Rand researchers first discovered wide variations in 123 medical procedures for Medicare patients in various parts of the country. The rate at which the procedures were performed varied by as much as 6, 7, or 8 to 1, with no apparent explanation. Areas that were high in performing one procedure were often low in performing another. The Rand study was consistent with other studies of non-Medicare patients, which have found widespread variations in medical practice across geographical areas for several decades. Just knowing about the variations, however, did not reveal whether some patients were being shortchanged and others overtreated.
Consequently, a follow-up Rand study collected medical records for 5,000 Medicare patients and convened a panel of experts to judge the appropriateness of three procedures. The results (as reported by the national news media) are shown in Table I. As the table shows, in more than a fifth of the cases, the procedure performed was judged to be inappropriate and therefore unnecessary. For carotid endarterectomy (the removal of plaque in major arteries to the brain), the procedure was judged to be appropriate only about one-third of the time.
Before we jump on the doctors, itís worth noting that in some ways the Rand study was unfair. Suppose we convened experts in your field and asked them to review decisions you have made. Would they agree with every decision? You might respond, as the doctors do, that you didnít have the opportunity to consult with a panel of experts before you made your decisions. A second problem with the Rand results is the way in which they have been reported. What Rand means by "equivocal" is that a majority of the experts couldnít agree. But in newspaper reports, the word "equivocal" often became "questionable" (as in "40 percent of the procedures were either inappropriate or questionable"), which is not the same thing. Equivocal means that not performing the procedure is just as problematic as performing it. A third problem is that media reports of the Rand study obscured the actual extent of disagreement and uncertainty in the medical community. The reason why Rand had to convene a panel of experts was that researchers could not answer questions about appropriateness by merely consulting the medical literature. Once the experts were convened, they were far less unified than is commonly known.
Table II presents a different way of looking at the Rand study, showing the number of times that 7 of 9 experts agreed (the two opinions ignored are the two most extreme views, on either side of the middle). As the table shows, when 7 of 9 experts were asked to agree, they found only 12 percent of the procedures to be inappropriate, not 22 percent. And even this degree of consensus is misleading. In the Rand study, each expert initially expressed a personal judgment. Then they met in group discussions (where group pressure had an opportunity to forge a consensus), after which several members often changed their minds. Indeed, the most remarkable fact about the Rand study was that even with all of those efforts to arrive at a definitive judgment, 7 of 9 experts could agree that the procedures were either definitely appropriate or definitely inappropriate less than half the time.
So when Rand spokesmen state that 40 percent of surgery is "unnecessary," thatís a personal point of view, not the unanimous conclusion of experts. Rand researchers have adopted the viewpoint that, if physicians canít agree that surgery should be performed, it should not be. Rand research, however, shows something different. The fact that the experts couldnít agree in half the cases tells us much more about the state of medical science than about the state of medical practice. A fourth problem is that the data from the Rand study were for 1981, more than a decade ago. Generalizing about todayís health care based on what happened in 1981 is clearly wrong. The 1980s produced major changes in the way hospitals are run. For better or for worse, American physicians today are scrutinized more closely by peers and third-party payers than physicians anywhere else in the world. There are probably still cases in which experts would agree that the surgery promised more harm than good, but those cases are likely to be presented to disciplinary committees and are probably far less than the 12 percent shown in Table II.
A fifth problem is that Rand researchers tried to develop a purely medical test to determine whether surgery should have been performed, when the real problem they were trying to address was an economic one. Their study compared the medical benefits of procedures with the medical harm. However, as noted above, cases in which the expected medical harm exceeds the expected medical benefit are probably quite rare. The real question is whether procedures are worth their cost. And the only people who can accurately answer that question are patients who can choose to spend that same money on other goods and services.
A final problem is that the Rand research is frequently used by advocates of socialized medicine to criticize the U.S. system of private medical care. What the critics fail to mention is that countries with national health insurance also experience wide variations in medical practice. In Britain, for example, physicians have no direct financial interest in performing any medical procedure. But the rate at which British general practitioners refer patients to hospital physicians varies by at least 4 to 1 and, according to one study, by 25 to 1. Moreover, there is a high correlation between patient referrals and subsequent hospital admissions.
What can we conclude? The most important implication from the Rand study is that medical practice is still more an art than a science, and that when physicians are faced with difficult choices, they may not be able to get firm direction either from the medical literature or from a national panel of experts. A second implication is that a small number of physicians may be systematically putting their patients at risk by using clearly inappropriate procedures - and, if that is the case, something should be done about it. A third implication is that patients who are told they need a medical procedure should ask questions before agreeing to it, especially if the procedure is expensive or risky.
Can medicine be made more scientific? Some apparently think so. The American Medical Association and the Rand Corporation are working to develop "practice guidelines" for physicians considering certain procedures, and Congress has mandated that the Department of Health and Human Services draw up similar guidelines. The goal is the development of "computerized protocols" that will let physicians know what they should do when confronted with certain patient symptoms and conditions. Will the guidelines work? Thatís not clear. Many people believe they will be a waste of money. Some argue that their development is such a lengthy process that medical science will have outpaced them by the time they are available. In other words, computerized protocols will always be years behind scientific developments in medicine. Others raise the philosophical objection that computerized protocols assume correct medical procedures usually can be determined by a computer program, which obviously has never met or talked to the patient. Studies have not borne out that assumption, however. In one test, judgments of general practitionersí were matched with three different computerized protocols in the treatment of patients with abdominal pain; the GPs outperformed the protocols in every test.
If workable computerized protocols were available to physicians, they might prove to be valuable tools. A physician could consult the computer, then substitute his own judgment where appropriate. Less complicated protocols might become available to patients for use on their home computers, giving advice on whether to see a physician, for example.
Other the other hand, if computerized protocols and practice guidelines were used to control the behavior of physicians and patients, they could threaten the quality of medical care. And, unfortunately, that threat is real. Researcher Robert Brook has argued that the Rand Corporationís techniques can be used to ration health care under the Medicare system, if Medicare funds run short.
Lurking behind the public discussion of practice guidelines is a fundamental difference of philosophy that is rarely discussed in print. The bureaucratic view of health care is usually also a technocratic view. Its more extreme proponents are fundamentally antiphysician and antipatient, in the sense that they believe the attitudes and judgments of individuals are largely irrelevant. Ultimately, the technocrats do not see the computer as an aid to physicians and patients but as a substitute. They envision medical practice being literally dictated for the country as a whole from a central location in " well, Washington.
Although the discussion of practice guidelines frequently is couched in terms of helping physicians make good decisions, the technocrats also see the guidelines as a means of exerting control. In their view, physicians who substitute their own judgment for the computerís should have to prove that they are right, which would require them to use cumbersome and costly bureaucratic procedures. As a result, instances in which the guidelines are not followed would be rare and unusual events.
In an ideal system, patients would become far more involved in the decisionmaking process. They would have new opportunities to learn about the potential costs and benefits of medical procedures and to make decisions based on their evaluations of those costs and benefits. The role of the physician would be to help patients make these difficult decisions, based on their own values. To the extent that computer programs can be a real aid to patients and physicians, they would be used for that purpose. But people would not surrender their decisionmaking authority. Most importantly, patients would not be told there is only one correct way of treating a condition when the opinions of physicians vary. When professional opinions diverge, patients would be the first to know. Most unnecessary surgery today is unnecessary only in the sense that itís not worth the cost. In an ideal health care system, patients would be encouraged to weigh its benefits against its costs and have the option of forgoing the procedure and spending the money elsewhere.
WHEN NO TREATMENT IS BEST
The best course of treatment for a disease in some cases is no treatment at all, say experts. Asymptomatic prostate cancer is a good example: a 50-year-old American male has about a 10 percent lifetime risk of prostate cancer, but the lesions grow so slowly that many elderly men die with the disease, but not because of it.
While the disease can be deadly and is more curable if detected early, the potential harm of screening all American men over 50 includes the risks associated with unnecessary surgery for thousands who are disease free or whose malignancy is clinically insignificant. Experts say that the billions of dollars required for widespread screening would likely divert resources from health care services with proven benefits.
Studies of tests commonly used to screen for prostate cancer in the absence of symptoms show that:
The cost-effectiveness and risk-benefit of alternative treatments for prostate cancer that hasn't spread has also been measured. Studies have found that:
For prostrate cancer that has spread to distant organs, another study found that survival increased by 4.3 quality-adjusted months by using drugs and/or surgery, at an incremental cost of $25,000 per quality-adjusted life-year saved.
Source: Marilyn Dix Smith, et al., "Prostate Cancer: Sometimes It Pays to Do Nothing," Business & Health, January 1997.
Home | Support Us | All Issues | Social Security | Debate Central | Contact UsDallas Headquarters: 12770 Coit Rd., Suite 800 - Dallas, TX 75251-1339 - 972/386-6272 - Fax 972/386-0924
Washington Office: 601 Pennsylvania Ave. NW, Suite 900 South Building - Washington, DC 20004 - 202/220-3082 - Fax 202/220-3096
© 2001 NCPA
