Breast cancer treatment across health care systems: linking electronic medical records and state registry data to enable outcomes research

This is a review of Kurian's article "Breast cancer treatment across health care systems: linking electronic medical records and state registry data to enable outcomes research".

Introduction/Background

Cancer outcome research has been limited by the fragmentation and lack of detail in variable data. Linking EMR-derived data across health care systems and population-based cancer registries offers the promise of more complete information. This paper analyzed the breast cancer diagnosis and treatment data from three sources: electronic medical records (EMRs) from two health care systems and the state registry. This 3-way data linkage offered a practical approach and provided additional information regarding the variability in cancer care.

Materials and Methods

Breast cancer diagnosis test and treatment data from January 1, 2000 through January 1, 2010 were extracted from University (Stanford University Medical Center) and Community (Palo Alto Medical Foundation) EMRs. The authors requested California Cancer Registry (CCR), a SEER component. CCR and EMR records were linked using names, social security numbers, medical records numbers and birthdates. They generated two separate University and Community patient Cohorts and then they linked these 2 EMR-CCR cohorts to identify patients treated at both institutions.

Results

The authors initially identified 8210 University analytical patients and 5770 Community patients; linked records identified 12,109 records in combined analytical cohort where 1902 records (16%) are overlapped records. This group of patients were treated at both University and Community facilities (defined as “Both” patients). Increasing percentage of Community patients but not University patients fell into “Both” patients category as prognostic factors such as decreasing age, increasing stage of disease, increasing grade and less favorable receptor subtype worsened. Before linking the data sets, Community patients appeared to receive less intervention than University patients (mastectomy: 37.6% vs. 43.2%; chemotherapy: 35% vs 41.7%; magnetic resonance imaging: 10% vs 29.3%; and genetic testing: 2.5% vs 9.2%). Linked data sets revealed that patients treated at both facilities received more interventions (mastectomy: 55.8%; chemotherapy: 47.2%; magnetic resonance imaging: 54%; and genetic testing: 10.9%).

Conclusions

This 3-way data linkage revealed 16% patient overlap between two health care systems and reduced missing data and generated unique insights. The “Both” patents received more intensive treatment than others despite comparable prognostic factors. Integrating cancer data from EMRs and population-based registries broadened the understanding of cancer care beyond what could be achieved from just one or two data sources. This approach can advance comparative effectiveness and outcome oncology research.