Without a doubt, the EMR makes our lives easier. Notes are legible and in chronological order. They can be sent electronically to other providers. Most systems can link notes to imaging and laboratory results. There is another major advantage for using electronic records: the opportunity to harness the data for research. Take, for example, the use of patient data mining using search-engine technology. Work done in this field by Kenneth Mandl, MD, MPH, associate professor of pediatrics, and Isaac Kohane, MD, PhD, professor of pediatrics, at Harvard Medical School in Boston, has demonstrated the utility of this technology. They studied whether data mining could identify adverse medical events occurring in large populations by searching the electronic medical records of patients seen during the previous ten years at Brigham and Women’s and Massachusetts General Hospitals, both in Boston. After searching the full database, they then focused on patients with coronary heart disease. Analyzing demographic information, billing codes, visit dates, medication histories, and diagnostic data, and using natural language processing and other search tools, they identified those patients who had taken rofecoxib and whether they had suffered myocardial infarctions (MIs). They observed a nearly 20% jump in the number of MIs just eight months after the release of rofecoxib, an effect that vanished within a month of the drug’s withdrawal from the market.
Data mining as a tool in rheumatology research is moving forward, too. Some of my rheumatology colleagues at Brigham and Women’s—Katherine P. Liao, MD, Robert Plenge, MD, PhD, Elizabeth Karlson, MD, and Soumya Raychaudhuri, MD, PhD—have developed computer algorithms that can accurately identify patients with rheumatoid arthritis (RA) in very large databases, even across different EMR systems. When such information is linked to biospecimen repositories, the opportunities for research are endless. Some of their current studies in RA patients include the identification of genetic predictors of the response to anti–tumor necrosis factor (TNF) therapies and the genetic predictors of lipid levels. By combining billing data with clinical data from the EMR, Sonali Desai, MD, has led a study within our division that can assess the pneumococcal vaccination rates of immunosuppressed patients. Using this data she created a quality-improvement intervention that has increased vaccination rates from 50% in 2008 to more than 80% in 2012.
Back To the Future
The tension between the need to document the patient’s medical condition versus the constraints of the medical record persists. Dr. Siegler described how changes in medical record structure—in particular, the compromises required to manage data and improve efficiency—dramatically altered what physicians wrote more than a century ago. Faced with a rigid system, the physicians who created these records responded by conforming generally to the structural constraints but, when necessary, they found ways to break free of these constraints. As Dr. Siegler stresses, the transformation of these records mirrors the challenge that we face today when using the EMR—managing information in a way that does not discourage expression and thoughtful analysis while fulfilling the more mundane aspects of data collection and record documentation. We need to become the custodians of the medical record who will be remembered for keeping the art of medicine alive in every chart.