Risk and Regulatory Policy Assessment Based on Complex Data Sources

• Activity Number: 525
• Type: Invited
• Date/Time: Thursday, August 2, 2007 : 10:30 AM to 12:20 PM
• Sponsor: Section on Risk Analysis
• Author(s): Michael E. Tarter*+ and Joseph P. Brown and Andrew G. Salmon
• Companies: University of California, Berkeley and State of California, Air Toxicology, and Risk Assessment Unit and State of California, Air Toxicology and Risk Assessment Unit
• Keywords: Animal experiments ; Best Linear Estimation ; Order statistics ; Small sized samples ; Subsurvival ; Toxicity

Abstract

It is convenient to estimate a 95% tolerance bound for the time T beyond which 90% or more subjects had not, as yet, died with targeted tumor present. The quantity 95% applies to what is often called uncertainty, and the quantity 90% is associated with the curve property, variability. Two factors influence uncertainty, experiment quality and data quantity. A bound's low value makes a putative toxic substance appear harmful. Hence experiments conducted with an insufficient number of animals, say n, are penalized for this false economy by a low-valued bound which, in a well thought out experiment would have, as n increased, approached T. That T, not the median or some other measure of central tendency, is targeted, takes experimental subject variability into account. The less the variability, the closer T will be to the population median or to some other measure of central tendency.