Biostatistical issues and concepts in epidemiologic research.

This paper describes recent developments in failure time regression, and the application of such methods to epidemiologic data. Relative risk and odds ratio estimation procedures are described under cohort and case-control sampling, and under some less standard designs for sampling within a cohort. Some special issues that arise in the analysis of vital data are also mentioned. While methods for descriptive analyses are emphasized some attention is also given to mechanistic models for carcinogenesis and their uses. Topics meriting further methodologic development are delineated throughout.