The Effect of Environmental Factors on the Association between Airborne Particulate Matter and Mortality in East Asia

Abstract

Background: Worldwide, air pollution is responsible for large numbers of deaths, and substantial epidemiological research has provided evidence for the association between air pollution and mortality. There has been growing demand from policymakers for a better understanding of the relationship between air pollution and adverse health effects, including mortality. Methods: First, we examined seasonal patterns in the short-term association of airborne particulate matter (PM) smaller than 10 μm (PM10) with daily mortality in 29 cities in three East Asian countries. Stratified time-series models were used to determine whether seasons alter the effect of PM10 on mortality. Furthermore, this effect was first quantified for each season and at each location using a time-series model, after which city-specific estimates were pooled using a hierarchical Bayesian model. Second, we investigated the effects of temperature on the relationship between PM10 and mortality due to non-accidental, cardiovascular, and respiratory death in seven cities in South Korea. We applied stratified time-series models to the datasets in order to examine whether the effects of PM10 on mortality were modified by temperature. The effect of PM10 on daily mortality was estimated for different temperature ranges at each location using a time-series model, then the estimates were pooled through a random-effects meta-analysis using the maximum likelihood method. Lastly, we estimated the durational effect on mortality of consecutive days with a daily mean PM10 concentration of ≥ 75 µg/m3. A standard time-series Poisson model was fitted in each location with duration as the main variable of interest while controlling for daily mean PM10 concentration, meteorological variables, seasonal trend, and day of the week. Moreover, the duration-mortality relationships were estimated and then a meta-analysis of the country-specific estimates was performed using the maximum likelihood method. In addition, the additional percent increase in deaths were calculated considering the consecutive days of elevated PM10 levels for each country. Findings: For seasonal analysis, a 10 µg/m3 increase in PM10 was significantly associated with increases in non-accidental mortality of 0.44% (95% confidence interval [CI]: 0.03%, 0.8%) in spring and 0.42% (95% CI: 0.02%, 0.82%) in the fall for Japan. In South Korea, a 10 µg/m3 increase in PM10 was significantly associated with increases in non-accidental mortality of 0.51% (95% CI: 0.01%, 1.01%) in summer and 0.45% (95% CI: 0.03%, 0.87%) in the fall, in cardiovascular disease mortality of 0.96% (95% CI: 0.29%, 1.63%) in the fall, and in respiratory disease mortality of 1.57% (95% CI: 0.40%, 2.75%) in the fall. In China, a 10 µg/m3 increase in PM10 was associated with increases in non-accidental mortality of 0.33% (95% CI: 0.01%, 0.66%) in summer and 0.41% (95% CI: 0.09%, 0.73%) in winter, in cardiovascular disease mortality of 0.41% (95% CI: 0.08%, 0.74%) in spring and 0.33% (95% CI: 0.02%, 0.64%) in winter, and in respiratory diseases mortality of 0.78% (95% CI: 0.27%, 1.30%) in winter. For the modifying effect of temperature, a total of 828,787 non-accidental deaths were registered from 2000-2009 from 7 cities in South Korea. The highest overall risk between PM10 and non-accidental or cardiovascular mortality was observed on extremely hot days (daily mean temperature: > 99th percentile) in individuals aged < 65 years. In those aged ≥ 65 years, the highest overall risk between PM10 and non-accidental or cardiovascular mortality was observed on very hot days (daily mean temperature: 95-99th percentile) but not on extremely hot days. There were strong harmful effects from PM10 on non-accidental mortality with the highest temperature range (> 99th percentile) in men and a very high temperature range (95-99th percentile) in women. For the durational effect, the mortality risk is significantly higher overall when the elevated PM10 concentration lasts multiple days in all three countries. Estimated non-accidental mortality was increased by 0.68% (95% CI: 0.35, 1.01) for Japan, 0.48% (95% CI: 0.30, 0.66) for South Korea, and 0.24% (95% CI: 0.14, 0.33) for China for an additional consecutive day of PM10 ≥ 75 µg/m3. For the annual maximum duration of high PM10 (≥ 75 µg/m3) in Japan (2.40 days), South Korea (6.96 days), and China (42.26 days) corresponded to increases in non-accidental death of 1.64% (95% CI: 1.31, 1.98), 3.37% (95% CI: 3.19, 3.56) and 10.43% (95% CI: 10.33, 10.54), respectively. Interpretation: Our analyses suggest that the acute effect of particulate air pollution varies seasonally and geographically, temperature affects the relationship between the PM10 levels and cause-specific mortality, and there are additional mortality effects when high PM10 levels last for several days while accounting for the effect of each days PM10 concentration. These findings have important implications for the planning of public health interventions to minimize the health burden of air pollution.