Göreli Yaşam Çözümlemesi
Özet
In observational survival analysis studies with long follow-up periods, cause of death information may not be available or there may be some lack of information. In survival analysis, it may not be possible to directly predict the probability of survival due to the disease when the cause of death is unclear or the available information is unreliable. Therefore, the probability of survival due to the disease is sometimes evaluated by a measure of the relative probability of survival between a group diagnosed with the disease and the larger population. In such cases, relative survival analysis methods should be used. Cancer researches that only patient follow-up times and vital status at the end of follow-up are recorded, is an area where relative survival analysis is frequently used.
Creating life tables and obtaining survival probabilities have an important place in relative survival analysis. For the estimation of these probabilities Ederer I, Ederer II, Hakulinen and Pohar Perme methods are used. The age variable is one of the main covariates used in the life tables. Standardization methods are used to eliminate the effect of age distribution on survival probabilities. In relative survival analysis, survival probabilities can also be obtained by using age standardization methods.
Relative survival models are concerned with modeling the excess hazard ratio whereas the classical hazard ratio obtained in survival models. The approaches of additive model and generalized linear model are mostly used to model the excess hazard ratio.
In this study, relative survival probabilities and excess hazard models were examined in detail and the related methods were applied on the prostate cancer dataset in the literature which includes competing risks. Life tables containing relative survival probabilities for different age groups were obtained. Ederer I, Ederer II, Hakulinen methods were used for the relative survival and Pohar Perme method was used for the net survival and no significant difference was observed between the results obtained. In all methods, the five year relative survival for male patients with prostate cancer was 44% for patients were aged 44-59, 49% for patients were aged 60-74, and 36% for patients were aged 75 or over. Similarly, life tables were created using age standardization, no significant difference was observed in survival probabilities as a result of age standardization. In addition, crude death probabilities due to both prostate cancer and competing risks were obtained and interpreted.
