Abstract PS7-63: Cause of death discordance between death certificates and medical files: Impact on cancer survival assessment in a Belgian case study

Introduction Accurate information on cause of death is essential for correct breast cancer-specific mortality assessment. However, registration and coding of cause of death is prone to error since determining the exact underlying condition related to the death is challenging. In this study, an expert review of medical files was done to determine the principal cause of death for breast cancer patients of a Belgian tertiary hospital. The retrieved cause of death was compared to death certificate information to assess concordance between both sources. Secondly, the impact of discordant reporting on cause-specific survival (CSS) and other net survival approaches were examined. Methods Breast cancer patients diagnosed and treated at University Hospitals Leuven (UHL) between 2009 and 2014 with follow-up until December 31st, 2016, were included in the study. Information on cause of death was obtained from death certificates (following ICD-10 rules) and medical files. The latter were reviewed by a board of experts at UHL. Agreement was calculated using Cohen’s kappa coefficient, and reasons for discordant reporting were assessed. CSS was calculated based on cause of death information from both sources using the Kaplan-Meier method. These survival estimates were compared to the relative survival probability (RS) using the Ederer II and Pohar Perme method. Results A total of 2,862 patients were included, of whom 354 died after a median follow-up of 54.6 months. We found overall substantial agreement (kappa-value of 0.69 (95% C.I.: 0.62-0.77)) between cause of death reported by death certificates and medical files (Table 1). In 84.8% of cases, there was concordance between both methods. When comparing to medical files, misattribution of breast cancer-specific death in death certificates (4.5% of cases) was linked to the presence of comorbidities (43.7%), metastases (37.5%), or unspecified causes (18.8%). Five-year CSS based on medical files (93.1% (95% C.I.: 91.9-94.1)) was only slightly higher compared to CSS based on death certificates (92.3% (95% C.I.: 91.2-93.4)). RS measures using Ederer II and Pohar Perme were comparable to CSS measures. Conclusions Overall, substantial agreement of cause of death was seen between death certificates and medical files. Attribution of cause death to comorbidities was the most common reason for discordant reporting of breast cancer-specific death. Five-year breast cancer-specific survival was slightly higher based on cause of death information from medical files, compared to death certificates. Periodic reviews and implementation of ICD-10 guidelines for classification of cause of death could improve accuracy in cause of death annotation. Citation Format: Hava Izci, Tim Tambuyzer, Jessica Vandeven, Jerome Xicluna, Hans Wildiers, Kevin Punie, Nynke Willers, Eva Oldenburger, Els Van Nieuwenhuysen, Patrick Berteloot, Ann Smeets, Ines Nevelsteen, Liesbet Van Eycken, Harlinde De Schutter, Patrick Neven, Geert Silversmit, Freija Verdoodt. Cause of death discordance between death certificates and medical files: Impact on cancer survival assessment in a Belgian case study [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS7-63.