Statistical analytical methods for comparing the incidence of tumors to the historical control data.

Abstract

Statistical analysis for comparing the incidence of tumors in treated groups to historical control data is not generally performed. In the present study, a number of data exhibiting the lowest, low, moderate and highest incidence of tumors from long-term rodent bioassay studies has been compared with the historical control data. In studies exhibiting the lowest incidence (less than a few percent) of tumors, the Kastenbaum and Bowman test was found to be relevant since it takes into account the sample size of both the historical control data base and each treated group in the study. In studies where a wider range of tumor incidence was exhibited, a statistical method which employs a rejection limits based on the range of incidence in the historical data is recommended. When malignant tumors are evident in treated groups, no matter how low the incidence, they should be analyzed statistically and compared with the incidence in historical control data as well as those in the concurrent control group. Statistical analytical comparisons of study results to historical control data may contribute to more meaningful evaluations in carcinogenicity studies by eliminating possible false positive or false negative results.