Air Pollution and Hospital Admissions for Respiratory Disease in Certain Areas of Korea

Transcription

1 J Occup Health 2000; 42: Journal of Occupational Health Air Pollution and Hospital Admissions for Respiratory Disease in Certain Areas of Korea Belong CHO 1, Jaewook CHOI 2 and Yong-Tae YUM 2 1 Department of Family Medicine, Clinical Research Institute, Seoul National University Hospital, Seoul National University, Seoul, Korea 2 Department of Preventive Medicine, School of Medicine, Korea University, Seoul, Korea Abstract: Air Pollution and Hospital Admissions for Respiratory Disease in Certain Areas of Korea: Belong CHO, et al. Department of Family Medicine, Clinical Research Institute, Seoul National University Hospital, Seoul National University Recently studies reporting the incidence of diseases at air pollution levels below the recognized standard levels are increasing and the issue is becoming a matter of concern. We therefore examined the relationship between the level of air pollution in three areas of Korea and the incidence of respiratory diseases. Poisson regression models were used to account for day-of-the-week effects and nonparametric smoothing to make adjustments for season and weather in this time series analysis. The levels of air pollution across the three selected areas were characterized by the total suspended particle (TSP) concentration, which in Daejon was ± 29.22, in Ulsan ± 25.99, and in Suwon was ± The total number of hospitalized cases due to respiratory diseases during the study period was 5,446. The relative risk of hospitalization due to respiratory diseases caused by air pollutants after accounting for seasonal and temperature effects were as follows:- CO (R.R.; 1.21, 95% C.I.; ) in a residential area (Daejon), and NO 2 (R.R.; 1.47, 95% C.I.; ) and CO (R.R; 2.51, 95% C.I.; ) in a heavily industrialized area (Ulsan), and were statistically significant. When the TSP concentration was manipulated as a quintile dummy variable, the relative risk of admission increased by 2.48% (95% C.I.: 1.82% 3.15%) for every quintile increase. In conclusion, respiratory disease admissions are related to NO 2, CO, and TSP concentrations below the environmental standard, but the significance of this relationship was area dependent. (J Occup Health 2000; 42: ) Received Dec 22, 1999; Accepted March 31, 2000 Correspondence to: B. Cho, Department of Family Medicine, Clinical Research Institute, Seoul National University Hospital, Seoul National University, Yeongeon-dong 28, Chongro-gu, Seoul, Korea Key words: Air pollution, Respiratory disease, Time series analysis, Hospital admission, TSP, NO 2, CO Air pollution disasters in London 1), the Mesus Valley 2) and Donora, PA, USA 3) have provided indisputable evidence that high levels of air pollution have major shortterm effects on mortality and morbidity. As a result each country has instituted maximum air pollution concentration standards, but concern continues to be raised because measured air pollution levels below the standard do not guarantee the absence of health hazards. During the last decade, a number of studies 4 6) have indicated the existence of short-term health effects at air pollution levels below the national and international norms. Therefore, research 7) is being conducted to reset these environmental air pollution standards. For these reasons it is important that information about the quantitative effects of air pollution on health be made readily available. We quantitatively investigated the relationship to the incidence of respiratory disorders, which had occurred at levels below that of the national standard, by using air pollution data and medical records of respiratory disease incidence in three areas of Korea. Data and Methods Many of the existing studies have focused on the association between environmental pollution and the morbidity or mortality in one area. We set out to determine if these established associations occur predictably in different environments and under different conditions. Three areas, which were presupposed to have different levels of air pollution, were chosen for the exercise. They were Ulsan, which represented a heavily industrialized area, Daejeon, which is a residential area and Suwon, which is a mix of the two characters. The incidence of target respiratory diseases was estimated from the number of bronchial asthma, chronic obstructive pulmonary disease, and bronchitis admissions

2 186 J Occup Health, Vol. 42, 2000 to every general hospital in each pertinent area. To determine the number of patients admitted, we referred to the admission records of the 16 general hospitals involved. When the patients were admitted repeatedly, every admission was counted, and those admitted between January 1 and December 31 of 1996 were included in the study. Five monitored pollution criteria were obtained from the Ministry of the Environment; these were: total suspended particles (TSP), SO 2, NO 2, O 3, and CO, which were continuously measured by automated measuring units in the three selected areas. We calculated the hourly and daily average concentrations of TSP, SO 2, NO 2, O 3, and CO in Daejeon, Suwon, and Ulsan, respectively, and the maximum concentration of O 3 per hour. Meteorological information concerning temperature and relative humidity were obtained from the Automatic Weather Stations located at meteorological observation sites in the selected areas. Poisson regression analysis was used to analyze the number of daily-admitted respiratory disease patients. To account for other controlling factors, which affected the change in the daily number of respiratory patients admitted, such as time (trends and seasonal fluctuations), weather (association between the temperature and relative humidity and respiratory diseases), and holidays etc., we used the General Additive Model (GAM) 8). Loess s nonparametric method 9) to investigate the interaction between any two factors. Most of the acute respiratory diseases are known to have immediate relationships with the effects of the pollution variables, but some reports 10) have claimed that delay effects introduce real confounding factors. Therefore, lag effect analysis was undertaken simultaneously to identify such occurrences. The relative risk ratios corresponding to each quintile of the TSP, which is recognized as the most sensitive indicator of air pollution, from pertinent areas were plotted and examined for evidence of the presence of a dose-response relationship. From these results, the relative risk ratios and their corresponding 95% confidence intervals, for an increase of 100 ppb (or µg/ m 3 ) in the pollutant concentration, were estimated with respect to its effects on the number of respiratory disease patents admitted 11). Data analysis was undertaken with SAS version 6.12, and the General Additive Poisson Model with S-plus 4.0 for Windows. Results General characteristics of the data In total 2,151 hospitalized asthma cases were involved, 1,052 cases in Daejeon, 747 cases in Suwon, and 352 cases in Ulsan. The admission rate due to asthma was highest in the under 5s and the over 70s in all three areas. The total number of hospitalized cases due to bronchitis was 2,003 in all three areas, the under 5 age group showed the highest incidence in terms of the population adjusted admission rate. The total number of cases hospitalized due to chronic obstructive pulmonary diseases was 1,292, and there was a clear pattern of increase in the number of hospitalizations and the population size adjusted admission rates with increasing age (Table 1). As can be seen in Table 2, the average daily number of hospitalized patients was 6.68 in Daejeon, 5.4 in Suwon, and 2.96 in Ulsan. Suwon had the highest TSP, SO 2, and NO 2 concentrations of, µg/m 3, ppb, and ppb, respectively, though O 3 and CO were highest in Daejeon. Although the data are not shown in this report, the admission rate, SO 2 and TSP levels were higher in winter, whereas, O 3 was higher in the summer. Analyses of the association between variables The results of calculated Pearson correlation coefficients for the pollution and environmental criteria, and number of admission cases are shown in Table 3. The number of admission cases and the concentrations of NO 2 and TSP correlated positively. Ozone was negatively correlated, except in Ulsan, and SO 2 correlated positively in Suwon only. The coefficients of correlation between TSP and SO 2, NO 2 lay between 0.4 to 0.6, which shows a close relationship. The temperature was negatively correlated with the number of admissions and pollution indicators, except ozone. Analyses of relative risk of respiratory diseases due to air pollutants Initially, residual analysis was used to examine whether the effects of long term fluctuations and trends were adequately controlled by the model. Table 4 shows the result of an analysis done with the single pollutant model in each area. It shows that the admission rate was highly correlated with CO (R.R.; 1.25, 95% C.I.; ) in Daejeon, and with NO 2 (R.R.; 1.50, 95% C.I.; ) and CO (R.R.; 3.55, 95% C.I.; ) in Suwon. Other environmental pollution indicators were divided into quartiles and were transformed into dummy variables, creating a multipollutant model. This model still showed the same statistically significant results. Table 5 is the result obtained from a multi-pollutant model, which does not apply dummy variables, but the actual measured integrated environmental pollution indicator values across the three areas. TSP (R.R; 1.04, 95% C.I; ) and CO (R.R; 1.34, 95% C.I; ) were significant in Daejeon, but on the other hand, CO (R.R; 3.55, 95% C.I; ) alone was found to be statistically significant in Ulsan. Analyses of the number of the respiratory disease admission cases and the concentration of TSP Instead of the actual concentration of TSP, the quintiles

3 Belong CHO, et al.: Air Pollution and Hospital Admissions for Respiratory Ds. 187 Table 1. Number of admissions for respiratory disease in areas studied Area Age Number of admissions group Asthma Bronchitis COPD Subtotal Population , ,663 Daejeon , , ,561 Subtotal 1, ,361 1,293, , , ,723 Suwon , , ,201 Subtotal ,998 1,297, , ,846 Ulsan , , ,747 Subtotal , ,285 Total 2,151 2,003 1,292 5,446 3,582,571 Table 2. Daily data on air pollutions and hospital admissions for respiratory diseases in period studied Variable Mean ± SD (Range) Daejeon Ulsan Suwon Admission* 6.68 ± 3.82 ( ) 2.96 ± 2.15 ( ) 5.40 ± 3.40 ( ) TSP (µg/m 3 ) ± ( ) ± ( ) ± ( ) SO 2 (ppb) ± 6.51 ( ) ± 9.38 ( ) ± 9.45 ( ) NO 2 (ppb) ± ( ) ± 6.91 ( ) ± ( ) O 3 (ppb) ± 8.31 ( ) ± 6.07 ( ) ± 7.98 ( ) O 3MAX (ppb) ± ( ) ± ( ) ± ( ) CO (ppm 10) ± 6.11 ( ) 9.50 ± 2.11 ( ) ± 5.49 ( ) Temperature ( C) ± ( ) ± 8.98 ( ) ± ( ) Humidity (%) ± ( ) ± ( ) ± ( ) *Total number of daily admissions for respiratory diseases in all hospitals studied. were transformed into dummy variables and the relative risk ratio of the admission rates were calculated (Figs. 1, 2 and 3). The relative risk ratio (2.48%, 95% C.I.: 1.82% 3.15%) showed a linear increase(,) which was dependent upon the TSP concentration, and this trend was applicable in all three areas. Discussion In a time series analysis based method as used in this study, we observed fluctuations in the same population over time, rather than attempting to compare different areas. Therefore, the first aim of this study was to conduct a within-community comparison, but we also tried to determine whether similar results would be obtained in three areas with different environments. The results obtained show that respiratory disease admissions are related to NO 2, CO, and TSP concentrations below Korean environmental air quality standards, but these significances were not consistent in the different areas. We believe that the reason for this is that we failed to

4 188 J Occup Health, Vol. 42, 2000 Table 3. Pearson coefficients of correlation between variables according to areas Tempera- Humidity SO 2 NO 2 TSP O 3 O 3 Max CO Admisture ( C) (%) (ppb) (ppb) (µg/m 3 ) (ppb) (ppb) (ppm ( 10)) sion* Temperature ( C) Daejeon Ulsan Suwon/Ansan Humidity (%) Daejeon Ulsan Suwon/Ansan SO 2 (ppb) Daejeon Ulsan Suwon/Ansan NO 2 (ppb) Daejeon Ulsan Suwon/Ansan TSP (µg/m 3 ) Daejeon Ulsan Suwon/Ansan O 3 (ppb) Daejeon Ulsan Suwon/Ansan O 3 Max (ppb) Daejeon Ulsan Suwon/Ansan CO (ppm ( 10)) Daejeon Ulsan Suwon/Ansan Admission case* Daejeon Ulsan Suwon/Ansan *: Total number of cases admitted per day at all hospitals studied. : P<0.05. : P<0.01. correct for another major factor which affects the occurrence of respiratory disease. In the time series analysis method, we presumed that the same population is continuously studied. A new inflow of patients susceptible to respiratory diseases, a decrease in the number of patients due to the development of new therapeutic methods, the establishment of new hospitals, regular migration of the population caused by changes in roads or administrations can break down the assumptions and cause problems, but forfunately luckily there were no such events recorded during our period of study. During the process of analysis, the inclusion of weather factors, such as temperature and humidity, directly into the model was not appropriate; instead we used the non-parametric smoothing method of Lowess to account for these effects. One of the factors that could influence the relationship between air pollution and the number of hospitalized patients is an outbreak of influenza, but if seasonal factors are properly controlled, its relative importance declines. Fortunately during our study no such event occurred. To eliminate the day and seasonal effects, they were manipulated as dummy variables, and included in the model. The model analysis showed that the relative risk ratio of admission per day was 1.04 with an increase in TSP of 20 µg/m 2. These results compare well with the RR for death caused by respiratory disease of 1.17, 1.12, and 1.13, reported by Pope 12) Vigotti MA et al. 13) from Milan, Italy, and Saldiva PH et al. 14) from San Paulo, respectively. Very significant differences have been reported in epidemiological studies, which have examined the effects of air pollution on the respiratory system. These have been assigned to the study population, the study method, and air pollutant concentrations. Nevertheless, many studies 14, 15) claim that TSP is more closely associated with chronic obstructive pulmonary diseases and bronchial

5 Belong CHO, et al.: Air Pollution and Hospital Admissions for Respiratory Ds. 189 Table 4. Relative risk and 95% confidence interval of hospital admissions for respiratory disease in relation to each pollutant Pollutant Relative risk (individually) 95% CI* Relative risk (controlled for other pollutants) 95% CI* Daejeon TSP µg/m SO 2 ppb NO 2 ppb O 3 ppb CO ppm ( 10) Suwon TSP µg/m SO 2 ppb NO 2 ppb O 3 ppb CO ppm ( 10) Ulsan TSP µg/m SO 2 ppb NO 2 ppb O 3 ppb CO ppm ( 10) *RRs and their corresponding 95% CIs apply to 100 ppb (or 100 µg/m 3 (TSP), 1000 ppm (CO)) increase in the pollutant concentration in relation to the number of patients admitted for respiratory diseases. Table 5. Relative risk (RR) and 95% confidence intervals of hospital admission for respiratory disease in relation to each pollutant Air Relative risk (95% C.I.)* pollutant Daejeon Ulsan Suwon TSP µg/m ( ) 1.03 ( ) 1.05 ( ) SO 2 ppb 0.08 ( ) 0.59 ( ) 0.87 ( ) NO 2 ppb 1.02 ( ) 1.13 ( ) 0.95 ( ) O 3 ppb 1.10 ( ) 0.89 ( ) 1.02 ( ) CO ppm ( 10) 1.34 ( ) 3.55 ( ) 1.27 ( ) *RRs and their corresponding 95% CIs apply to 100 ppb (or 100 µg/m 3 (TSP), 1000 ppm (CO)) increase in the pollutant concentration in relation to the number of patients admitted for respiratory diseases. Fig. 1. Relative risk of hospital admission for respiratory diseases and TSP concentrations in Daejeon. Fig. 2. Relative risk of hospital admission and TSP concentrations in suwon.

6 190 J Occup Health, Vol. 42, 2000 Fig. 3. Relative risk of hospital admission for respiratory diseases and TSP concentrations in Ulsan. asthma than air pollutants in the gas phase. In our study, although statistically non-significant, there was a uniform association between chronic respiratory disorders and the TSP concentration, and a certain fraction of cases seemed to be attributable to TSP. Characteristically, in addition to TSP, CO showed a high association with chronic respiratory tract disease and bronchial asthma. In many other studies, the effect of CO was neglected and further investigation is warranted. In interpreting the influence of CO, confounding variables such as smoking, heating, and classes of cooking fuel etc., should be taken into consideration. CO can act as the cause of such respiratory diseases as bronchitis and asthma, but in the context of public health, it is more closely related to admission rates or mortality due to cardiovascular diseases. Recently, some studies have proposed that among the air pollution indicators, CO should be interpreted as a surrogate marker of other pollution indicators. More extensive studies should be conducted on this topic. In addition, the high level of association between TSP and SO 2 suggests that the use of diesel oil in factories and large vehicles, coupled with the intimate interdependence between TSP and NO 2, and SO 2, demonstrates that special care should be taken in excluding specific pollutants while estimating admission rates. Nevertheless, in both multiple analysis models and including all the air pollutants and quintile analysis, a uniform association between the admission rate and TSP concentration persisted. We therefore believe that TSP can be used as a representative predictive indicator amongst the air pollution indicators, but additional complementary studies are needed to confirm this hypothesis. The limitations of our study are as follows. First, there is a danger of misclassification when evaluating the exposure level, which is common to all ecologic studies. Until we discover an indicator that can directly measure the exposure of target organs to air pollutants, this will remain an unanswered question in this field of study. This is especially true in the case of NO 2, in which indoor air pollution plays a greater role, so that interpretation of the results should be reserved. Second, there is the possibility of error in the use of medical admission records. Hospitalizing, just like the state of health, can depend upon differences in culture and customary medical practices either between areas or countries and in addition, the degree of diagnostic accuracy varies among hospitals. Moreover, because health indicators also seem to be fine or sub-clinical, such as respiratory symptoms not requiring treatment, there seems to be a gap between the exposure and admission rates. We believe this error would have acted randomly, and would have had the effect of diluting the association between air pollution and respiratory illness, and therefore confirm our overall results. Third, the hospitalized patients in the selected hospitals do not truly represent the inhabitants of the area. This is especially true in this study as target hospitals were general hospitals, which opens up the possibility of an influx of population from neighboring areas. In this case, since the patients are not exposed to the measured air pollution level, they act as a bias in the study. But most of the target diseases in our study are acute illnesses, and this reduces the likely effects that this situation is likely to have on the results. Last, we included only the general hospitals of each area, the reason being that we wanted to utilize data from higher level hospitals to increase diagnostic accuracy and to increase our level of control by allowing us to collect primary data directly, but this is likely to reduce the completeness of our study, compared to those that have employed secondary data. Nevertheless, we believe that the Time series analysis technique used throughout this study resolves shortcomings caused by this lack of breadth. References 1) Her Majesty s Public Health Service. Mortality and morbidity during the London fog of December 1952: HMSO, 1954 Report No 95 in Public Health and Medical Subjects. 2) Firket M. Fog along the Meuse valley. Transaction of the Faraday Society 1936; 32: ) Shrenk HH, Heimann H, Clayton GD, et al. Air pollution in Donora PA: epidemiology of the unusual smog episode of October Preliminary report. Washington, DC: US Public Health Service, Public Health Bulletin no ) Schwarz J, Marcus A. Mortality and air pollution in London: A time series analysis. Am J Epidemiol 1990; 131: ) Sunyer J, Anto JM, Murillo C, Saez M. Effects of Urban air pollution on emergency room admissions for chronic obsructive pulmonary disease. Am J Epidemiol

7 Belong CHO, et al.: Air Pollution and Hospital Admissions for Respiratory Ds ; 134: ) Brunekreef 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): ) Krzyzanowski M. Methods for assessing the extent of exposure and effects of air pollution. Occup Environ Med 1997; 54: ) Schwartz J, Spix C, Touloumi G, et al. Methodological issues in studies of air pollution and daily counts of deaths or hospital admissions. J Epidemiol Community Health 1996; 50 (Suppl 1): S3 S11. 9) Cleveland WS, Devlin SJ. Robust locally-weighted regression and smoothing scatterplots. J Am Stat Assoc 1979; 74: ) Vigotti M.A., Rossi G., Bisanti L., et al. Short term effects of urban air pollution on respiratory health in Milan, Italy, J Epidemiol Commun H 1996; 50 (suppl): S71 S75. 11) Kuhn L, Davidson LL, Durkin MS. Use of Poisson regression and time series analysis for detecting changes over time in rates of child injury following a prevention program. Am J Epidemiol 1994; 140 (10): ) Pope CA, Respiratory hospital admission associated with PM 10 pollution in Utah, Salt Lake and Cache Valleys. Arch Environ Health 1991; 46: ) Vigotti MA, Rossi G, Bisanti L, et al. Short term effects of urban air pollution on respiratory health in Milan, Italy, J Epidemiol Commun H 1996; 50 (suppl 1): S71 S75. 14) Saldiva PHN, Pope III CA, Schwartz J, et al. Air pollution and mortality in elderly people: a time-series study in San Paulo, Brazil. Arch Environ Health 1995; 50 (2): ) Cody RP, Weisel CP. The effect of ozone associated with summertime photochemical smog on the frequency of asthma visits to hospital emergency departments. Environ Res 1992; 58: