Abstract P1-08-31: Comparing estimates of survival for breast cancer patients

Abstract

Abstract Introduction Net survival can be considered as the most accurate evaluation of survival from cancer, as it encompasses the survival that would occur if the only possible underlying cause of death was the cancer under study. Both relative survival (RS) and cause-specific survival (CSS) have been developed as approaches to estimate net survival at the population level. Given the debated accuracy of death certificates in identifying the underlying cause of death, RS measures are most commonly used by cancer registries. However, RS calculations can also be biased as it is sometimes unavoidable to match a group of cancer patients with a non-comparable disease-free population group. In this study, we aim to compare RS and CSS using causes of death obtained from death certificates in a cohort of Belgian breast cancer patients by evaluating them to CSS using causes of death obtained from medical files. Methods A total of 3,205 breast cancer patients diagnosed and treated in University Hospitals Leuven between 2009-2014 were included in this study. RS was calculated with the Ederer II method, dividing the observed survival by the expected survival for females of the same age and region. CSS was calculated using cause of death information either gathered from death certificates or collected from medical files. The estimates for RS and CSS as obtained from death certificates were compared to CSS obtained from medical files. Follow-up was guaranteed until 31th of December 2014. Results From the included cohort of breast cancer patients, 255 were deceased. Cause of death was available for 254 patients from death certificates and for 191 patients from medical files. By considering the available cause of death information, 3,141 patients were included to calculate the survival estimates. The 1-year relative survival estimate was 99.3% (95%CI, [98.7, 99.8]), while 1-year cause-specific survival based on death certificates was 99.0% [98.5, 99.3]. The 1-year survival estimate according to CSS based on medical files was 98.1% [97.5, 98.5]. The 5-year survival estimates were 95.8% [94.0, 97.4] for RS, 93.3% [91.8, 94.4] for CSS based on death certificates and 89.7% [88.0, 91.2] for CSS based on medical files. Conclusion This study used causes of death information as derived directly from medical files to calculate CSS in order to compare it with RS and CSS derived from death certificates. This allowed an evaluation of the RS and CSS estimates commonly used at the population level against the presumably more reliable CSS as obtained from medical files. Both RS and CSS from death certificates were divergent from CSS from medical files, but the CSS from death certificates appeared to be a closer estimate to CSS from medical files. After longer follow-up, both survival estimates seemed to deviate more from the CSS from medical files. Citation Format: Izci H, De Schutter H, Wildiers H, Neven P. Comparing estimates of survival for breast cancer patients [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P1-08-31.