Abstract PS5-05-10: A Comparison of Participant-Reported and Medical Record Stage and Pathology Information in the Susan G. Komen ShareForCuresRegistry

Abstract Background: ShareForCures is Komen’s crowd-sourced breast cancer research registry that collects and integrates data from and on behalf of its participants. Participants living in the U.S. with a history of breast cancer consent to share their medical records and complete surveys. In the U.S. primary care literature, patients and their medical records are concordant at rates ranging from 65% to 93%. However, there are limited data on how patients with breast cancer self-report details of their diagnoses. Here, we leveraged a cohort of ShareForCures participants to assess whether the recollection of their diagnosis information matches the information in their medical records. Not all researchers can access medical records to extract diagnosis information, and a high concordance could allow researchers to rely more heavily on patient-reported clinical details. Methods: We compared data for participants in ShareForCures who had completed the enrollment survey and had abstracted medical records (requested from the oncology and primary care practices) available by April 30, 2025. We had asked participants for the following variables in the intake survey: 1) stage at diagnosis, 2) estrogen-receptor (ER) and HER2 status (e.g., positive or negative), 3) receptor combination (e.g., triple-negative or triple positive), and 4) histology (e.g., ductal or lobular). These same variables are abstracted from the medical record. For staging, we only examined stages 1 - 4 and did not assess the accuracy of sub-stages (e.g., IB). We used the AJCC anatomic/clinical staging system in effect at the time of diagnosis. A board-certified medical oncologist specializing in breast cancer (JCD) adjudicated any unclear cases. We used descriptive statistics to summarize the data. We define concordance as the fraction of participant-reported details that match the abstracted medical records. Results: Data were obtained from 151 participants enrolled between 2022 and 2024. The majority (78%) had a diagnosis between 2010 - 2024. On median, 292 (IQR, 129 - 579) pages of records were available for abstraction per participant. The median age at diagnosis was 45 (IQR, 39 -52). The cohort was predominantly white (80%), and 82% of the participants had at least a college degree. The median time from diagnosis to enrollment in ShareForCures was 6 years (IQR, 2 - 11). 14% of the cohort reported as living with metastatic disease at enrollment. The diagnostic stage was concordant in 79% of the cases. ER status and HER2 status were concordant in 84% and 83% of the cases, respectively. When receptor statuses were combined (e.g., triple-negative, triple-positive), concordance was 75%. Histology was concordant in 73% of cases. We observed concordance for all these variables in 50% of the cases. Conclusions: Despite the inherent challenges of abstracting information from medical records and the complexities of breast cancer diagnoses, half of the responses from ShareForCures participants were concordant with all the diagnostic details abstracted from their medical records, with concordance rates ranging from 73% to 84% for individual variables. Although we may have anticipated higher concordance rates in a highly educated, young population, further research is needed to explore factors influencing concordance, such as the completeness of medical records and the time elapsed since diagnosis. This preliminary analysis highlights the importance of integrating both participant-reported and medical record-derived information to ensure the accurate reporting of breast cancer diagnosis information for research purposes. Citation Format: J. C. Dickerson, J. Epps, A. Gyden, A. Cleveland, K. Sabelko, A. W. Kurian, J. Jourquin, M. Bondy. A Comparison of Participant-Reported and Medical Record Stage and Pathology Information in the Susan G. Komen ShareForCuresRegistry abstract. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS5-05-10.