Once guidelines are embedded in the work flow, it will be much easier to apply them than it is when physicians must essentially take them off the shelf and look them up.
When he was 10, Jim Schibanoff watched a physician save his father's life by correctly diagnosing and treating a ruptured appendix. That act of healing inspired him to become a physician. When he graduated from medical school in 1969, Schibanoff looked forward to a career as an office-based internist.
Today, about three careers later, Schibanoff is in his eleventh year as editor-in-chief of Milliman Care Guidelines, a Seattle-based unit of Milliman, the consulting company that has produced evidence-based clinical guidelines since 1990. At his San Diego office, Schibanoff supervises editorial and peer review of guidelines and analysis of the evidence base that underpins them. He also appoints and coordinates expert clinician panels, plans new guidelines, and revises existing ones.
Between leaving his IM practice and joining Milliman, Schibanoff's other careers included 18 years as a hospital intensivist, after which he was chief medical officer of a health system in San Diego, and, for three years, CEO of two hospitals. Schibanoff is a graduate of Princeton University and the University of Southern California School of Medicine, and received his postgraduate training at the University of California–San Diego. He spoke recently with Senior Contributing Editor Patrick Mullen.
MC: Is there much doubt remaining among physicians, health plan executives, or other health care stakeholders that wise use of clinical guidelines, Milliman's or others, makes good sense?
SCHIBANOFF: Guidelines are uniformly accepted, but only when they're built on a valid evidence base and have strong functionality that enables them to be translated into action at the point of care. Validity is totally contingent upon the quality of the evidence, and we're becoming increasingly aware of significant gaps in the evidence. The most extreme examples involve patients with multiple chronic diseases. It's technically difficult to write a guideline that addresses all of those contingencies, and the evidence base itself is lacking. Randomized controlled trials — which tend to exclude these complex patients — are the gold standard of evidence. A randomized controlled trial for a common condition ends up excluding most people that have heart failure because they have one or more comorbidities. Research studies always seek to compare apples to apples, which isn't possible with these different subgroups, who happen to be the patients who require the most extensive and expensive care. It's a real dilemma.
MC: What's the current state of the art in the technology behind guidelines?
SCHIBANOFF: Right now, electronic health records are good clinical data repositories, but that's not enough and it needs to continue to evolve. When a provider wants the results of a chest X-ray or blood test, he queries the system and receives that result right away. That's the clinical repository function. Another stage is documentation, so that things like vital signs get entered into the electronic health record. That is moving along well. Beyond that — and this is more promise than reality — are the decision support tools that are being developed. Guidelines are an important part of the guts of those tools. Once guidelines are embedded in the work flow, it will be much easier to apply them than it is when physicians must essentially take them off the shelf and look them up as a reference tool. Having guidelines as part of a personal health record will be a great advance in guideline adherence. The issues with PHRs are interoperability, so that data can be shared, and portability, so if you are getting your insurance from Aetna this year but your employer switches to UnitedHealth next year, you still keep your own PHR and all that information.
MC: Which health systems or managed care plans are leaders in developing these tools?
SCHIBANOFF: The usual suspects: Group Health Cooperative of Puget Sound, Kaiser, the Deparment of Veterans Affairs, academic centers like Harvard Medical School. Innovation is occurring in a lot of places, but not yet broadly across the health care system.
MC: How might use of clinical guidelines evolve over the next several years?
SCHIBANOFF: The hope for the future is with the concepts of a learning health care organization and patient-centered research, so that organizations can capitalize on the huge amount of data being captured in electronic health records. There are substantial methodological challenges and it will take a lot of research to know what data actually are valid in reaching decisions about patients, but it offers great promise. The promise of technology is in embedding guidelines in the artificial intelligence at the core of a computerized decision support tool within an electronic health record at the point of care. Kaiser is starting to exploit its huge database, which is how they helped discover the association between cox-2 inhibitors and thrombotic events.
MC: Is Milliman working with Kaiser or is Kaiser keeping its research in-house?
SCHIBANOFF: They're doing it in-house. We'd love to get access to that data but we don't have that.
MC: What concerns about guidelines are you hearing from practicing physicians?
SCHIBANOFF: Physicians are learning that guidelines do not replace clinical judgment. They require clinical judgment because there are deficits in the evidence. Physicians have to be able to make inference leaps and apply the available evidence to the care of an individual patient. I like to point out to people that between 2,000 and 4,000 references are added daily to Medline through the National Library of Medicine's library service. One key role of guidelines is to take all that evidence and translate it because there's no human way to sift through that much information. What I used to hear from multiple sources was, "Where'd that guideline come from?" We have gone to great pains to disclose and analyze our evidence sources. It is definitely not black box technology. We provide evidence summaries and footnotes and an evidence hierarchy. Inevitably there will be disagreements on interpretation of the evidence. That's normal and healthy and to be expected. When I was in medical school, we always argued about whether this or that article conclusively proved a point. That's to be desired. The most common question I hear now is, "Can you write a guideline for this or that?" This has to do with the explosion of technology and the fact that physicians have to make complex decisions around the use of new technology such as imaging and injectables.
MC: To what do you attribute the change in physicians' acceptance of guidelines? Is it generational?
SCHIBANOFF: I wouldn't call it generational. It's a maturation of several different trends. We have probably gone through a change process not unlike the stages of grief, from denial and anger through bargaining and depression. Now we're getting to acceptance, driven by improvements in the science of measurement. Clearly, use of guidelines will be integrated with pay-for-performance efforts, so there's a financial incentive.
MC: How quickly can you turn around a valid guideline?
SCHIBANOFF: How good is the available evidence? In the case of a new technology that's FDA approved but for which the evidence is not strong enough to write a definitive guideline, we often will say, "Here's a summary of the evidence, but we're still not sure of the current role," and we have to leave it at that. The evidence is just not conclusive to say this will be a definitive therapy or will replace some other treatment.
MC: So will there be treatments targeted at fairly small groups of potential patients for which there will never be enough evidence to build guidelines?
SCHIBANOFF: Absolutely. High-cost injectables are the current biggest example.
MC: What are some other barriers to physician acceptance of guidelines?
SCHIBANOFF: One barrier is physicians' perception that they already practice superb health care. So another exciting development is an emerging consensus around which outcomes should be measured, so physicians can see how accurate that perception is. This consensus started with groups like the NCQA, and includes the National Quality Forum, a clearinghouse for outcome measures. The Joint Commission is doing similar things. The Ambulatory Quality Alliance, AQA [pronounced aqua], is working with individual specialties so that physicians can decide which outcomes should be measured. I attended an AQA meeting as an observer and I was impressed with the sense of purpose and the collegiality with which the discussions went on. I think this is a real change. So for example the American Academy of Neurology would develop and recommend outcome measures, which then get blessed and endorsed by AQA, which has ties to the Agency for Health Care Quality and Research, and to Medicare and Medicaid, which integrate these measures.
MC: Once Medicare and Medicaid are on board, there's your standard.
SCHIBANOFF: The federal government is the world's largest health insurance company, yes.
MC: Describe how information systems that can record the results of guideline application could change the practice of medicine.
SCHIBANOFF: It would close the feedback loop. Any physician applying a guideline for heart failure whose outcomes are undesirable would want to re-think the guideline. Believe me, physicians live with bad decisions for the rest of their lives. I think about my bad decisions a lot more than I think of my triumphs. When it comes to trusting outcome measures, there's a difference between applying what's in the literature to your group of patients, which is what happens today, and knowing how that evidence actually works on your population of patients, which will happen in the future.
MC: So within a given system of care, physicians will look at actual outcomes of clinical decisions made by their colleagues, people they know and trust, and see what they've done and how it's turned out?
SCHIBANOFF: Precisely. Physicians will have the current best information at their fingertips and have a feedback loop to tell them how well each treatment worked.
MC: Thank you.
The interviewer, Patrick Mullen, is a former managing editor and editor of Managed Care. He writes from his home outside Cleveland.