Abstract |
Epidemiological data have often had a key role in the assessment of risks associated with exposure to chemicals and pollutants and for development of regulatory limits in the environment and in occupational settings. Examples include the development of occupational health standards and health risk values in the United States for benzene (OSHA 1987; U.S. EPA 2003), as well as for the U.S. National Ambient Air Quality Standards for Particulate Matter (U.S. EPA 2006). The strengths and weaknesses, as well as the overall value of the use of epidemiological data in regulatory risk assessment have been widely discussed in the scientific and public health policy literature. Epidemiological data have been criticized as too often flawed by quality issues and incompletely controlled sources of bias (see, for example, Graham et al. 1995). Epidemiological studies of environmental agents involving typical ambient levels of exposure have been particularly characterized as uninformative or especially susceptible to bias and uncontrolled confounding because the target for estimation is often relatively small risk ratios that are dismissed as 'weak associations' (Gamble and Lewis 1996). Proponents of use of epidemiological data, while acknowledging the limitations of observational studies, advance its strengths; the investigation of the effects of real exposures as received by the general population, the characterization of an effect across the full range of susceptibility in the population and, most significantly, the direct relevance of epidemiologic evidence to public health (Whittemore 1986; Gordis 1988; Hertz-Picciotto 1995; Burke 1995; Samet, Schnatter, and Gibb 1998). In addition, the ability to ascertain relatively low relative risks has improved with advances in exposure assessment and study design methodologies. |