A summary of the workshop was provided by Dr. Michael McGinnis, who is a leader of IOM’s initiative on evidence-based medicine. He began his summary by noting that traditional linear learning has to be replaced by the rapid network learning that typifies the current computer-driven world. “We live in a network world in which we can gain information in a much more rapid fashion, and we can vet that information in a much more rapid fashion too. So it is clearly fundamentally important we move our learning process in health care considerably into the twenty-first century with the development of a learning healthcare system,” he said. This is a system in which learning is a natural by-product of the care process and is applied as a routine part of the care structure, Dr. McGinnis added. It’s also a system where there is a global collection of data from multiple sources—the “notion of a global clinical data trust, the theoretical notion that eventually we would be able to capture data not just from one institution or another or from Kaiser or Mayo or Harvard or the VA, but from many institutions globally”—that will tremendously accelerate the learning process. Patients have the potential to be leaders in moving this learning process along, Dr. McGinnis said, and need to be vital partners in that respect because they are clearly motivated.
There are barriers and challenges in developing the appropriate level of confidence going forward, including justifiable concerns about how reliable observational data can be. Dr. McGinnis raised the question, How do we
understand exactly when we have arrived at a credible and an appropriate lesson in the learning process to apply it to patient care? It is also important to structure the data-gathering process to anticipate, plan for, and execute the collection of data in a systematic fashion to capture the lessons learned and feed those lessons back into the learning enterprise to improve the system’s performance, he said.
However this endeavor has numerous challenges, Dr. McGinnis pointed out, including legal, regulatory, fiscal, and professional. He reiterated the potential of regulatory agencies, such as Medicare and the FDA, to spur the development of a cancer RLHS. Medicare has the ability to “transform our mind-set about the way in which every clinical encounter ought to add to the learning process,” he said, and added that it is encouraging that the FDA, which has traditionally limited itself to the premarket domain, “is now working hard to engage the postmarket domain. So we have the right perspectives, insights, and inclinations on the part of the leadership to act on some of the exciting activities that we heard about in the course of the meeting.”
He then described six basic elements of the continuous learning process:
Capturing the experiences of every clinical encounter
Developing consensus and guidelines based on the experiences that have been captured
Validating the various guidelines that have been developed
Delivering care based on those guidelines
Ensuring that care is standardized and harmonized, while controlling variation that allows for innovation and the generation of new information
Creating natural feedback loops so the results captured are evaluated and fed back into the system for learning and improvement purposes
Because the modern network approach to learning is nonlinear, as opposed to the more traditional linear approach, “one of our biggest challenges is to force ourselves to look specifically at those intersecting dynamics, at each of those points in the feedback process and ensure that the elements necessary for the success of the activities at each of those points are given consideration,” Dr. McGinnis said.
Dr. McGinnis provided some take-home messages, noting that we
have the capacity to gather large amounts of data and use the EHR in a fashion that is systematic, sensitive, and appropriate. “The three Rs of learning now are not reading, writing, and arithmetic, but research-ready records so that we are thinking ahead about the kinds of information that are needed to ensure the learning process.” He pointed to the need we have as a society to give much more energy and attention to the processes of care and the adaptive dynamics in those processes. “If we are going to develop the incentives that will change the state of play on the front line, we have to spend a lot more time understanding those [care delivery] dynamics,” Dr. McGinnis said. He also emphasized the importance of improving the value of health care, since we live in a world that is starkly constrained in terms of resources.
Dr. McGinnis agreed with the need for public policy changes and noted that nongovernmental groups could also “take ownership of the learning obligation.” “You have in the cancer community the patients, the providers, and the motivations to serve as the leading edge of progress and to engage decision makers in an explicit exploration of the issues, opportunities, and responsibilities when it comes to improving access to data, which is a fundamental issue for a learning healthcare system. With the passionate testimony that you have all witnessed firsthand of patients with whom you deal, you really have the opportunity to lay out in a much more compelling fashion than any other professional discipline, how shared decision making can make a difference in the learning process and turn the process of exploration [of a RLHS] into a process for action,” Dr. McGinnis said.