Applying the realized probability of dying to cancer survival

Age comparisons of survival in cancer cohorts generally utilize relative survival rates, which are based on indicators of the probability of survival for a given number of years after diagnosis. Cancer relative survival rates for the same number of years tend to decline as age at diagnosis increases. However, the same number of years of survival reflects higher relative longevity at older ages than at younger ages. The realized probability of dying (RPD) is a survival measure that expresses individual survival time after diagnosis relative to the survival distribution of an age-, race-, and sex-specific reference population, in effect weighing individual survival time more heavily as age at diagnosis increases. The purpose of this study was to apply the RPD as a survival measure in cancer epidemiology. Two cohorts of cancer patients, white males with prostate cancer and white females with breast cancer, aged 55 years and over at diagnosis, were followed for 15 years. Although older subjects survived less time after diagnosis than younger subjects, they achieved more favorable RPD values. We present survival analysis methods for analyzing the RPD in this population, an approach not previously used with this measure. The implications for use of the RPD in cancer epidemiology are discussed.