Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Economic Incentives and Clinic Demons Harold S. Luft The recent growth of for-profit activities- in medical care has led to concern about the growth of a "new medical-industrial complex" (Rel- man, 1980; see also Caper, 1982; Saward and Sorensen, 1982~. These concerns range from the fear that such enterprises will skim off all the profitable patients and leave the voluntary and public sectors with mounting bad debts to the fear that patients will lose their trust in the medical profession. There is also an emerging controversy over the possibility that for-profit health care entails some economic in- centives that may affect day-to-day clinical decisions. Will physicians who own hospitals, laboratories, or other for-profit enterprises, or who work for a national chain of hospitals or urgent care clinics, practice differently from those who do not? These concerns are based on a more general question of the extent to which economic incentives affect physicians' decisions about patient care. This paper is about clinical or patient care decisions, such as whether to order an X-ray for an ankle injury, rather than about production decisions, such as whether needed equipment should be leased or pur- chased. No one seems to question the notion that economic factors do (and probably should) influence production decisions, nor do people I am indebted to Bradford Gray, Susan Maerki, Victor Rodwin, Anne Scitovsky, Jonathan Showstack, Joan Trauner, and anonymous referees for their helpful comments on earlier drafts. 103
104 HAROLD S. LUFT seem very concerned about the propriety of for-profit enterprises in this arena. Profit incentives seem to be acceptable as Tong as they are limited to how certain services are provided and as long as the profes- sionals who order the services and evaluate their quality are insulated from the proprietary system. Such incentives are being questioned, however, when they potentially affect physicians who decide what services are needed and whether the care delivered is of appropriate quality. To help set the stage for a larger investigation of ethical and profes- sional concerns about physician involvement in for-profit enterprises in health care, this paper addresses a relatively simple question: In the current environment, to what extent do economic or financial incentives influence physicians' clinical or patient care decisions? Unfortunately, physicians and economists offer markedly different answers to the question. Unlike the situation of the 8-ounce glass containing 4 ounces of water, the responses reflect more than seman- tics. Physicians typically argue that such incentives have little influ- ence, except in a few obvious and distressing cases, such as fee splitting and kickbacks and, furthermore, that economic incentives should have no place in clinical decisions. Economists maintain the opposite view: Such incentives are ubiquitous and have a major influence. Some economists would argue that expanding incentives and freeing phy- sicians to use them would cure the ills of the medical care sector. For instance, some economists would like to see the physician more free to serve as the patient's agent, taking into account both the medical and economic consequences of alternative clinical decisions. Other economists think that a greater role of economic incentives could ex- acerbate current problems. Although the physicians' and economists' positions have been over- drawn for the sake of exposition, the perceptions are so divergent that an examination of the models of clinical decision making used by the physician and the economist is warranted. The differences in these models partially explain why the answers to the simple question of the influence of economic incentives are so divergent. The following section reviews the evidence on the effects of economic incentives on physician behavior, prefaced by a discussion of what is considered acceptable evidence by physicians and economists. The third section offers a synthesis ofthe two views, suggesting how economic incentives can have an important role in decision making despite being invisible to the clinician. A final section provides a brief summary and conclu- s~ons.
Economic Incentives and Clinical Decisions 105 Models of Clinical Decision Making The usual medical mode! of decision making involves a complex and largely intuitive process whereby the physician considers signs, symp- toms, and a variety of test results and, based upon scientific knowledge and clinical experience, arrives at a diagnosis and chooses the best treatment (Eddy, 19821. The classic biomedical mode} presumes a sin- gle, potentially identifiable cause of a disease for which there is a single best treatment. Only in recent years has the existence of mul- tiple factors in the causation of disease been recognized by some phy- sicians, along with the recognition that patients may differ in their preferences for or responses to alternative treatments for the same condition (McNeil et al., 19821. The traditional, reductionist single- cause/single-cure model, which long has been at the root of biomedical research and medical education, makes the physician a seeker oftruth, who must vigorously resist any deviation from the one right path, for economic or other reasons. This model of behavior has other important implications. The physician is clearly in authority, and the patient must wait to find out the correct course of action. The authority of the physician also implies the responsibility for making the correct diagnosis and for choosing the correct treatment. The implicit re- sponsibility probably contributes to the large volume of malpractice suits. Diagnosis and treatment decisions often are not clear-cut. However, in practice, many physicians act as if things were clear-cut and develop "standard operating rules" or "clinical policies" that dictate what should be done (Eddy, 1982~. These clinical policies may be highly complex, such as: "If signs A, B. and C are present, test X is negative, and there is no history of Y. the appropriate diagnosis is Q and the treatment is R." Furthermore, as will be seen below, experienced physicians may often have different clinical policies. Some physicians, trained in the techniques of decision analysis, argue for the explicit consideration of alternative causes and treatments (McNeil and Adelstein, 1975; Pau- ker and Kassirer, 19751. Explicit choice making, however, highlights the uncertainties in medicine that the typical physician would often rather ignore. For example, suppose that the available information allows the physician to reduce the problem to the following: "Every- thing points to the conclusion that the patient has disease Q and the appropriate treatment is R. but there is ~ chance in 100 that it is disease S and the treatment should be T." Most physicians seem to prefer not to deal explicitly with the probabilities and potential out-
106 HAROLD S. LUFT comes, Instead focusing on statements such as: "In my experience the problem seems to be Q. and the correct treatment is R." Because most such clinical policies are based on extensive experience and the de- cisions generally are accepted by the patients, the results usually are not substantially different from what would be found after a careful decision analysis exercise. And the physician usually sleeps more soundly by ignoring the overwhelming number of implicit probabili- ties, valuations, and choices that arise each day. In contrast to medical training, which emphasizes the single best course, economists are trained to believe that there are an infinite number of potential solutions, the selection of which should depend on individual preferences, and that the most efficient allocation of resources will be achieved if everyone pursues his or her own self- interest in a market economy. Furthermore, although physicians have traditionally seen medical problems in terms of immutable laws of chemistry and physics, economists have been expanding the realm of economic analysis, with its emphasis on individual choice and trade- offs, to include politics, the family, and natural selection (Becker, 1981; Wilson, 1978~. (The holistic approach in medicine takes a much wider view and recognizes the importance of multiple factors, but it is still far from the mainstream.) This difference in perspective has clear implications for the way that economic incentives are perceived. Under the traditional physi- cian mode] the medical problem and its potential solutionsis dealt with independently from all else. Moreover, although the physician is primarily concerned with the patient's well-being, the evaluation of what is best is usually from the perspective of the physician rather than the patient. The extreme economic view is to include everything in the choice set. For example, Grossman's mode} of the demand for health views the body as a machine that depreciates yearly until it breaks down and is overhauled (medical intervention) or scrapped (death) and for which preventive maintenance decisions are consid- ered relative to other ways that the owner can spend his or her time and money (Grossman, 1972, 1982~. From the economist's perspective, physicians are like auto mechanics, who want to turn every ordinary family car into a luxury machine without considering whether the family would like to spend its time or money on something else, such as a summer vacation. The analogy to an auto mechanic may distress some physicians, but it incorporates the economist's recognition that many of the technical details of medicine are too complicated for patients to evaluate di- rectly. However, just as few consumers understand the complexities
Economic Incentives and Clinical Decisions 107 of auto repair, although most can determine whether a car is running better, patients often can evaluate the results of medical care without understanding the disease processes or the therapeutic alternatives. The problem for the patient is to have someone determine what is wrong, to have the treatment choices identified and explained, and to choose the appropriate people to carry out the desired interventions. In this regard the economist views the ideal primary care physician as the patient's agent, providing the relevant information and select- ing the appropriate specialists (Feldstein, 1974; Pauly, 19801. (Note the parallel to a trusted mechanic who can diagnose a transmission problem and then identify a competent specialty shop to do the work.) A perfect physician-agent would lead the consumer to precisely the same decision as the consumer would have reached given all of the physician's expertise. This decision may well differ from traditional "best medical practice" because the patient is likely to take into con- sideration the cost of the services, the time involved in treatment, and other factors not usually involved in choosing the best medical out- come. Although the notion of a perfect agent is a very attractive theoretical concept, there are few perfect agents because of the conventional meth- ods of organizing care and paying physicians. It is often the case that most of the work is in arriving at the diagnosis, so the same physician provides both the diagnosis and the treatment. The dominant mode of payment is fee-for-service, and, more important, fees are heavily weighted toward laboratory tests and diagnostic and therapeutic pro- cedures in contrast to time spent talking with the patient (Schroeder and Showstack, 1978; Showstack et al., 1979~. The hypothetical "phy- sician as a perfect agent" would be available and willing to spend time with the patient, investigating the problem, pondering the di- agnosis, and calculating the alternatives. The best physician-agent would have no personal economic incentives either to encourage or discourage additional tests and procedures or to prefer one course of treatment over another. In practice, however, except for psychiatrists, fee-for-time arrangements for physicians are uncommon.) Further- more, primary physicians are rarely only counselors, and even the diagnostic function involves many highly profitable tests. The crucial issue here is not the method of paying the physician- fee-for-service, capitation, or salary but the linkage between the Surgeons often charge a fee that includes pro- and postoperative visits, obstetricians have prenatal/maternity packages, and pediatricians sometimes have a single fee for the first year of well-baby care. In most of these cases, however, tests and treatments for complications are handled on a standard fee-for-service basis.
108 HAROLD S. LUFT physician as agent and the physician as provider. For instance, many medical school faculty are on a straight salary, yet they know that their department's revenues are dependent on fee-for-service billings and that a revenue shortfall will affect salaries, promotions, and per- quisites. Similarly, the medical group in a health maintenance or- ganization (HMO) may receive a capitation payment covering the annual primary care of its enrollees, but if the group orders too much hospital care, its share of the plan's net income will be smaller (Luft, 19811. The incentives to provide services are reversed in some sys- tems fee-for-service has a bias toward more services while the fixed budget of a HMO sets up a bias toward fewer services but in each case economic factors are present that could influence clinical deci- sions. Whether physicians respond to such incentives is another ques- tion. Incentives from the Perspective of the Physician and the Economist Before reviewing the evidence concerning the role of economic incen- tives, we must consider what would be recognized as an "influence on clinical decisions." Although the medical literature generally ignores the possibility of such influences, when they are discussed, it seems to be in terms of conscious behavior on the part of the physician. For instance, physicians in prepaid plans have identified as an advantage the fact that all their patients have comprehensive coverage, so the physicians need never be concerned that a proposed treatment would bankrupt the patient (Cook, 19711. Clinicians in fee-for-service prac- tice have mentioned being aware of the gross revenues associated with a procedure while making clinical recommendations (personal com- munications). Most physicians, however, seem to claim that financial incentives do not influence their patient care decisions. Economists take a much broader perspective, and, although their language may suggest conscious decision making, they typically care little about what people say they do (or why they do it), focusing their attention instead on behavior. if financial incentives would reward a certain behavior, everything else being equal, then if the behavior is observed, the role of incentives is deemed empirically supported. In this type of analysis the economist typically ignores (or attempts to hold constant statistically) all but the economic variables such as prices and incomes. It is understood that in any particular case, clin- ical, personal, professional, or other factors may be present and even dominant, but such factors are seen as essentially random. That is, if
Economic Incentives and Clinical Decisions 109 one considers a large number of cases, these noneconomic factors will cancel out, leaving one able to observe the pattern left by the consis- tent impact of economic incentives. By contrast, the clinician is trained to focus on precisely those noneconomic factors that the economist dismisses as random and will believe and argue that each case is handled individually, with attention only, or almost only, to the clin- ical problem. It may be the case that 95 percent of the decision is based on clinical factors and 5 percent on economic factors. The phy- sician will fee] that the economic factors are inconsequential, and the economist will respond that if one examines many similar cases, ab- stracting from the random clinical factors, economics dominates and patterns emerge that cannot be explained by clinical factors. Yet another difference in approach helps explain the different per- ceptions of the economist and the physician. The economist tends to be denominator oriented, focusing on the influence of economic var- iables on various decisions, such as whether individuals experiencing a given set of symptoms decide to see a physician. The physician is typically numerator oriented, focusing on persons who come to the office for care. The physician notes that fees do not influence his or her patients, although the economist responds that fees will determine (at least in part) how many people decide to present themselves as patients. In most instances the economist attempts to demonstrate a statis- tically significant effect of an economic variable. The interpretation of such a finding is generally open to question on two grounds. The first is whether the observed correlation really implies causation or whether other, unobserved factors may be causing the measured re- lationship. The second is whether the statistically significant effect is substantively important. Large samples and sophisticated economet- ric models often allow very small effects to be measured, but such differences may be of no policy import. From the physician's perspec- tive, subtle tendencies, regardless of the statistical significance or aggregate importance, are inconsequential unless one can identify clear instances in which the economic incentives can be shown to have led to an altered clinical decision. Given the different orientations and tools, the physician's microscope and the economist's telescope, it is not surprising that the two cannot easily agree on what evidence is appropriate. Largely because physician-researchers have not consid- ered the role of incentives a fruitful research area, most of the avail- able evidence uses the economist's approach of searching for tendencies across large numbers of cases. There are, however, a few exceptions that are clinically oriented.
110 HAROLD S. LUFT Economic incentives can potentially influence clinical decisions in a wide variety of ways, and it may be useful to consider two broad categories that may bear on subsequent discussions of for-profit en- terprise in health care. The first type of incentive or situation involves the physicians' ability to hire other workers (and equipment) and to make more money by owning a larger enterprise. Although the di- viding line is not clear, most would perceive a qualitative difference between a physician who hires a nurse practitioner and one who runs a chain of weight reduction clinics. At one end of the spectrum the physician is still primarily a clinician but one who must give some attention to the economic realities of his or her practice. At the other extreme the physician is primarily a business-operator with little day- to-day clinical involvement. The second category of incentives pertains to the way the physician is paid for patient care activities. This in- cludes methods such as fee-for-service, salary, or prepayment with physician responsibility for the costs of prescribed services. It also includes the incentives inherent in each type of payment system, such as relatively higher fees for procedures and tests compared with pay- ment for the physician's time. Incentives and the Use of Technology Much of the rapid growth in the use of various medical technologies may stem not just from their clinical efficacy but also from the high returns physicians can get by using such technology. A primary care internist can increase his or her net income by a factor of almost three by prescribing a wide but not unreasonable set of tests (Schroeder and Showstack, 19781. The term not unreasonable is a reflection that the use of such tests is so common as to be almost standard practice; yet some clinicians would argue that few of the tests are actually nec- essary (Griper et al., 1981; Martin, 19821. Some diagnostic technol- ogies, such as endoscopy, have been studied in detail. They are highly profitable, have proliferated rapidly, yet rarely result in a definitive change in treatment outcome (Showstack and Schroeder, 19811. Is this evidence of economic incentives influencing practice patterns? From the economist's perspective, the answer is yes, but the clinician might quickly point out such factors as the low risk of the procedure and the importance of the reassurance it can provide to the patient (and the physician). A study by Childs and Hunter (1972) of diagnostic X-ray use pro- vides an example of the role of multiple factors in clinical patterns. They examined the use of X-rays for persons under old-age assistance
Economic Incentives and Clinical Decisions 111
112 HAROLD S. LUFT Payment and Practice Setting Incentives In contrast to the near dearth of studies on incentives and technology, there is considerable evidence concerning the role of direct payment incentives and practice patterns. The clearest distinction is between physicians paid on a fee-for-service basis, in which case there is a clear incentive to do more, and those paid a fixed sum in salary or capitation. Bunker (1970) found that certain discretionary surgical procedures were performed twice as frequently per capita in the United States as in England and Wales. Although the difference in mode of pay- ment fee-for-service in the United States and salaried in Britain- might explain this difference, the number of surgeons in the United States was also proportionately higher. International comparisons are fraught with difficulty, so studies of different practice settings within the United States may be more conclusive. In almost all comparisons of persons enrolled in HMOs of the prepaid group practice (POP) type and those obtaining care in conventional fee-for-service settings, the hospitalization rate for the HMO enrollees is lower (Luft, 19811. There is some evidence that people who switch into PGPs from conventional plans have previously been lower uti- lizers of hospital care than those who do not switch into a PGP (Berki and Ashcraft, 1980; Eggers and Prihoda, 1982; Luft, 19811. Despite this, the average PGP enrollee is not noticeably more healthy than enrollees in conventional insurance plans (Blumberg, 19801. (This is possible because relatively recent enrollees make up only a relatively small proportion of PGP members at any point in time.) Therefore, it is unlikely that differential health status accounts for all the observed differences in hospitalization rates between enrollees in conventional plans and prepaid group practice HMOs (Luft, 1981~. However, the observed differences in hospitalization rates do not necessarily reflect physicians' decisions whether to treat patients. Some of the changes in utilization rates reflect differences in the ways that treatments are provided. For instance, the design of most PGPs involves comprehen- sive coverage of diagnostic services in and out of the hospital, incen- tives to reduce hospital use, and physically convenient ambulatory facilities. Thus, for example, patient stays may be shortened by having the patient arrive the morning of the operation rather than the night before. Similarly, Kaiser-PortIand reports that 35 percent of all its operative procedures were performed on a come-and-go basis, i.e., in the operating room but without a hospital admission unless compli- cations occur (Marks et al., 19801. Such practices are becoming in- creasingly common in the fee-for-service sector, but the different
Economic Incentives and Clinical Decisions 113 incentives in prepaid and nonprepaid settings may explain why this cost saving technique was more quickly adopted by HMOs (Lavin, 1982~. More important, things such as ambulatory diagnostic work- ups, same day (come-and-go) surgery, and come-and-stay surgery (i.e., the patient is admitted on the day of the operation) really involve minimal changes in clinical practice; they are primarily production process decisions concerning the most efficient way to carry out a specific task. Another issue to be considered in the HMO studies is the extent to which differences may be attributable to group practice, rather than to the economic incentives resulting from prepayment. The relative performance of independent practice associations, which involve some financial risk sharing by independent, primarily fee-for-service prac- titioners, is much less impressive than that of PGPs (Luft, 19811. On the other hand, some fee-for-service groups seem to have hospitali- zation rates for their patients comparable to those of prepaid groups (Broida et al., 1975; Nobrega et al., 1982; Scitovksy, 19811. Why this is the case is not clear, but speculating on the cause may help clarify the different perspectives physicians and economists have on the role of incentives. One explanation that has been offered for the low hospitalization rate in certain group practices is that the number of specialists relative to generalists is so low that the specialists are occupied with clearly necessary admissions and do not have time for the more discretionary cases. This implies that different decision criteria are used, such that the same patient would be treated differently by the specialist in such a group than by a similar specialist in solo practice. If solo practi- tioners are less busy (in general this is the case, with surgeons pre- ferring to do more procedures than they actually do), then their patients may have more extensive tests and workups, followed by hospitali- zation. By contrast, the patients in a group setting might be more likely to be told to monitor the condition over the next few months, and, if it does not improve, more aggressive treatment will be under- taken. Notwithstanding the differences in hospitalization, patient out- comes in both styles of practice may be similar because many medical problems are self-limiting. Practitioners in both settings see their own practice styles as clinically successful. But one may ask: If both the solo and group physicians are in a fee-for-service environment, why do they not develop similar practice patterns? Put another way: What prevents the group practice from adding more specialists who, pre- sumably, would do more discretionary procedures?
114 HAROLD S. LUFT We must now move back from the economist's mode] to something closer to clinical practice. Procedures often seen as discretionary, such as cholecystectomy, hysterectomy, and hemorrhoidectomy, are prob- ably seen as more mundane and less challenging if only because they are so common and the patient is not in a crisis situation. If the specialists can keep busy with interesting cases by limiting the num- ber of physicians in the group, then they probably will do so, rather than expand the group just for the sake of bigness at the cost of diluting the clinical case mix. (Note that if the group expands by adding pri- mary physicians and thus enlarges its patient base, dilution is not an issue.) Although this scenario is plausible, one should note that the empirical base for these observations is extraordinarily thin, being limited to a handful of studies focusing on large, well-respected mul- tispecialty group practices, often with large numbers of referral pa- tients, such as the Mayo Clinic. Specialists in such settings may well establish rather stringent criteria for hospitalization because they have so much experience with sicker patients. The notion of different criteria for hospitalization is very close to our original question about the impact of incentives on clinical deci- sion making. Observational studies suffer from an inability to control for case mix, so the standard retort to the differences between HMO and fee-for-service settings is that in some subtle way HMO enrollees were healthier at the outset. HIatky et al. (1981) undertook an im- portant, although limited, study that controls for this problem. They sent a series of case histories of patients with various types of heart problems to a sample of board-certified cardiologists. Each physician was asked a series of questions about how he or she would manage the case and, in particular, whether certain diagnostic tests or bypass surgery would be recommended. Physicians in independent fee-for- service practice were significantly more likely than those in a prepaid group practice to recommend the tests and surgery. This finding sup- ports the notion that clinical decision-making patterns in prepaid groups are different. Interestingly, the recommendations of the HMO physicians were similar to those of university cardiologists, making it more difficult to say that the POP practice pattern represents in- ferior care. As has been noted, one cannot separate the prepayment from the group practice effects. More important, as we will soon dis- cuss, the data do not indicate why or how the difference occurred. Incl:ividual versus Collective Patterns of Practice This brief review suggests that, despite the physician's general view that economic incentives do not influence clinical decisions, various
Economic Incentives and Clinical Decisions 115 bits of evidence at least are consistent with the notion that economic incentives do have an impact. The physician's perspective may be based largely on the absence of incentives in a conscious choice process. Most clinicians develop preferred ways to handle particular clinical situations and, when presented with a case, may not give much thought to alternatives or at least to the role that nonclinical factors, such as price, might have on the selection among alternatives. This section will take the issue one step further, to examine whether the influence of such incentives seems to result in clinically inappropriate choices. After all, the concerns about for-profit enterprises in medicine stem largely from the notion that care will suffer.2 The first step in this examination is the recognition that medicine abounds with situations in which alternative clinical strategies are available with no scientific evidence indicating which is preferable. The second step is the rec- ognition that despite this physicians may have strong preferences concerning these alternatives and that there may be a correlation between economic incentives and these preferences. A careful review of the medical literature indicates a wide range of situations in which adequate scientific evidence does not exist to es- tablish one treatment as definitively superior. For instance, Wennberg et al. (1980) found substantial controversies surrounding nine common surgical procedures. A great debate continues over whether certain types of coronary diseases are best managed surgically or medically (Carr et al., 1982; Mcintosh, 19811. Yet in each situation individual physicians tend to prefer and to use one mode of treatment and do not behave as though there is a gray area that research evidence does not resolve. Definitive clinical trials to narrow the gray area are extremely difficult and costly because the patient's outcome in any particular case is dependent on a host of factors in addition to the one under consideration. (Even major clinical trials often provide ambiguous results.) Thus, very large samples and sophisticated methods may be required to determine the specific gray area situations in which treat- ment A is superior to treatment B. Individual clinicians cannot un- dertake such studies in a systematic way. Yet many act as if the evidence were clear. ~ think the reason for this is threefold. First, medical training generally lacks training in research design, epide- 2Another concern is that the rise of for-profit enterprises in medicine will change the physician's perceived role so that it will no longer be that of a professional. This may entail a loss of prerogatives, status, and credibility. Furthermore, to the extent that the physician's credibility has a beneficial placebo effect on the patient, the loss of status may indirectly affect patient outcomes.
116 HAROLD S. LUFT miology, and other analytic methods. Case reports and uncontrolled trials abound in the medical literature (McKinIay, 19811. Second, re- ports of new techniques are generally offered by their innovators, and the early "evaluation" is usually done by strong advocates of the tech- nique. Although this may not lead to intentional bias in the results, various studies indicate that subsequent controlled trials often are far less supportive of the technique. (It is important to note that in most cases the technique is not found to be worse than the alternative, only not superior; i.e., it is in the gray area.) Third, although the medical literature offers little useful guidance, the practitioner constantly makes ad hoc observations that tend to support and reinforce whichever view is initially held. Suppose the decision concerns a service that, given the available research, is truly in the gray area, such as bypass surgery for two diseased coronary arteries. For more severe disease there is clear evidence of improved survival with surgery, but the available studies are less clear for intermediate levels of obstruction. Survival rates for medical and surgical management are roughly comparable. While death rates tend to be low for both treatments, the morbidity (and costs) associated with each method differs. A physician choosing one method will tend to focus on the good outcomes, recalling that the failure rate is really no higher than for the alternative. Because pa- tients' beliefs often are significant factors in improved outcomes (viz., the placebo effect), a physician who strongly recommends one alter- native as being "superior in my experience" may well be correct be- cause of the expectation of improvement. More important, those patients who demand a careful evaluation of the alternatives are likely to lead their physicians to react in one of two ways.3 If they eventually follow their physician's recommendation and do less well than might be expected, their poor outcome may be blamed on insufficient trust, and the physician's preferences for not giving the patient explicit choices will be reinforced. This is often referred to as reducing cognitive dis- sonance. (If they do well, it merely confirms the physician's original view about the correct treatment.) However, some patients will decide against the recommended alternative and will change physicians, 3Some physicians encourage patients to ask questions and make choices; the increased control their patients experience may actually result in better outcomes. Careful studies of this hypothesis are not available. As has been noted earlier, this model of patient decision making reduces the physician's power, takes more time, and may be perceived as more risky. Moreover, it is probably the case that some patients want their physicians to make all the choices, although others want the reverse, and this, too, may result in self-selection (O'Donnell, 1982).
Economic Incentives and Clinical Decisions 117
118 HAROLD S. LUFT recommended surgery for patients with left main artery and triple vessel disease, a recommendation clearly supported by the research literature. The variability was concentrated among the less severe cases in which the research is not definitive, further supporting the notion of differences in the gray area. Although the wide variation in patterns within practice settings may have idiosyncratic origins, such as the teachings of an influential professor or a memorably bad experience with an alternative strategy, there also seem to be consistent patterns of care related to the method of payment and other economic incentives. The reasons for a statistical relationship between economic incentives and practice patterns are not well established. Two explanations may be offered. First, it may be that economics directly shapes the clinical patterns, so that, for example, a new physician, even one trained in a conservative, watch- and-wait style, who enters fee-for-service practice quickly recognizes that the loan will not be paid off and the yacht will be long in arriving unless he or she does more tests and procedures. (One can describe a counter example for a new partner in an HMO.) This explanation is incorporated in much of the rhetoric about fee-for-service and prepaid systems, but it rarely is reported by physicians in those settings.4 The second explanation focuses on the selection behavior of patients and physicians. Just as patients select a physician they think will provide the advice they desire and with whom they feel comfortable, physicians select practice patterns. By the time residency is completed, an aggressive physician probably knows an HMO is not the most conducive setting for that style of practice. In many cases choice may not be conscious. Such a physician's mentors are much more likely to be in a fee-for-service setting, and the new physician's perception is merely of following a style that clearly works. Likewise, a conserva- tively oriented physician may find the HMO environment more com- fortable. Decisions of this type may be made primarily on the basis of collegial support; it may be difficult always to be different in one's clinical recommendations. Economic incentives probably also have a more direct influence, even if not influencing clinical decisions. Given the current structure of medical fees, a conservative practitioner in a fee-for-service environment will typically earn less than his or her Occasional stories of this type do appear, but they seem limited to Medicaid mills or other extreme settings. There is certainly not enough evidence of this type to explain the different practice patterns in large, mature HMOs such as Kaiser and Group Health Co- operative. The truth may be concealed because such a gross influence of economic incentives runs counter to the hallowed view of medical ethics. However, given the emotionally charged debates in this area, if this were a common problem one would have expected more evidence to have surfaced.
Economic Incentives and Clinical Decisions 119 peers, so a switch to an HMO setting might be attractive because less use of expensive services might yield a larger year-end bonus. The selection hypothesis also helps explain the observed positive correlation between the supply of surgeons and the incidence of sur- gery without resorting to a crude demand-generating mode] (McClure, 19821. If physicians have some implicit income target, this income level can be reached by aggressive practitioners with a small but intensively treated population base or by conservative practitioners with a larger and less intensively treated population.5 This could result in a natural sorting process through which areas happening to have conservative practitioners are in equilibrium with low-intensity care, while areas with aggressive practitioners reach an equilibrium with high-intensity care. Of course, such a situation requires con- sumer insensitivity to costs (a result of extensive third-party coverage) and lack of knowledge or relative indifference to alternative treatment options. This brings us back to the question of the effectiveness of the physician as the patient's agent. However, it is important to recall that as long as we are discussing decisions in the gray area, individual physicians may firmly believe that they are following appropriate practice and that this has nothing to do with economic incentives. Furthermore, most clinicians appear to be unaware of costs or to be- lieve that a third-party payer, not the patient, will foot the bill. Summary and Conclusions Much of this paper has been devoted to an attempted reconciliation of the apparently opposing opinions of physicians and economists con- cerning the influence of economic incentives on clinical decisions. Dif- ferent approaches to empirical research and different criteria for acceptable evidence are partial explanations of the different percep- tions. Perhaps snore important is the central role given choice and adjustments at the margin in the economist's world view and the tendency by clinicians to view a problem as a challenge to find the one correct solution. Given such widely divergent starting points, it is difficult for economists and physicians to agree even on terminology and to discuss their differences without becoming convinced the other is totally missing the point. On the empirical side, there is certainly evidence concerning the 5The target-income hypothesis is hotly debated by economists who seem unable to reach a definitive conclusion on this issue yet continue to hold strong beliefs about it. Gray areas exist in medical economics as well as in medicine. See Fuchs and Newhouse (1978), Hixon (1980), Richardson (1981), Wilensky and Rossiter (1981).
120 HAROLD S. LUFT adoption and use of medical technology and different practice pat- terns that is consistent with the notion that economic incentives matter. Such evidence may be sufficient to convince the economist that we should examine the effects of the types of economic incentives created by different types of practices and payment settings, but it lacks the power of a randomized controlled trial to convince skeptical physicians. One of the difficulties is that from the clinician's per- spective the observational studies are missing an explanation of how economic incentives alter practice patterns, particularly when they do not see such factors playing a role in their own experience. A possible explanation for both sets of evidence is that there is often a wide range of acceptable clinical practice, even though each clinician may believe in his or her own way. If clinicians sort themselves into different practice settings whose economic incentives are consistent with aggressive or conservative practice styles, we will observe clinical patterns that appear to be shaped by economics, although the clini- cians themselves see no such effects. The foregoing is a description of a relatively slow and passive pro- cess. Morever, because the medical care market has been relatively noncompetitive, there has been little active encouragement to the sorting out of physicians, let alone the evaluation of alternative clin- ical approaches. The gray area often is wide, but there has been rel- atively little exploration of how wide it might be. The situation is now beginning to change. More and more studies are being proposed or undertaken to evaluate new technologies (Bunker et al., 1982; Green- berg and Derzon, 1981; Towery and Perry, 19811. Simultaneously, the growth and development of HMOs, for-profit hospitals, health care corporations, and other organized systems provide both the means and incentive to evaluate the alternative clinical strategies in cost-effec- tiveness terms. This may lead to more active efforts by such organi- zations to use incentives or pressures to get their clinicians to alter their practice patterns. Some physicians already are beginning to view the world through the economist's eyes and to use the language of choices, trade-offs, and financial transactions (rein, 1982~. Whether such changes are desirable is a much larger question, but there can be little doubt that they are occurring.
EconomicIncentives and Cl~nicalDecisions References 121 Becker, Gary S. A Treatise on the Family. Cambridge: Harvard University Press, 1981. Berki, S. E., and Ashcraft, Marie L. F. "HMO Enrollment: Who Joins and Why: A Review of the Literature." Milbank Memorial Fund Quarterly/Health and Society 58:4 (Fall 1980):558-632. Blumberg, Mark S. "Health Status and Health Care Use by Type of Private Health Cov- erage." Milbank Memorial Fund Quarterly/Health and Society 58:4 (Fall 1980):633-55. Blumberg, Mark S. "Provider Price Charges for Improved Health Care Use." In Health Handbook. Edited by George K. Chako. Amsterdam: North-Holland, 1979. Broida, Joel et al. "Impact of Membership in an Enrolled Prepaid Population on Utilization of Health Services in a Group Practice." The New England Journal of Medicine 292:15 (10 April 1975):780-83. Bunker, John P. "A Comparison of Operations and Surgeons in the United States and in England and Wales. The New England Journal of Medicine 282:3 (15 January 1970):135- 44. Bunker, John P., Jinnet Fowles, and Ralph Schffarzick. "Evaluation of Medical Technology Strategies." The New England Journal of Medicine 306:10-11 (11, 18 March 1982):620- 624, 687-92. California Medical Association. 1974 Revision of the 1969 California Relative Value Stud ies. San Francisco: Sutter Publications, 1975. Caper, Philip. "Competition and Health Care: A New Trojan Horse?" The New England Journal of Medicine 306:15 (15 April 1982):928-29. Carr, Kenneth W., Robert L. Engler, and John Ross, Jr. "Do Coronary Artery Bypass Operations Prolong Life?" Western Journal of Medicine 136:4 (April 1982):295-308. Childs, Alfred W., and Hunter, E. Diane. "Non-Medical Factors Influencing the Use of Diagnostic X-ray by Physicians." Medical Care 10:4 (July/August 1972):323-35. Cook, Wallace H. "Profile of the Permanente Physician." In The Kaiser-Permanente Medical Care Program: A Symposium. Edited by Anne R. Somers. New York: Commonwealth Fund, 1971. Eddy, David M. "Clinical Policies and the Quality of Clinical Practice." The New England Journal of Medicine 307:6 (5 August 1982):343-47. Eggers, Paul W., and Prihoda, Ronald. "Pre-Enrollment Reimbursement Patterns of Med- icare Beneficiaries Enrolled in 'At Risk' HMOs." Health Care Financing Review 4:1 (Sep- tember 1982):55-74. Fein, Rashi. "What is Wrong with the Language of Medicine." The New England Journal of Medicine 306:14 (8 April 1982):863-64. Feldstein, Martin S. "Econometric Studies of Health Economics." In Frontiers of Quantitative Economics. Edited by M. Intriligator and D. Kendrick. Amsterdam: North-Holland, 1974. Fineberg, Harvey V. "Clinical Chemistries: The High Cost of Low-Cost Diagnostic Tests." In Medical Technology: The Culprit Behind Health Care Costs? Proceedings of the 1977 Sun Valley Forum on National Health. National Center for Health Services Research, DHEW Publ. No. (PHS) 79-3216, Washington, D.C., 1979. Freeborn, Donald K. et al. "Determinants of Medical Care Utilization: Physician's Use of Laboratory Services." American Journal of Public Health 62:6 (June 1972):846-53. Fuchs, Victor R., and Newhouse, Joseph P. "The Conference and Unresolved Problems." Journal of Human Resources 13 Suppl. (1978):5-20. Greenberg, Barbara, and Derzon, Robert A. "Determining Health Insurance Coverage of Technology: Problems and Options." Medical Care 19:10 (October 1981):967-78. Griner, Paul F. et al. "Selection and Interpretation of Diagnostic Tests and Procedures." Annals of Internal Medicine 94:5 (Part 2), (May 1981):553-600. Grossman, Michael. The Demand for Health: A Theoretical and Empirical Investigation. New York: Columbia University Press for the National Bureau of Economic Research, 1972.
122 HAROLD S. LUFT Grossman, Michael. "The Demand for Health After a Decade." Journal of Health Economics 1:1 (1982):1-3. Hixon, Jesse S., ed. The Target Income Hypothesis. Bureau of Health Manpower, DHEW Publ. No. (HRS) 80-27. Washington, D.C.: U.S. Government Printing Office, 1980. Hlatky, Mark A., E. Botvinick, and B. Brundage. "A Controlled Comparison of Cardiac Diagnostic Test Use in a Health Maintenance Organization." Presented at the Annual Meeting of the Robert Wood Johnson Clinical Scholars. San Antonio, Tex., 11-14 Novem- ber 1981. Lavin, John H. "Same-Day Surgery: Why Everyone Is Learning to Love It." Medical Eco- nomics 59:12 (7 June 1982):110 If. Lewin, Lawrence S., Robert A. Derzon, and Rhea Margulies. "Investor-Owned and Non- profits Differ in Economic Performance." Hospitals 55 (1 July 1981):52-58. Luft, Harold S. Health Maintenance Organizations: Dimensions of Performance. New York: Wiley-Interscience, 1981. Lyle, Carl B. et al. "Practice Habits in a Group of Eight Internists." Annals of Internal Medicine 84:5 (May 1976):594-601. Marks, Sylvia D. et al. "Ambulatory Surgery in an HMO: A Study of Costs, Quality of Care and Satisfaction." Medical Care 18:2 (February 1980):127-46. Martin, Albert R. "Common and Correctable Errors in Diagnostic Test Ordering." Western Journal of Medicine 136:5 (May 1982):456-61. McClure, Walter. "Toward Development and Application of a Qualitative Theory of Hospital Utilization." Inquiry 19:2 (Summer 1982): 117-35. McIntosh, H. D. Overview of Aortocoronary Bypass Grafting for the Treatment of Coronary Artery Disease: An Internist's Perspective. Washington, D.C.: National Center for Health Care Technology, 1981. McKinlay, John B. "From 'Promising Report' to 'Standard Procedure': Seven Stages in the Career of a Medical Innovation." Milbank Memorial Fund Quarterly/Health and Society 59:3 (Summer 1981):374-411. McNeil, Barbara J. et al. "On the Elicitation of Preferences for Alternative Therapies." The New England Journal of Medicine 306:21 (27 May 1982):1259-62. McNeil, Barbara J., and Adelstein, S. James. "The Values of Case Finding in Hypertensive Renovascular Disease." The New England Journal of Medicine 293:5 (31 July 1975):221- 26. Moloney, Thomas W., and Rogers, David E. "Medical Technology A Different View of the Contentious Debate Over Costs." The New England Journal of Medicine 301:26 (27 De- cember 1979):1413-19. Nobrega, Fred T. et al. "Hospital Use in a Fee-for-Service System." Journal of the American Medical Association 247:6 (12 February 1982):806-10. O'Donnell, Walter E. "Let's Stop Ducking Decisions on Patient Care." Medical Economics 59:14 (5 July 1982):32ff. Pauker, Stephen G., and Kassirer, Jerome P. "Therapeutic Decision Making: A Cost-Benefit Analysis." The New England Journal of Medicine 293:5 (31 July 1975):229-34. Pauly, Mark V. Doctors and Their Workshops: Economic Models of Physician Behavior. Chicago: University of Chicago Press, 1980. Relman, Arnold S. "The New Medical-Industrial Complex." The New England Journal of Medicine 303:17 (23 October 1980):963-70. Richardson, J. "The Inducement Hypothesis: That Doctors Generate Demand for Their Own Services." In Health, Economics and Health Economics. Edited by Jacques Van der Gaag and Mark Perlman. Amsterdam: North-Holland, 1981. Roney, James G., and Estes, Hilliard D. "Automated Health Testing in a Medical Group Practice." Public Health Reports 90:2 (March/April 1975):126-32. Saward, Ernest, and Sorensen, Andrew. "Competition, Profit and the HMO." The New England Journal of Medicine 306:15 (15 April 1982):929-31.
Economic Incentives and Clinical Decisions 123 Schroeder, Steven A., and Showstack, Jonathan A. "financial Incentives to Perform Medical Procedures and Laboratory Tests: Illustrative Models of Office Practice." Medical Care 16:4 (April 1978):289-98. Schroeder, Steven A. et al. "Use of Laboratory Tests and Pharmaceuticals: Variation Among Physicians and Effects of Cost Audit on Subsequent Use." Journal of theAmerican Medical Association 225:8 (20 August 1973):969-73. Scitovsky, Anne A. "The Use of Medical Services Under Prepaid and Fee-for-Service Group Practice." Social Science and Medicine 15C (1981):107-16. Showstack, Jonathan A., and Schroeder, Steven A. '`The Cost-Effectiveness of Upper Gas- trointestinal Endoscopy." Background Paper No. 2, Case Studies of Medical Technolo- gies Case Study 8: Implications of Cost-Effectiveness Analysis of Medical Technology. Washington, D.C.: Office of Technology Assessment, U.S. Congress, May 1981. Showstack, Jonathan A. et al. "Fee-for-Service Physician Payment: Analysis of Current Methods and Their Development." Inquiry 16:3 (Fall 1979):230-46. Towery, O. B., and Perry, Seymour. "The Scientific Basis for Coverage Decisions by Third- Party Payers." Journal of the American Medical Association 245:1 (2 January 1981):59- 61. Wennberg, John E., John P. Bunker, and Benjamin Barnes. "The Need for Assessing the Outcome of Common Clinical Practices." Annual Review of Public Health 1 (1980):277- 95. Wennberg, John E., and Gittelsohn, Alan. "Health Care Delivery in Maine I: Patterns of Use of Common Surgical Procedures." Journal of Maine Medical Association 66:5 (May 1975):123-30,49. Wennberg, John E., and Gittelsohn, Alan. "Small Variations in Health Care Delivery." Science 182:4117 (14 December 1973):1102-8. Wennberg, John E., and Gittelsohn, Alan. "Variations in Medical Care Among Small Areas." Scientific American 246:4 (April 1982):120ff. Wilensky, Gail Roggin, and Rossiter, Louis F. "The Magnitude and Determinants of Phy- sician-Initiated Visits in the United States. In Health, Economics and Health Economics. Edited by Jacques van der Gaag and Mark Perlman. Amsterdam: North-Holland, 1981. Wilson, Edward O. "The Ergonomics of Caste in the Social Insects." American Economic Review 68:6 (December 1978):25-36.