Avtor/Urednik		Stare, Janez; Henderson, Robin; Pohar, Maja
Naslov		New developments in relative survival
Tip		članek
Vir		In: ROeS seminar 2005; Sep 25-29; Graz. Graz: Medical university,
Leto izdaje		2005
Jezik		eng
Abstrakt		Relative survival techniques are used to compare survival experience in a study cohort with that expected should population rates apply. The techniques are especially useful when causespecific death information is not accurate or not available as they provide a measure of excess mortality in a group of patients with a certain disease. When we want to model the dependence of the observed hazard on measured covariates while correcting for population differences, we need to assume some form of relation between the observed and population hazard. This assumption influences the fit of the model and this infiuence is difficult, or even impossible, to distinguish from the influence of assumptions about the effects of covariates. We have recently introduced a method that solves this problem, and in this paper I briefly review our approach. Another, related, problem is that there are presently no methods to check the fit of the additive model, which is the model most commonly used in relative survival analysis. To this purpose we have introduced residuals, akin to Schoenfeld residuals for the Cox model, and developed methods for checking the goodness of fit based on these residuals. The methods enable the analyst to check the assumptions about the coefficients graphically, and test the null hypothesis of a good fit. An important side effect of the methods is that they can be used, with proper adjustments, for very different models in survival analysis, and are thus not limited to relative survival models.
Deskriptorji		MYOCARDIAL INFARCTION SURVIVAL RATE PROPORTIONAL HAZARDS MODELS COHORT STUDIES