Using Relative Survival to Estimate the Burden of Kidney Failure

Estimates of mortality from kidney failure are misleading because the mortality from kidney failure is inseparable from the mortality attributed to comorbid conditions. We sought to develop an alternative method to reduce the bias in estimating mortality due to kidney failure using life table methods. Longitudinal cohort study. Using data from the US Renal Data System and the Medicare 5% sample, we identified an incident cohort of patients, age 66+, who first had kidney failure in 2009 and a similar general population cohort without kidney failure. Kidney failure. Death. We created comorbidity, age, sex, race, and year-specific life tables to estimate relative survival of patients with incident kidney failure and to attain an estimate of excess kidney failure-related deaths. Estimates were compared with those based on standard life tables (not adjusted for comorbidity). The analysis included 31,944 adults with kidney failure with a mean age of 77±7 years. The 5-year relative survival was 31% using standard life tables (adjusted for age, sex, race, and year) versus 36% using life tables also adjusted for comorbidities. Compared with other chronic diseases, patients with kidney failure have among the lowest relative survival. Patients with incident kidney failure ages 66-70 and 76-80 have a survival comparable to adults without kidney failure roughly 86-90 and 91-95 years old, respectively. Relative survival estimates can be improved by narrowing the specificity of the covariates collected (eg, disease severity and ethnicity). Estimates of survival relative to a matched general population partition the mortality due to kidney failure from other causes of death. Results highlight the immense burden of kidney failure on mortality and the importance of disease prevention efforts among older adults. Estimates of death due to kidney failure can be misleading because death information from kidney failure is intertwined with death due to aging and other chronic diseases. Life tables are an old method, commonly used by actuaries and demographers to describe the life expectancy of a population. We developed life tables specific to a patient’s age, sex, year, race, and comorbidity. Survival is derived from the life tables as the percentage of patients who are still alive in a specified period. By comparing survival of patients with kidney failure to the survival of people from the general population, we estimate that patients with kidney failure have one-third the cha