Part 5. Public health and air pollution in Asia (PAPA): a combined analysis of four studies of air pollution and mortality

BACKGROUND: In recent years, Asia has experienced rapid economic growth and a deteriorating environment caused by the increasing use of fossil fuels. Although the deleterious effects of air pollution from fossil-fuel combustion have been demonstrated in many Western nations, few comparable studies have been conducted in Asia. Time-series studies of daily mortality in Asian cities can contribute important new information to the existing body of knowledge about air pollution and health. Not only can these studies verify important health effects of air pollution in local regions in Asia, they can also help determine the relevance of existing air pollution studies to mortality and morbidity for policymaking and environmental controls. In addition, the studies can help identify factors that might modify associations between air pollution and health effects in various populations and environmental conditions. Collaborative multicity studies in Asia-especially when designed, conducted, and analyzed using a common protocol-will provide more robust air pollution effect estimates for the region as well as relevant, supportable estimates of local adverse health effects needed by environmental and public-health policymakers.

SPECIFIC OBJECTIVES: The Public Health and Air Pollution in Asia (PAPA*) project, sponsored by the Health Effects Institute, consisted of four studies designed to assess the effects of air pollution on mortality in four large Asian cities, namely Bangkok, in Thailand, and Hong Kong, Shanghai, and Wuhan, in China. In the PAPA project, a Common Protocol was developed based on methods developed and tested in NMMAPS, APHEA, and time-series studies in the literature to help ensure that the four studies could be compared with each other and with previous studies by following an established protocol. The Common Protocol (found at the end of this volume) is a set of prescriptive instructions developed for the studies and used by the investigators in each city. It is flexible enough to allow for adjustments in methods to optimize the fit of health-effects models to each city's data set. It provides the basis for generating reproducible results in each city and for meta-estimates from combined data. By establishing a common methodology, factors that might influence the differences in results from previous studies can more easily be explored. Administrative support was provided to ensure that the highest quality data were used in the analysis. It is anticipated that the PAPA results will contribute to the international scientific discussion of how to conduct and interpret time-series studies of air pollution and will stimulate the development of high-quality routine systems for recording daily deaths and hospital admissions for time-series analysis.

METHODS: Mortality data were retrieved from routine databases with underlying causes of death coded using the World Health Organization (WHO) International Classification of Diseases, 9th revision or 10th revision (ICD-9, ICD-10). Air quality measurements included nitrogen dioxide (NO2), sulfur dioxide (SO2), particulate matter with aerodynamic diameter < or = 10 microm (PM10), and ozone (O3) and were obtained from several fixed-site air monitoring stations that were located throughout the metropolitan areas of the four cities and that met the standards of procedures for quality assurance and quality control carried out by local government units in each city. Using the Common Protocol, an optimized core model was established for each city to assess the effects of each of the four air pollutants on daily mortality using generalized linear modeling with adjustments for time trend, seasonality, and other time-varying covariates by means of a natural-spline smoothing function. The models were adjusted to suit local situations by correcting for influenza activity, autocorrelation, and special weather conditions. Researchers in Hong Kong, for example, used influenza activity based on frequency of respiratory mortality; researchers in Hong Kong and Shanghai used autoregressive terms for daily outcomes at lag days; and researchers in Wuhan used additional smoothing for periods with extreme weather conditions.

RESULTS AND DISCUSSION: For mortality due to all natural (nonaccidental) causes at all ages, the effects of air pollutants per 10-microg/m3 increase in concentration was found to be higher in Bangkok than in the three Chinese cities, with the exception of the effect of NO2 in Wuhan. The magnitude of the effects for cardiovascular and respiratory mortality were generally higher than for all natural mortality at all ages. In addition, the effects associated with PM10 and O3 in all natural, cardiovascular; and respiratory mortality were found to be higher in Bangkok than in the three Chinese cities. The explanation for these three findings might be related to consistently higher daily mean temperatures in Bangkok, variations in average time spent outdoors by the susceptible populations, and the fact that less air conditioning is available and used in Bangkok than in the other cities. However, when pollutant concentrations were incorporated into the excess risk estimates through the use of interquartile range (IQR), the excess risk was more comparable across the four cities. We found that the increases in effects among older age groups were greater in Bangkok than in the other three cities. After excluding data on extremely high concentrations of PM10 in Bangkok, the effect estimate associated with PM10 concentrations decreased in Bangkok (suggesting a convex relationship between risk and PM10, where risk levels off at high concentrations) instead of increasing, as it did in the other cities. This leveling off of effect estimates at high concentrations might be related to differences in vulnerability and exposure of the population to air pollution as well as to the sources of the air pollutant. IMPLICATIONS OF THE STUDY: The PAPA project is the first coordinated Asian multicity air pollution study ever published; this signifies the beginning of an era of cooperation and collaboration in Asia, with the development of a common protocol for coordination, data management, and analysis. The results of the study demonstrated that air pollution in Asia is a significant public health burden, especially given the high concentrations of pollutants and high-density populations in major cities. When compared with the effect estimates reported in the research literature of North America and Western Europe, the study's effect estimates for PM10 were generally similar and the effect estimates for gaseous pollutants were relatively higher. In Bangkok, however, a tropical city where total exposures to outdoor pollution might be higher than in most other cities, the observed effects were greater than those reported in the previous (i.e., Western) studies. In general, the results suggested that, even though social and environmental conditions across Asia might vary, it is still generally appropriate to apply to Asia the effect estimates for other health outcomes from previous studies in the West. The results also strongly support the adoption of the global air quality guidelines recently announced by WHO.