- Email: [email protected]
Predictors of Methacholine Responsiveness in a General Population
Predictors of Methacholine Responsiveness in a General Population* Joel Schwartz, PhD; Christian Schindler, PhD; Elisabeth Zemp, MD; Andre´ P. Perruchoud, MD, FCCP; Jean-Pierre Zellweger, MD; Brunello Wu¨thrich, MD; Philippe Leuenberger, MD; Ursula Ackermann-Liebrich, MD; and SAPALDIA Team†
Objective: Methacholine responsiveness is an end point widely used in epidemiologic studies of asthma. This study aims to quantify the relative importance of different predictors of responsiveness such as age, sex, airway caliber, smoking and atopic status, and potential interactions deserving further investigation. Methods: Methacholine challenge was performed in 7,126 participants (aged 18 to 60 years) of the Swiss Study on Air Pollution and Lung Diseases in Adults according to the European Respiratory Health Survey protocol. Responsiveness was quantified by the slope between percentage decrements in FEV1 and cumulative methacholine dose. Variation of slopes according to sex, smoking, and atopy was then examined separately by multivariate regression models that controlled for baseline FEV1. Results: We found a nonlinear relationship between methacholine slope and baseline FEV1 for both sexes, which could be well described by a quadratic function. The corresponding curves were almost identical in the region of overlap for male and female neversmokers. Methacholine responsiveness declined with age. The slope of this decline was less steep among nonatopic persons and nonsmokers compared with atopic neversmokers. Methacholine responsiveness increased with the number of cigarettes smoked per day and with the number of positive skin-prick test results (except among heavy smokers). Conclusions: Our multiple regression results show that bronchial responsiveness (BR) varies with age, FEV1, and smoking and atopic status. They suggest that there is a physiologic basis for the univariate sex difference in BR. Secondly, they show that while smaller airways are more responsive than larger ones, the reduction of responsiveness diminishes with each increase of lung size. The quantification of the relative influence of the different factors examined should help in the interpretation of BR. (CHEST 2002; 122:812– 820) Key words: atopy; methacholine responsiveness; sex; smoking Abbreviations: ATS ⫽ American Thoracic Society; BR ⫽ bronchial responsiveness; CI ⫽ confidence interval; ECRHS ⫽ European Respiratory Health Survey; PD20 ⫽ provocative dose of methacholine causing a 20% fall in FEV1; SAPALDIA ⫽ Swiss Study on Air Pollution and Lung Diseases in Adults; SPT ⫽ skin-prick test
responsiveness is often used to conM ethacholine firm asthmatic status in patients, and as a predictor of later development of respiratory diseases.1,2 It is *From the Institute of Social and Preventive Medicine (Drs. Schindler, Zemp, and Ackermann-Liebrich), University of Basel, Basel, Switzerland; Harvard School of Public Health (Dr. Schwartz), Boston, MA; Department of Internal Medicine (Dr. Perruchoud), University Clinic Basel, Basel, Switzerland; Division of Pulmonology (Drs. Zellweger and Leuenberger), University of Lausanne, Lausanne, Switzerland; and Department of Dermatology (Dr. Wu¨thrich), University Hospital Zurich, Zurich, Switzerland. †A complete list of participants is given in the Appendix. Apart from the Swiss National Science Foundation and the Federal Office for Science and Education, logistic and financial support was received from cantonal governments (Basel Geneva, Graubu¨nden, Vaud, Ticino, Zurich, and Valais) and from the cantonal offices for air hygiene measurements. Further financial support was received from the Swiss Society of 812
widely used in epidemiologic studies, as in the European Respiratory Health Survey (ECRHS),3 where a standardized tool for measurements of bronchial responsiveness (BR) to methacholine has been developed to estimate variation in prevalence of increased BR and Pulmonology, the Lega Ticinese contra le tubercolosi e le malattie polmonari, and various cantonal offices of public health. SAPALDIA is part of Swiss National Research Program 26A, supported by grant Nos. 4026-28099, 32-042532.94, and 3252720.97 from the Swiss National Science Foundation, and by the Federal Office of Education and Science. SAPALDIA Basel is part of the European Respiratory Health Survey. Manuscript received July 9, 2001; revision accepted April 29, 2002. Correspondence to: Ursula Ackermann-Liebrich, MD, Institute of Social and Preventive Medicine, Steinengraben 49, CH– 4051, Basel, Switzerland Clinical Investigations
predictors of asthma in different European countries. Apart from a general overview of variation in BR,4 participating countries and also authors from other geographic regions reported findings regarding several predictors. Besides FEV1 and symptom status, female sex,5–12 smoking,5–7,10,12 atopy,12–19 occupational exposures,13,20,21 and geographic region14 were found to be associated with increased responsiveness. The discussion on how best to report results on BR and its most important predictors has not yet come to a satisfactory conclusion. Several of the abovementioned articles used the provocative dose of methacholine causing a 20% fall in FEV1 (PD20) as a measure of BR.5,6,12,13,20 Others have argued that the slope of the decline of FEV1 with increasing dose of methacholine is a better way of measuring responsiveness because a value can be assigned to all subjects,22 and it was shown to be a more sensitive measure for center comparison in the ECRHS4 than PD20. The influence of the airway caliber on responsiveness,11,12,22 and the resulting differences between men and women, the relative importance of differences between smokers and neversmokers,5,6 or the influence of atopy still deserve investigation. We decided to use the slope of BR in the Swiss Study on Air Pollution and Lung Diseases in Adults (SAPALDIA)23 to analyze whether symptom status, sex, smoking, and atopy independently influence BR. Since the prevalence of atopy, smoking, and asthma vary in men and women in different age groups and, moreover, effects may vary across subgroups or as a function of the other risk factors, we decided to stratify analyses by sex and different levels of risk factors. This was possible due to the large sample of participants who have undergone methacholine testing in the SAPALDIA study (n ⫽ 7,126), and the relatively broad age range (18 to 62 years) of the population studied. Materials and Methods Study Population SAPALDIA is a multicenter study designed to examine potential associations between air pollution and respiratory outcomes.23–27 The eight study areas (Aarau, Basel, Davos, Geneva, Lugano, Montana, Payerne, and Wald) were chosen to represent a range of urbanization, altitude, air pollution, and meteorology. A random sample of adults aged 18 to 60 years was drawn from the registry of inhabitants of each town. Successfully recruited subjects (59%) underwent a standardized computerized interview with an expanded version of the ECRHS questionnaire.3 A total of 9,651 subjects were recruited for the study. Allergy tests included skin-prick tests (SPTs), in addition to blood samples for total IgE. SPTs were performed on the volar forearm using Phazet (Pharmacia; Uppsala, Sweden) allergen-coated steel lancets, corresponding to a prick test extract of 100,000 biological units per milliliter.19 The inhalant allergens tested for were www.chestjournal.org
chosen according to the known prevalence of specific sensitization in Switzerland and in accordance with the ECRHS allergy testing protocol3: cat fur; mold (Cladosporium herbarum); Timothy grass pollen; pellitory-of-the-wall pollen (Parietaria); house dust mite (Dermatophagoides pteronyssinus); mold (Alternaria tenuis); birch pollen; dog epithelia. The test also included histamine (positive control) and a noncoated lancet (negative control). Spirometry measurements were done using a pulmonary function system (2200SP; SensorMedics; Yorba Linda, CA) using a mass flow anemometer, displaying and recording flow-volume curves for each trial. The testing was performed by trained fieldworkers according to the recommendations on standardization of spirometry of the American Thoracic Society (ATS).27 Three ATS acceptability criteria (volume of extrapolation ⬍ 5% FVC or ⬍ 100 mL, expiratory time of at least 6 s, and end of test [no volume change for the last 2 s of the maneuver]) were displayed and recorded with each maneuver. The two ATS reproducibility criteria (largest two FVC and FEV1 values within 5% of one another) were recorded with the set of best values, and guided the technician’s decisions for additional maneuvers. At least three and up to eight maneuvers were required, until the five criteria described were met. Subjects were excluded from methacholine testing if they had an FEV1/FVC ratio ⬍ 80% of predicted, had an FEV1 ⬍ 70% of predicted according to European standards, or had difficulty performing the spirometry maneuver. This reduced the sample size of subjects to 7,126.1 Methacholine chloride (Provocholine; Roche; Nutley, NJ) was prepared in 25.0, 6.25, 1.56, and 0.39 mg/mL solutions in a phosphate buffer without phenol. Methacholine was administered through an aerosol dosimeter (Mefar MB3; Boveso, Italy). The schedule in nonreactive subjects was then four inhalations at 0.39 mg/mL and three inhalations in each of the other concentrations. If FEV1 decreased ⬎ 10% from the baseline level, smaller increments were introduced. Testing continued until the final dose, or until a 20% reduction in FEV1 was reached. Methacholine responsiveness was quantified by calculating a dose-response slope for each subject similar to the one suggested by O’Connor et al.22 In order to avoid negative slopes, the absolute decline in FEV1 was defined as the difference between the maximum FEV1 over all levels tested and FEV1 at the last level measured, irrespective of the dose level at which the maximum occurred. The percentage decline was computed using the maximum of FEV1 as the reference value. The slope was then defined as the ratio between the percentage decline in FEV1 and the total cumulative dose of methacholine (in micromoles) administered. As the distribution of these slopes was quite skewed, a logarithmic transformation was used to obtain a more symmetrical distribution better suited for statistical analysis. In order not to lose slope values of 0, a small constant (ie, 0.01) was added before taking the logarithm. We first examined variation between slopes according to sex, smoking, and atopy. In a second step, multivariate regression models controlling for baseline FEV1 (ie, by a linear and a quadratic term) were recomputed. To examine potential interactions between different factors, analyses were stratified or interaction terms were used. Definition of Symptom Categories Asthma was defined as a positive response to both parts of the question, “Have you ever had asthma, and was this confirmed by a doctor?” Bronchitis symptoms were defined as a positive response to the question, “Do you usually cough during the day or night?” or the question, “Do you usually bring up any phlegm from your chest during the day or night?” If the subjects reported CHEST / 122 / 3 / SEPTEMBER, 2002
813
Table 1—Characteristics of the Sampled Subjects With and Without Methacholine Slopes* Without Methacholine Slopes†
With Methacholine Slopes
Characteristics
Men (n ⫽ 1,138)
Women (n ⫽ 1,387)
Men (n ⫽ 3,605)
Women (n ⫽ 3,521)
Age, yr Asthma Wheeze apart from cold Bronchitic symptoms Hay fever Current smoker Former smoker Neversmoker Positive SPT result FEV1, L
44.6 ⫾ 11.6 9.6 12.5 14.7 18.2 41.9 26.6 31.5 26.1 3.63
43.3 ⫾ 11.5 9.3 9.2 10.7 17.8 27.3 18.5 54.2 19.7 2.86
39.6 ⫾ 11.6 6.0 7.4 10.1 16.6 37.4 25.5 37.1 25.3 4.18
40.6 ⫾ 11.4 5.5 5.8 7.8 16.2 29.3 19.9 50.7 21.6 3.09
*Data are presented as mean ⫾ SD or % unless otherwise indicated. †Excluded from methacholine test.
a physician diagnosis of asthma (regardless of whether the subject had bronchitis symptoms), they were classified as asthmatic.
Results Table 1 shows the characteristics of the examined persons, and of those without methacholine slopes. As expected, the prevalence of asthma and wheezing was higher in the excluded group and FEV1 was lower. There was little difference between the samples in the prevalence of atopy by SPT or in self-report of hay fever. The prevalence of neversmoking was higher among tested men, but lower among tested women. The prevalence of increased BR in terms of the PD20 was 11.7% in men and 21.8% in women. Table 2 shows the geometric mean of methacholine slope (percentage decrease in FEV1 per micromole of methacholine) by smoking category, sex, and level of atopy (defined as none vs one or two and more positive SPT results). Women had higher slopes in all categories. Moreover, methacholine responsiveness showed a trend to increase with increasing number of positive SPT results. However, smoking was associated with an increase in responsiveness only among subjects with negative SPT results. Figure 1 shows the medians and interquartile ranges of slopes for eight different subgroups (ie, male and female neversmokers and current smokers with and without positive SPT results, respectively). The medians exhibit a similar pattern as the geometric means in Table 2. Again, a clear association of slopes with smoking can only be seen among subjects with negative SPT results. In addition, Figure 1 shows that the shape of the distributions changes with atopic status (ie, slopes have a more skewed distribution in the subgroups of subjects with positive SPT results). 814
Table 3 shows the geometric mean slope by smoking category, sex, and symptom category. If the subject reported neither a physician’s diagnosis of asthma nor bronchitis symptoms, the symptom status was classified as “none.” Again, women had higher average slopes in all categories considered. Consistently, the geometric means of slope were lowest among nonsymptomatic subjects, intermediate among subjects with bronchitis symptoms, and highest among subjects with asthma symptoms. As before, associations with smoking were less consistent, being positive only among subjects without asthma symptoms. None of these results are adjusted for continuous risk factors, such as level of lung function or age. The next sets of results are from multivariate analyses, and are reported by major categories of risk factors. The Influence of the Level of Lung Function and Sex on Methacholine Responsiveness Women are known to have greater BR than men.5,6 BR also varies with level of pulmonary Table 2—Geometric Means of Methacholine Slope in Percentage Decrease in FEV1 per Micromole of Methacholine by Smoking, Sex, and Atopy Categories* Positive SPT Results, No. Smoking Status
None
One
Men (n ⫽ 3,485) n ⫽ 2,621 n ⫽ 483 Neversmoker (n ⫽ 1,285) 0.65 0.94 Former smoker (n ⫽ 884) 0.63 1.01 Current smoker (n ⫽ 1,316) 0.91 1.01 Women n ⫽ 2,683 n ⫽ 436 Neversmoker (n ⫽ 1,730) 1.11 1.72 Former smoker (n ⫽ 682) 1.29 1.47 Current smoker (n ⫽ 994) 1.35 1.99
Two or More n ⫽ 381 1.55 1.83 1.19 n ⫽ 287 2.83 3.22 2.14
*To avoid values of zero, a constant of 0.01 was added to each slope before computing the geometric means. Clinical Investigations
Table 3—Geometric Means of Methacholine Slopes in Percentage Decrease in FEV1 per Micromole of Methacholine by Smoking, Sex, and Symptom Status* Symptom Status† Smoking Status
None
Men (n ⫽ 3,594) Neversmoker (n ⫽ 1,334) Former smoker (n ⫽ 917) Current smoker (n ⫽ 1,343) Women (n ⫽ 3513) Neversmoker (n ⫽ 1,783) Former smoker (n ⫽ 700) Current smoker (n ⫽ 1,030)
n ⫽ 3,014 0.71 0.64 0.87 n ⫽ 3044 1.18 1.23 1.38
Bronchitis
Asthma
n ⫽ 363 0.96 0.85 1.06 n ⫽ 275 1.37 1.66 1.68
n ⫽ 217 2.34 4.62 2.18 n ⫽ 194 5.93 5.74 3.53
*To avoid values of zero, a constant of 0.01 was added to each slope before computing the geometric means. †Symptom categories are “none” if the subject had neither physician-diagnosed asthma or bronchitis symptoms, “bronchitis” if the subject had bronchitis symptoms without a diagnosis of asthma, and “asthma” if the subject had physician-diagnosed asthma with or without bronchitis symptoms.
function.11,12,22 This section examines the extent to which the sex difference in responsiveness is explained by the level of pulmonary function. The functional form of the dependence of methacholine slope on level of lung function is also not clear. To test this, separate regression analyses were conducted for each sex, controlling for area of study, age, and the number of positive SPT results. The analyses began with neversmokers. FEV1 was divided into quintile classes within each sex, and a
dummy variable for each quintile class was entered into the regression models for each sex. Figure 2 shows the geometric mean of methacholine slope (adjusted for covariates) in each quintile class, for male and female neversmokers, plotted against the mean FEV1 in that quintile class. Two patterns emerged. First, within each sex, the relationship with FEV1 appears nonlinear. Second, the two sets of points appear to lie almost on the same curve. When both sexes were combined, and a linear
Figure 1. Medians and interquartile ranges of methacholine slopes (expressed as percentage change in FEV1 per micromole of methacholine) in male neversmokers (mNS) and female neversmokers (fNS), and male current smokers (mCS) and female current smokers (fCS) with negative (⫺) and positive (⫹) SPT results, respectively.
Figure 2. Geometric means of methacholine slopes (expressed as percentage change in FEV1 per micromole of methacholine) for different quintile classes of baseline FEV1 in male and female neversmokers (NS) and current smokers (CS), respectively, after group-specific adjustment for area of study, age, number of positive SPT results, and daily number of cigarettes smoked.
www.chestjournal.org
CHEST / 122 / 3 / SEPTEMBER, 2002
815
Table 4 —Decline in Methacholine Responsiveness With Age by Smoking and Atopic Status* SPT Result Negative
Positive
Smoking Status
Coefficient
95% CI
Coefficient
Neversmokers Former smokers Current smokers
⫺0.0095† ⫺0.0072 ⫺0.0059
⫺0.0146 to ⫺0.0044 ⫺0.0161 to 0.0017 ⫺0.0122 to 0.0004
⫺0.0200† ⫺0.026‡ ⫺0.0032
95% CI ⫺0.0319 to ⫺0.0081 ⫺0.0466 to ⫺0.0054 ⫺0.0155 to 0.0091
*Natural logarithm of methacholine responsiveness, to avoid values of zero, a constant of 0.01 was added to each slope before taking the logarithm. This Table shows the coefficient of age (1 year) in each strata, after adjustment for covariates (area, FEV1, [linear and quadratic term], sex, No. of cigarettes per day [smokers]), and No. of positive SPT results (subjects with at least one positive SPT result). †p ⬍ 0.001 for regression coefficients. ‡p ⬍ 0.05 for regression coefficients.
and quadratic term for FEV1 used, sex was no longer a significant predictor of methacholine responsiveness (p ⫽ 0.55). Figure 2 also shows the results of the analogous analysis in current smokers. Here, in the regression model, cigarettes per day were additionally controlled for. The results are similar to those in neversmokers. However, in current smokers, sex was still significant as a predictor of BR after control with a linear and a quadratic term of FEV1 (p ⫽ 0.03).
Amount of Smoking and Methacholine Responsiveness Cigarettes per day, but not pack-years of cigarettes smoked, was a significant predictor of BR in regression models for current cigarette smokers. To test whether a linear term for number of cigarettes smoked was appropriate, smokers were ranked into quartile classes of cigarettes per day. Figure 4 shows the covariate-adjusted geometric means of methacholine slope in each quartile class, plotted against the mean number of cigarettes per day in that class.
Age and Methacholine Responsiveness Separate analyses were conducted by smoking category and by positive SPT result (any vs none) to examine the relationship between BR and age. While BR appeared to decline with age, this decline was found to differ across these strata. Figure 3 shows the mean methacholine slope by quartile of age in neversmokers and current smokers with and without positive SPT results, after controlling, via regression, for area, FEV1 (linear and quadratic terms), sex, number of positive SPT results (subgroups of atopics), and daily number of cigarettes smoked (subgroups of smokers). While atopic neversmokers have substantially higher slopes than nonatopic neversmokers, their slopes decline more rapidly with age. However, the same plot for current smokers shows only very weak rates of decline in methacholine slope with age in both subgroups. Hence, the effect of age on BR may vary both with atopic status and smoking status. Based on these plots, separate regression models were fit in smoking and atopy strata. These results are shown in Table 4. Similar regression coefficients were found across the smoking categories for subjects with negative SPT results; however, the fall in BR with age was considerably smaller in all nonatopic categories and in atopic smokers than in atopic neversmokers and former smokers. 816
Figure 3. Geometric means of methacholine slopes (expressed as percentage change in FEV1 per micromole of methacholine) for different quartile classes of age in neversmokers and current smokers with negative and positive SPT results, respectively, after group-specific adjustment for sex, baseline FEV1 (linear and quadratic term), area of study, number of positive SPT results, and daily number of cigarettes smoked. See legends for Figures 1 and 2 for expansion of abbreviations. Clinical Investigations
Figure 4. Geometric means of methacholine slopes (expressed as percentage change in FEV1 per micromole of methacholine) for different quartile classes of daily cigarettes smoked in male and female current cigarette smokers, after adjusting for age, area of study, baseline FEV1 (linear and quadratic term), and number of positive SPT results.
A monotonic and approximately linear relationship is seen for both men and women. However, the association appears to be stronger in women than in men, although the difference in gradients does not reach statistical significance. Influence of Degree of Atopy on Methacholine Responsiveness Separate regression models were fit for nonsmokers and different categories of smokers (ie, 1 to 10 cigarettes per day, 11 to 20 cigarettes per day, 21 to 30 cigarettes per day, ⬎ 30 cigarettes per day) to estimate covariate-adjusted geometric means of methacholine slope by number of positive SPT results. Regression models included the number of positive SPT results and its square along with age and FEV1 (linear and quadratic term) and dummy variables for sex and for the study areas. Figure 5 shows the results for zero to three positive SPT results (in the two upper smoking categories, no subjects with more than three positive SPT results were observed). Methacholine responsiveness showed an increasing trend with the number of positive SPT results in nonsmokers and in three of the four smoking categories. In the highest smoking category, however, the trend was not only weaker or absent but appeared to be reversed (p ⫽ 0.06). To test these results formally, the regression www.chestjournal.org
Figure 5. Estimated geometric means of methacholine slopes (expressed as percentage change in FEV1 per micromole of methacholine) by number of positive SPT results (assuming a parabolic dose-response relationship) in nonsmokers and in different categories of smokers (ie, 1 to 10, 11 to 20, 21 to 30, and ⬎ 30 cigarettes [cigs] per day, respectively), after group-specific adjustment for sex, age, baseline FEV1 (linear and quadratic term), and area of study.
model for neversmokers was estimated using the number of positive SPT results as a covariate. The number of positive SPT results was a significant predictor of BR ( ⫽ 0.330; 95% confidence interval [CI], 0.393 to 0.267; p ⬍ 0.0001), and had greater explanatory power than a simple dummy variable for any positive SPT result. In former smokers, a similar result was found ( ⫽ 0.373; 95% CI, 0.475 to 0.271; p ⬍ 0.0001). In current smokers, both the number of positive SPT results ( ⫽ 0.353; 95% CI, 0.505 to 0.201; p ⬍ 0.0001), and an interaction term with cigarettes per day ( ⫽ ⫺ 0.0082; 95% CI, 0.0013 to 0.0151; p ⬍ 0.05) were significant. Adding an interaction term between age and number of positive SPT results had little impact on these results, but the association between methacholine responsiveness and number of positive SPT results appeared to become slightly weaker with age in all three smoking categories. Discussion Our results indicate that BR varies by atopy, by symptom status and by smoking habits, but that the sex difference in neversmokers disappears when differences in lung size are taken into account. Influences of most of these factors have been deCHEST / 122 / 3 / SEPTEMBER, 2002
817
scribed before,6,11,12,18,19 but the sample sizes of most of these studies did not allow for stratified analyses. While sex appears to be a strong predictor of BR in univariate analyses, our multiple regression results suggest that the sex difference in responsiveness is essentially explained by the difference in level of FEV1, at least among neversmokers and former smokers. This confirms the findings of Kanner et al11 and Britton et al12 that the sex difference in methacholine responsiveness disappeared after adjusting the FEV1 (percent predicted) and FVC. We believe that our results are important for several reasons. First, they indicate a physiologic basis for the sex difference in BR. Second, the nonlinear relationship between responsiveness and FEV1 shows that while smaller airways are more responsive than larger ones, the reduction in responsiveness diminishes with each increment in FEV1. This is an interesting finding in itself, and may stimulate some useful thought about why this is true, and what implications this has for understanding the relationship between reactivity and level of lung function. In addition, it suggests that control for this nonlinear dependence on FEV1 will be necessary in epidemiologic studies, in order to avoid potential confounding when examining risk factors for BR that are correlated with FEV1. This is illustrated by Leynaert et al,6 who calculated, in their data set, that the percentage of female reactors at a methacholine dose of ⬍ 0.5 mg was the same as that of male reactors at 4 mg. It also underlines the importance of the adjustment for FEV1 in clinical test settings in order to avoid overdiagnosis of increased BR in women. For current smokers, sex remained a predictor also after controlling with a linear and quadratic term of FEV1. Smoking (rather than methacholine) might have a different effect on the airways in men and women. This is illustrated in Figure 4, which shows that the strong relationship observed between number of cigarettes smoked per day and percentage decrease in FEV1 per micromole of methacholine is mainly based on the effect in women. BR appeared to decline more rapidly with age in atopic nonsmokers than in nonatopic nonsmokers in this study. In nonatopic subjects, little decline with age was seen in current or former smokers. Compared to never-smoking subjects without atopy, both smoking and atopy were associated with increased BR. However, subjects with higher methacholine slopes due to smoking had less reduction in slope with age than subjects with higher methacholine slopes due to atopy. Subjects with both risk factors showed a similar rate of reduction in slope with age when expressed on the logarithmic scale (Table 4). As expected, we found methacholine responsive818
ness to be higher in asthmatic subjects than in subjects with no symptoms, or with symptoms of bronchitis. It is interesting to note that in current smokers, the geometric mean of methacholine slope would not quite produce a 20% decrease in FEV1, at a dose of 10 mol. While the average decline is larger in neversmokers, it is clear that if methacholine tests are coded positive or negative based on PD20s for usual doses, a substantial number of persons with physician-diagnosed asthma would be classified as nonreactive, particularly among current smokers. While misdiagnosis is possible for some of these subjects, that is unlikely to completely explain these results. It is likely that like asthma symptoms, methacholine responsiveness varies also across time, and that a single test will fail to pick up subjects who may be, for example, responsive only during part of the year. However, some persons may have reversible airway obstruction but would not respond to methacholine under normal conditions (eg, in the absence of a respiratory infection). Clearly, care must be taken in using one-time tests of methacholine responsiveness to classify asthmatic status.2 Smoking and atopy showed a complex pattern of interaction when considered by degree of atopy as well. In neversmokers and former smokers, a linear term for number of positive SPT results was more significant than a dummy variable for any positive SPT result. Figure 5 supports this result, showing a monotonic increase in BR with number of positive SPT results in nonsmokers and subjects smoking ⬍ 30 cigarettes per day. A similar pattern was seen in former smokers (results not shown). Hence in most subjects, there appears to be an increase in BR with increasing number of positive SPT results. The pattern is different in the highest category of smokers. Their BR in the absence of any atopy is considerably higher than in the average atopic nonsmoker. This may reflect the high degree of responsiveness heavy smokers are demonstrating even in the absence of an atopic disposition. If both smoking and atopy influence responsiveness by stimulating underlying inflammation, it is possible that the effects would be subadditive. Support for this hypothesis comes as well from the observed trend reversal in heavy smokers. It was not only heavy smokers who showed elevated BR in this sample. The degree of responsiveness increased in a dose-dependent manner with the number of cigarettes smoked as shown in Figure 4, but this increase was more marked in women than in men. Current smokers might be a selected subpopulation. Subjects who respond greatly to tobacco smoke are more likely to quit smoking. Here it is noteworthy that the former smokers with atopy have greater Clinical Investigations
responsiveness than the current smokers (Table 2), and again we have a stronger reaction in women than in men. Again, this finding is important both in terms of what it tells us about smoking, and also because it indicates that control for level of smoking within smoking categories is important in epidemiologic studies of methacholine responsiveness. It is important to recognize several limitations of this analysis. First, it is a cross-sectional analysis. This makes it difficult to determine whether some of the interaction between degree of smoking and degree of atopy in predicting methacholine responsiveness is due to selection of more responsive subjects out of the smoking pool, a selection that could be differential with respect to atopic category. The age dependence of BR in smokers may similarly be affected. If true, this suggests that the dose-dependent effects of smoking on BR reported here are underestimates. The follow-up of this population will help to clarify these issues. The results within neversmokers are less subject to this problem. They show clear evidence that the sex differences in responsiveness are due to differences in level of FEV1, that responsiveness increases in a dose-dependent manner with degree of atopy, and that BR declines more slowly with age in nonatopics than in atopics. Another limitation is the selection of the sample. Persons with serious COPD were excluded from testing. Hence, these results cannot describe the dependency of BR on age, level of lung function, or degree of atopy in those subjects. Yet they might be the ones for whom such information may be the most useful clinically. Persons who declined to participate in the survey may also differ in this regard. Replication of these findings will be important in order to exclude the possibility of these results to be a characteristic of their particular sample. Nevertheless, this large study population has allowed the investigation of a number of issues that would be difficult to examine in smaller samples; it showed that for neversmokers, the difference in responsiveness between men and women is essentially explained by different levels of FEV1, and that responsiveness decreases with each increment in lung size. Furthermore, it shows that responsiveness declines with age but more slowly in nonatopic subjects than in atopic subjects. The interaction between atopy and smoking is complex; further results of follow-up studies might clarify the role of atopy in BR of heavy smokers. Appendix The SAPALDIA Team: P. Leuenberger (Study Director), U. Ackermann-Liebrich (Program Director), P. Alean, K. Blaser, www.chestjournal.org
G. Bolognini, J. P. Bongard, O. Bra¨ ndli, P. Braun, C. Bron, M. Brutsche, C. Defila, G. Domenighetti, S. Elsasser, L. Grize, P. Guldimann, P. Hufschmid, W. Karrer, H. Keller-Wossidlo. R. Ku¨ nzli, J. C. Luthy, B. W. Martin, T. C. Medici, C. Monn, A. G. Peeters, A. P. Perruchoud, A. Radaelli, C. Schindler, J. Schwartz, G. Solari, M. H. Scho¨ ni, J. M. Tschopp, B. Wu¨ thrich, J. P. Zellweger, and E. Zemp. ACKNOWLEDGMENT: This study could not have been done without the help of the fieldworkers of the local medical teams: Aarau: C. Persoz-Borer, C. Wettstein, G. Giger, J. Lohmu¨ ller, K. Ha¨ feli, and U. Rippstein; Basel: V. Fluri, M. Herrous, G. Imboden, and L. Joos; Davos: K. D’Alberti and A. So¨ nnichsen; Geneva: I. Barbey, K. Gegerle, and N. Penay; Lugano: M. Astone, E. Haechler, E. Riesen, and B. Viscardi; Montana: Dr. Ch. Hollenstein, E. Borgeat, and I. Clivaz; Payerne: S. Mene´ treyJaques, C. Gilomen-Pages, and M. C. Coullaud; Wald: B. Salzmann, V. Kienast, H. Astone, V. Keller, and Ch. Schwalm. Throughout the study, the investigators received advice from Professor Frank Speizer of the Harvard Medical School; and the authors thank those who participated in this study, and Nora Bauer for secretarial work.
References 1 Laprise C, Boulet LP. Asymptomatic airway hyperresponsiveness: a three-year follow-up. Am J Respir Crit Care Med 1997; 156:403– 409 2 Pattemore PK, Asher MI, Harrison AC, et al. The interrelationship among bronchial hyperresponsiveness, the diagnosis of asthma, and asthma symptoms. Am Rev Respir Dis 1990; 142:549 –554 3 Burney PGJ, Luczynska C, Chinn S, et al. The European Community Respiratory Health Survey. Eur Respir J 1994; 7:954 –960 4 Chinn S, Burney PGJ, Jarvis D, et al. Variation in bronchial responsiveness in the European Community Respiratory Health Survey (ECRHS). Eur Respir J 1997; 10:2495–2501 5 Norrman E, Plaschke P, Bjo¨ rnsson E, et al. Prevalence of bronchial hyper-responsiveness in the southern, central and northern parts of Sweden. Respir Med 1998; 92:480 – 487 6 Leynaert B, Bousquet J, Henry C, et al. Is bronchial hyperresponsiveness more frequent in women than in men? A population-based study. Am J Respir Crit Care Med 1997; 156:1413–1420 7 Paoletti P, Carrozzi L, Viegi G, et al. Distribution of bronchial responsiveness in a general population: effect of sex, age, smoking, and level of pulmonary function. Am J Respir Crit Care Med 1995; 151:1770 –1777 8 Malo JL, Pineau L, Cartier A, et al. Reference values of the provocative concentrations of methacholine that cause 6% and 20% changes in forced expiratory volume in one second in a normal population. Am Rev Respir Dis 1983; 128:8 –11 9 Cerveri I, Bruschi C, Zoia MC, et al. Distribution of bronchial nonspecific reactivity in the general population. Chest 1988; 93:26 –30 10 Bakke PS, Baste V, Gulsvik A. Bronchial responsiveness in a Norwegian community. Am Rev Respir Dis 1991; 143:317– 322 11 Kanner RE, Connet JE, Altose MD, et al. Gender difference in airway hyperresponsiveness in smokers with mild COPD: the Lung Health Study. Am J Respir Crit Care Med 1994; 150:956 –961 12 Britton J, Parvod I, Richards K, et al. Factors influencing the occurrence of airway hyperreactivity in the general population: the importance of atopy and airway calibre. Eur Respir J 1994; 7:881– 887 CHEST / 122 / 3 / SEPTEMBER, 2002
819
13 Soriano JB, Tobias A, Kogevinas M, et al. Atopy and nonspecific bronchial responsiveness: a population-based assessment; Spanish Group of the European Community Respiratory Health Survey. Am J Respir Crit Care Med 1996; 154:1636 –1640 14 Nowak D, Heinrich J, Jo¨ rres R, et al. Prevalence of respiratory symptoms, bronchial hyperresponsiveness and atopy among adults: West and East Germany. Eur Respir J 1996; 9:2541–2552 15 Annema JT, Sparrow D, O’Conner GT, et al. Chronic respiratory symptoms and airway responsiveness to methacholine are associated with eosinophilia in older men: the normative aging study. Eur Respir J 1995; 8:62– 69 16 Omenaas E, Bakke P, Eide GE, et al. Serum house dust mite antibodies: predictor of increased bronchial responsiveness in adults of a community. Eur Respir J 1996; 9:919 –925 17 Sunyer J, Anto JM, Castellsague J, et al. Total serum IgE is associated with asthma independently of specific IgE level: The Spanish Group of the European Study of Asthma. Eur Respir J 1996; 9:1880 –1884 18 Chinn S, Burney P, Sunyer J, et al. Sensitization to individual allergens and bronchial responsiveness in the ECRHS: European-Community Respiratory Health Survey. Eur Respir J 1999; 14:876 – 884 19 Rijcken B, Schouten JP, Weiss ST, et al. The relationship between airway responsiveness to histamine and pulmonary function level in a random population sample. Am Rev Respir Dis 1988; 137:826 – 832 20 Sunyer J, Kogevinas M, Kromhout H, et al., Pulmonary ventilatory defects and occupational exposures in a populationbase study in Spain: Spanish Group of the European Com-
820
21 22 23
24
25
26
27
munity Respiratory Health Survey. Am J Respir Crit Care Med 1998; 157:512–517 Leuenberger P, Schindler C, Schwartz J, et al. Occupational exposure to inhalative irritants and methacholine responsiveness. Scand J Work Environ Health 2000; 26:146 –152 O’Connor G, Sparrow D, Taylor D, et al. Analysis of doseresponse curves to methacholine. Am Rev Respir Dis 1987; 136:1412–1417 Martin BW, Ackermann-Liebrich U, Leuenberger PH, et al. SAPALDIA: methods and participation in the cross-sectional part of the Swiss Study on Air pollution and Lung Diseases in Adults (SAPALDIA-Team). Soz Pra¨ ventivmed 1997; 42: 67– 84 Leuenberger P, Schwartz J, Ackermann-Liebrich U, et al. Passive smoking exposure in adults and chronic respiratory symptoms (SAPALDIA Study): Swiss Study on Air Pollution and Lung Disease in Adults, SAPALDIA-Team. Am J Respir Crit Care Med 1994; 150:1222–1228 Ackermann-Liebrich U, Leuenberger P, Schwartz J, et al. Lung function and long term exposure to air pollutants in Switzerland: Study on Air Pollution and Lung Disease in Adults (SAPALDIA) Team. Am J Respir Crit Care Med 1997; 155:122–129 Zemp E, Elsasser S, Schindler C, et al. Long-term ambient air pollution and respiratory symptoms in adults (SAPALDIA Study): The SAPALDIA Team. Am J Respir Crit Care Med 1999; 159:1257–1266 Ku¨ nzli N, Ackermann-Liebrich U, Keller R, et al. Variability of FVC and FEV1 due to technician, team, device and subject in an eight centre study: three quality control studies in SAPALDIA; Swiss Study on Air Pollution and Lung Disease in Adults. Eur Respir J 1995; 8:371–376
Clinical Investigations
Objective: Methacholine responsiveness is an end point widely used in epidemiologic studies of asthma. This study aims to quantify the relative importance of different predictors of responsiveness such as age, sex, airway caliber, smoking and atopic status, and potential interactions deserving further investigation. Methods: Methacholine challenge was performed in 7,126 participants (aged 18 to 60 years) of the Swiss Study on Air Pollution and Lung Diseases in Adults according to the European Respiratory Health Survey protocol. Responsiveness was quantified by the slope between percentage decrements in FEV1 and cumulative methacholine dose. Variation of slopes according to sex, smoking, and atopy was then examined separately by multivariate regression models that controlled for baseline FEV1. Results: We found a nonlinear relationship between methacholine slope and baseline FEV1 for both sexes, which could be well described by a quadratic function. The corresponding curves were almost identical in the region of overlap for male and female neversmokers. Methacholine responsiveness declined with age. The slope of this decline was less steep among nonatopic persons and nonsmokers compared with atopic neversmokers. Methacholine responsiveness increased with the number of cigarettes smoked per day and with the number of positive skin-prick test results (except among heavy smokers). Conclusions: Our multiple regression results show that bronchial responsiveness (BR) varies with age, FEV1, and smoking and atopic status. They suggest that there is a physiologic basis for the univariate sex difference in BR. Secondly, they show that while smaller airways are more responsive than larger ones, the reduction of responsiveness diminishes with each increase of lung size. The quantification of the relative influence of the different factors examined should help in the interpretation of BR. (CHEST 2002; 122:812– 820) Key words: atopy; methacholine responsiveness; sex; smoking Abbreviations: ATS ⫽ American Thoracic Society; BR ⫽ bronchial responsiveness; CI ⫽ confidence interval; ECRHS ⫽ European Respiratory Health Survey; PD20 ⫽ provocative dose of methacholine causing a 20% fall in FEV1; SAPALDIA ⫽ Swiss Study on Air Pollution and Lung Diseases in Adults; SPT ⫽ skin-prick test
responsiveness is often used to conM ethacholine firm asthmatic status in patients, and as a predictor of later development of respiratory diseases.1,2 It is *From the Institute of Social and Preventive Medicine (Drs. Schindler, Zemp, and Ackermann-Liebrich), University of Basel, Basel, Switzerland; Harvard School of Public Health (Dr. Schwartz), Boston, MA; Department of Internal Medicine (Dr. Perruchoud), University Clinic Basel, Basel, Switzerland; Division of Pulmonology (Drs. Zellweger and Leuenberger), University of Lausanne, Lausanne, Switzerland; and Department of Dermatology (Dr. Wu¨thrich), University Hospital Zurich, Zurich, Switzerland. †A complete list of participants is given in the Appendix. Apart from the Swiss National Science Foundation and the Federal Office for Science and Education, logistic and financial support was received from cantonal governments (Basel Geneva, Graubu¨nden, Vaud, Ticino, Zurich, and Valais) and from the cantonal offices for air hygiene measurements. Further financial support was received from the Swiss Society of 812
widely used in epidemiologic studies, as in the European Respiratory Health Survey (ECRHS),3 where a standardized tool for measurements of bronchial responsiveness (BR) to methacholine has been developed to estimate variation in prevalence of increased BR and Pulmonology, the Lega Ticinese contra le tubercolosi e le malattie polmonari, and various cantonal offices of public health. SAPALDIA is part of Swiss National Research Program 26A, supported by grant Nos. 4026-28099, 32-042532.94, and 3252720.97 from the Swiss National Science Foundation, and by the Federal Office of Education and Science. SAPALDIA Basel is part of the European Respiratory Health Survey. Manuscript received July 9, 2001; revision accepted April 29, 2002. Correspondence to: Ursula Ackermann-Liebrich, MD, Institute of Social and Preventive Medicine, Steinengraben 49, CH– 4051, Basel, Switzerland Clinical Investigations
predictors of asthma in different European countries. Apart from a general overview of variation in BR,4 participating countries and also authors from other geographic regions reported findings regarding several predictors. Besides FEV1 and symptom status, female sex,5–12 smoking,5–7,10,12 atopy,12–19 occupational exposures,13,20,21 and geographic region14 were found to be associated with increased responsiveness. The discussion on how best to report results on BR and its most important predictors has not yet come to a satisfactory conclusion. Several of the abovementioned articles used the provocative dose of methacholine causing a 20% fall in FEV1 (PD20) as a measure of BR.5,6,12,13,20 Others have argued that the slope of the decline of FEV1 with increasing dose of methacholine is a better way of measuring responsiveness because a value can be assigned to all subjects,22 and it was shown to be a more sensitive measure for center comparison in the ECRHS4 than PD20. The influence of the airway caliber on responsiveness,11,12,22 and the resulting differences between men and women, the relative importance of differences between smokers and neversmokers,5,6 or the influence of atopy still deserve investigation. We decided to use the slope of BR in the Swiss Study on Air Pollution and Lung Diseases in Adults (SAPALDIA)23 to analyze whether symptom status, sex, smoking, and atopy independently influence BR. Since the prevalence of atopy, smoking, and asthma vary in men and women in different age groups and, moreover, effects may vary across subgroups or as a function of the other risk factors, we decided to stratify analyses by sex and different levels of risk factors. This was possible due to the large sample of participants who have undergone methacholine testing in the SAPALDIA study (n ⫽ 7,126), and the relatively broad age range (18 to 62 years) of the population studied. Materials and Methods Study Population SAPALDIA is a multicenter study designed to examine potential associations between air pollution and respiratory outcomes.23–27 The eight study areas (Aarau, Basel, Davos, Geneva, Lugano, Montana, Payerne, and Wald) were chosen to represent a range of urbanization, altitude, air pollution, and meteorology. A random sample of adults aged 18 to 60 years was drawn from the registry of inhabitants of each town. Successfully recruited subjects (59%) underwent a standardized computerized interview with an expanded version of the ECRHS questionnaire.3 A total of 9,651 subjects were recruited for the study. Allergy tests included skin-prick tests (SPTs), in addition to blood samples for total IgE. SPTs were performed on the volar forearm using Phazet (Pharmacia; Uppsala, Sweden) allergen-coated steel lancets, corresponding to a prick test extract of 100,000 biological units per milliliter.19 The inhalant allergens tested for were www.chestjournal.org
chosen according to the known prevalence of specific sensitization in Switzerland and in accordance with the ECRHS allergy testing protocol3: cat fur; mold (Cladosporium herbarum); Timothy grass pollen; pellitory-of-the-wall pollen (Parietaria); house dust mite (Dermatophagoides pteronyssinus); mold (Alternaria tenuis); birch pollen; dog epithelia. The test also included histamine (positive control) and a noncoated lancet (negative control). Spirometry measurements were done using a pulmonary function system (2200SP; SensorMedics; Yorba Linda, CA) using a mass flow anemometer, displaying and recording flow-volume curves for each trial. The testing was performed by trained fieldworkers according to the recommendations on standardization of spirometry of the American Thoracic Society (ATS).27 Three ATS acceptability criteria (volume of extrapolation ⬍ 5% FVC or ⬍ 100 mL, expiratory time of at least 6 s, and end of test [no volume change for the last 2 s of the maneuver]) were displayed and recorded with each maneuver. The two ATS reproducibility criteria (largest two FVC and FEV1 values within 5% of one another) were recorded with the set of best values, and guided the technician’s decisions for additional maneuvers. At least three and up to eight maneuvers were required, until the five criteria described were met. Subjects were excluded from methacholine testing if they had an FEV1/FVC ratio ⬍ 80% of predicted, had an FEV1 ⬍ 70% of predicted according to European standards, or had difficulty performing the spirometry maneuver. This reduced the sample size of subjects to 7,126.1 Methacholine chloride (Provocholine; Roche; Nutley, NJ) was prepared in 25.0, 6.25, 1.56, and 0.39 mg/mL solutions in a phosphate buffer without phenol. Methacholine was administered through an aerosol dosimeter (Mefar MB3; Boveso, Italy). The schedule in nonreactive subjects was then four inhalations at 0.39 mg/mL and three inhalations in each of the other concentrations. If FEV1 decreased ⬎ 10% from the baseline level, smaller increments were introduced. Testing continued until the final dose, or until a 20% reduction in FEV1 was reached. Methacholine responsiveness was quantified by calculating a dose-response slope for each subject similar to the one suggested by O’Connor et al.22 In order to avoid negative slopes, the absolute decline in FEV1 was defined as the difference between the maximum FEV1 over all levels tested and FEV1 at the last level measured, irrespective of the dose level at which the maximum occurred. The percentage decline was computed using the maximum of FEV1 as the reference value. The slope was then defined as the ratio between the percentage decline in FEV1 and the total cumulative dose of methacholine (in micromoles) administered. As the distribution of these slopes was quite skewed, a logarithmic transformation was used to obtain a more symmetrical distribution better suited for statistical analysis. In order not to lose slope values of 0, a small constant (ie, 0.01) was added before taking the logarithm. We first examined variation between slopes according to sex, smoking, and atopy. In a second step, multivariate regression models controlling for baseline FEV1 (ie, by a linear and a quadratic term) were recomputed. To examine potential interactions between different factors, analyses were stratified or interaction terms were used. Definition of Symptom Categories Asthma was defined as a positive response to both parts of the question, “Have you ever had asthma, and was this confirmed by a doctor?” Bronchitis symptoms were defined as a positive response to the question, “Do you usually cough during the day or night?” or the question, “Do you usually bring up any phlegm from your chest during the day or night?” If the subjects reported CHEST / 122 / 3 / SEPTEMBER, 2002
813
Table 1—Characteristics of the Sampled Subjects With and Without Methacholine Slopes* Without Methacholine Slopes†
With Methacholine Slopes
Characteristics
Men (n ⫽ 1,138)
Women (n ⫽ 1,387)
Men (n ⫽ 3,605)
Women (n ⫽ 3,521)
Age, yr Asthma Wheeze apart from cold Bronchitic symptoms Hay fever Current smoker Former smoker Neversmoker Positive SPT result FEV1, L
44.6 ⫾ 11.6 9.6 12.5 14.7 18.2 41.9 26.6 31.5 26.1 3.63
43.3 ⫾ 11.5 9.3 9.2 10.7 17.8 27.3 18.5 54.2 19.7 2.86
39.6 ⫾ 11.6 6.0 7.4 10.1 16.6 37.4 25.5 37.1 25.3 4.18
40.6 ⫾ 11.4 5.5 5.8 7.8 16.2 29.3 19.9 50.7 21.6 3.09
*Data are presented as mean ⫾ SD or % unless otherwise indicated. †Excluded from methacholine test.
a physician diagnosis of asthma (regardless of whether the subject had bronchitis symptoms), they were classified as asthmatic.
Results Table 1 shows the characteristics of the examined persons, and of those without methacholine slopes. As expected, the prevalence of asthma and wheezing was higher in the excluded group and FEV1 was lower. There was little difference between the samples in the prevalence of atopy by SPT or in self-report of hay fever. The prevalence of neversmoking was higher among tested men, but lower among tested women. The prevalence of increased BR in terms of the PD20 was 11.7% in men and 21.8% in women. Table 2 shows the geometric mean of methacholine slope (percentage decrease in FEV1 per micromole of methacholine) by smoking category, sex, and level of atopy (defined as none vs one or two and more positive SPT results). Women had higher slopes in all categories. Moreover, methacholine responsiveness showed a trend to increase with increasing number of positive SPT results. However, smoking was associated with an increase in responsiveness only among subjects with negative SPT results. Figure 1 shows the medians and interquartile ranges of slopes for eight different subgroups (ie, male and female neversmokers and current smokers with and without positive SPT results, respectively). The medians exhibit a similar pattern as the geometric means in Table 2. Again, a clear association of slopes with smoking can only be seen among subjects with negative SPT results. In addition, Figure 1 shows that the shape of the distributions changes with atopic status (ie, slopes have a more skewed distribution in the subgroups of subjects with positive SPT results). 814
Table 3 shows the geometric mean slope by smoking category, sex, and symptom category. If the subject reported neither a physician’s diagnosis of asthma nor bronchitis symptoms, the symptom status was classified as “none.” Again, women had higher average slopes in all categories considered. Consistently, the geometric means of slope were lowest among nonsymptomatic subjects, intermediate among subjects with bronchitis symptoms, and highest among subjects with asthma symptoms. As before, associations with smoking were less consistent, being positive only among subjects without asthma symptoms. None of these results are adjusted for continuous risk factors, such as level of lung function or age. The next sets of results are from multivariate analyses, and are reported by major categories of risk factors. The Influence of the Level of Lung Function and Sex on Methacholine Responsiveness Women are known to have greater BR than men.5,6 BR also varies with level of pulmonary Table 2—Geometric Means of Methacholine Slope in Percentage Decrease in FEV1 per Micromole of Methacholine by Smoking, Sex, and Atopy Categories* Positive SPT Results, No. Smoking Status
None
One
Men (n ⫽ 3,485) n ⫽ 2,621 n ⫽ 483 Neversmoker (n ⫽ 1,285) 0.65 0.94 Former smoker (n ⫽ 884) 0.63 1.01 Current smoker (n ⫽ 1,316) 0.91 1.01 Women n ⫽ 2,683 n ⫽ 436 Neversmoker (n ⫽ 1,730) 1.11 1.72 Former smoker (n ⫽ 682) 1.29 1.47 Current smoker (n ⫽ 994) 1.35 1.99
Two or More n ⫽ 381 1.55 1.83 1.19 n ⫽ 287 2.83 3.22 2.14
*To avoid values of zero, a constant of 0.01 was added to each slope before computing the geometric means. Clinical Investigations
Table 3—Geometric Means of Methacholine Slopes in Percentage Decrease in FEV1 per Micromole of Methacholine by Smoking, Sex, and Symptom Status* Symptom Status† Smoking Status
None
Men (n ⫽ 3,594) Neversmoker (n ⫽ 1,334) Former smoker (n ⫽ 917) Current smoker (n ⫽ 1,343) Women (n ⫽ 3513) Neversmoker (n ⫽ 1,783) Former smoker (n ⫽ 700) Current smoker (n ⫽ 1,030)
n ⫽ 3,014 0.71 0.64 0.87 n ⫽ 3044 1.18 1.23 1.38
Bronchitis
Asthma
n ⫽ 363 0.96 0.85 1.06 n ⫽ 275 1.37 1.66 1.68
n ⫽ 217 2.34 4.62 2.18 n ⫽ 194 5.93 5.74 3.53
*To avoid values of zero, a constant of 0.01 was added to each slope before computing the geometric means. †Symptom categories are “none” if the subject had neither physician-diagnosed asthma or bronchitis symptoms, “bronchitis” if the subject had bronchitis symptoms without a diagnosis of asthma, and “asthma” if the subject had physician-diagnosed asthma with or without bronchitis symptoms.
function.11,12,22 This section examines the extent to which the sex difference in responsiveness is explained by the level of pulmonary function. The functional form of the dependence of methacholine slope on level of lung function is also not clear. To test this, separate regression analyses were conducted for each sex, controlling for area of study, age, and the number of positive SPT results. The analyses began with neversmokers. FEV1 was divided into quintile classes within each sex, and a
dummy variable for each quintile class was entered into the regression models for each sex. Figure 2 shows the geometric mean of methacholine slope (adjusted for covariates) in each quintile class, for male and female neversmokers, plotted against the mean FEV1 in that quintile class. Two patterns emerged. First, within each sex, the relationship with FEV1 appears nonlinear. Second, the two sets of points appear to lie almost on the same curve. When both sexes were combined, and a linear
Figure 1. Medians and interquartile ranges of methacholine slopes (expressed as percentage change in FEV1 per micromole of methacholine) in male neversmokers (mNS) and female neversmokers (fNS), and male current smokers (mCS) and female current smokers (fCS) with negative (⫺) and positive (⫹) SPT results, respectively.
Figure 2. Geometric means of methacholine slopes (expressed as percentage change in FEV1 per micromole of methacholine) for different quintile classes of baseline FEV1 in male and female neversmokers (NS) and current smokers (CS), respectively, after group-specific adjustment for area of study, age, number of positive SPT results, and daily number of cigarettes smoked.
www.chestjournal.org
CHEST / 122 / 3 / SEPTEMBER, 2002
815
Table 4 —Decline in Methacholine Responsiveness With Age by Smoking and Atopic Status* SPT Result Negative
Positive
Smoking Status
Coefficient
95% CI
Coefficient
Neversmokers Former smokers Current smokers
⫺0.0095† ⫺0.0072 ⫺0.0059
⫺0.0146 to ⫺0.0044 ⫺0.0161 to 0.0017 ⫺0.0122 to 0.0004
⫺0.0200† ⫺0.026‡ ⫺0.0032
95% CI ⫺0.0319 to ⫺0.0081 ⫺0.0466 to ⫺0.0054 ⫺0.0155 to 0.0091
*Natural logarithm of methacholine responsiveness, to avoid values of zero, a constant of 0.01 was added to each slope before taking the logarithm. This Table shows the coefficient of age (1 year) in each strata, after adjustment for covariates (area, FEV1, [linear and quadratic term], sex, No. of cigarettes per day [smokers]), and No. of positive SPT results (subjects with at least one positive SPT result). †p ⬍ 0.001 for regression coefficients. ‡p ⬍ 0.05 for regression coefficients.
and quadratic term for FEV1 used, sex was no longer a significant predictor of methacholine responsiveness (p ⫽ 0.55). Figure 2 also shows the results of the analogous analysis in current smokers. Here, in the regression model, cigarettes per day were additionally controlled for. The results are similar to those in neversmokers. However, in current smokers, sex was still significant as a predictor of BR after control with a linear and a quadratic term of FEV1 (p ⫽ 0.03).
Amount of Smoking and Methacholine Responsiveness Cigarettes per day, but not pack-years of cigarettes smoked, was a significant predictor of BR in regression models for current cigarette smokers. To test whether a linear term for number of cigarettes smoked was appropriate, smokers were ranked into quartile classes of cigarettes per day. Figure 4 shows the covariate-adjusted geometric means of methacholine slope in each quartile class, plotted against the mean number of cigarettes per day in that class.
Age and Methacholine Responsiveness Separate analyses were conducted by smoking category and by positive SPT result (any vs none) to examine the relationship between BR and age. While BR appeared to decline with age, this decline was found to differ across these strata. Figure 3 shows the mean methacholine slope by quartile of age in neversmokers and current smokers with and without positive SPT results, after controlling, via regression, for area, FEV1 (linear and quadratic terms), sex, number of positive SPT results (subgroups of atopics), and daily number of cigarettes smoked (subgroups of smokers). While atopic neversmokers have substantially higher slopes than nonatopic neversmokers, their slopes decline more rapidly with age. However, the same plot for current smokers shows only very weak rates of decline in methacholine slope with age in both subgroups. Hence, the effect of age on BR may vary both with atopic status and smoking status. Based on these plots, separate regression models were fit in smoking and atopy strata. These results are shown in Table 4. Similar regression coefficients were found across the smoking categories for subjects with negative SPT results; however, the fall in BR with age was considerably smaller in all nonatopic categories and in atopic smokers than in atopic neversmokers and former smokers. 816
Figure 3. Geometric means of methacholine slopes (expressed as percentage change in FEV1 per micromole of methacholine) for different quartile classes of age in neversmokers and current smokers with negative and positive SPT results, respectively, after group-specific adjustment for sex, baseline FEV1 (linear and quadratic term), area of study, number of positive SPT results, and daily number of cigarettes smoked. See legends for Figures 1 and 2 for expansion of abbreviations. Clinical Investigations
Figure 4. Geometric means of methacholine slopes (expressed as percentage change in FEV1 per micromole of methacholine) for different quartile classes of daily cigarettes smoked in male and female current cigarette smokers, after adjusting for age, area of study, baseline FEV1 (linear and quadratic term), and number of positive SPT results.
A monotonic and approximately linear relationship is seen for both men and women. However, the association appears to be stronger in women than in men, although the difference in gradients does not reach statistical significance. Influence of Degree of Atopy on Methacholine Responsiveness Separate regression models were fit for nonsmokers and different categories of smokers (ie, 1 to 10 cigarettes per day, 11 to 20 cigarettes per day, 21 to 30 cigarettes per day, ⬎ 30 cigarettes per day) to estimate covariate-adjusted geometric means of methacholine slope by number of positive SPT results. Regression models included the number of positive SPT results and its square along with age and FEV1 (linear and quadratic term) and dummy variables for sex and for the study areas. Figure 5 shows the results for zero to three positive SPT results (in the two upper smoking categories, no subjects with more than three positive SPT results were observed). Methacholine responsiveness showed an increasing trend with the number of positive SPT results in nonsmokers and in three of the four smoking categories. In the highest smoking category, however, the trend was not only weaker or absent but appeared to be reversed (p ⫽ 0.06). To test these results formally, the regression www.chestjournal.org
Figure 5. Estimated geometric means of methacholine slopes (expressed as percentage change in FEV1 per micromole of methacholine) by number of positive SPT results (assuming a parabolic dose-response relationship) in nonsmokers and in different categories of smokers (ie, 1 to 10, 11 to 20, 21 to 30, and ⬎ 30 cigarettes [cigs] per day, respectively), after group-specific adjustment for sex, age, baseline FEV1 (linear and quadratic term), and area of study.
model for neversmokers was estimated using the number of positive SPT results as a covariate. The number of positive SPT results was a significant predictor of BR ( ⫽ 0.330; 95% confidence interval [CI], 0.393 to 0.267; p ⬍ 0.0001), and had greater explanatory power than a simple dummy variable for any positive SPT result. In former smokers, a similar result was found ( ⫽ 0.373; 95% CI, 0.475 to 0.271; p ⬍ 0.0001). In current smokers, both the number of positive SPT results ( ⫽ 0.353; 95% CI, 0.505 to 0.201; p ⬍ 0.0001), and an interaction term with cigarettes per day ( ⫽ ⫺ 0.0082; 95% CI, 0.0013 to 0.0151; p ⬍ 0.05) were significant. Adding an interaction term between age and number of positive SPT results had little impact on these results, but the association between methacholine responsiveness and number of positive SPT results appeared to become slightly weaker with age in all three smoking categories. Discussion Our results indicate that BR varies by atopy, by symptom status and by smoking habits, but that the sex difference in neversmokers disappears when differences in lung size are taken into account. Influences of most of these factors have been deCHEST / 122 / 3 / SEPTEMBER, 2002
817
scribed before,6,11,12,18,19 but the sample sizes of most of these studies did not allow for stratified analyses. While sex appears to be a strong predictor of BR in univariate analyses, our multiple regression results suggest that the sex difference in responsiveness is essentially explained by the difference in level of FEV1, at least among neversmokers and former smokers. This confirms the findings of Kanner et al11 and Britton et al12 that the sex difference in methacholine responsiveness disappeared after adjusting the FEV1 (percent predicted) and FVC. We believe that our results are important for several reasons. First, they indicate a physiologic basis for the sex difference in BR. Second, the nonlinear relationship between responsiveness and FEV1 shows that while smaller airways are more responsive than larger ones, the reduction in responsiveness diminishes with each increment in FEV1. This is an interesting finding in itself, and may stimulate some useful thought about why this is true, and what implications this has for understanding the relationship between reactivity and level of lung function. In addition, it suggests that control for this nonlinear dependence on FEV1 will be necessary in epidemiologic studies, in order to avoid potential confounding when examining risk factors for BR that are correlated with FEV1. This is illustrated by Leynaert et al,6 who calculated, in their data set, that the percentage of female reactors at a methacholine dose of ⬍ 0.5 mg was the same as that of male reactors at 4 mg. It also underlines the importance of the adjustment for FEV1 in clinical test settings in order to avoid overdiagnosis of increased BR in women. For current smokers, sex remained a predictor also after controlling with a linear and quadratic term of FEV1. Smoking (rather than methacholine) might have a different effect on the airways in men and women. This is illustrated in Figure 4, which shows that the strong relationship observed between number of cigarettes smoked per day and percentage decrease in FEV1 per micromole of methacholine is mainly based on the effect in women. BR appeared to decline more rapidly with age in atopic nonsmokers than in nonatopic nonsmokers in this study. In nonatopic subjects, little decline with age was seen in current or former smokers. Compared to never-smoking subjects without atopy, both smoking and atopy were associated with increased BR. However, subjects with higher methacholine slopes due to smoking had less reduction in slope with age than subjects with higher methacholine slopes due to atopy. Subjects with both risk factors showed a similar rate of reduction in slope with age when expressed on the logarithmic scale (Table 4). As expected, we found methacholine responsive818
ness to be higher in asthmatic subjects than in subjects with no symptoms, or with symptoms of bronchitis. It is interesting to note that in current smokers, the geometric mean of methacholine slope would not quite produce a 20% decrease in FEV1, at a dose of 10 mol. While the average decline is larger in neversmokers, it is clear that if methacholine tests are coded positive or negative based on PD20s for usual doses, a substantial number of persons with physician-diagnosed asthma would be classified as nonreactive, particularly among current smokers. While misdiagnosis is possible for some of these subjects, that is unlikely to completely explain these results. It is likely that like asthma symptoms, methacholine responsiveness varies also across time, and that a single test will fail to pick up subjects who may be, for example, responsive only during part of the year. However, some persons may have reversible airway obstruction but would not respond to methacholine under normal conditions (eg, in the absence of a respiratory infection). Clearly, care must be taken in using one-time tests of methacholine responsiveness to classify asthmatic status.2 Smoking and atopy showed a complex pattern of interaction when considered by degree of atopy as well. In neversmokers and former smokers, a linear term for number of positive SPT results was more significant than a dummy variable for any positive SPT result. Figure 5 supports this result, showing a monotonic increase in BR with number of positive SPT results in nonsmokers and subjects smoking ⬍ 30 cigarettes per day. A similar pattern was seen in former smokers (results not shown). Hence in most subjects, there appears to be an increase in BR with increasing number of positive SPT results. The pattern is different in the highest category of smokers. Their BR in the absence of any atopy is considerably higher than in the average atopic nonsmoker. This may reflect the high degree of responsiveness heavy smokers are demonstrating even in the absence of an atopic disposition. If both smoking and atopy influence responsiveness by stimulating underlying inflammation, it is possible that the effects would be subadditive. Support for this hypothesis comes as well from the observed trend reversal in heavy smokers. It was not only heavy smokers who showed elevated BR in this sample. The degree of responsiveness increased in a dose-dependent manner with the number of cigarettes smoked as shown in Figure 4, but this increase was more marked in women than in men. Current smokers might be a selected subpopulation. Subjects who respond greatly to tobacco smoke are more likely to quit smoking. Here it is noteworthy that the former smokers with atopy have greater Clinical Investigations
responsiveness than the current smokers (Table 2), and again we have a stronger reaction in women than in men. Again, this finding is important both in terms of what it tells us about smoking, and also because it indicates that control for level of smoking within smoking categories is important in epidemiologic studies of methacholine responsiveness. It is important to recognize several limitations of this analysis. First, it is a cross-sectional analysis. This makes it difficult to determine whether some of the interaction between degree of smoking and degree of atopy in predicting methacholine responsiveness is due to selection of more responsive subjects out of the smoking pool, a selection that could be differential with respect to atopic category. The age dependence of BR in smokers may similarly be affected. If true, this suggests that the dose-dependent effects of smoking on BR reported here are underestimates. The follow-up of this population will help to clarify these issues. The results within neversmokers are less subject to this problem. They show clear evidence that the sex differences in responsiveness are due to differences in level of FEV1, that responsiveness increases in a dose-dependent manner with degree of atopy, and that BR declines more slowly with age in nonatopics than in atopics. Another limitation is the selection of the sample. Persons with serious COPD were excluded from testing. Hence, these results cannot describe the dependency of BR on age, level of lung function, or degree of atopy in those subjects. Yet they might be the ones for whom such information may be the most useful clinically. Persons who declined to participate in the survey may also differ in this regard. Replication of these findings will be important in order to exclude the possibility of these results to be a characteristic of their particular sample. Nevertheless, this large study population has allowed the investigation of a number of issues that would be difficult to examine in smaller samples; it showed that for neversmokers, the difference in responsiveness between men and women is essentially explained by different levels of FEV1, and that responsiveness decreases with each increment in lung size. Furthermore, it shows that responsiveness declines with age but more slowly in nonatopic subjects than in atopic subjects. The interaction between atopy and smoking is complex; further results of follow-up studies might clarify the role of atopy in BR of heavy smokers. Appendix The SAPALDIA Team: P. Leuenberger (Study Director), U. Ackermann-Liebrich (Program Director), P. Alean, K. Blaser, www.chestjournal.org
G. Bolognini, J. P. Bongard, O. Bra¨ ndli, P. Braun, C. Bron, M. Brutsche, C. Defila, G. Domenighetti, S. Elsasser, L. Grize, P. Guldimann, P. Hufschmid, W. Karrer, H. Keller-Wossidlo. R. Ku¨ nzli, J. C. Luthy, B. W. Martin, T. C. Medici, C. Monn, A. G. Peeters, A. P. Perruchoud, A. Radaelli, C. Schindler, J. Schwartz, G. Solari, M. H. Scho¨ ni, J. M. Tschopp, B. Wu¨ thrich, J. P. Zellweger, and E. Zemp. ACKNOWLEDGMENT: This study could not have been done without the help of the fieldworkers of the local medical teams: Aarau: C. Persoz-Borer, C. Wettstein, G. Giger, J. Lohmu¨ ller, K. Ha¨ feli, and U. Rippstein; Basel: V. Fluri, M. Herrous, G. Imboden, and L. Joos; Davos: K. D’Alberti and A. So¨ nnichsen; Geneva: I. Barbey, K. Gegerle, and N. Penay; Lugano: M. Astone, E. Haechler, E. Riesen, and B. Viscardi; Montana: Dr. Ch. Hollenstein, E. Borgeat, and I. Clivaz; Payerne: S. Mene´ treyJaques, C. Gilomen-Pages, and M. C. Coullaud; Wald: B. Salzmann, V. Kienast, H. Astone, V. Keller, and Ch. Schwalm. Throughout the study, the investigators received advice from Professor Frank Speizer of the Harvard Medical School; and the authors thank those who participated in this study, and Nora Bauer for secretarial work.
References 1 Laprise C, Boulet LP. Asymptomatic airway hyperresponsiveness: a three-year follow-up. Am J Respir Crit Care Med 1997; 156:403– 409 2 Pattemore PK, Asher MI, Harrison AC, et al. The interrelationship among bronchial hyperresponsiveness, the diagnosis of asthma, and asthma symptoms. Am Rev Respir Dis 1990; 142:549 –554 3 Burney PGJ, Luczynska C, Chinn S, et al. The European Community Respiratory Health Survey. Eur Respir J 1994; 7:954 –960 4 Chinn S, Burney PGJ, Jarvis D, et al. Variation in bronchial responsiveness in the European Community Respiratory Health Survey (ECRHS). Eur Respir J 1997; 10:2495–2501 5 Norrman E, Plaschke P, Bjo¨ rnsson E, et al. Prevalence of bronchial hyper-responsiveness in the southern, central and northern parts of Sweden. Respir Med 1998; 92:480 – 487 6 Leynaert B, Bousquet J, Henry C, et al. Is bronchial hyperresponsiveness more frequent in women than in men? A population-based study. Am J Respir Crit Care Med 1997; 156:1413–1420 7 Paoletti P, Carrozzi L, Viegi G, et al. Distribution of bronchial responsiveness in a general population: effect of sex, age, smoking, and level of pulmonary function. Am J Respir Crit Care Med 1995; 151:1770 –1777 8 Malo JL, Pineau L, Cartier A, et al. Reference values of the provocative concentrations of methacholine that cause 6% and 20% changes in forced expiratory volume in one second in a normal population. Am Rev Respir Dis 1983; 128:8 –11 9 Cerveri I, Bruschi C, Zoia MC, et al. Distribution of bronchial nonspecific reactivity in the general population. Chest 1988; 93:26 –30 10 Bakke PS, Baste V, Gulsvik A. Bronchial responsiveness in a Norwegian community. Am Rev Respir Dis 1991; 143:317– 322 11 Kanner RE, Connet JE, Altose MD, et al. Gender difference in airway hyperresponsiveness in smokers with mild COPD: the Lung Health Study. Am J Respir Crit Care Med 1994; 150:956 –961 12 Britton J, Parvod I, Richards K, et al. Factors influencing the occurrence of airway hyperreactivity in the general population: the importance of atopy and airway calibre. Eur Respir J 1994; 7:881– 887 CHEST / 122 / 3 / SEPTEMBER, 2002
819
13 Soriano JB, Tobias A, Kogevinas M, et al. Atopy and nonspecific bronchial responsiveness: a population-based assessment; Spanish Group of the European Community Respiratory Health Survey. Am J Respir Crit Care Med 1996; 154:1636 –1640 14 Nowak D, Heinrich J, Jo¨ rres R, et al. Prevalence of respiratory symptoms, bronchial hyperresponsiveness and atopy among adults: West and East Germany. Eur Respir J 1996; 9:2541–2552 15 Annema JT, Sparrow D, O’Conner GT, et al. Chronic respiratory symptoms and airway responsiveness to methacholine are associated with eosinophilia in older men: the normative aging study. Eur Respir J 1995; 8:62– 69 16 Omenaas E, Bakke P, Eide GE, et al. Serum house dust mite antibodies: predictor of increased bronchial responsiveness in adults of a community. Eur Respir J 1996; 9:919 –925 17 Sunyer J, Anto JM, Castellsague J, et al. Total serum IgE is associated with asthma independently of specific IgE level: The Spanish Group of the European Study of Asthma. Eur Respir J 1996; 9:1880 –1884 18 Chinn S, Burney P, Sunyer J, et al. Sensitization to individual allergens and bronchial responsiveness in the ECRHS: European-Community Respiratory Health Survey. Eur Respir J 1999; 14:876 – 884 19 Rijcken B, Schouten JP, Weiss ST, et al. The relationship between airway responsiveness to histamine and pulmonary function level in a random population sample. Am Rev Respir Dis 1988; 137:826 – 832 20 Sunyer J, Kogevinas M, Kromhout H, et al., Pulmonary ventilatory defects and occupational exposures in a populationbase study in Spain: Spanish Group of the European Com-
820
21 22 23
24
25
26
27
munity Respiratory Health Survey. Am J Respir Crit Care Med 1998; 157:512–517 Leuenberger P, Schindler C, Schwartz J, et al. Occupational exposure to inhalative irritants and methacholine responsiveness. Scand J Work Environ Health 2000; 26:146 –152 O’Connor G, Sparrow D, Taylor D, et al. Analysis of doseresponse curves to methacholine. Am Rev Respir Dis 1987; 136:1412–1417 Martin BW, Ackermann-Liebrich U, Leuenberger PH, et al. SAPALDIA: methods and participation in the cross-sectional part of the Swiss Study on Air pollution and Lung Diseases in Adults (SAPALDIA-Team). Soz Pra¨ ventivmed 1997; 42: 67– 84 Leuenberger P, Schwartz J, Ackermann-Liebrich U, et al. Passive smoking exposure in adults and chronic respiratory symptoms (SAPALDIA Study): Swiss Study on Air Pollution and Lung Disease in Adults, SAPALDIA-Team. Am J Respir Crit Care Med 1994; 150:1222–1228 Ackermann-Liebrich U, Leuenberger P, Schwartz J, et al. Lung function and long term exposure to air pollutants in Switzerland: Study on Air Pollution and Lung Disease in Adults (SAPALDIA) Team. Am J Respir Crit Care Med 1997; 155:122–129 Zemp E, Elsasser S, Schindler C, et al. Long-term ambient air pollution and respiratory symptoms in adults (SAPALDIA Study): The SAPALDIA Team. Am J Respir Crit Care Med 1999; 159:1257–1266 Ku¨ nzli N, Ackermann-Liebrich U, Keller R, et al. Variability of FVC and FEV1 due to technician, team, device and subject in an eight centre study: three quality control studies in SAPALDIA; Swiss Study on Air Pollution and Lung Disease in Adults. Eur Respir J 1995; 8:371–376
Clinical Investigations