A "Rapid-Learning" Health Care System

    • January 18, 2007

The Rapid Learning Project, based at George Washington University and funded by the Robert Wood Johnson Foundation, is exploring national strategies to accelerate the pace at which we learn about best uses of new biomedical technologies, products and treatments. As more patient data are housed in electronic health records, and the power of the Internet and computing systems continue to grow, experts believe the U.S. can create a national system of real-time reporting of millions of patients' clinical experiences to national databases.

The ability to network across medical databases, thus accessing vast numbers of anonymized patient records, holds enormous potential for advancing collaborative health policy and clinical research. Ultimately, the aim is to gather information on new treatments, drugs and medical technologies much faster than is currently possible so that physicians can immediately apply the findings in medical practice and better tailor care to patients.

In the January 26, 2007, online issue of Health Affairs, researcher Lynn Etheredge, a consultant with the Rapid Learning Project, writes about the potential of a Rapid-Learning Health System and ways it can accelerate our ability to fill in gaps in the clinical evidence base. Here, Etheredge offers additional views on a Rapid-Learning Health System.

Q: What exactly is a Rapid-Learning Health System and how is different from what we have now?
Compatible electronic health records from the private and public sector can be merged into research databases and rapid-learning networks that contain clinical information on tens of millions of patients. With large databases and networks, we can learn from real-world experience and eventually do it in real time. In other words, we can rapidly develop new evidence, learn from it, and apply these learnings to medical practice and health policy.

Right now, physicians are trying to use an inadequate evidence base-built on a patchwork of small-sample studies and proprietary databases-to determine how to treat patients. Often, existing studies don't pertain to the 'typical patient' who walks into their offices.

Q: What is the ultimate goal for a Rapid-Learning Health System?
At the macro level, the long-term national goal for our health care system should be to learn about the best uses of new technologies and therapies as quickly as we develop them. Private-sector leadership organizations and the federal government will be the primary drivers for a Rapid-Learning Health System and national standards for the inter-operability of databases. As a health policy, the federal government should aim for a health care system that automatically learns about the best uses of new technologies and drugs-and procedures-as fast as they are developed. Currently, there is no system for evaluating new procedures and they represent a much larger portion of annual health care costs than do drugs.

At the micro level, a Rapid-Learning Health System makes it possible for physicians and patients to answer the question, "What does this drug or technology mean for a patient like me?" We can come very close to personalizing treatment decisions around drugs, procedures and devices.

Q: What can large research databases built around electronic health records tell us that we can not learn from randomized clinical trials?
Electronic health records contain information on all kinds of patients-including population groups not well-represented in randomized clinical trials: children, women of child-bearing age, people with multiple chronic conditions, as well as the Medicare and Medicaid populations, which include 85 million enrollees and account for $600 billion in federal expenditures annually.

There is an inferential gap in the evidence base. Patients come in who are very different from those in the clinical trials. They want to know how well a particular drug or procedure will work for them. Unfortunately, many drugs and procedures vary widely in how well they work in large population groups. This kind of information is not collected in randomized clinical trials. Clinical trials get to the average. In addition, the randomized clinical trial system is focused on market entry. It uses small groups of carefully selected patients to examine the effectiveness and safety of a new drug or device before it is brought to market. FDA [Food and Drug Administration] requirements on clinical trials for new drugs usually stop by 18 months, so we don't know the long-term effect of many of the most popular drugs people are taking.

With large databases created from electronic health records, we can understand how well new drugs and technologies work in the real world. We will be able to identify who they work for and who they don't.

Q: Are there organizations that are currently using electronic health records and are well on their way to becoming Rapid-Learning Health Systems?
In the private sector, Kaiser Permanente, and in the public sector, the U.S. Department of Veterans Affairs each has amassed databases containing records for 8 million-plus patients. There will be professional and marketplace competition to provide the best medical care using the best clinical evidence and information. Most large HMOs and many multi-specialty group clinics are also building electronic health record databases. Those who are most successful will be those who invest in this new information technology.