At last week’s AcademyHealth’s Concordium 2015 meeting in Washington, D.C., several themes emerged around how to better include communities of patients in research, how to use patient-reported outcome data, and how to move toward the systematic incorporation of research and quality improvement into the organization.
These may seem like academic topics to some, but given the strength of the “data liberation movement” and the increasing focus on patient engagement at the federal level, I predict CIOs at all levels will soon be required to think more about these emerging elements of a learning health system.
A workshop focused on “research participation in a learning health system” featured a great talk by Peter Embi, M.D., of Ohio State University’s College of Medicine, where he is an associate professor and vice chair of biomedical informatics as well as associate dean for research informatics and the chief research information officer at the Wexner Medical Center. Dr. Embi outlined some of the challenges with the traditional approach to evidence-based medicine: basically that it is a research/practice paradigm where the information flow is unidirectional, and clinical practice and research are distinct activities, with the research design as an afterthought. “We want to leverage information at the point of care and in engagements with patients so we can systematically learn. That is what the learning health system is all about,” Embi said.
But in the current model, he noted, there is little consideration of research during planning of health systems. That limits the ability to invest in and leverage clinical resources to advance research. Also there are no financial incentives for non-researchers to engage in research. Research as an afterthought also leads to regulatory problems and wasted investments.
Embi and others have been arguing for moving from “evidence-based medicine” to an “evidence-generating medicine” approach, which he defined as the systematic incorporation of research and quality improvement into the organization. Rather than findings flowing only from research done looking back at historical data, this approach creates a virtuous cycle where clinical practice is not distinct from research.
He gave an example from an effort at Ohio State to create a local learning health system, and solve some of these issues at the micro level. In Embi’s own rheumatology department, one of the first steps was to standardize on approaches to collecting data. They had approximately a dozen rheumatologists gathering data in 12 different ways. “It was not allowing us to compare how we are caring for patients with rheumatoid arthritis, much less cull that data for research purposes,” he said.
They pared down the necessary data elements to 50, and 43 of those were already in the EHR. So they added seven data fields and adjusted work flows so that those working at the top of their license, not just physicians, could gather and enter the necessary data. By addressing regulatory and cultural issues, they enabled data collection as an expectation among the team. He said that although each specialty would have different data elements and work flows to consider, the approach his department took could be standardized.
Embi said he has also run into examples of clinicians who say helping to recruit patients into clinical research projects is not their job. His first reaction was that OSU is an academic research center with a research mission. But he realized that these clinicians are right in one way. They are paid for RVUs or relative value units. That is what they get judged on. “If we want them to spend time on research, shouldn’t we pay them?” Embi asked. Maybe a relative research unit is needed to measure the activity of recruiting patients.
“We as researchers need to find better ways to do these activities and design systems that enable them," Embi said.
Another panel session, called “Bringing Engagement to Scale: Patients, Partners, and Participants” also raised some interesting questions about how difficult it still is for patients to get access to their medical data when they want to participate in research. Megan O’Boyle is principal investigator of the Phelan-McDermid Syndrome Data Network and the Phelan-McDermid Syndrome International Registry, part of the Patient-Centered Outcomes Research Network (PCORnet). Her daughter has Phelan-McDermid Syndrome, which is at the severe end of the autism spectrum.
She said the idea for the network was to get patient records on a Flash drive, send them to Harvard, where they would use natural language processing (NLP) to pull the relevant data into a registry. “The NLP and registry have worked beautifully,” she said. Getting the records from the 16 healthcare facilities her daughter visited proved much more difficult for her. Every organization had different rules, and wanted to charge 50 cents to $1.50 per page. “I gave up,” she said. She and others in the network eventually contracted with a third-party company called CareSync to act on their behalf getting the data from providers. “I don't understand why in 2015 I'm hiring a third party to get medical records for a chronically ill child,” O’Boyle said.