Abstract

The causal interpretation of reported associations between daily air pollution and daily admissions requires consideration of residual confounding, correlation between pollutants, and effect modification. If results obtained in Hong Kong and London--which differ in climate, lifestyle, and many other respects--were similar, a causal association would be supported. We used identical statistical methods for the analysis in each city. Associations between daily admissions and pollutant levels were estimated using Poisson regression. Nonparametric smoothing methods were used to model seasonality and the nonlinear dependence of admissions on temperature, humidity, and influenza admissions. For respiratory admissions (> or = 65 years of age), significant positive associations were observed with particulate matter < 10 microm in aerodynamic diameter (PM(10), nitrogen dioxide, sulfur dioxide, and ozone in both cities. These associations tended to be stronger at shorter lags in Hong Kong and at longer lags in London. Associations were stronger in the cool season in Hong Kong and in the warm season in London, periods during which levels of humidity are at their lowest in each city. For cardiac admissions (all ages) in both cities, significant positive associations were observed for PM(10), NO(2), and SO(2) with similar lag patterns. Associations tended to be stronger in the cool season. The associations with NO(2) and SO(2) were the most robust in two-pollutant models. Patterns of association for pollutants with ischemic heart disease were similar in the two cities. The associations between O(3) and cardiac admissions were negative in London but positive in Hong Kong. We conclude that air pollution has remarkably similar associations with daily cardiorespiratory admissions in both cities, in spite of considerable differences between cities in social, lifestyle, and environmental factors. The results strengthen the argument that air pollution causes detrimental short-term health effects.