What does it cost, and how does it help the patient?

What Does It Cost, and Does It Help the Patient?

Dr. Eddy's presentation covered topics ranging from variations in diagnoses (observer variation) to the complexity of medical decisions, the limitations of clinical judgment, the lack of experimental evidence, poor design and reporting of research, variations in practice patterns, and discrepancies between the actions of practitioners and the recommendations of experts. Dr. Eddy argued that without good information on the health and economic consequences of alternative interventions, it is impossible to make intelligent choices about how to allocate resources or how to deliver optimal care to patients. He emphasized that until the medical profession develops a much better information infrastructure, efforts to improve the quality and efficiency of medical care will be severely hampered.

You paint a grim picture of the information system for clinical decision making. Both the information and the information-using processes seem essentially irrational. How could things have reached this state of affairs? How much whistle-blowing will be needed to turn things around?

It is puzzling that the information infrastructure for medicine could have remained so poor for so long. No doubt there are many reasons. I believe that one of the more important ones is that for centuries there has been an assumption that clinical judgment is a sufficient basis for medical decisions. Until very recently, we have assumed that clinicians, at least in the aggregate, are capable of synthesizing all the important information about patients, interventions, and outcomes and of determining intuitively what the appropriate practices are. If this assumption is accepted, there is little need for well-designed research, because clinicians can simply sense the best practices in their heads. Indeed, under this assumption, research might even be considered an impediment, because it takes up time and resources and can delay change. The "revolution" we are currently undergoing in the intellectual foundation of medicine is driven in large part by the realization that this assumption is false. The evidence is the growing body of research on wide variations in physicians' perceptions, the inability of the human mind to deal with the complexity of modern medical practice, wide variations in actual practices, and wide discrepancies between the judgments of experts versus the judgments of practicing clinicians. The net message of this research is that it is impossible for each clinician's judgment to be correct, if the judgments of each clinician are so different. When the vulnerability of clinical judgment is combined with the uncontrollable rise in costs, and the poverty of research evidence, the need to change is clear.

With respect to how much whistle-blowing will be needed to turn things around, I hope not much. The problems are so severe that I would hope that clinicians, researchers, and administrators will appreciate the need to improve the information base for clinical decisions. It is possible the profession will be stubborn on this point, but I do not believe that will be the case. The reaction thus far from the medical leadership has been extremely responsible and encouraging.

One more point. The term whistle-blowing implies that the current deficiencies in medical information have been deliberate and that someone is at fault. I do not believe that to be the case. The problems we currently face are due to a long tradition and forces that run far deeper than the motivations or actions of any particular individuals. This problem is no one's fault, and there is no role for pointing fingers or finding villains.

Outcomes management would seem to require even greater reliance on scientific information than has heretofore been the case in health care delivery. How can we be certain that past errors in research will not be visited on new research on the value of medical interventions?

In fact, outcomes management is an attempt to improve the scientific information. It calls for the collection of better information, not just better use of existing information. This said, it is still possible that past problems in research methods will affect new efforts to improve information about medical outcomes. There are several reasons to be optimistic, however. Most important, outcomes management is inherently different from the traditional approach to research in that outcomes management is a concerted, coordinated effort. There is more leadership, both in the conduct of the research, and in the selection of methods. While we will still encounter methodological problems, and while no research will ever be perfect, the research methods of outcomes management do promise to improve on what is currently happening.

Physician executives can play a role by appreciating the need for data collection on outcomes as an integral part of delivering high-quality care, and by providing the institutional support for better data collection. Physician executives can also help coordinate research across institutions. It is likely that, some time in the coming year, physician executives will be asked to have their institutions participate in a concerted national effort to measure outcomes. I would hope that they say, "yes."

How do your proposals and those of Paul Ellwood, MD, compare and contrast?

Dr. Ellwood and I are concerned about the same things--the quality of information available for decisions, and the processing of that information accurately to make better decisions. If we differ at all, it is that currently he is working with various organizations to put in place the institutional mechanisms to collect data on outcomes. My emphasis has been more on determining precisely what data should be collected, and how it should be processed after it is collected.

An example might help make the distinction. Suppose the problem is to develop policies for the treatment of hypercholesterolemia. An integral part of the solution is to collect data on the outcomes of large numbers of patients who do and do not receive drug therapy, in a variety of settings. But to design a policy for a population, data must also be collected on the demographics of the population, the prevalence of various risk factors, specific characteristics of the patients who do or do not receive treatment, the costs of treatment, the cost of coronary artery disease (CAD), and so forth. Furthermore, models must be built to predict the risk of CAD as a function of patient characteristics, the number of events that can be expected in a population, the afbect of specific treatment strategies on the expected number of events, and so forth. Additional models are needed to determine an "optimal" policy that identifies the particular patient characteristics and risk factors that maximize the effect of treatment for a specified budget. The analytical process begins with the collection of data, but raw data by themselves are like the cloth from which a tailor makes a suit; the value depends critically on how it is processed.

The public is now bombarded with reports on medical techniques and equipment. They also receive a lot of information about lifestyle issues. There are conflicts in that information. What responsibility does the medical profession have in ensuring that the public is lroperly informed about health care matters? What role should the profession play in convincing the public that it has no need to know everything?

Conflicts in information are a very important problem. To a certain extent, discrepancies in the information and recommendations provided by different authors are due to differences in philosophies or values. However, to a very great extent, the discrepancies in messages are due to poor information. That is, the range of uncertainty about important medical problems is frequently so wide that there is plenty of room for reasonable people to come to different conclusions, with no evidentiary basis for deciding who is correct. The solution, once again, is to improve the information basis, and the methodological tools used to analyze medical practices and make recommendations to the public.

With respect to how much information the public should be given, that is a complex issue that has to be addressed on a case-by-case basis--disease-by-disease or intervention-by-intervention. In some cases, the public will want to know everything about a decision, warts and all. In other cases, the decision will be so complex and emotion-laden that the public will need considerable guidance.

While there is a lot of talk about quality these days, the main issue, particularly for payers, is cost. How would your proposals affect health care costs?

I believe that a good information infrastructure is essential to controlling health care costs. The principal problem of the coming decade will be to determine how to preserve the quality of care to the greatest extent possible while reducing costs or at least controlling costs to acceptable rates of increase. I cannot see any way to achieve this until we know the actual quality (health outcomes) and the actual costs of different medical interventions. At present, we have that information for exceedingly few interventions. (They can be counted on two hands.) Imagine trying to decide whether to build a new wing of a hospital without knowing how many more patients could be served, the population projection for the community, what new staff would be required, the projected revenues or the cost. Yet we make national recommendations about medical practices without ever estimating the comparable information. The starting point for solving the cost problem must be to get information on the consequences of different medical interventions. Without that information, we are doomed to continue recommending interventions simply if they "might" have benefit, without knowing how much benefit, how much cost, or whether our resources could be put to better use somewhere else.

What specific involvement do you see in all this for physician executives?

Clearly physician executives can play an important role. First, they can help practitioners, experts, and researchers within their organizations understand the need for and value of better information. One simple way to do this is to require people who ask for resources to support some project to describe the expected health and economic consequences of what they are requesting. Ask them to develop the equivalent of a business plan, or an environmental impact statement. This will force the requesters to explore the evidence and analysis needed to justify their project. When they do that, they will appreciate in a hurry how little they can actually say and how difficult it is to actually document the value of projects they would prefer to take for granted. It is only when people understand at a gut level how poor the information base is that they will begin to support the need for better information.

Physician executives can also help support coordinated, statewide or nationwide programs to collect and process data on outcomes. They might well be asked to provide money from their institutions to support a national effort.

Finally, physician executives might be in a position to change the incentives that determine what practitioners, experts, and clinical researchers do, and to help them appreciate the limitations of what they know. Most clinical researchers are still judged by the number of papers they've managed to get their names on, rather than on whether the research actually improves anyone's health. Most experts still believe that a mere statement by them is sufficient reason to swing millions of dollars into their pet projects. Most practitioners still believe that their personal perceptions are accurate. I have often pondered the irony that as a society our procedures for determining whether Mr. Smith or Mrs. Smith should get the silverware after a divorce, or whether a vacant lot can be turned into a parking lot, are far more rigorous than our procedures for determining whether billions of dollars should be spent on some treatment. I'll bet physicians executives can do something about that.

Wesley Curry is Editorial Director of Physician Executive.