EXPOSURE TO POLLUTANTS IN AMBIENT AIR has been associated with short-term impairment of respiratory health. Twelve epidemiological ecological studies (1-12) have dealt with hospital admissions for chronic obstructive pulmonary disease (COPD). Significant increases in such admissions were related to levels of particulate matter (PM), (1,2,7-9,11,12) ozone ([O.sub.3]), (1-4,6,7) sulfur dioxide (S[O.sub.2]), (1,4,11,12) nitrogen dioxide (N[O.sub.2]), (1) and carbon monoxide (CO) (12) in inhaled air.
Only 4 panel studies have been conducted on patients suffering from COPD, and the results are contradictory. For the patients studied, PM with aerodynamic diameters less than 10 [micro]m (P[M.sub.10]) had an effect on nocturnal cough (13) and on forced expiratory volume in 1 sec (FE[V.sub.1.0]). (14) Two teams (15,16) described no effect of N[O.sub.2] on patient state, whereas another team (13) revealed an association between N[O.sub.2] exposure and increased use of bronchodilators.
We studied a panel of adults who suffered from COPD to test their sensitivity to air pollution in Paris. In this study, we followed this panel of individuals over a 14-mo period to determine the short-term effects of both winter air pollutants (i.e., P[M.sub.10], S[O.sub.2], and N[O.sub.2]) and summer photochemical air pollutants (i.e., [O.sub.3] and N[O.sub.2]). This study was based on medical data collected by pulmonary physicians at the time of clinical examinations.
Study design. The panel study was carried out from October 1995 to November 1996 on Parisian patients with COPD. The follow-up of this panel was based on regular physician appointments and patient-initiated consultations. The panel comprised patients selected by 2 pulmonary physicians who completed a standardized inclusion questionnaire for each participant, and who subsequently used a follow-up form to report episodes of exacerbation each time a patient came in for consultation. A new event could be diagnosed only if at least 1 wk had passed from the time of the most-recent consultation. Patients were admitted into the study until December 1995. The study was approved by the National Ethics Committee (National Commission of Data Processing and Freedoms, opinion no. 413385).
Study population. Participants were recruited from patients who regularly attended the Center for Treatment of Respiratory Diseases (CTAR) in Paris. Upon entry, patients were asked to consult their pulmonary physician at the CTAR when they experienced exacerbations of their symptoms. Patients were also to report if they were hospitalized. Eligibility criteria for the study included (a) moderate-to-severe physician-diagnosed COPD, determined on the basis of a history of airway obstruction (defined as an FE[V.sub.1.0]/VC [vital capacity] ratio lower than 80% and a total lung capacity [TLC] that exceeded 80% of the predicted value); (b) oxygen pressure (Pa[O.sub.2]), at rest or after effort, lower than 65 mm Hg, or right cardiac insufficiency with systolic pulmonary artery pressure higher than 45 mm Hg; and (c) dwelling in Paris or its suburbs (6 Departments proximal to Paris). Thirty-nine adults gave written consent that evidenced their agreement to participate in the study.
Demographic and health data. The inclusion questionnaire contained demographic (age, gender, residence, occupation) and medical management data. Baseline pulmonary function parameters (FE[V.sub.1.0], VC, and TLC) were noted and compared with predicted values in accordance with age, gender, and height. Pa[O.sub.2], carbon dioxide pressure (PaC[O.sub.2]), and systolic pulmonary artery pressure were also determined. Regularly scheduled medications, such as inhaled or systemic [beta]2 agonists and steroids, were noted, as were smoking habits. We used the following indicators to determine disease severity: (a) Sadoul's dyspnea measuring scale (with 5 grades, from 0 [no dyspnea] to 5 [breathing discomfort when removing clothing]); (b) ventilation; and (c) oxygen treatment.
The physician noted whether the patient was clinically enduring an exacerbation of his or her health state, and/or was functionally worse, at the time of consultation. The physician defined "exacerbation of the patient's health state" by considering both clinical and functional variations from baseline status--each patient being his or her own "control." Hence, every acute exacerbation of the disease was confirmed by 1 or more of the following: (a) decrease in "vesicular" breath sound, (b) bronchial obstruction, (c) tachycardia or arrhythmia, and (d) cyanosis. The 2 CTAR physicians agreed to use the same definitions in their medical practices. The date of the 1st day of an attack and data on potential aggravating factors (e.g., infection, change in smoking habits, stress, unusual indoor pollution [such as use of a solvent or insecticide]) were reported.
Air pollution and weather data measurements. Information regarding ambient air pollution was obtained from daily measurements by urban background stations within the existing monitoring network (AIRPARIF), all of which were located in the Greater Paris area. The stations operated continuously and were calibrated every 2 wk. Air pollution data recorded included values for S[O.sub.2], P[M.sub.10], N[O.sub.2], and [O.sub.3]. Daily values were given by 28 stations for S[O.sub.2], 7 stations for P[M.sub.10], 15 stations for N[O.sub.2], and 6 stations for [O.sub.3]. S[O.sub.2] was measured by ultraviolet (UV) fluorescence, [O.sub.3] by UV photometry, P[M.sub.10] by [beta]-radiometry, and N[O.sub.2] by chemiluminescence. We obtained ambient concentrations of air pollutants from the station closest to each participant's home, and 24-hr average levels were calculated from midnight to midnight every day for S[O.sub.2], P[M.sub.10], and N[O.sub.2]. Eight-hour average levels (10 A.M. to 6 P.M.) were used for [O.sub.3]. The 1-hr maximum value was also recorded for N[O.sub.2] and [O.sub.3]. Daily average temperature and relative humidity were measured by the Paris weather station, Meteo France.
Statistical analysis. The association between air pollutants and health outcomes--that is, episodes of exacerbation (as described earlier)--was examined with marginal logistic-regression analyses on the basis of the Generalized Estimating Equations (GEEs) proposed by Liang and Zeger. (17) These models account for repeated measurements in response data. The correlation structure selected was exchangeable. We performed all analyses with the SAS GEE procedure, GENMOD (SAS institute [Cary, North Carolina]).
In this study, we performed the analysis step-by-step. initially, we determined, by bivariate analysis and synthesis of the literature, the covariates (i.e., confounders or effect modifiers) that must belong to the regression models: subject characteristics, and temporal and meteorological variables. Different lags (up to 5 days) for temperature and relative humidity were investigated. The final model included the following individual characteristics: (a) ratio of FE[V.sub.1.0] to its expected value, classified in accordance with the median value of the distribution (32%); (b) smoking habits (current or former smokers vs. nonsmokers); (c) PaC[O.sub.2], classified according to the median value of the distribution (43 mm Hg); (d) oxygen treatment or ventilation (yes or no); (e) Sadoul's dyspnea, classified according to the median value of the distribution (grade 3); (f) weather (daily temperature and relative humidity mean); (g) season; and (h) holiday periods (yes or no).
We estimated the effect of each pollutant on health by entering it separately into the models described earlier, as well as into 2-pollutant models. We considered the possibility of a nonlinear relationship between pollution and health outcomes by examining logarithmic transformation of pollutants. Inasmuch as the analysis of log-transformed data did not improve the model, untransformed data were used for further evaluation. We examined lags of up to 5 days--an approach that comports with those described in the epidemiological studies on the impact of air pollution on health. We also explored the effect of cumulative exposure--defined as a mean of the previous days (up to 5 days). Odds ratios (ORs) for exacerbation were given for a 10-[micro]g/[m.sup.3] increase in pollutant concentration. Interaction terms among variables were also tested. We checked the fit to the model by deviance, scaled deviance, and Pearson's [chi square]. Analyses were conducted for the entire study period, as well as by season (winter: October-March; summer: April-September). Only [O.sub.3] was studied in summer.
Subjects and visits. The panel comprised 10 women and 29 men, whose average age was 67 yr (standard deviation [SD] = 9 yr). Of these 39 patients, 18% had never smoked, 23% were current smokers, and 59% were former smokers; their consumption averaged 63 packs-years (SD = 13 packs-years) and 57 packs-years (SD = 17 packs-years), respectively. All subjects had bronchoconstriction, as evidenced by an average FE[V.sub.1.0] of 34% (SD = 11) of the expected value and a decreased VC of 67% (SD = 15) of the expected value. Mean values of Pa[O.sub.2], PaC[O.sub.2], and systolic pulmonary artery pressure were 62 mm Hg (SD = 7 mm Hg), 43 mm Hg (SD = 6 mm Hg), and 55 mm Hg (SD = 10 mm Hg), respectively. The severity of COPD in these patients was, on average, grade 3 ("unable to walk like a healthy subject in the street") on Sadoul's dyspnea scale. The majority of subjects used inhaled [beta]2 agonists (95%) and steroids (74%) as maintenance medications, 26% took scheduled oral steroids, and 39% had oxygen treatments.
The 14-mo study period permitted collection of data during 290 physician visits (average of 21 visits/mo) that were distributed regularly throughout the week (Monday-Saturday) and year. Fifty episodes of symptom exacerbation were noted (an average of 4/mo). Thirty-two patients experienced at least 1 exacerbation of symptoms during the study period, but they were not hospitalized. A suspected cause for exacerbation of symptoms was noted in only 5 crises: cardiovascular pathology (2 crises), infection (1), stress (1), and indoor pollution (1).
Pollution levels. The average levels of P[M.sub.10] and N[O.sub.2] were 26 [micro]g/[m.sup.3] and 54 [micro]g/[m.sup.3], respectively; daily concentrations were homogeneous throughout the year. Daily variations in S[O.sub.2] and [O.sub.3] levels followed seasonal patterns. S[O.sub.2] concentrations increased during the winter (average 24-hr level = 19 [micro]g/[m.sup.3] vs. 7 [micro]g/[m.sup.3] in summer) and [O.sub.3] concentrations increased in summer (average 8-hr level = 41 [micro]g/[m.sup.3] vs. 11 [micro]g/[m.sup.3] in winter) (Table 1). Environmental data were strongly intercorrelated. A highly significant positive correlation was seen among P[M.sub.10], N[O.sub.2], and S[O.sub.2], regardless of the season. In summer, [O.sub.3] was correlated with P[M.sub.10] and N[O.sub.2], but not with S[O.sub.2]. The temperature was generally associated negatively with the atmospheric pollutants S[O.sub.2], P[M.sub.10], and N[O.sub.2], and positively with [O.sub.3]. Relative humidity was usually associated negatively with all the other variables.
Relationship between air pollution and episodes of exacerbation. No association was evident between episodes of symptom exacerbation and mean 24-hr concentrations of S[O.sub.2], P[M.sub.10], and N[O.sub.2], and the maximum 1-hr level of N[O.sub.2] (Table 2), regardless of the lag. Analysis by season did not highlight an OR significantly different from 1 (results not given).
The 8-hr average concentration of [O.sub.3] was related significantly with episodes of exacerbation for a 24-hr lag (Table 2). An increase of 10 [micro]g/[m.sup.3] above the maximum 1-hr [O.sub.3] level was associated with an increase in the incidence of exacerbation, after lags of 2 (p = .003) or 3 days (p = .01). These relationships also appeared with the maximum 1-hr concentration recorded during the same days (OR = 1.44; 95% CI = 1.14, 1.82; p = .003).
The study of interactions made it possible for us to show that the influence of [O.sub.3] was stronger in some groups of patients (Table 3). The action of [O.sub.3] appeared greater among smokers or former smokers (OR = 2.61; 95% CI: 1.85, 3.70; p < .00001) than among nonsmokers (OR = 1.32; 95% CI: 1.00, 1.75; p = .05). The OR of symptom exacerbation approximated 2 in patients with a PaC[O.sub.2] that exceeded 43 mm Hg, whereas patients with a lower PaC[O.sub.2] were less sensitive to [O.sub.3]. The effect of [O.sub.3] was unchanged, regardless of the severity of the obstructive syndrome, which was determined on the basis of Pa[O.sub.2], Sadoul's dyspnea, oxygen treatment or ventilation, and/or maintenance by oral [beta]2 agonists or steroids.
The effect of [O.sub.3] remained constant and statistically significant when both [O.sub.3] and another pollutant entered the model (Table 4). When we conducted the analysis and excluded the 5 crises that had 1 suspected cause other than outdoor air pollution, the same results were obtained.
The methodology we used in our study had several advantages: (a) homogeneity of patients with regard to their age and the severity of their COPD, (b) evaluation of the medical variables on an individual scale, and (c) the variable of exacerbation of COPD was always verified by 1 of the 2 CTAR physicians, who standardized their medical practices. The aforementioned item (c) offered a real advantage, compared with a diary, which depends on patient subjectivity and goodwill--with the main risk being the potential for declaration biases. On the other hand, we considered only crises verified by a physician; therefore, our study could have suffered from a lack of power. The requirement for access to a doctor did not introduce bias, because the patients visited the CTAR regularly.
Confounding and modifying factors--particularly subject characteristics--were also studied individually. Even if these characteristics did not vary daily at the same time as the pollutants (and might not, therefore, be a confounding factor for the short-term effect of air pollution in panel studies), they might modify the way patients react to air pollution, as has been proven by analysis of the interactions. It was also important that we collect information on the factors that were likely responsible for episodes of symptom exacerbation (e.g., infection, change in smoking habits, stress, unusual indoor pollution) to exclude possibilities other than air pollution. Knowledge of these individual characteristics provided a significant advantage over the "semi-individual" studies in which Kunzli and Tager (18) conferred an inferential validity equivalent on individual studies.
A statistically significant association between [O.sub.3] and the exacerbation of symptoms in our Parisian COPD patients was observed in the summer of 1996. This relationship was established with relatively weak concentrations of [O.sub.3] (maximum 1-hr [O.sub.3] concentration = 163 [micro]g/[m.sup.3]); the first level of the Parisian air pollution warnings system determined the information level for the sensitive group, fixed at 180 [micro]g/[m.sup.3], was never reached. The aforementioned relationship was highly significant for an increase in the daily 1-hr maximum level of [O.sub.3]. The patients most severely affected by increased PaC[O.sub.2] also showed a greater sensitivity to [O.sub.3]. Indeed, an increase in PaC[O.sub.2] with hypercapnia marks a significant turning point in the evolution of pathology, resulting from physiopathological modifications. Smokers with COPD were more sensitive to [O.sub.3] than nonsmokers, whereas healthy smoking volunteers did not respond to [O.sub.3] in some chamber exposure studies. (19,20) However, the 2 studies we have cited were conducted on asymptomatic smokers who did not suffer from COPD, and who might, therefore, have reacted differently to [O.sub.3] exposure than smokers in our panel, who were more severely ill and, therefore, were more sensitive to [O.sub.3] exposure. Nevertheless, the impact of [O.sub.3] remained identical, regardless of the level of basic pulmonary function and Pa[O.sub.2], whether patients were under oxygen treatment or ventilation in their residence, or if they were undergoing treatment with oral [beta]2 agonists or steroids. Inasmuch as all patients in our panel tended to have a very low FE[V.sub.1.0] (34% [SD = 11] of the theoretical value), it is not surprising that this criterion did not discriminate for sensitivity to [O.sub.3]. The treatment did not seem sufficient for the prevention of exacerbation of COPD among these patients following [O.sub.3] exposure. The benefits of long-term steroid therapy are currently under discussion. (21)
In no other panel study have investigators examined the effect of [O.sub.3] among subjects suffering from COPD. Many results in the literature on the impact of [O.sub.3] on subjects suffering from COPD have been provided by ecological epidemiological studies in which the authors analyzed hospital admissions involving this pathology. The effects of [O.sub.3] on these admissions were highlighted in 4 of 7 studies. (1,4-9) In the meta-analysis performed in the Air Pollution and Health-A European Approach (APHEA) program, (4) this relationship seemed very homogeneous in all European cities. It appeared after a lag of 0-2 days, which is similar to the lag time of 1-3 days found in our study.
Our results are consistent with those of toxicological studies that have shown the inflammatory mechanism of [O.sub.3]. The recruitment of inflammatory cells into the lung presents a risk of tissue damage through the release of toxic mediators by activated inflammatory cells. (22,23) Perhaps this phenomenon would be more serious among patients-suffering from COPD, in whom a preexistent inflammation of the small or large airways would be constant.
In our study, no association was found between COPD and PM, S[O.sub.2], or N[O.sub.2]. In contrast, an effect for all of these pollutants, except S[O.sub.2], was observed in a New Zealand panel, (13) with an increase in night cough in relation to P[M.sub.10], and a greater use of inhalers after an increase in N[O.sub.2]. Pope and Kanner, (14) who studied 624 smokers who suffered from light COPD, noted a reduction in FE[V.sub.1.0] and FE[V.sub.1.0]/VC ratio--conversely associated with increasing P[M.sub.10] concentrations. Ambient levels of particles were higher in these 2 studies than in our study. In New Zealand, (13) the 24-hr average concentration exceeded 5 times the recommended concentration of 120 [micro]g/[m.sup.3]; in Salt Lake City, Utah, (14) the 24-hr average concentration was 55 [micro]g/[m.sup.3] vs. 26 [micro]g/[m.sup.3] in Paris. The only other panel study that dealt with N[O.sub.2] (15) reported no relationship between this pollutant and COPD.
More associations are described between admissions for COPD and air pollutant indicators (mainly PM) in ecological studies (1,2,7-9,11,12) than in other types of studies. Results of ecological studies are contradictory for N[O.sub.2]--being positive, (1) negative, (5) or unclear. (10) The only significant results with ambient S[O.sub.2] concentrations were produced in ecological studies in Europe.(1,4,11,12) It is important to note that some of the associations highlighted, particularly in ecological studies, resulted from 1-pollutant models that did not test the simultaneous influence of several pollutants. There is a risk of attributing an effect to 1 pollutant that should, in fact, be ascribed to another because of the colinearity that exists among several urban air pollutants.
The apparent contradiction between the results of ecological studies and individual studies might be explained by the difference in disease severity of the subjects included in the various studies. Patients who participate in controlled human exposures, or in a panel study, are voluntary subjects for whom COPD has already been diagnosed--often because the disease is at an advanced stage. We could, therefore, assume that these severely ill patients who receive a treatment adapted to the severity of their pathology are less sensitive to air pollution than subjects who suffer from COPD that (a) has evolved fairly recently, (b) has often been ignored by the individual, and (c) has not been treated. Indeed, for each patient diagnosed with COPD there are 2 not yet diagnosed. (24,25) Ambient levels of pollution could result in hospital admissions of these individuals for respiratory disease or COPD. For ethical reasons, patients who suffer from severe COPD do not participate in controlled human exposure or panel studies; however, as suggested by Anderson et al., (1) the health of these weakened patients is easily exacerbated, and every exacerbation will be accompanied by a hospitalization recorded during ecological studies.
In summary, although [O.sub.3] was related to COPD exacerbation in our panel, no association was observed for the other pollutants studied. The severely affected patients in our panel were exposed to low levels of ambient outdoor pollutants that resulted from their reduced mobility and ventilatory flow. Furthermore, all of these patients are monitored regularly at the CTAR by their pulmonary physicians, and they benefit from specialized treatment, which often includes oxygen treatment. They are always offered respiratory rehabilitation. Perhaps their therapeutic treatment is sufficient for the prevention of main disorders that could be caused by slight fluctuations in ambient air pollution concentrations--with the possible exception of [O.sub.3], which could be more aggressive.
The authors are grateful to Dr. Fabien Squinazi from the Hygiene Laboratory of Paris city for the environmental data he gave us and for his advice. The authors also thank Dr. Jean Bourcereau for his help in medical data collection, Pr. Jean Pierre Daures for his statistical advice, and Verane Roig for her comments on the manuscript.
Submitted for publication September 28, 2000; revised; accepted for publication April 2, 2001.
Requests for reprints should be sent to Pr. Isabelle Momas, Laboratoire d'Hygiene et de Sante Publique, Faculte de Pharmacie, 4, avenue de I'Observatoire 75270 Paris cedex 06, France.
(1.) Anderson H, Spix C, Medina S, et al. Air pollution and daily admissions for chronic obstructive pulmonary disease in 6 European cities: results from the APHEA project. Eur Respir J 1997; 10:1064-71.
(2.) Burnett RT, Dales RE. Associations between ambient particulate sulfate and admissions to Ontario hospitals for cardiac and respiratory disease. Am J Epidemiol 1995; 142:15-22.
(3.) Burnett RT, Dales RE, Raizenne ME. Effects of low ambient levels of [O.sub.3] and sulfates on the frequency of respiratory admissions to Ontario hospitals. Environ Res 1994; 65:172-94.
(4.) Dab W, Medina S, Quenel P, et al. Short term respiratory health effects of ambient air pollution: results of the APHEA project in Paris. J Epidemiol Community Health 1996; 50(suppl 1): S42-S46.
(5.) Moolgavkar S, Luebeck E, Anderson E. Air pollution and hospital admissions for respiratory causes in Minneapolis-St. Paul and Birmingham. Epidemiology 1997; 8:364-70.
(6.) Schwartz J. Air pollution and hospital admissions for respiratory disease. Epidemiology 1996; 7:20-28.
(7.) Schwartz J. Air pollution and hospital admissions for the elderly in Detroit, Michigan. Am J Respir Crit Care Med 1994; 150: 648-55.
(8.) Schwartz J. Air pollution and hospital admissions for the elderly in Birmingham, Alabama. Am J Epidemiol 1994; 139:589-98.
(9.) Schwartz J. P[M.sub.10], ozone, and hospital admissions for the elderly in Minneapolis-St. Paul, Minnesota. Arch Environ Health 1994; 49:366-74.
(10.) Shouten J, Vonk J, de Graaf A. Short-term effects of air pollution on emergency hospital admissions for respiratory disease: results of the APHEA project in two major cities in The Netherlands, 1977-89. J Epidemiol Community Health 1996; 50(suppl 1): S22-S29.
(11.) Sunyer J, Saez M, Murillo C, et al. Air pollution and emergency room admissions for chronic obstructive pulmonary disease: a 5-year study. Am J Epidemiol 1993; 137:701-05.
(12.) Sunyer J, Anto JM. Effect of urban air pollution on emergency room admission for chronic obstructive pulmonary disease. Am J Epidemiol 1991; 134:277-86.
(13.) Harre ES, Price PD. Respiratory effects of air pollution in COPD: a three month prospective study. Thorax 1997; 52:1040-44.
(14.) Pope C, Kanner R. Acute effects of P[M.sub.10] pollution on pulmonary function of smokers with mild to moderate chronic obstructive pulmonary disease. Am Rev Respir Dis 1993; 147:1336-40.
(15.) Hackney J, Linn W, Avol E, et al. Exposures of older adults with chronic respiratory illness to nitrogen dioxide. A combined laboratory and field study. Am Rev Respir Dis 1992; 146:1480-86.
(16.) Rokaw SN, Massey F. Air pollution and chronic respiratory disease. Am Rev Respir Dis 1962; 86:703-04.
(17.) Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986; 73:13-22.
(18.) Kunzli N, Tager IB. The semi-individual study in air pollution epidemiology: a valid design as compared to ecologic studies. Environ Health Perspect 1997; 105:1078-83.
(19.) Schephard RJ, Urch B, Silverman F, et al. Interaction of ozone and cigarette smoke exposure. Environ Res 1983; 31:125-37.
(20.) Emmons K, Foster WM. Smoking cessation and acute airway response to ozone. Arch Environ Health 1991; 46:288-94.
(21.) Rees PJ. Bronchodilators in the therapy of chronic obstructive disease. Eur Respir Mon 1998; 7:135-49.
(22.) Bhalla DK. Ozone-induced lung inflammation and mucosal barrier disruption: toxicology, mechanism, and implications. J Toxicol Environ Health 1999; 2:31-86.
(23.) Bouthillier L, Renaud V, Goegan P, et al. Acute effects of inhaled urban particles and ozone, lung morphology, macrophage activity, and plasma endothelin-1. Am J Pathol 1998; 153:1873-84.
(24.) Morrow P, Utell M, Bauer M, et al. Pulmonary performance of elderly normal subjects and subjects with chronic obstructive pulmonary disease exposed to 0.3 ppm nitrogen dioxide. Am Rev Respir Dis 1992; 145:291-300.
(25.) Littlejohns P, Ebrahim S, Anderson R. Prevalence and diagnosis of chronic respiratory symptoms in adults. Br Med J 1989; 298: 1556-60.
COPYRIGHT 2002 Heldref Publications
COPYRIGHT 2003 Gale Group