Why do estimates of the acute and chronic effects of air pollution on mortality differ?

Several investigators have discussed various reasons for the large and consistent difference in the relative risk estimates per unit exposure to air pollution derived from time-series and cohort study designs. This article considers whether the two study designs are estimating fundamentally different parameters even though both are commonly expressed in terms of relative risk coefficients. A frailty model for air pollution mortality is proposed to illustrate some of the theoretical behaviors of heterogeneous populations exposed to time-varying pollution patterns and their implications for the kinds of summary measures widely used in air pollution risk assessment.