Urban air pollution and cardiopulmonary ill health

OBJECTIVES: To examine possible associations between daily concentrations of urban air pollutants and hospital emergency admissions and mortality due to cardiac and pulmonary disease. METHODS: A time series study was conducted in the City of Edinburgh, which has a population of about 450,000. Poisson log linear regression models were used to investigate the relation of the daily event rate with daily air pollution concentrations of sulphur dioxide (SO2) and black smoke from 1981 to 1995, and of nitrogen dioxide (NO2), ozone (O3), carbon monoxide (CO), and particulate matter (PM10) from 1992 to 1995. Adjustments were made for seasonal and weekday variation, daily temperature, and wind speed. RESULTS: The most significant findings were positive associations over the period 1981-95 between black smoke as a mean of the previous three days and daily all cause mortality in people aged > or = 65, and respiratory mortality also in this age group (3.9% increase in mortality for a 10 micrograms/m3 increment in black smoke). For hospital emergency admissions between 1992 and 1995 the two most significant findings (p < 0.05) were for cardiovascular admissions of people aged > or = 65 which showed a positive association with PM10 as a mean of the 3 previous days, and a negative association with O3 as a mean of the previous three days. Analyses of outcomes based on linkage with previous cardiorespiratory emergency admissions did not show substantially different results. CONCLUSION: These data suggest that in the City of Edinburgh, after correction for confounders, there was a small but significant association between concentrations of black smoke and respiratory mortality in the older age group, probably attributable to higher pollution levels in the early part of the study period. There were also generally weak and variable associations between day to day changes in concentrations of urban air pollutants at a single central point and emergency hospital admission rates from cardi