMOG cohort is not a general population sample, and none of its members currently smoke. Thus, the risk factors for lung obstruction determined in this cohort may not be the same as in the general population, especially among current smokers. Former smokers are included in the AHSMOG Study, but the prevalence rate of ever smoking is much lower than in the general population. It had been at least 16 years since any subject had smoked. Those who were tested were a selected sample of survivors through the study period, and may differ from those in the initial cohort. More healthy subjects may also have been preferentially selected, because previous questionnaires were mailed, while performance of PFTs required travel to a central testing location. Together, these may cause differential loss to follow-up of those with worse PFT results, and thus bias against finding significant risk factors when they exist, similar to the survivor effect 57 or the healthy worker effect. 58 Because PFTs were not performed initially, longitudinal decrements of lung function cannot be assessed. Thus, only the prevalence of obstructive disease by PFTs in 1993 can be determined, and not incidence. This study may have failed to detect significant risk factors for obstruction by PFTs in the multiple logistic regressions due to lack of power, with the relatively small sample size and small number of cases of CAD. For example, PM 10 (100) was of borderline significance in men. Previous results from this cohort with larger numbers have shown significant results with PM 10 (100) and symptomatic CAD. 13,21,22 Many alternative definitions of obstruction based on PFTs have been used. 59,60 The definition used in this paper was selected a priori and is consistent with definitions frequently used in clinical settings; it is more selective than some but more inclusive than others. 59 Another method for defining obstruction based on PFTs is to use the ratio of FEV 1 /FVC less than the fifth percentile of the lower limit of normal in a reference population. 60 Such a definition of obstructive disease for this cohort has been created. 15 Use of this definition gave a somewhat smaller number with obstructive disease than the definition used in this paper, although in the majority of subjects with obstruction (86 of 116), the definitions overlapped. Using this alternative definition in analysis of variance comparing categories of CAD based on symptom algorithms gave similar results to those presented in this paper. For logistic regressions with this alternative definition, odds ratios were similar to those presented here. However, most variables showed less significance, although prior smoking was slightly more significant. If this previous definition is revised to additionally include any subjects with PPFEV 1 less than 75%, then more subjects will be included. Logistic regressions using this definition of obstruction showed significance for years of prior smoking (RR 1.34 for 10 years of exposure; p 0.023) and history of past exposure to ETS as an adult (RR 1.47 for exposure averaging more than 1 h daily; p 0.020), and also borderline significance for mean concentration of PM 10 exposure (RR 1.29 for an exposure increment equal to the PM 10 interquartile range of 25 g/m 3 ;p 0.065). The use of self-reporting of both symptoms and physician diagnoses for disease assessment is an additional limitation of this study, because subjects may be inconsistent in their recall of symptoms or diagnoses. However, questions on whether subjects had ever had asthma and emphysema showed little inconsistency (less than 1%) when answers from 1992 were compared with 1993 answers. Separating asthma, chronic bronchitis, and emphysema by questionnaire is a weakness, but follows ATS guidelines. 20,26 Previously, self-report of diagnosis of asthma has been verified to agree with actual physician diagnosis by abstracting of physician records, performed by Greer et al. 9 Other studies have evaluated eosinophilia and IgE levels as risk factors for CAD, and particularly for asthma. 54,61 These results are not available on the AHSMOG population, so they cannot be assessed for association with the risk of obstruction. Bronchial CHEST / 115 / 1/ JANUARY, 1999 57

challenge tests were not performed. However, questionnaire responses regarding history of wheezing may correlate better with physician assessment of asthma than bronchial challenge tests do. 19,62 The PEF index of lability was obtained, and it correlated very well with asthma. 15 Nevertheless, a wider battery of physiologic and immunologic tests would have been of great benefit in assessing CAD in this study. 25,53 CAD remains an important cause of morbidity and mortality, despite the decline in smoking in this country. 63 Large numbers of individuals may have undiagnosed CAD or airway obstruction, which may have few early symptoms. Thus, further study of CAD and airway obstruction and their determinants in nonsmokers is needed. 5 References 1 Turkeltaub PC, Gergen PJ. Prevalence of upper and lower respiratory conditions in the US population by social and environmental factors: data from the second National Health and Nutrition Examination Survey, 1976 to 1980 (NHANES II). Ann Allergy 1991; 67:147 154 2 Enright PL, Kronmal RA, Higgins MW, et al. Prevalence and correlates of respiratory symptoms and disease in the elderly. Chest 1994; 106:827 834 3 Boezen HM, Schouten JP, Postma DS, et al. Distribution of peak expiratory flow variability by age, gender, and smoking habits in a random population sample aged 20 70 years. Eur Respir J 1994; 7:1814 1820 4 Krzyzanowski M, Camilli AE, Lebowitz MD. Relationships between pulmonary function and changes in chronic respiratory symptoms: comparison of Tucson and Cracow longitudinal studies. Chest 1990; 98:62 70 5 Whittemore AS, Perlin SA, DiCiccio Y. Chronic obstructive pulmonary disease in lifelong nonsmokers: results from NHANES. Am J Public Health 1995; 85:702 706 6 Collins JG. Prevalence of selected chronic conditions: United States, 1986 88. Vital Health Stat [10] 1993; 182:1 87 7 Institute of Medicine Committee on the Health Effects of Indoor Allergens. Indoor allergens: assessing and controlling adverse health effects. Pope AM, Patterson R, Burge H, eds. Washington, DC: National Academy Press, 1993; 44 85 8 Weiss KB, Gergen PJ, Wagener DK. Breathing better or wheezing worse? The changing epidemiology of asthma morbidity and mortality. Annu Rev Public Health 1993; 14:491 513 9 Greer JR, Abbey DE, Burchette RJ. Asthma related to occupational and ambient air pollutants in nonsmokers. J Occup Environ Med 1993; 35:909 915 10 Mannino DM, Homa DM, Pertowski CA, et al. Surveillance for asthma United States, 1960 1995. MMWR Morb Mortal Wkly Rep 1998; 47(SS-1):1 27 11 Lebowitz MD, Burrows B. Respiratory symptoms related to smoking habits of family adults. Chest 1976; 69:48 50 12 Robbins AS, Abbey DE, Lebowitz MD. Passive smoking and chronic respiratory disease symptoms in non-smoking adults. Int J Epidemiol 1993; 22:809 817 13 Abbey DE, Hwang BL, Burchette RJ. Estimated long-term ambient concentrations of PM 10 and development of respiratory symptoms in a nonsmoking population. Arch Environ Health 1995; 50:139 150 14 Pope CA, Bates DV, Raizenne ME. Health effects of particulate air pollution: time for reassessment? Environ Health Perspect 1995; 103:472 480 15 Enright PL, Burchette RJ, Peters JA, et al. Peak flow lability: association with asthma and spirometry in an older cohort. Chest 1998; 112:895 901 16 Quanjer PH, Lebowitz MD, Gregg I, et al. Peak expiratory flow: conclusions and recommendations of a Working Party of the European Respiratory Society. Eur Respir J Suppl 1997; 24:2S 8S 17 Helsing KJ, Comstock GW, Speizer FE, et al. Comparison of three standardized questionnaires on respiratory symptoms. Am Rev Respir Dis 1979; 120:1221 1231 18 Burr ML. Diagnosing asthma by questionnaire in epidemiological surveys. Clin Exp Allergy 1992; 22:509 510 19 Jenkins MA, Clarke JR, Carlin JB, et al. Validation of questionnaire and bronchial hyperresponsiveness against respiratory physician assessment in the diagnosis of asthma. Int J Epidemiol 1996; 25:609 616 20 Hodgkin JE, Abbey DE, Euler GL, et al. COPD prevalence in nonsmokers in high and low photochemical air pollution areas. Chest 1984; 86:830 838 21 Abbey DE, Lebowitz MD, Mills PK, et al. Long-term ambient concentrations of particulates and oxidants and development of chronic disease in a cohort of non-smoking California residents. Inhalat Toxicol 1995; 7:19 34 22 Abbey DE, Nishino N, McDonnell WF. Development of chronic productive cough as associated with long-term ambient inhalable particulate pollutants (PM 10 ) in non-smoking adults: the AHSMOG Study. Appl Occup Environ Hyg 1998; 13(6):444 452 23 Enright PL, Johnson LR, Connett JE, et al. Spirometry in the lung health study: 1. Methods and quality control. Am Rev Respir Dis 1991; 143:1215 1223 24 McDonnell WF, Enright PL, Abbey DE, et al. Spirometric prediction equations for older adults. Respir Med 1998; 92:914-921 25 Quackenboss JJ, Lebowitz MD, Krzyzanowski M. The normal range of diurnal changes in peak expiratory flow rates: relationship to symptoms and respiratory disease. Am Rev Respir Dis 1991; 143:323 330 26 Ferris BG. Epidemiology Standardization Project (American Thoracic Society). Am Rev Respir Dis 1978; 118(6):1 120 27 Abbey DE, Euler GL, Moore JK, et al. Applications of a method for setting air quality standards based on epidemiological data. JAPCA 1989; 39:437 445 28 Abbey DE, Moore J, Petersen F, et al. Estimating cumulative ambient concentrations of air pollutants: description and precision of methods used for an epidemiological study. Arch Environ Health 1991; 46:281 287 29 Euler GL, Abbey DE, Hodgkin JE, et al. Chronic obstructive pulmonary disease symptom effects of long-term cumulative exposure to ambient levels of total oxidants and nitrogen dioxide in California Seventh-day Adventist residents. Arch Environ Health 1988; 43:279 285 30 Abbey DE, Petersen FF, Mills PK, et al. Long-term ambient concentrations of total suspended particulates, ozone, and sulfur dioxide and respiratory symptoms in a non-smoking population. Arch Environ Health 1993; 48:33 46 31 Environmental Protection Agency. Respiratory health effects of passive smoking: lung cancer and other disorders. EPA/ 600/6 90/006F. Washington, DC: Environmental Protection Agency, Office of Research and Development, Office of Air and Radiation, 1992 32 Kawachi I, Colditz GA, Speizer FE, et al. A prospective study of passive smoking and coronary heart disease. Circulation 1997; 95:2374 2379 58 Clinical Investigations

33 Glantz SA, Parmley WW. Passive smoking and heart disease: mechanisms and risk. JAMA 1995; 273:1047 1053 34 Jinot J, Bayard S. Respiratory health effects of passive smoking: EPA s weight-of-evidence analysis. J Clin Epidemiol 1994; 47:339 349 35 Schilling RSF, Letai AD, Hui SL, et al. Lung function, respiratory disease, and smoking in families. Am J Epidemiol 1977; 106:274 283 36 Kentner M, Triebig G, Weltle D. The influence of passive smoking on pulmonary function: a study of 1351 office workers. Prev Med 1984; 13:656 669 37 Lebowitz MD. Influence of passive smoking on pulmonary function: a survey. Prev Med 1984; 13:645 655 38 White JR, Froeb HF. Small-airways dysfunction in nonsmokers chronically exposed to tobacco smoke. N Engl J Med 1980; 302:720 723 39 Kauffmann F, Tessier JF, Oriol P. Adult passive smoking in the home environment: a risk factor for chronic airflow limitation. Am J Epidemiol 1983; 117:269 280 40 Pron GE, Burch JD, Howe GR, et al. The reliability of passive smoking histories reported in a case-control study of lung cancer. Am J Epidemiol 1988; 127:267 273 41 Coultas DB, Peake GT, Samet JM. Questionnaire assessment of lifetime and recent exposure to environmental tobacco smoke. Am J Epidemiol 1989; 130:338 347 42 Dayal HH, Khuder S, Sharrar R, et al. Passive smoking in obstructive respiratory diseases in an industrialized urban population. Environ Res 1994; 65:161 171 43 Brunekreef B, Fischer P, Remijn B, et al. Indoor air pollution and its effect on pulmonary function of adult non-smoking women: III. Passive smoking and pulmonary function. Int J Epidemiol 1985; 14:227 230 44 Hole DJ, Gillis CR, Chopra C, et al. Passive smoking and cardiorespiratory health in a general population in the west of Scotland. BMJ 1989; 299:423 427 45 Kalandidi A, Trichopoulos D, Hatzakis A, et al. The effect of involuntary smoking on the occurrence of chronic obstructive pulmonary disease. Soz Praventivmed 1990; 35(1):12 16 46 Masi JA, Hanley JA, Ernst P, et al. Environmental exposure to tobacco smoke and lung function in young adults. Am Rev Respir Dis 1988; 138:296 299 47 Masjedi MR, Kazemi H, Johnson DC. Effects of passive smoking on the pulmonary function of adults. Thorax 1990; 45:27 31 48 Lesmes GR, Donofrio KH. Passive smoking: the medical and economic issues. Am J Med 1992; 93(1A):38S 42S 49 Spitzer WO, Lawrence V, Dales R, et al. Links between passive smoking and disease: a best-evidence synthesis. A report of the Working Group on Passive Smoking. Clin Invest Med 1990; 13:17 42 50 Murray AB, Morrison BJ. The decrease in severity of asthma in children of parents who smoke since the parents have been exposing them to less cigarette smoke. J Allergy Clin Immunol 1993; 91:102 110 51 Anthonisen NR, Connett JE, Kiley JP, et al. Effects of smoking intervention and the use of an inhaled anticholinergic bronchodilator on the rate of decline of FEV 1 : the Lung Health Study. JAMA 1994; 272:1497 1505 52 Abbey DE, Burchette RJ, Knutsen SF, et al. Long-term particulate and other air pollutants and lung function in nonsmokers. Am J Respir Crit Care Med 1998; 158:289 298 53 Lebowitz MD, Postma DS, Burrows B. Adverse effects of eosinophilia and smoking on the natural history of newly diagnosed chronic bronchitis. Chest 1995; 108:55 61 54 Postma DS, Lebowitz MD. Persistence and new onset of asthma and chronic bronchitis evaluated longitudinally in a community population sample of adults. Arch Intern Med 1995; 155:1393 1399 55 Lebowitz MD, Holberg CJ. Comparisons of spirometric reference values and the proportions of abnormal subjects among male smokers and those symptomatic in a community population. Am Rev Respir Dis 1990; 141:1491 496 56 Renwick DS, Connolly MJ. Prevalence and treatment of chronic airways obstruction in adults over the age of 45. Thorax 1996; 51:164 168 57 Burrows B, Lebowitz MD, Camilli AE, et al. Longitudinal changes in forced expiratory volume in one second in adults: methodologic considerations and findings in healthy nonsmokers. Am Rev Respir Dis 1986; 133:974 980 58 Eisen EA, Wegman DH, Louis TA, et al. Healthy worker effect in a longitudinal study of one-second forced expiratory volume (FEV 1 ) and chronic exposure to granite dust. Int J Epidemiol 1995; 24:1154 1161 59 Quadrelli SA, Roncoroni AJ, Porcel G. Analysis of variability in interpretation of spirometric tests. Respiration 1996; 63: 131 136 60 American Thoracic Society. Lung function testing: selection of reference values and interpretative strategies. Am Rev Respir Dis 1991; 144:1202 1218 61 Burrows B, Lebowitz MD, Barbee RA, et al. Findings before diagnosis of asthma among the elderly in a longitudinal study of a general population sample. J Allergy Clin Immunol 1991; 88:870 877 62 Pattemore PK, Asher MI, Harrison AC, et al. The interrelationship among bronchial hyperresponsiveness of asthma, and asthma symptoms. Am Rev Respir Dis 1990; 142:549 554 63 Davis RM, Novotny TE. The epidemiology of cigarette smoking and its impact on chronic obstructive pulmonary disease. Am Rev Respir Dis 1989; 140(3 pt 2):S82 S84 CHEST / 115 / 1/ JANUARY, 1999 59