Identifying women at high-risk for breast cancer using data from the electronic health record compared to self-report

Authors: Jiang Xinyi, Li Xinyi, McGuinness Julia E., Vanegas Alejandro, Colbeth Hilary, Vargas Jennifer, Sandoval Rossy, Kukafka Rita, Crew Katherine D.

Category: Electronic Health Records (EHRs), Electronic Health Records (EHRs)
Conference Year: 2018

Abstract Body:
Purpose: A barrier to chemoprevention uptake among high-risk women is the lack of routine breast cancer risk assessment in the primary care setting. We calculated breast cancer risk using the Breast Cancer Surveillance Consortium (BCSC) model, accounting for age, race/ethnicity, first-degree family history of breast cancer, benign breast disease, and mammographic density using data collected from the electronic health record (EHR) and self- report.  Patients and Methods: Among women undergoing screening mammography, we enrolled women, age 35-74, without a prior history of breast cancer, mastectomy, or breast augmentation. We extracted data on demographics, structured first-degree family history, breast radiology and pathology reports from the EHR. We assessed agreement in breast cancer risk information between the EHR and a self-administered questionnaire.  Results: Among 13,764 women, 2719 women (19.7%) met high-risk criteria, based upon a 5-year invasive breast cancer risk 1.67% according to the BCSC model. From the EHR, data was missing on 31% for race/ethnicity and 85% for family history. Among 2303 women with both EHR and self-report data, more first-degree family history (14% vs. 3%) and prior breast biopsies (18% vs. 11%) were identified by self-report vs. EHR, respectively. However, more women with atypia and lobular carcinoma in situ were identified from the EHR. More high-risk women (20% vs. 16%) were identified with EHR vs. survey data, respectively, with a moderate agreement (kappa of 0.41).  Conclusion: Among women undergoing screening mammography, we identified 20% who met high-risk criteria according to the BCSC model based upon EHR data with a moderate agreement to self-report data, despite missing data on race/ethnicity and family history. This may serve as an initial screen for identifying women eligible for breast cancer chemoprevention.

Keywords: breast cancer risk, electronic health record