ABSTRACT
This chapter presents a general sample survey approach to survival analysis of case-control data. It explains how the case-control data may be viewed as a two-phase sample, where the cohort data correspond to the Phase I sample and the case-control data to the Phase II sample. The chapter describes the statistical methods to estimate relative and absolute risks when fitting the Cox model to full cohort data and to two-phase samples. It reviews how estimation is performed in the Lin-Ying additive hazards (AH) model. The fact that dose actually was known for the entire cohort allowed results obtained from the case-control analyses to be compared, for illustrative purposes, with results of fitting the same models to the entire cohort. The chapter also describes in general terms how authors estimate the parameters of the AH model from cohort and case-control data. With these new tools, survival analysis for case-control studies has become more transparent and convenient.
