Influence of coefficient of variation in determining significant difference of quantitative values obtained from 28-day repeated-dose toxicity studies in rats.

ABSTRACT

In order to understand the influence of coefficient of variation (CV) in determining significant difference of quantitative values of 28-day repeated-dose toxicity studies, we examined 59 parameters of 153 studies conducted in accordance with Chemical Substance Control Law in 12 test facilities. Sex difference was observed in 12 parameters and 10 parameters showed large CV in females. The minimum CV was 0.74% for sodium. CV of electrolytes was comparatively small, whereas enzymes had large CV. Large differences in CV were observed for major parameters among 7-8 test facilities. The changes in CV were grossly classified into 11. Our study revealed that a statistical significant difference is usually detected if there is a difference of 7% in mean values between the groups and the groups have a CV of about 7%. A parameter with a CV as high as 30% may be significantly different, if the difference of the mean between the groups is 30%. It would be ideal to use median value to assess the treatment-related effect, rather than mean, when the CV is very high. We recommend using CV of the body weight as a standard to judge the adverse effect level.