Air Pollution Epidemiology: Considerations in Time-Series Modeling

Abstract

Recent epidemiological studies have indicated that ambient air pollution, including PM-10, is associated with excess mortality and morbidity. These studies have included both cross-sectional comparisons across communities and time-series analyses over time in single communities. Time-series analysis offers certain advantages, primarily in that the study population is the same over time, so that it acts as its own “control.” However, modeling such data is complicated by the fact that other environmental factors and other causes of illness can confound the results unless they are adequately addressed. For example, wintertime influenza epidemics cause long-wave peaks in respiratory mortality, and variations in emissions, dispersion, and atmospheric chemistry can cause seasonal cycles in pollution. Such superimposed long-wave variations in both health outcomes and pollutant concentrations can undermine the statistical validity of time-series models by inducing autocorrelation, and can create long-wave “noise” signals that can overwhelm a short-term “signal” of interest. Also, model specification can strongly affect the results of a time-series model. For example, analyses focusing on only one routinely collected pollution metric, to the exclusion of other possibly more influential pollution components, can cause the effects of the overlooked pollutants to be ascribed to the studied pollutant. In addition, the potential effects of nonnormal (e.g., Poisson) data distributions on time-series results need to be considered. It is concluded that how these various time-series modeling factors are, or are not, addressed can have a large influence on the study conclusions, or the “message” resulting from such analyses. Sensitivity analyses incorporating multiple modeling methods and model specifications are therefore recommended as part of such an analysis. Moreover, in this article exploratory and diagnostic procedures are recommended that may aid the modeler in assessing and avoiding the noted problems and that will allow the validity of such studies to be more easily documented and intercompared.