P Martus1, S Birkenhake, R Sauer. 1. Institut für Medizinische Statistik und Dokumentation, Universität Erlangen-Nürnberg.
Abstract
PURPOSE: Survival of elder oncological patients may depend on the tumor disease, on reasons not related to the underlying disease or simply on patient's age. Therefore patient' age has to be taken into account if the objective of an oncological study is the disease-related survival. Basically, there are 3 different approaches: multivariate statistical modelling, estimation of disease-related survival after censoring of cases with cause of death not due to the tumor disease, and consideration of demographical data in the general population. PATIENTS AND METHODS: Our paper deals with the age correction using the data from the 1986/88 census in Germany. We have written a computer program based on SPSSWIN, which allows an automated age correction. Using clinical data of radiotherapy after surgical therapy of bladder cancer, we illustrate the method. RESULTS: The estimated 5-year-survival using the age correction was--dependent on patient's age in the sample--10% up to 15% higher than the uncorrected rate. In the whole sample of 333 unselected patients, 5-year-survival was corrected from 47% to 59%. In the normal population adjusted for age- and sex-distribution, the expected survival rate was 80%. CONCLUSIONS: The age correction has substantial influence on survival analysis of male patients elder than 60 years and female patients elder than 70 years. It should therefore be used in oncological studies investigating long-term survival of patients in these age-groups.
PURPOSE: Survival of elder oncological patients may depend on the tumor disease, on reasons not related to the underlying disease or simply on patient's age. Therefore patient' age has to be taken into account if the objective of an oncological study is the disease-related survival. Basically, there are 3 different approaches: multivariate statistical modelling, estimation of disease-related survival after censoring of cases with cause of death not due to the tumor disease, and consideration of demographical data in the general population. PATIENTS AND METHODS: Our paper deals with the age correction using the data from the 1986/88 census in Germany. We have written a computer program based on SPSSWIN, which allows an automated age correction. Using clinical data of radiotherapy after surgical therapy of bladder cancer, we illustrate the method. RESULTS: The estimated 5-year-survival using the age correction was--dependent on patient's age in the sample--10% up to 15% higher than the uncorrected rate. In the whole sample of 333 unselected patients, 5-year-survival was corrected from 47% to 59%. In the normal population adjusted for age- and sex-distribution, the expected survival rate was 80%. CONCLUSIONS: The age correction has substantial influence on survival analysis of male patients elder than 60 years and female patients elder than 70 years. It should therefore be used in oncological studies investigating long-term survival of patients in these age-groups.