ABSTRACT

Daily time-series studies show that exposure-response relationships for the health effects of air pollutants are often observed at low concentrations, so that setting air quality objectives (AQO) on the basis of a risk free level is unrealistic.To develop a method of modeling daily time-series which could present decision-makers with choices for setting AQOs. Air pollutant concentrations were first transformed to a percentile scale and were modeled with daily counts of hospital admissions for cardio-respiratory diseases. A linear exposure-response curve with a gradient ß in terms of change in hospital admissions per 100 percentile of air pollutant concentration was derived. In setting an AQO at a concentration level corresponding to a percentile ±, an expected number of days exceeding AQO is expected to be 100(1–±)% and reduction in the number of hospitalizations is estimated to be 0.5 ß (1–± ² ).We illustrate that ± is a reference value in terms of the number of days exceeding the AQO for any given level of air quality, which is also directly related to the costs of pollution to government and society in terms of action needed to prevent exceedances of AQO and the benefits of avoiding illness episodes and premature deaths. Modeling daily health outcomes from local studies on the percentile values of pollutant concentrations provides a cost-benefit approach to setting AQO and a framework for surveillance.