Abstract P4-10-06: Factors impacting the accuracy of self-reported breast procedures among women with and without breast cancer

Abstract

Background: Clinical/epidemiologic observational studies frequently rely on participants9 recall for information about breast procedures. However, there is limited data on the accuracy of self-reported breast procedures. To address this knowledge gap and inform future study design and collection and interpretation of similar data, we investigated the impact of type, diagnosis, age, time, and other patient characteristics on the accuracy of self-report in a prospective cohort. Methods: All benign breast biopsies, lumpectomies, and mastectomies for breast cancer treatment among women enrolled in the BOSS Cohort, a prospective study of women and men with a familial risk of breast/ovarian cancer, were identified. Study staff obtained pathology reports for 93% of women from self-reported breast procedure locations. For this analysis, we focused on 577 women who had at least one ascertained pathology report, and who self-reported at least one breast procedure at baseline. We estimated the percentage of self-reports (95% confidence interval (CI)) with matching pathology report within 6 months (+/- 6 months), and agreement between self-reported procedures and pathology-confirmed diagnoses (normal/benign vs. atypical hyperplasia vs. LCIS, and DCIS vs. invasive cancer) with the Kappa statistic. We also examined predictors of an accurate biopsy self-report, including age at baseline, personal and family history of breast cancer, educational attainment, and time between biopsy and baseline, using logistic regression models. Results: At baseline, 158 women reported having at least one benign biopsy, 193 women reported having a lumpectomy for cancer treatment, and 174 women reported having a mastectomy for cancer treatment. The median time between biopsy, lumpectomy, mastectomy, and baseline was 9 years, 2 years, and 2 years, respectively. Fifty-seven percent (95% CI: 49-64.5%) of benign biopsy self-reports, 90.7% (95% CI: 85.6-94.1%) of lumpectomy self-reports, and 85.1% (95% CI: 78.9-89.7%) of mastectomy self-reports had a matching pathology report within 6 months. Further diagnostic agreement was moderate for biopsies, lumpectomies, and mastectomies with Kappa statistics of 0.65, 0.66, 0.65, respectively. Age at baseline (p-interaction =0.01) and time (p-interaction = 0.03) were independent and joint predictors of accurate biopsy self-reports. Women less than 49 years old had the largest reduction in odds of having an accurate self-report (26%) for every additional year between biopsy and baseline [adjusted odds ratio = 0.74 (95% CI: 0.63-0.88)]. Similarly, women with a biopsy within 4 years prior to baseline had a 10% reduction in the odds of having an accurate self-report with increasing age [adjusted odds ratio = 0.9 (95% CI: 0.84-0.97)]. Conclusions: In this highly-educated cohort, the overall accuracy of self-report of benign biopsies was only modest, and the accuracy of self-report of lumpectomies and mastectomies was lower than expected. This study suggests that age at baseline and time between procedure and baseline are important predictors of accuracy of self-report and should be considered when utilizing self-reported information. Furthermore, where possible, prospective collection of breast procedure data should be prioritized. Citation Format: Schaeffer ML, May B, Cimino-Mathews A, Orellana M, McCullough M, Hogan B, Armstrong D, Visvanathan K. Factors impacting the accuracy of self-reported breast procedures among women with and without breast cancer [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 P4-10-06.