Statistical test of the shortening of latent periods in experimental radiation carcinogenesis under competing risks.

The mean life span of animals killed by a specific type of tumor is influenced by competing risks. The parametric inference method of Noda et al. was applied to a set of mouse experimental data in order to correct for the appearance of observed life span distribution under competing risks and to test statistical significances of difference between corrected mean life spans. Data were from an experiment on tumors induced in mice by X-rays: whole body irradiated with 600 R; head with 800 R; trunk or lower body with 800 R. The statistically significant changes were found: Whole body irradiation shortened the mean latent period of malignant lymphomas whereas the head irradiation shortened the latent periods of none of tumors. Trunk irradiation shortened mean latent periods of lung tumors. Lower body irradiation showed a tendency to shorten the mean latent periods of mammary tumors. Results that differed from the above were obtained when there was no correction for competing risks.