点击显示 收起
Adrian G. Barnett, Gail M. Williams, Joel Schwartz, Anne H. Neller, Trudi L. Best, Anna L. Petroeschevsky and Rod W. Simpson
School of Population Health, University of Queensland, Herston; Faculty of Science, Health and Education, University of the Sunshine Coast, Maroochydore DC, Queensland, Australia; and Exposure, Epidemiology, and Risk Program, Harvard School of Public Health, Harvard University, Boston, Massachusetts
ABSTRACT
Rationale: The strength of the association between outdoor air pollution and hospital admissions in children has not yet been well defined. Objectives: To estimate the impact of outdoor air pollution on respiratory morbidity in children after controlling for the confounding effects of weather, season, and other pollutants. Methods: The study used data on respiratory hospital admissions in children (three age groups: < 1, 1eC4, and 5eC14 years) for five cities in Australia and two in New Zealand. Time series of daily numbers of hospital admissions were analyzed using the case-crossover method; the results from cities were combined using a random-effects meta-analysis. Measurements and Main Results: Significant increases across the cities were observed for hospital admissions in children for pneumonia and acute bronchitis (0, 1eC4 years), respiratory disease (0, 1eC4, 5eC14 years), and asthma (5eC14 years). These increases were found for particulate matter with a diameter less than 2.5 e蘭 (PM2.5) and less than 10 e蘭 (PM10), nephelometry, NO2, and SO2. The largest association found was a 6.0% increase in asthma admissions (5eC14 years) in relation to a 5.1-ppb increase in 24-hour NO2. Conclusions: This study found strong and consistent associations between outdoor air pollution and short-term increases in childhood hospital admissions. A number of different pollutants showed significant associations, and these were distinct from any temperature (warm or cool) effects.
Key Words: air pollutants Australasia meta-analysis respiration disorders
Many studies have found associations between selected air pollutants and adverse health effects in children (1). These adverse health effects include the following: childhood hospital admissions (2eC6), school absences (7), physician visits for upper and lower respiratory illness (8), deficits in lung function growth rates (9), bronchitis and chronic cough (10, 11), and increased infant mortality (12eC14). A recent review in Europe strongly recommended a reduction in children's exposure to air pollution (15). However, the strength of the association is still not well defined because of the small number of studies of hospital admissions (which represent a common and serious outcome in children), and the complexity of the time-series modeling. In addition, there are few studies that have been able to examine a range of pollutants. When multiple pollutants have been examined, the independent effect of each pollutant is usually addressed in multipollutant models, but these are sensitive to the assumptions inherent in the time-series models. If the association with one pollutant is nonlinear, or varies by season, then a two-pollutant model assuming a linear relationship with each pollutant might not give the independent effect of the second pollutant. This suggests that an approach less sensitive to model assumptions is desirable.
This study used the case-crossover method to examine the effects of air pollution exposure on daily counts of hospital admissions for children. The case-crossover design (16) is a method for investigating the effects of acute exposures that can also examine multiple exposures and interactions between exposures. This approach has been applied to the analysis of the acute effects of environmental exposures, especially air pollution (17eC19), and additional detail on this method is provided in the online data supplement.
Many of the previous studies investigating the health effects of air pollution in children used an age group of 0 to 14 years. However, this combines infants and teenage children, two groups that have a very different lung function and immune system, and that are also in very different environments (prior to school and school). This study examined three age bands: 0, 1 to 4, and 5 to 14 years.
The emphasis here is on identifying which air pollutants have significant impacts on the respiratory health of children using hospital admissions data, and on assessing the consistency of such associations across cities in Australia and New Zealand. Some of the results of this study have been previously reported in the form of an abstract (20).
METHODS
Respiratory Health Data and Air Pollutant Data
Daily hospital and pollution data were collected for the period 1998eC2001 in five of the largest cities in Australia (Brisbane, Canberra, Melbourne, Perth, and Sydney) and the two largest cities in New Zealand (Auckland, Christchurch). In 2001, these cities comprised 53% of the Australian population and 44% of the New Zealand population.
Health data for all respiratory admissions of children aged 14 years or younger were collected from state government health departments in Australia and the New Zealand Health Information Service (Ministry of Health), and details, including International Classification of Disease codes, are provided in the online supplement.
The air pollutants considered were particulate matter less than 2.5 e蘭 in diameter (PM2.5; e蘥 · meC3) and less than 10 e蘭 in diameter (PM10; e蘥 · meC3), coefficient of light-scattering by nephelometry (an indicator of fine particles 0.1eC2 e蘭 in diameter ; in 10eC4 · meC1), nitrogen dioxide (NO2; in ppb), carbon monoxide (CO; in ppm), sulfur dioxide (SO2; in ppb), and ozone (O3; in ppb). Additional information on how these pollutants were measured is provided in the online supplement.
Statistical Methods
The case-crossover method controlled for long-term trend, seasonal changes, and respiratory epidemics by design. Matched case-crossover analyses were also used to investigate whether some pollutant effects were related to those of other pollutants (21). The fixed 28-day-window case-crossover approach was used, with a 1-day exclusion period around the case day (22). Using covariates, there were additional controls for the following: temperature, current minus previous day's temperature, relative humidity, pressure, extremes of hot and cold (coldest and warmest 1% of days), day of the week, public holiday (yes/no), and day after a public holiday (yes/no).
To estimate the average effect for all cities, the estimates were combined across cities using a random effects meta-analysis (23), and the differences (heterogeneity) between cities were quantified using the I-squared statistic (24). Analyses were conducted using the SAS package (SAS System for Windows, version 8; SAS Institute, Inc., Cary, NC).
In the absence of an a priori opinion of which pollutants were important to child health in Australia, we used a statistical significance level of 5% with no correction for multiple comparisons. Although this increased the chances of finding spurious associations, it reduced the chances of missing any important associations during this early stage of investigation of the effects of air pollution in Australia and New Zealand.
To test whether one city had an undue influence on the meta-analysis, the meta-analyses were repeated with each city left out in turn (a leave-one-city-out sensitivity analysis) (23).
RESULTS
Summary statistics of hospital admissions and demographic data for each city are given in Table 1. Summary statistics of the air pollutant and meteorology data in each city are given in Table 2.
Pollutant exposures used were the average of the current and previous day. Estimates of the percentage increase in morbidity are shown for an interquartile range increase (using the mean interquartile range across cities). This made the increases from different pollutants more comparable because the results showed the changes to be expected for the cities under study, and allowed the largest impacts to be identified.
The statistically significant increases in hospital admissions for all cities are shown in Table 3, together with the estimated differences (heterogeneity) between cities and the leave-one-city-out sensitivity analyses (see online Tables E1eCE3 for a complete set of results). Statistically significant increases were found for PM2.5, PM10, bsp, NO2, and SO2, but not for CO or O3. Results are shown for an interquartile increase to facilitate comparisons across pollutants. These interquartile ranges were as follows: 3.8 e蘥 · meC3 for 24-hour PM2.5, 7.5 e蘥 · meC3 for 24-hour PM10, 0.18 10eC4 · meC1 for 24-hour bsp, 9.0 ppb for 1-hour NO2, 5.1 ppb for 24-hour NO2, 5.4 ppb for 1-hour SO2, and 9.8 ppb for 1-hour O3.
As with most cities around the world, there were strong correlations between some pollutants (see Table E4), because they often arise from the same emission source (e.g., motor vehicles). Given these correlations, matched pollutant models were run where significant increases were found with more than one pollutant (Table 4) to identify whether the pollutant impacts are different or are related to each other (or some other pollutant arising from the same emission sources). Matched control days were defined as follows: 24-hour PM2.5 within 2 e蘥 · meC3, 24-hour PM10 within 3 e蘥 · meC3, 24-hour NO2 within 1 ppb, 1-hour SO2 within 1 ppb, and temperature within 1°C.
Differences between the Cities and Countries
The I-squared statistic represents the proportion of total variation in the estimated increase that is caused by heterogeneity between cities and was used to identify any differences between cities. This measure allowed the estimation of the average effect of an air pollutant on hospital admissions using all the data, and to identify if this effect is the same across all cities (positive or negative) or whether the results for some cities are different from others. In this way, we can conclude whether any identified impact is the same for all cities or whether there are different results arising from factors not identified in the analyses. The I-squared statistic was generally very low, except for NO2 (Table 3), where all seven cities were available for analysis. The contrasting increases between cities are apparent in Figure 1, which shows three of the significant meta-analysis increases for respiratory admissions (age groups: 0, 1eC4, and 5eC14 years).
To look for differences between countries, separate meta-analyses were run for the Australian and New Zealand cities. The differences in the associations with NO2 between the five Australian cities and the two New Zealand cities are shown in Table 5. The increases are quite different, but clearly not all of the heterogeneity between the increases for all the cities is caused by the difference in countries.
Multiple pollutant models showed that the results for NO2 were often independent of the effects of other pollutants, although some impacts caused by particles and SO2 could not be separated from those found for NO2.
Differences by Season
To test whether any pollutant effects were dependent on season, case-crossover analyses were separated into a (southern hemisphere) cool season of May to October, and a warm season of November to April. The significant effects of pollutants were analyzed separately for cool and warm seasons (Table 6), because some pollutants (e.g., O3) peak significantly in the warm periods (because of the formation of photochemical smog) and these "smog" impacts may affect the results. The increase in respiratory admissions in the younger-than-1-year age group associated with bsp (Australian cities only) was a cool-season effect, indicating these results are not caused by summer smog episodes. However, the increases in respiratory admissions in the 1- to 4-year age group that were associated with PM2.5 and PM10 were warm-season effects, and the increases in respiratory admissions in the 5- to 14-year age group associated with NO2 were larger in the warm season.
Effect Modification
Where differences in impacts were found for different cities, possible reasons for this were explored by examining whether the effect estimates for each city were related to different city characteristics. These reasons may include some cities having more pollution than others, having a larger proportion of children younger than 15 years, or being hotter (or colder) than others. A hierarchic model in which the increases in admissions in each city were regressed against potential city-level effect modifiers was used to incorporate variables that may differ between cities and therefore modify the results (effect modifiers). The effect modifiers examined included the following: average pollutant level, number of monitors, temperature, and percentage of the population younger than 15 years. Statistical significance was assessed using the estimate of the regression slope. The only differences between the cities that could be related to the different effect estimates were related to climate. The only significant effect modification found was that cities with higher average temperatures had greater increases in hospital respiratory admissions in the 1- to 4-year age group for increases in 1-hour NO2.
DISCUSSION
This study has shown statistically significant relationships between outdoor air pollution and child health in Australia and New Zealand. Levels of air pollution in these two countries are generally low compared with countries in Europe or North America where similar associations have been found (1).
When studying the health effects of a hazardous exposure, children are often considered as if they were small adults; however, children represent the largest subpopulation susceptible to the adverse health effects of air pollution (25). Infants and children inhale and retain larger amounts of air pollution per unit of body weight than adults; the air intake of a resting infant is twice that of an adult. As children grow, their organ systems are still developing, and their normal growth may be affected when exposed to pollutants at critical periods (25). Children also spend more time outdoors than adults, and concentrations of pollutants of an ambient origin are higher outdoors than indoors. Children playing outdoors also engage in exercise that increases ventilation. This is particularly true in afternoons, when photochemical pollutant concentrations (O3 and particulate sulfate) are highest. This study found an association with respiratory admissions in the 1- to 4-year age group and 1-hour ozone exposure during warm months.
We can only identify associations here, not causes, but other studies have reviewed the air pollution effects on human respiratory diseases (26eC28). It has been suggested that NO2 effects arise from the NO2 exposure, making people more susceptible to respiratory viral infections that exacerbate asthma (27), and that the direct effects of NO2 on health are less important than its role as a precursor to the onset of photochemical smog leading to the formation of ozone and secondary particles. SO2 effects have been related to decreases in pulmonary function in controlled human exposure studies (28) but may also contribute to acid sulfate aerosol formation (26, 28). Diesel exhaust particles have been found to increase airway inflammation and exacerbate asthma (28), and there have been a number of studies on the inflammatory impacts of O3 on respiratory disease (28). The emissions from motor vehicles increase the concentrations of all the pollutants under study, making it difficult to separate the differing impacts.
Pollutants Confounding Each Other
If a health outcome was significantly associated with more than one pollutant, there was always the question as to whether one pollutant effect is actually showing the impact of another with which it is correlated. To examine whether some of the impacts were interrelated, multipollutant models were run using a matched case-crossover method, a traditional approach to control for potential confounding in epidemiology. By choosing control days for each subject that are both nearby in time and also matched on the level of another pollutant, the effect estimate cannot be confounded by the other pollutant (21).
For respiratory admissions in the 5- to 14-year age group, the significant association with PM10 disappeared after matching on NO2, indicating that this result could not be separated from that for NO2. However, the association with NO2 remained after matching on PM10. Similar results were found in a southern California study, which concluded that NO2 deserved greater attention as a potential cause of the chronic symptoms of bronchitis in children with asthma (29). For respiratory admissions in the 1- to 4-year age group, the effects of both PM10 and PM2.5 (only four Australian cities involved) became larger when matched with each other, indicating separate effects. Also, the PM2.5 and SO2 impacts could not be separated and may indicate pollution from the same emission source. None of the significant associations found were confounded by temperature, because associations matched to within 1°C of temperature changed very little when compared with the unmatched result. Hence, the pollutant impacts are not related to temperature effects but are separate and different impacts. However, both of the associations with PM10 increased by approximately 50% when matched on temperature.
Heterogeneity and Comparison between Australian and New Zealand Cities
The differences (heterogeneity) between cities for the significant associations were low except for associations with NO2 (although the number of cities involved for this pollutant was always larger than the others). Figure 1 shows that the four largest Australian cities (Brisbane, Melbourne, Perth, and Sydney) usually showed different results to the New Zealand cities and Canberra (the coldest Australian city). The only significant effect modification found was that cities with higher average temperatures had greater increases in hospital respiratory admissions in the 1- to 4-year age group for increases in 1-hour NO2. The average temperatures of Christchurch and Canberra are lower than the other cities (Table 2). Auckland's average temperature is slightly higher than Melbourne's, but its maximum is much lower. There is evidence of seasonal impacts on asthma and respiratory admissions (5- to 14-year age group) for NO2; on respiratory admissions (1- to 4-year age group) for PM2.5, PM10, and O3; and for pneumonia and acute bronchitis admissions (1- to 4-year age group) for PM2.5 and SO2.
In examining the effects of pollution, it is important to be able to identify the sources of pollutants. Most of the nitrogen oxide pollution in Australian and New Zealand cities comes from motor vehicle exhausts in summer and a combination of motor vehicle exhausts and home heating in winter. Diurnal patterns for NO2 show the morning peak-hour traffic times to be particularly notable, especially on weekdays. Also, the NO2 levels are generally much higher in winter, especially at night, and the differences between summer and winter are much more marked in the New Zealand cities than in most of the Australian cities.
A recent meta-analysis of three European studies of respiratory admissions in children aged 0 to 14 years reported a 1.0% increase in admissions for a 10 e蘥 · meC3 increase in PM10 (95% confidence interval, eC0.2, 2.1%) (30). In this study, we found a relationship between PM10 and respiratory admissions in the 5- to 14-year age group of 1.9% (95% confidence interval, 0.1, 3.8) for a 7.5 e蘥 · meC3 increase in PM10.
Unfortunately, not all pollutants were available in all cities. This reduced the power for some comparisons, and means that comparisons across pollutants are also (to a varying degree) comparisons across cities.
Conclusions
This study has shown strong and consistent associations between children's hospital admissions and outdoor air pollutants in the urban centers of Australia and New Zealand for NO2, particles, and SO2. The largest association found in this study for all cities was a 6.0% increase in asthma admissions in the 5- to 14-year age group related to a 5.1-ppb increase in 24-hour NO2. These impacts also appeared to be distinct from any temperature (warm or cold) effects. NO2 effects were often the strongest and appeared to be generally independent of the impact of other pollutants. However, there were differences in the impacts between cities with different climates (impacts were higher in the warmer cities). There were also differing impacts in summer and winter because of summer smog events.
Acknowledgments
The authors acknowledge the contributions made to this project by Queensland Health, NSW Health, EPA Victoria, Western Australian Department of Environmental Protection, Environment ACT, New Zealand Ministry for the Environment, Auckland Regional Council, Environment Canterbury, and the project steering committee. Daily weather data were provided by the Australian Bureau of Meteorology and the New Zealand National Climate Database. This study used monitoring data provided by the relevant monitoring agency in each city. The datasets have been used without extensive analysis or corrections beyond the basic quality control needed to ensure data validity for the case-crossover analysis. Some datasets have not been fully used (e.g., the PM10 data from Auckland), as they did not fully meet the strict requirements of the study, but they are still regarded as valid datasets for the purposes for which they were gathered.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
REFERENCES
Schwartz J. Air pollution and children's health. Pediatrics 2004;113:1037eC1043.
Pope CA III. Respiratory hospital admissions associated with PM10 pollution in Utah, Salt Lake, and Cache Valleys. Arch Environ Health 1991;46:90eC97.
Pope CA III. Respiratory disease associated with community air pollution and a steel mill, Utah Valley. Am J Public Health 1989;79:623eC628.
Braga A, Conceicao G, Pereira L, Kishi H, Pereira J, Andrade M, Goncalves F, Saldiva P, Latorre M. Air pollution and pediatric respiratory hospital admissions in Sao Paulo, Brazil. J Environ Med 1999;1:95eC102.
Chew F, Goh D, Ooi B, Saharom R, Hui J, Lee B. Association of ambient air-pollution levels with acute asthma exacerbation among children in Singapore. Allergy 1999;54:320eC329.
Atkinson R, Anderson H, Sunyer J, Ayres J, Baccini M, Vonk J, Boumghar A, Forastiere F, Forsberg B, Touloumi G, et al. Acute effects of particulate air pollution on respiratory admissions: results from APHEA 2 project. Am J Respir Crit Care Med 2001;164:1860eC1866.
Ransom M, Pope CA III. Elementary school absences and PM10 pollution in Utah Valley. Environ Res 1992;58:204eC219.
Ostro B, Eskeland G, Sanchez J, Feyzioglu T. Air pollution and health effects: a study of medical visits among children in Santiago, Chile. Environ Health Perspect 1999;107:69eC73.
Gauderman W, McConnell R, Gilliland F, London S, Thomas D, Avol E, Vora H, Berhane K, Rappaport E, Lurmann F, et al. Association between air pollution and lung function growth in southern California children. Am J Respir Crit Care Med 2000;162:1383eC1390.
Braun-Fahrlander C, Vuille JC, Sennhauser FH, Neu U, Kunzle T, Grize L, Gassner M, Minder C, Schindler C, Varonier HS, et al. Respiratory health and long-term exposure to air pollutants in Swiss schoolchildren. SCARPOL Team. Swiss Study on Childhood Allergy and Respiratory Symptoms with Respect to Air Pollution, Climate and Pollen. Am J Respir Crit Care Med 1997;155:1042eC1049.
Dockery D, Speizer F, Stram D, Ware J, Spengler J, Ferris B Jr. Effects of inhalable particles on respiratory health of children. Am Rev Respir Dis 1989;139:587eC594.
Loomis D, Castillejos M, Gold D, McDonnell W, Borja-Aburto V. Air pollution and infant mortality in Mexico City. Epidemiology 1999;10:118eC123.
Ha E, Lee J, Kim H, Hong Y, Lee B, Park H, Christiani D. Infant susceptibility of mortality to air pollution in Seoul, South Korea. Pediatrics 2003;111:284eC290.
Woodruff T, Grillo J, Schoendorf K. The relationship between selected causes of postneonatal infant mortality and particulate air pollution in the United States. Environ Health Perspect 1997;105:608eC612.
Binkove?B, Bobak M, Chatterjee A, Chauhan A, Dejmek J, Dockery D, Everard M, Forastiere F, Gilliand F, Holgate S, et al. The effects of air pollution on children's health and development: a review of the evidence. Geneva, Switzerland: World Health Organization; 2004. WHO Monograph.
Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol 1991;133:144eC153.
Neas L, Schwartz J, Dockery D. A case-crossover analysis of air pollution and mortality in Philadelphia. Environ Health Perspect 1999;107:629eC631.
Sunyer J, Schwartz J, Tobias A, Macfarlane D, Garcia J, Anto JM. Patients with chronic obstructive pulmonary disease are at increased risk of death associated with urban particle air pollution: a case-crossover analysis. Am J Epidemiol 2000;151:50eC56.
Bateson T, Schwartz J. Selection bias and confounding in case-crossover analyses of environmental time-series data. Epidemiology 2001;12:654eC661.
Simpson R, Williams G, Barnett A, Neller A, Best T, Petroeschevsky A, Denison L, Neville G, Sheppeard V, Runnion T. Multi-city studies in Australia and New Zealand: the impact of air pollution on daily mortality and morbidity . Epidemiology 2004;106:S54.
Schwartz J. Is the association of airborne particles with daily deaths confounded by gaseous air pollutants An approach to control by matching. Environ Health Perspect 2004;112:557eC561.
Levy D, Lumley T, Sheppard L, Kaufman J, Checkoway H. Referent selection in case-crossover analyses of acute health effects of air pollution. Epidemiology 2001;12:186eC192.
Normand S-LT. Meta-analysis: formulating, evaluating, combining, and reporting. Stat Med 1999;18:321eC359.
Higgins J, Thompson S. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539eC1558.
Mathieu-Nolf M. Poisons in the air: a cause of chronic disease in children. J Toxicol Clin Toxicol 2002;40:483eC491.
Katsouyanni K. Ambient air pollution and health. Br Med Bull 2003;68:143eC156.
Chauhan A, Johnston S. Air pollution and infection in respiratory illness. Br Med Bull 2003;68:95eC112.
Bernstein J, Alexis N, Barnes C, Bernstein L, Nel A, Peden D, Diaz-Sanchez D, Tarlo S, Williams P. Health effects of air pollution. J Allergy Clin Immunol 2004;114:1116eC1123.
McConnell R, Berhane K, Gilliland F, Molitor J, Thomas D, Lurmann F, Avol E, Gauderman W, Peters J. Prospective study of air pollution and bronchitic symptoms in children with asthma. Am J Respir Crit Care Med 2003;168:790eC797.
Anderson H, Atkinson R, Peacock J, Marston L, Konstantinou K. Meta-analysis of time-series studies and panel studies of particulate matter (PM) and ozone (O3). Geneva, Switzerland: World Health Organization; 2004. EUR/04/5042688.