Recommended housing conditions and test procedures can interact to obscure a significant experimental effect.

Routine animal husbandry variables, such as group housing of mice and the order of testing of cage-mates, are currently viewed to be essentially neutral with respect to the outcome of most, if not all, animal-based experiments, including those that utilize behavioral measurements. During the course of experiments that have utilized the elevated plus-maze to examine the ability of a bacterial challenge of mice to induce anxiety-like behavior, due to the activation of various cytokine pathways, we followed the recommendation of laboratory animal care staff to house the mice in pairs. Whenwe testedthe members of the pairs successively, it was found, for the first experimental set, that the behavior that reflects anxiety (time in closed arms) of the first-tested animal differed from that of the second-tested animal for both the experimental and the control animals and, critically, that these changes were in the opposite directions for the controls and the experimental animals, thus obscuring the effect of the experimental manipulation. A second, independent experimental set also obtained a significant effect for the order of testing effect in the bacterial-challenged group, but not in the saline control group, although a similar trend was evident in this group as well. These results indicate that special care should to be taken in implementing housing recommendations and that preliminary tests may be necessary to ensure that housing conditions do not interact with tests of the phenomenon under experimental investigation.