Time‐series analyses and cohort studies to investigate relationships between particulate matter and mortality—two approaches to one endpoint

Abstract

Time‐series analyses and cohort studies provide different methods of looking at the health effects of air pollution. With respect to mortality, the relative risk associated with a defined increase in the concentration of particulate matter is numerically much higher if derived from cohort studies than if based on time‐series analyses. This discrepancy is often explained by arguing that chronic effects of air pollutants can only be assessed in cohort studies. However, with respect to changes in life expectancy, the results of the two study types can be shown to be consistent. In calculations of pollution‐related health costs, mortality plays a dominant role. Applying the relative risk derived from cohort studies directly as a multiplier on daily or annual mortality leads to a considerable overestimation of effects. Copyright © 1999 John Wiley & Sons, Ltd.