Short term variations in hospital admissions and mortality and particulate air pollution

Objectives--To determine the presence and magnitude of any relation between short term variations in ambient concentrations of particulate matter under 10 mu m in diameter (PM(10)) and hospital admissions and mortality in Birmingham, United Kingdom. To find the relative risk associated with various concentrations of PM(10), and to estimate the potential public health benefit of reducing PM(10) to below various thresholds. Methods--Retrospective ecological study. Air pollution data were taken from a national network monitoring station between 1 April 1992 and 31 March 1994, and weather data for the same period from the University of Birmingham Weather Service. Daily total hospital admissions for the same period for asthma, bronchitis, pneumonia, chronic obstructive pulmonary disease (COPD), acute ischaemic heart disease, acute cerebrovascular disease, all respiratory conditions, and all circulatory conditions were obtained from the West Midlands Regional Health Authority, as well as daily total deaths from 1 April 1992 to 31 December 1994 for chronic obstructive pulmonary disease, pneumonia, all respiratory diseases, all circulatory diseases, and all causes. Multiple linear regression models were constructed after adjusting for confounding factors (day of week, month, linear trend, relative humidity, and temperature). Relative risk of admission at various thresholds of PM(10) was calculated with the model, by comparing risk of admission over the threshold with mean risk of admission over the whole period. Potential public health benefits at various thresholds were calculated with the model to predict the number of admissions of deaths that could be saved if, on each day that the PM(10) had exceeded that threshold, it had instead been kept at the threshold level. Results--Significant associations were found between all respiratory admissions, cerebrovascular admissions, and bronchitis admissions and PM10 on the same day. Pneumonia, all respiratory admissions, and asthma admissions were significantly associated with the mean PM(10) values for the past three days. Deaths from COPD, all circulatory deaths, and all causes mortality were significantly associated with PM(10) 24 hours previously, and COPD deaths also with PM,, on the same day. The effect of a 10 mu g/m(3) rise in PM(10) was estimated to represent a 2.4% increase in respiratory admissions, a 2.1% increase in cerebrovascular admissions, and a 1.1% increase in all causes mortality. In a population of 1 million, this would represent 0.5 extra respiratory admissions and 0.3 extra deaths. The increase in relative risk was linear without evidence of a threshold. The impact of reducing PM(10) to below 70 mu g/m(3) would be small, representing less than 0.1% of respiratory admissions and 0.2% all causes mortality. The impact would be greater at lower thresholds. Conclusion--Ambient outdoor concentrations of PM(10) in the United Kingdom are significantly associated with several indicators of acute health effect. These associations are similar to and consistent with other studies. However, the estimated size of the public health effect is small, accounting for only a small proportion of hospital admissions and mortality over a two year period.