The search for non-linear exposure-response relationships at ambient levels in environmental epidemiology.

Environmental exposures to ambient air particulate matter (PM), ozone (O(3)), environmental tobacco smoke (ETS), and to dioxin and related compounds are of considerable public health concern, and risk assessments for them have generally been based on linear, non-threshold models derived from epidemiological study data. While the epidemiological databases for PM, O(3), and ETS have been sufficient to show that adverse health effects are occurring, the relative risks have been quite low, and it has not been possible, to date, to identify thresholds or non-linear relationships for them. For dioxin and related compounds, the evidence for excess cancer risks has been inadequate to establish causality, and there is suggestive evidence that hormesis may have occurred.