Long Term Outdoor Air Pollution and Cause-Specific Mortality in a Pooled Analysis of Asian Cohorts
George S Downward1, Gerard Hoek1, Perry Hystad2, Eiko Sato3, Manami Inoue3, Sarah K Abe3, Md Shafiur Rahman3, Nat Rothman4, Qing Lan4, Zhu Tong5, Lutzen Portengen1, Roel Vermeulen1
1Utrecht University, Utrecht, The Netherlands; 2Oregon State University, Corvalis, Oregon, USA; 3National Cancer Center, Tokyo, Japan; 4National Cancer Institute, Bethesda, Maryland, USA; 5Peking University; Beijing, China
Background: Ambient air pollution represents a major risk factor for death and disability, being responsible for approximately 4 million deaths per year. The bulk of the research evaluating disease burdens of air pollution have been performed in North America and Europe. However, there are likely to be larger impacts of air pollution within Asia, where absolute levels of pollution are generally higher. The ability to understand this impact and burden is however complicated by a relative lack of air pollution monitoring and highly variable cultural and socio-economic differences between (and within) Asian nations. The objective of this project is to assess the association between long-term exposure to ambient PM2.5 and NO2 and all-cause/cause-specific mortality in an analysis of multiple Asian cohorts.
Methods: Cohorts have been recruited from the Asia Cohort Consortium, a multi-center consortium of cohort studies across Asia. Ambient PM2.5 and NO2 has been assigned to cohorts across India, Iran, Japan, Bangladesh, representing over 300,000 people and millions of person-years of follow-up. Ambient PM2.5 and NO2 has been assigned to the residences of study participants via the application of satellite-derived global models. The association between ambient air pollution and cause-specific mortality was assessed via Cox proportional hazard models after adjustment for common covariates including gender, smoking, and socio-economic determinants. Individual hazard ratios (HR) were pooled via random effects meta-analysis, which is being expanded by examination of additional studies utilizing the same predictive models in their work.
Results: Notable variation in patterns of PM2.5 and NO2 was observed across participating cohorts. For example, mean PM2.5¬ ranged from 7.9 µg/m3 (Taiwan) to 51 µg/m3 (Bangladesh). Individual hazard ratios similarly showed variations in findings, including both positive and null findings. For example, the adjusted all-cause mortality HR for a 5 µg/m3 increase in PM2.5 in Japan is 1.06 (CI: 1.03, 1.09) and in India it is 1.08 (1.01, 1.16), whereas in Iran it is 0.98 (0.94, 1.03). A pooled, random effects meta-analysis has identified a positive relationship between NO2 and all cancer (pooled HR for a 10 µg/m3 increase: 1.18; 95% CI: 1.13, 1.23) and lung cancer mortality (HR: 1.13, 95% CI:1.01, 1.26) as well as PM2.5 and cardiovascular mortality (HR for a 5 µg/m3 increase: 1.04, 95%CI: 1.00, 1.09).
Conclusions: Our study observed variation in exposure and risk parameters across multiple cohorts. In pooled analysis, a positive relationship with cancer, lung cancer, and cardiovascular disease was observed.