Comparing distributions of time to onset of disease in animal tumorigenicity experiments |
| |
Authors: | Ralph L Kodell James J Chen Gary E Moore |
| |
Institution: | 1. Division of Biometry and Risk Assessment , National Center for Toxicological Research , Jefferson, 72079, Arkansas;2. Computer Based Systems, Inc , National Center for Toxicological Research , Jefferson, 72079, Arkansas |
| |
Abstract: | The cause-of-death test of Peto et al.(1980)pools information from a Hoel-Walburg test on incidental tumors with information from a logrank test on fatal tumors in order to compare the tumor rate of a group of rodents exposed to a carcinogen against the tumor rate of a group of unexposed animals. The cause-of-death test, which can arise as a partial likelihood score test from a model that assumes proportional odds for tumor prevalence and proportional hazards for tumor mortality, is not, in general, a direct test for equality of tumor onset distributions for occult tumors that are observed in both fatal and incidental contexts. This paper develops a direct cause-of-death test for comparing distributions of time to onset of occultumors. The test is derived as a partial likelihood score test under an assumed proportional hazards model for tumor onset distributions. The size and power of the proposed test are compared in a Monte Carlo simulation study to the size and power of competitive procedures, including procedures that do not require cause-of-death information. |
| |
Keywords: | carcinogenicity bioassay cause-of-death peto et al test poly-3 test score test |
|
|