Utilizing Data from Multiple Studies (Meta-Analysis) To Determine Effective Dose Levels: Example of Mortality in Rats and Mice Exposed to Hydrogen Sulfide. K. G. Brown and J. A. Strickland, KBinc, Integrated Laboratory Systems, Inc.
Most chemical-specific risk assessments use a single best study to identify a toxicity marker, such as a NOAEL, benchmark dose, or LD50. The objective of this exercise is to incorporate as much data as possible from multiple studies to derive a chemical-specific dose-duration-response curve from which to identify toxicity markers. This has the advantage of incorporating more information than single-study assessments. The concept of an ED-box (or EC for concentration) is introduced. An ED05 box for exposure duration of 1 hour, for example, has coordinates on the dose scale corresponding to a 95% confidence interval for the ED05 at 1 hour, and coordinates on the response scale corresponding to the range of response estimates over the confidence interval. An ED-box is more informative than a confidence interval alone because it includes the sensitivity of response over the confidence interval. For fixed exposure durations, dose-response curves can be augmented by adding ED-boxes for ED values of interest; for fixed doses, a plot of ED boxes for different durations illustrates the effect of exposure duration. Meta-analysis and the applications of ED-boxes are illustrated by an example for mortality of rats and mice from acute exposure to hydrogen sulfide. Methods of statistical analysis are used to determine when data from different studies can be pooled. A response curve for mortality is fit to the rat data simultaneously for exposures of 5 min., 10 min., 30 min., and 1 hour, using a logistic curve additive in log(concentration) and log(duration). Separate curves are fit at 2, 4, and 6 hours. The curves for rats fit the data exceedingly well and exhibit a duration-dependent threshold-like response, followed by a steep incline. The response in mice is less predictable. The NOAELs for mortality from individual studies fall within 95% confidence bounds for the EC01 (except lower at 5 min. for rats and higher at 30 min. for mice). The authors gratefully acknowledge the financial support of the American Chemistry Council.
Go to . . .
2002 SRA Annual Meeting Table of Contents
2002 SRA Annual Meeting Author Index
Main Abstracts Menu Page
RiskWorld Home Page