Linking Hospital EMR and Cancer Registry Data to Compare Breast Cancer Treatment and Patient-Centered Outcomes

BACKGROUND: Hospital electronic medical records (EMRs) contain patient health, diagnostic and treatment information beyond the level of detail collected by central cancer registries. Linking hospital EMRs with cancer registry data enhances existing registry records and provides evidence for patient-centered outcomes research. These analyses have important implications for treatment effectiveness, especially as it relates to high risk cancer patient sub-groups, and comparative effectiveness research investigations are made more robust with the collection of patient biomarker data such as Estrogen Receptor (ER), Progesteron Receptor (PR), and Human Epidermal Growth Factor Receptor 2 (HER2) status.

METHODS: A large Florida hospital system shared EMR records with the Florida Cancer Data System (FCDS) for patients diagnosed with and/or treated for invasive breast cancer between 2007 and 2011. Standard medical ICD-9-CM diagnosis codes were used as a trigger mechanism to identify breast cancer patients. EMR patient encounter data included patient discharge summaries, clinical history, medication orders, pathology reports, and diagnostic and treatment procedure codes. These data were linked with registry data and used to perform analysis of treatment outcomes with regard to tumor biomarker status.

RESULTS: A total of 12,804 tumor records from FCDS were linked with hospital EMRs, resulting in data from 11,506 unique patients with breast cancer which include patient chemotherapy, radiation and surgical information. Biomarker results for ER, PR and HER2 test results were available for 1,090 patients, 13% of whom had “triple negative” status, representing a clinical predisposition for needing aggressive chemotherapy.

CONCLUSIONS: Differing patient sub-groups have malignancies that may pose a range of special challenges to diagnosis and treatment. Hospital EMR data provide detailed patient treatment information that may not be included in routine cancer reporting. These clinical data are useful for analyzing treatment effectiveness as well as understanding opportunities for improving outcomes for patients with hormone-positive tumors and patients in other high-risk sub-groups requiring specialized therapies.