analysis could be performed, at least for a few facilities and years of operation, for the case-control study.

Nevertheless, at the very least, any epidemiologic study will need to address uncertainty, at least qualitatively. Such an analysis should:

  • Identify, evaluate, and rank all potential sources of major uncertainty and identify site-to-site and temporal differences;
  • Identify potential bias versus random errors in the dose calculations that could affect interpretation of the epidemiologic findings; and
  • Identify shared errors23 as opposed to stochastic variability to properly evaluate the risk from radiation exposure should any increased risk of cancer be identified.

Although the reported environmental monitoring data for almost all sites and times was either below minimum detectable levels or, for external radiation, not distinguishable from background, an epidemiologic study could still use these data to set upper limits on the reported effluents by back-calculating from the minimum detection levels. This would at least place upper bounds on effluent releases.


This chapter provides the committee’s assessment of methodological approaches for assessing offsite radiation doses to populations living near nuclear plants and fuel-cycle facilities to support an epidemiologic study. Based on this assessment, the committee finds that:

  1. Absorbed dose—the energy deposited by ionizing radiation per unit mass of tissue in specific organs of interest—is the appropriate dose quantity for use in an epidemiologic study. Other dose quantities, for example effective dose, equivalent dose, and collective dose, are designed for regulatory purposes and are not appropriate for epidemiologic studies (see Section 3.4.1). The dose to a maximally exposed individual (MEI) is also not an appropriate quantity for an epidemiologic study because it provides a high-sided estimate at

23 As discussed in NCRP (2009b), uncertainties that are common to many individuals (for example, error in the amount of effluents from a facility) can introduce bias (systematic uncertainty) in estimated doses compared to uncertainties that are unshared and represent stochastic variability in true doses among individuals. When uncertainties are shared among individuals in a population, the degree of variability in true doses among individuals is less than would be estimated by assuming that uncertainties in each individual’s dose are purely random. An overestimation of the variability in true doses among individuals results in a suppression of dose-response relationships derived in an epidemiologic study, i.e., the true dose response is flattened (Schafer and Gilbert, 2006).

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement