With increasing sensitivity of the scientific community and the public at large concerning experimentation using animals, comes an increasing imperative to minimize the number of animals entered into experiments. For data grouped by time of failure, we will compare the test for comparing survival in two groups(Mantel Haenszel) with the binomial test which is based on simply comparing the proportions of animals with tumors. Conditions for which the Mantel Haenszel test requires an appreciably smaller number of animals than the binomial will be formulated. We expect that this research will demonstrate that the test for survival analysis based on the assumption of equal odds ratios (or for continuous time, proportional hazards) does not fare well when there is a time lag before the treatment shows toxicity. Based on this premise, the use of """"""""optimal"""""""" tests for grouped survival data will be explored for the case in which an assumption of proportional hazards does not hold. The use of an overall test procedure based on the maximum of two or three test statistics will be explored. The implications of the findings to the timing of interim sacrifices will be evaluated. Animals can also be saved by the method of stochastic curtailment, which-allows an experiment to be stopped early if the results becomes apparent early in the trial. Work previously done on stochastic curtailment with respect to continuous data will be generalized to handle dichotomous or survival data . Methods to measure agreement with respect to the reading of slides for specific sampling designs will be developed for the case in which the response is either dichotomous or graded. This work will allow investigators to better set up studies to evaluate the accuracy in interpreting pathology specimens. Alternate frameworks for analyzing interval censored data will be explored.
| Bodian, C A (1994) Intraclass correlation for two-by-two tables under three sampling designs. Biometrics 50:183-93 |
| Wallenstein, S; Wittes, J (1993) The power of the Mantel-Haenszel test for grouped failure time data. Biometrics 49:1077-87 |
| Berger, A; Gomez, G; Wallenstein, S (1988) A homogeneity test for follow-up studies. IMA J Math Appl Med Biol 5:101-12 |
| Wallenstein, S; Bodian, C (1987) Epidemiologic programs for computers and calculators. Inferences on odds ratios, relative risks, and risk differences based on standard regression programs. Am J Epidemiol 126:346-55 |
| Van Ryzin, J; Rai, K (1987) A dose-response model incorporating nonlinear kinetics. Biometrics 43:95-105 |
| Davis, D L; Mandula, B; Van Ryzin, J V (1985) Assessing the power and quality of epidemiologic studies of asbestos-exposed populations. Toxicol Ind Health 1:93-110 |
| Van Ryzin, J (1985) Risk assessment for fetal toxicity. Toxicol Ind Health 1:299-310 |
