The Simulated Population Method of Analysis of Animal Painting Experiments in Cancer Research

In many comparative carcinogenesis experiments in animals the nontumour mortality experience of the different groups of animals varies considerably. Allowing for this differing non-tumour mortality is a major problem in the analysis of such experiments. Many methods have been suggested to overcome this difficulty, the majority of which depend on an analysis of the age-specific tumour rates. Twort and Twort in their pioneer work [1930; 1931; 1933] used measures based on arbitrary standard populations. These measures were considered in detail by Irwin [1946], but he preferred to estimate the mean of the hypothetical distribution of time to tumour in the absence of non-tumour mortality. Pike and Roe [1963] also used this hypothetical distribution of tumourless life, but they considered the whole curve. Models assuming particular functional forms for this distribution have been suggested by Lea [1945], Pike [1966], and Paige (in Day [1967]). Standardisation has not been referred to widely in the literature, but Paige presented an alternative analysis based on Yule's [1934] direct method of age standardization. This method of analysis is essentially unsatisfactory, as it can lead to standardized numbers of tumour bearing animals in excess of the number of animals on test. Many investigators ignore the problem entirely and present only actual tumour counts. When the object of an experiment is merely to demonstrate the presence or absence of a carcinogenic effect, the method of analysis is not usually critical. In many fields, however, attempts are being made to estimate the relative carcinogenicities of substances and to relate these to chemical and physical properties. The method of analysis is then important. For the chemists, physicists, and biologists who use the results, the number of animals acquiring tumours is a simple and obvious measure of effect. A technique which corrects this count for differing non-tumour mortality will, therefore, have great intuitive appeal. This paper describes such a technique.