One motivation employed by us and others (e.g. Millimet and Maasoumi "Robust Inference Concerning Recent Trends in U.S. Environmental Quality,” Journal of Applied Econometrics 2005) is that there exists a "threshold" type effect where increasing concentrations have a non-linear effect on health.
The following planetark article therefore makes interesting reading:
NEW YORK - The risk of death due to stroke is associated with exposure to high concentrations of air pollution about 2 hours before death, Japanese investigators report.
Because this risk appears to be independent of 24-hour particulate matter levels, they suggest that air quality standards be based on hourly data, as well as 24-hour levels. Particulate matter, or PM, is the term used to describe the tiny particles emitted by automobiles, especially diesel vehicles.
Dr. Shin Yamazaki, an epidemiologist at Kyoto University, and associates collected data from the 13 largest cities in Japan regarding concentrations of suspended particulate matter 7 m diameters or higher (PM7), ambient temperature, plus other components of air pollution, from January 1990 to December 1994.