Survival statistics commonly reflect survival from the time of diagnosis but do not take into account survival already achieved after a diagnosis. The objective of this study was to provide conditional survival estimates for brain tumor patients as a more accurate measure of survival for those who have already survived for a specified amount of time after diagnosis.

Data on primary malignant and nonmalignant brain tumor cases diagnosed from 1985-2005 from selected SEER state cancer registries were obtained. Relative survival up to 15 years postdiagnosis and varying relative conditional survival rates were computed using the life-table method.

The overall 1-year relative survival estimate derived from time of diagnosis was 67.8% compared to the 6-month relative conditional survival rate of 85.7% for 6-month survivors (the probability of surviving to 1 year given survival to 6 months). The 10-year overall relative survival rate was 49.5% from time of diagnosis compared to the 8-year relative conditional survival rate of 79.2% for 2-year survivors. Conditional survival estimates and standard survival estimates varied by histology, behavior, and age at diagnosis. The 5-year relative survival estimate derived from time of diagnosis for glioblastoma was 3.6% compared to the 3-year relative conditional survival rate of 36.4% for 2-year survivors. For most nonmalignant tumors, the difference between relative survival and the corresponding conditional survival estimates were minimal. Older age groups had greater numeric gains in survival but lower conditional survival estimates than other age groups. Similar findings were seen for other conditional survival intervals.

Conditional survival is a useful disease surveillance measure for clinicians and brain tumor survivors to provide them with better 'real-time' estimates and hope.