THE EFFECT OF AIR POLLUTION on various health indicators has received considerable attention in the past few years during which investigations--particularly of the ecological time-series type (1)--have been conducted. Increased levels of some pollutants have been associated with increases in daily mortality (2,3) and with the number of hospital admissions for diverse illnesses, respiratory diseases, (4,5) and, recently, cardiovascular diseases. (6,7) The influence of air pollution on emergency room visits has been examined far less frequently than the aforementioned. Generally, the relationship with emergency room visits for asthma (8-13) has been examined more frequently than those for chronic obstructive pulmonary disease (COPD).(8,14-16)
In addition to the magnitude of the effects, it is important for investigators to analyze other aspects that are typically ignored, such as dose-response analysis, confounding and interaction between pollutants, and calculated changes in response to different analytical circumstances.
In the current investigation we sought to (1) calculate the short-term effect(s) of air pollution on emergency room visits for COPD within Valencia, Spain, and (2) analyze other aspects that are relevant to this relationship.
Materials and Method
Study setting. The investigation was conducted in Valencia, Spain, in which there is an average population of 750,000 inhabitants (1996 census). Valencia is located on the Mediterranean coast and possesses a mild climate. This analysis occurred in one of the city's main hospitals (i.e., Hospital Clinico Universitario [HCU]), with an urban catchment area of approximately 200,000 inhabitants. Inasmuch as the HCU is part of the public health service network, access to emergency health services is universal.
Health data. The individuals chosen were all daily cases of COPD in residents of the city; individuals were older than 14 yr of age, and they attended the HCU's emergency department during a 2-yr period between January 1, 1994, and December 31,1995. The data collection method had been previously validated and documented. (17) The reliability of this method, as used at the HCU, has been analyzed previously, (18) with good reliability (i.e., kappa intraobserver agreement indices of 0.79, 0.91 and a kappa interobserver index of 0.87).
Air pollution measurements. The pollution data were obtained from the air pollution monitoring network in the city. Pollutant indicators in the study were (1) daily mean levels of Black Smoke (reflectometry method), determined from manual stations; and from the automatic network: (2) 24-hr average levels of nitrogen dioxide N[O.sub.2] [chemiluminescence]), (3) sulfur dioxide S[O.sub.2] levels (ultraviolet fluorescence), and (4) carbon monoxide (CO) levels (infrared absorption). In addition, we obtained hourly maximum readings for ozone (O[sub.3] [UV absorption]), N[O.sub.2], S[O.sub.2], and CO. Only the data obtained from those stations with valid readings on more than 75% of days were used in this investigation. We used these criteria to finally select 6 manual and 5 automatic monitoring stations. The correlations between the measurements at different stations were from 0.40 to 0.68 for Black Smoke, 0.61 to 0.75 for CO (1 hr), 0.49 for S[O.sub.2] (24 hr), 0.87 for N[O.sub.2] (24 hr), and 0.75 for O[sub.3] (1 hr).
The daily data for temperature and relative humidity were obtained from Valencia's meteorological station, located within an urban "green area." The weekly influenza cases were extracted from the registers of notifiable diseases. We also considered local holidays and the period of strike action by medical staff during May of 1995.
Statistical methods. The relationship between pollution and health could be confounded by factors that changed over time (i.e., geophysical factors [seasonality and annual cycles], meteorological factors [temperature and humidity], and sociocultural factors [day of the week, feast days]). (19) To account for these factors, we used a Poisson autoregressive analysis and followed the methodological principles of the "Air Pollution and Health--a European Approach" (APHEA) project. (20,21) The possible delayed effect was assessed, with the inclusion of corresponding pollution terms up to 3 days prior to the emergency (5 days in the case of O[sub.3]). We calculated the effect of pollution with the relative risk (RR) that corresponded to an increase of 10 [micro]g/[m.sup.3] in air pollutant indicators (but 1 mg/[m.sup.3] for CO).
We evaluated the possible modification of the effect of pollution, depending on whether it was the warm or cold season of the year, by introducing an interaction term between the pollutant and a variable indicator for semester (cold semester--November to April; warm semester--May to October).
Sensitivity analysis. To prove the robustness of the calculations, we conducted several sensitivity analyses. Initially, the influence on the calculations of the various specifications of the basal model was determined. Finally, another approach was used that involved construction of generalized additive models. (22) With this approach, we used locally nonparametric methods to model the meteorological and tendency variables, thus avoiding the problems that originated from the nonlinearity of effect for these terms. We chose a lowess smoother, which had a span of 0.15 for tendency and 0.5 for temperature and humidity.
Dose-response analysis. We confirmed the potential linear relationship by comparing the deviance between the model with the pollutant introduced as a continuous variable (i.e., with the assumption that it had a linear effect) and the model with the pollutant modeled by a lowess smoother (i.e., a variety of spans from 0.1 to 0.5 were tested). A nonsignificant difference meant that the possibility of a linear relationship between pollution and risk of emergency cases could not be ruled out. When established, the value and significance of the linear relationship's slope corresponded to those calculated for the pollutant used as a continuous variable.
Two-pollutant models. The independence of the significant results was confirmed by the construction of models with 2 pollutants. We introduced no more than 2 pollutants to avoid problems resulting from multi-collinearity. The interaction between pollutants was assessed with the introduction of a multiplicative term between the pollutant in question and an indicator variable for the effect-modifying pollutant (i.e., 0 if its values were less than the median and 1 if they were greater).
During the 2 yr of the investigation, a total of 1,298 COPD patients who were 14+ yr of age and who resided in Valencia attended the HCU. The median of daily cases was 2 (interval = 0-8). Pollutants, such as Black Smoke, S[O.sub.2], and CO (Table 1), shared the seasonal distribution of COPD cases; more emergencies occurred during the colder semester. Ozone showed the opposite seasonal pattern as, given the photooxidation phenomena, its concentrations were higher during the warmer period of the year.
The correlations between the explanatory variables were as expected. Therefore, we observed a positive linear association between Black Smoke, S[O.sub.2], and CO, and a negative correlation was noted with [O.sub.3] (Table 2). Temperature produced a positive correlation with [O.sub.3] and a negative association with CO, Black Smoke, and S[O.sub.2]. N[O.sub.2] established only a weak correlation--both with other pollutants, as well as with temperature.
The RRs for the increase of 10 [micro]g/[m.sup.3] in air pollution indicators (1 mg/[m.sup.3] for CO) for emergency room visits at the Hospital Clinico of Valencia are shown in Table 3. [O.sub.3] and CO were associated positively with the daily number of emergency room visits for COPD. In the case of Black Smoke, a positive, though not significant, association was found. For S[O.sub.2] and N[O.sub.2], the effect was contrary to that expected, although not significantly so.
Of the 2 pollutants for which there existed a significant relationship with the daily number of COPD emergencies (i.e,.[O.sub.3] and CO), the effect appeared more delayed for [O.sub.3]. The relationship reached its maximum value within a narrow "time window" for a pair of adjacent lags (Fig. 1): lags 1 and 2 for CO and lags 4 and 5 for [O.sub.3].
[FIGURE 1 OMITTED]
In the sensitivity analysis (Table 4), the consistency of the associations can be seen. The generalized additive models produced computations similar to those carried out with the original basal models. In the dose-response analysis, there emerged a linear relationship for CO and [O.sub.3] in equal measure. During the linearity tests, the variation in the models' deviances was not significant (p > .1); therefore, a linear association between air pollution and emergency admissions resulting from COPD cannot be ruled out. The calculations for 2-pollutant models showed an effect independent of [O.sub.3] and CO inasmuch as they varied little from those models that included only 1 pollutant (Fig. 2).
[FIGURE 2 OMITTED]
The assessment of the interaction between pollutants clearly showed the modified effect of the [O.sub.3] under differing concentrations of CO and Black Smoke. For this reason, we decided to carry out a stratified analysis of [O.sub.3] at 3 different levels (terciles) of CO and Black Smoke. [O.sub.3] had the greatest effect when it coincided with moderately high levels of the other 2 pollutants (Table 5). When we introduced both interactions into the same model, only the interaction calculated for CO maintained its magnitude and degree of significance, whereas that for Black Smoke was smaller and was not significant.
In this study, we explored the short-term association between relatively low levels of air pollution and hospital emergency room visits for COPD in an urban setting. Neither the degree of association nor its significance was the same for all pollutants. The most consistent findings were those related to [O.sub.3] and CO; these findings showed a positive and significant effect over the entire period. An increase in 10 [micro]g/[m.sup.3] (or 1 mg/[m.sup.3] for CO) in the concentration of these pollutants was associated with an expected increase in COPD emergencies of 6.1% and 3.9%, respectively.
The relationship between air pollution and COPD has been analyzed during various investigations, both in the United States and in Europe. The investigation of COPD hospital admissions in various American cities(4,23-25) showed that the predominant pollution effect resulted from particulate matter that had a diameter of less than 10 [micro]m (i.e., measured as P[M.sub.10]). A significant pollutant effect was also found in Detroit for [O.sub.3].(24) S[O.sub.2] was not assessed because the research was conducted in locations that had very low levels of S[O.sub.2]. In the Ontario region of Canada, an investigation into the effects of [O.sub.3] and sulfates on admissions for COPD produced significant results for both pollutants during warmer months (i.e., May to August).(26) In the previously mentioned studies, investigators did not consider the incidence of influenza as a possible confounding variable, which could have led to an over- or infraestimation of the effects found. In our own investigation, the exclusion of influenza from the basal model would have caused a relatively large change in the assessment of CO (i.e., decrease of 11% in RR), although this would not have been the case for [O.sub.3] (i.e., increase less than 0.2%). In Europe, a meta-analysis of the relationship between air pollution and hospital admissions resulting from COPD has recently been undertaken in 6 European cities as part of the APHEA project. (5) The most consistent effects seen were for [O.sub.3], particularly during the warmer months, although significant results were also seen for S[O.sub.2], N[O.sub.2], and particles. These results are of special interest because of the use of a standardized methodology, the size of the study population, and the diversity of both climatological conditions and health service systems involved. In this investigation, the city of Barcelona exhibited the daily number of emergency room visits as a response variable. Barcelona has climatological characteristics similar to those of Valencia, but Barcelona has somewhat higher pollution levels than Valencia, particularly with respect to S[O.sub.2] levels. The main results were the significant effects for Black Smoke, S[O.sub.2], and [O.sub.3]. The effect of CO was not examined during the APHEA investigation, although it was assessed in a previous study, (14) during which time a significant association was found.
In New Jersey, the relationship between [O.sub.3] and S[O.sub.2] pollution and asthma/COPD emergencies at 9 of the city's hospitals during the months of May to August was studied during a 2-yr period. (8) In the case of COPD, the results were not significant statistically, although the effect of [O.sub.3] was analyzed up to a maximum of 2 days prior to emergencies.
Among the pollutants studied, [O.sub.3] and CO appeared to have the strongest association with hospital admissions. The effects of either of these 2 pollutants did not appear to have been confounded by any other pollutants, but [O.sub.3) showed a significant degree of interaction with CO and with Black Smoke. This interaction cannot be explained by any possible confounding by meteorological variables inasmuch as their relationship with [O.sub.3] is different from that with CO. It is unlikely that the interaction resulted from a defect in series modeling because the same effect was obtained when we used a generalized additive model as a basal model. The interaction seems to include more CO involvement than Black Smoke involvement, because the latter had no modifying effect whatsoever on [O.sub.3] when they were both included within the same model. We were unable to confirm any analogous effect in any other investigations that were similar to ours.
In panel-type longitudinal analyses, researchers have determined that patients with COPD are sensitive to variations in ambient concentrations of particulate matter, S[O.sub.2], and [O.sub.3]. (27,28) Despite this fact, in clinical experiments investigators have found it impossible to produce conclusive evidence of the susceptibility of these patients to [O.sub.3], (29) or to N[O.sub.2] (30)--even at higher-than-ambient concentrations. In the case of CO, there is no direct toxic mechanism, but its property of affecting a patient's ability to withstand physical exercise has been demonstrated. (31) On the whole, those experiments in which more than 1 pollutant is used (32) do not produce results that are significantly different from those obtained from exposure to [O.sub.3] alone. It should be noted that in none of the investigations did the mixture include CO. Additional research is, therefore, needed for mixtures that include [O.sub.3], CO, and particulate matter in volunteers under laboratory conditions that mimic the environmental conditions.
The quality of health indicator data was evaluated, and there was a good reliability. (18) This minimized the problems caused by misclassification between other diagnostic categories. There was no ambiguity in determining the day on which the visit to hospital took place, as can sometimes occur with admissions for which administrative formalities are delayed.
In conclusion, the results of this investigation, together with results of earlier research, demonstrate the significant impact of pollution on various health indicators within Valencia. (33,34) Moreover, the linear dose-response found hinders the establishment of a safety threshold below which it can be assumed that there is no adverse effect on the health of the population. We, therefore, consider that a reduction in levels of atmospheric pollution would have a positive effect on public health.
This work was supported in part by grant 95/0050-02 from the Fondo de Investigaciones Sanitarias.
Submitted for publication July 16, 1998; revised; accepted for publication December 29, 2000.
Requests for reprints should be sent to Dr. Ferran Ballester, Department of Epidemiology and Statistics, Institut Valencia d'Estudis en Salut Publica (IVESP), C/Joan de Garay 23, 46017, Valencia, Spain.
(1.) Brunekreeef B, Dockery DW, Krzyzanowski M. Epidemiologic studies on short-term effects of low levels of major ambient air pollution components. Environ Health Perspect 1995; 103(Suppl 2):3-13.
(2.) Dockery DW, Pope CA, Xu X, et al. An association between air pollution and mortality in six U.S. cities. N Engl J Med 1993; 329:1753-59.
(3.) Katsouyanny K, Touloumi G, Spix C, et al. Short-term effects of ambient sulphur dioxide and particulate matter on mortality in 12 European cities: results from time series data from the APHEA project. Br Med J 1997; 314:1658-63.
(4.) Schwartz J. Air pollution and hospital admissions for the elderly in Detroit, Michigan. Am J Respir Crit Care Med 1994; 150:648-55.
(5.) Anderson HR, 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.
(6.) Morris RD, Naumova EN, Munasinghe RL. Ambient air pollution and hospitalization for congestive heart failure among elderly people in seven large US cities. Am J Public Health 1995; 85:1361-45.
(7.) Burnett RT, Dales RE, Brook JR, et al. Association between ambient carbon monoxide levels and hospitalizations for congestive heart failure in the elderly in 10 Canadian cities. Epidemiology 1997; 8:162-67.
(8.) Cody RP, Weisel CP, Birnbaum G, et al. The effect of ozone associated with summertime photochemical smog on the frequency of asthma visits to hospital emergency departments. Environ Res 1992; 58:184-94.
(9.) Schwartz J, Slater D, Larson TV, et al. Particulate air pollution and hospital emergency room visits for asthma in Seattle. Am Rev Respir Dis 1993; 147:826-31.
(10.) Weisel CP, Cody RP, Lioy PJ. Relationship between summertime ambient ozone levels and emergency department visits for asthma in Central New Jersey. Environ Health Perspect 1995; 103(Suppl 2):97-102.
(11.) Castellsague J, Sunyer J, Saez M, et al. Short-term association between air pollution and emergency room visits for asthma in Barcelona. Thorax 1995; 50:1051-56.
(12.) Kesten S, Szalai J, Dzyngel B. Air quality and the frequency of emergency room visits for asthma. Ann Allergy Asthma Immunol 1995; 74:269-73.
(13.) Stieb DM, Burnett RT, Beveridge RC, et al. Association between ozone and asthma emergency room visits in Saint John, New Brunswick, Canada. Environ Health Perspect 1996; 104:1354-60.
(14.) Sunyer J, Anto JM, Murillo C, et al. Effects of urban air pollution on emergency room admissions for chronic obstructive pulmonary disease. Am J Epidemiol 1991; 134:277-86.
(15.) 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.
(16.) Guillen JJ, Guillen F, Medrano J, et al. Afluencia inusual por asma y enfermedad pulmonar obstructiva cronica en urgencias hospitalarias y contaminacion por S[O.sub.2] en Cartagena. [Unusual attendance at hospital emergency services for asthma and chronic obstructive pulmonary disease and S[O.sub.2] air pollution in Cartagena, Spain.] Rev Esp Salud Pub 1995; 69:305-14.
(17.) Martinez F, Sunyer J, Anto JM. Reliability of a monitoring system for respiratory emergency room admissions. Eur Respir J 1993; 6:337-41.
(18.) Merelles T, Rivera M, Ballester F, et al. Identification of emergencies for asthma and chronic obstructive pulmonary disease in the emergency room of a hospital. A reliability study. Gac Sanit 1999;13:30-37.
(19.) Goldsmith JR, Friger MD, Abramson M. Associations between health and air pollution in time series analysis. Arch Environ Health 1996; 51:359-67.
(20.) Schwartz J, Spix C, Toulouni G, et al. Methodological issues in studies of air pollution and daily counts of deaths or hospital admissions. J Epidemiol Commun Health 1996; 50(Suppl 1): 3-11.
(21.) Katsouyanni K, Schwartz J, Spix C, et al. Short-term effects of air pollution on health: a European approach using epidemiologic time series data: the APHEA protocol. J Epidemiol Commun Health 1996; 50(Suppl 1):12.
(22.) Hastie T, Tibshrani R. Generalized additive models. London, UK: Chapman and Hall, 1990.
(23.) Schwartz J. Air pollution and hospital admissions for the elderly in Birmingham, Alabama. Am J Epidemiol 1994; 139:589-98.
(24.) 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.
(25.) Schwartz J. Air pollution and hospital admissions for respiratory disease. Epidemiology 1996; 7:20-28.
(26.) Burnett RT, Dales RE, Raizenne MA, et al. Effects of low ambient levels of ozone and sulfates on the frequency of respiratory admissions to Ontario hospitals. Environ Res 1994; 65:172-94.
(27.) Pope CA, Kanner RE. 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; 1476:1336-40.
(28.) Higgins BG, Francis HC, Yates CJ, et al. Effects of air pollution on symptoms and peak expiratory flow measurements in subjects with obstructive airways disease. Thorax 1995; 50:149-55.
(29.) Kehrl HR, Hazuka MJ, Solic JJ, et al. Responses of subjects with chronic obstructive pulmonary disease after exposures to 0.3 ppm ozone. Am Rev Respir Dis 1985; 131:719-24.
(30.) Morrow PE, Utell MJ, Bauer MA, et al. Pulmonary performance of elderly normal subjects and subjects with chronic pulmonary disease exposed to 0.3 ppm nitrogen dioxide. Am Rev Respir Dis 1992; 145:291-300.
(31.) Calverley PMA, Legget RJE, Flenley DC. Carbon monoxide and exercise in chronic bronchitis and emphysema. Br Med J 1981; 283:878-80.
(32.) Bylin G. Controlled human studies of ozone exposure. Scand J Work Environ Health 1996; 22(suppl 3):52-71.
(33.) Ballester F, Corella D, Perez-Hoyos S, et al. Air pollution and mortality in Valencia, Spain: a study using the APHEA methodology. J Epidemiol Commun Health 1996; 50:527-33.
(34.) Tenias JM, Ballester F, Rivera M. Association between asthma emergency room visits and air pollution in Valencia, Spain. Occup Environ Med 1998; 55:541-47.
COPYRIGHT 2002 Heldref Publications
COPYRIGHT 2002 Gale Group