Evaluating the Evidence for Hormesis: A Statistical Perspective

Abstract

It is possible to account for hormesis under current regulatory guidelines by invoking criteria for departure from default risk assessment procedures. However, past experience suggests that it will be difficult to amass enough evidence for hormesis in an individual case to permit departure from default procedures. Accordingly, hormesis is likely to be important in agency risk assessments only if guidelines are modified to incorporate hormesis as a default assumption. This could be appropriate if hormesis is determined to be a universal or near-universal phenomenon. Although there is ample evidence that hormesis occurs in many specific situations, the overall prevalence of hormesis is very difficult to evaluate based on currently available data. The lack of a valid statistical test for hormesis is a major limitation when evaluating evidence for hormesis. The attempts at estimating the prevalence of hormesis reviewed herein did not adequately control for false positives and/or may have had inadequate power to detect hormesis. Some suggestions are made for constructing a database and analyzing the data therein that would provide more readily interpretable information on the prevalence of hormesis.

Keywords:

• hormesis
• meta analysis
• u-shaped curve
• false-positive
• false-negative.