Abstract:
|
In toxicity experiments, the "safe" level of exposure for a given species has traditionally been determined by NOAEL or No Observed Adverse Effects Level--recently, some advocate replacing such NOAELs by an estimated percent effect from a regression model. Another alternative step-down approach is offered here by utilizing the monotonicity of the dose-response that is generally expected biologically, making it more powerful and scientifically sounder than a pairwise comparison approach.
This MAXSD (Maximum Safe Dose) approach addresses the traditional power characteristics issued by incorporating the size effect it is desirable to detect into the test--starting from an assumption that the chemical is not safe until proved otherwise, the opposite of the usual presumption that the chemical is safe. By using this method, the risk of a false positive result is greatly reduced. At the same time, this approach eliminates the perceived reward for sloppy experimentation, since high variability will lower, not raise, the NOAEL. There is no need to assume a particular mathematical dose-response relationship in this method. The MAXSD approach will be illustrated with real-life data.
|