Abstract Background: Social support is an important predictor of breast cancer treatment adherence, quality of life, and survival. However, prior research has been predominantly representative of non-Latina white women and those with common breast cancer subtypes. We considered whether information from the electronic health record (EHR) could be developed and used to enhance representativeness of social support research in breast cancer. Methods: We conducted interviews with oncologists, breast care coordinators, nurse practitioners, and social workers to determine the nature and extent of social support data collection in patients with breast cancer and/or information in the EHR that might help to identify patients with low social support. We also reviewed EHR records for variables that have been used to measure structural and functional support. We included in our analysis 27,584 patients diagnosed with invasive breast cancer in Kaiser Permanente Northern California (KPNC) from 2006-2017. We evaluated concurrent validity of potential social support measures against levels of the gold standard Medical Outcomes Study Social Support measure, collected in the Pathways Study, a study of 4,505 women with stage I-IV invasive breast cancer. We further evaluated predictive validity of measures with overall mortality using National Death Index data. Finally, we evaluated the performance of an index based on the sum of widely available variables and mortality in the total population; scores ranged from 0 (low social support)-4 (high social support). Results: Variables widely available in the EHR included marital status, living status (i.e., whether living alone or with someone), religious status (i.e., indicator of religious involvement or not), availability of information about an emergency contact, and ICD V- and Z- codes for social problems. Variables determined to be not widely available included: support person on day of visit, transportation issues, and detailed descriptions of social support. Marital status, religious status, living status, presence of an emergency contact, and presence of ICD codes for social problems were each related to less than median levels of social support in Pathways. All indicators predicted subsequent patient mortality except living status which modified the association of the other predictors. Adjusted for age at diagnosis, race/ethnicity, and AJCC stage, compared to women with four points (high support) (reference), women with 3, odds ratio (OR)=1.31, 95% confidence interval (CI): 1.21-1.43; 2, OR=1.62 (1.46-1.80); and 0-1 (low support), OR=1.96 (1.54-2.49) points, had elevated risks of mortality; the association was larger in those who lived alone. We had adequate statistical power to explore associations in racial/ethnic subgroups and in women with triple negative breast cancer. Conclusions: Variables available in the EHR may be used to measure social support in breast cancer patients.

Citation Format: Candyce H Kroenke, Elizabeth Eldridge. Use of electronic health record data to improve generalizability of social support research in cancer: CRN pilot study [abstract]. In: Proceedings of the Twelfth AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2019 Sep 20-23; San Francisco, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(6 Suppl_2):Abstract nr A002.