A statistical procedure for modeling continuous toxicity data

Authors

  • Robert D. Bruce,

    1. The Procter & Gamble Company, Human and Environmental Safety Division, Miami Valley Laboratories, P.O. Box 398707, Cincinnati, Ohio 45239-8707
    Search for more papers by this author
  • Donald J. Versteeg

    Corresponding author
    1. The Procter & Gamble Company, Human and Environmental Safety Division, Ivorydale Technical Center, 5299 Spring Grove Avenue, Cincinnati, Ohio 45217-1087
    • The Procter & Gamble Company, Human and Environmental Safety Division, Ivorydale Technical Center, 5299 Spring Grove Avenue, Cincinnati, Ohio 45217-1087
    Search for more papers by this author

Abstract

Chronic aquatic toxicity test results are commonly analyzed with statistical hypothesis tests to generate summary statistics known as the no-observed-effect concentration (NOEC) and first-observed-effect concentration (FOEC). These procedures address statistical differences among treatments but suffer several critical limitations. Use of concentration-response statistics to estimate minimal effect concentrations (i.e., EC values) from quantal and continuous data has advantages over a hypothesis-testing approach for generating a biologically relevant end point and an estimate of variability from toxicity tests. Estimation of the concentration-response statistic (EC, effective concentration) for continuous data is not straightforward but is possible with a variety of approaches. A statistical method for estimating EC values described here is based on a nonlinear regression estimation procedure. The usefulness of this method is demonstrated with continuous data from chronic toxicity tests with algae, fish, and invertebrate populations.

Ancillary