Approaches to Uncertainty in Exposure Assessment in Environmental Epidemiology

Posted: 4 Jun 2010

See all articles by Donna Spiegelman

Donna Spiegelman

Harvard University - T.H. Chan School of Public Health

Date Written: April 2010

Abstract

Uncertainty in assessment of individual exposure levels leads to bias, often, but not always, toward the null in estimates of health effects, and to underestimation of the variability of the estimates, leading to anticonservative p-values. In the absence of data on the uncertainty in individual exposure estimates, sensitivity analysis, also known as uncertainty analysis and bias analysis, is available. Hypothesized values of key parameters of the model relating the observed exposure to the true exposure are used to assess the resulting amount of bias in point and interval estimates. In general, the relative risk estimates can vary from zero to infinity as the hypothesized values of key parameters of the measurement error model vary. Thus, we recommend that exposure validation data be used to empirically adjust point and interval estimates of health effects for measurement error. The remainder of this review gives an overview of available methods for doing so. Just as we routinely adjust for confounding, we can and should routinely adjust for measurement error.

Suggested Citation

Spiegelman, Donna, Approaches to Uncertainty in Exposure Assessment in Environmental Epidemiology (April 2010). Available at SSRN: https://ssrn.com/abstract=1601773 or http://dx.doi.org/10.1146/annurev.publhealth.012809.103720

Donna Spiegelman (Contact Author)

Harvard University - T.H. Chan School of Public Health ( email )

677 Huntington Avenue
Boston, MA MA 02115
United States

Do you have negative results from your research you’d like to share?

Paper statistics

Abstract Views
331
PlumX Metrics