|
2P Modelling ecotoxic effects (MOP/155) Avian acute oral toxicity: A statistical evaluation of sequential experimental designs. Chapman, Peter1, Collins, Brian2, Dark, Roberta1, Farrar, David 3, Green, John4, Healey, Graham5, Neyer, Barry6, Slob, Wout7, Springer, Tim8, 1 Syngenta, Bracknell, Berkshire, United Kingdom2 Environment Canada, Ottowa, Ontario, Canada3 United States Environmental Protection Agency, Washington, DC, United States4 DuPont, Wilmington, Delaware, United States5 Huntingdon Life Sciences, Huntingdon, Cambridge, United Kingdom6 PerkinElmer Optoelectronics, Miamisburg, Ohio, United States7 RIVM, Bilthoven, Utrecht, Netherlands8 Wildlife International, Easton, Maryland, United States ABSTRACT- OECD recently established an avian expert group to draft a new test guideline that reduced animal usage in avian acute studies. The expert group assembled a team of nine statisticians to help evaluate the adequacy of proposed designs and choose among possible alternatives. These statisticians collaborated on simulations that provided measures of the ability of the designs to provide estimates of LD50 values, dose-response slopes, and confidence interval for a large number of sequential test designs (including up-down methods) and variations on the current EPA guideline study. The various designs were evaluated by comparison to each other and to the current EPA acute avian oral toxicity test, which requires 60 birds in total. Dramatic reduction in animal use without loss of quality in LD50 and slope estimates were obtained by sequential designs. In these designs, the experiment is performed in stages so that results from stage 1 can be used to aid the design of stage 2, the results of stage 2 can be used to aid the design of stage 3, and so on. The simulations suggest that a two-stage design with twelve birds is good enough to estimate an LD50. Alternatively, if an estimate of the confidence interval of the LD50 or the slope of the dose response are also required, then a twenty-four bird, three-stage design performs as well as or better than the current sixty bird EPA design. Key words: optimal experimental design, dose response, acute avian oral toxicity, sequential experimental design |
