Evaluation of statistical tools used in short-term repeated dose administration toxicity studies with rodents

摘要

In order to know the different statistical tools used to analyze the data obtained from twenty-eight-day repeated dose oral toxicity studies with rodents and the impact of these statistical tools on interpretation of data obtained from the studies, study reports of 122 numbers of twenty-eight-day repeated dose oral toxicity studies conducted in rats were examined. It was found that both complex and easy routes of decision trees were followed for the analysis of the quantitative data. These tools include Scheffe's test, non-parametric type Dunnett's and Scheffe's tests with very low power. Few studies used the non-parametric Dunnett type test and Mann-Whitney's U test. Though Chi-square and Fisher's tests are widely used for analysis of qualitative data, their sensitivity to detect a treatment-related effect is questionable. Mann-Whitney's U test has better sensitivity to analyze qualitative data than the chi-square and Fisher's tests. We propose Dunnett's test for analysis of quantitative data obtainedfrom twenty-eight-day repeated dose oral toxicity tests and for qualitative data, Mann-Whitney's f/test. For both tests, one-sided test with p=0.05 may be applied.