Alternative Methods to Analyze Relative Survival Data  

Manca Jesenko, assistant
Faculty of Organizational Sciences, University of Maribor, Slovenia

Relative survival is usually defined as the ratio of the observed survival function and the expected survival function. Another approach, so called transformation approach, defines relative survival for each individual as the value which a subject reaches on the expected distribution function. Based on these two approaches different methods for modeling relative survival can be used. The standard approach is to use the additive model, only rarely the multiplicative model is used. Both models assume certain relationship between the observed and population hazard. With transformation approach any model for analyzing standard survival data can be used and the most obvious advantage of transformation approach is the absence of any kind of assumptions concerning the relationship between observed and population hazards. In this talk we will discuss two methods to analyze transformed survival data. The first method is the Buckley and James linear regression model which in contrast to other methods in survival analysis focuses on modeling observed (transformed) survival times, the second method is the Aalen linear regression model which models the hazard function. Both models could offer some new information and different insight into relative survival. We will shortly present theoretical backgrounds of both models, the possibilities to use them in relative survival and their application on a real data set.