Multistrain experiments for screening toxic substances

Abstract

The advantages of using homogeneous experimental groups (inbred animal strains) and of using multiple groups within an experiment, based on the power of the Mantel‐Haenszel test, were investigated. A simulation experiment was performed to empirically calculate the power of a one‐sided Mantel‐Haenszel test for multistrain experiments. In each case, the power of the multistrain experiment was compared to the (empirical) expected value, over strains, of the power where each strain is tested individually. In the simulation, use of subgroups, each having different response rates, resulted in an increase in power where a chemical exposure caused an average increase of effects in 10% or more of the animals across strains.