Brief Overview
Survival Analyses
The probability of survival is estimated by the product-limit procedure of Kaplan and Meier. Animals found dead of other than natural causes are censored from the survival analyses; animals dying from natural causes are not censored. Statistical analyses for possible dose-related effects on survival uses Cox's method for testing two groups for equality and Tarone's life table test to identify dose-related trends.
Calculation of Incidence
The incidences of neoplasms or nonneoplastic lesions are the numbers of animals bearing such lesions at a specific anatomic site and the numbers of animals with that site examined microscopically. A method is used to determine survival-adjusted neoplasm rate for each group and each site-specific neoplasm to account for differential mortality.
Analysis of Neoplasm and Nonneoplastic Lesion Incidences
The Poly-k test used to assess neoplasm and nonneoplastic lesion prevalence is a survival-adjusted procedure that takes survival differences into account. Tests of significance include pairwise comparisons of each exposed group with controls and a test for an overall exposure-related trend.
Analysis of Continuous Variables
Organ and body weight data have approximately normal distributions and are analyzed with parametric multiple comparison procedures. Hematology, clinical chemistry, urinalysis, urine concentrating ability, cardiopulmonary, immunotoxicologic, cell proliferation, tissue concentrations, spermatid, and epididymal spermatozoal data have skewed distributions and are analyzed using the nonparametric multiple comparison methods. Jonckheere's test is used to assess the significance of the dose-related trends. Prior to statistical analysis extreme values identified by an outlier test, are examined by NTP personnel and implausible values are eliminated from the analysis. Because vaginal cytology data are proportions, an arcsine transformation is used to bring the data into closer conformance with a normality assumption. Treatment effects are investigated by applying a multivariate analysis of variance to the transformed data to test for simultaneous equality of measurements across exposure concentrations.
