Correspondence

Self-Referral by Physicians

N Engl J Med 1993; 328:1274-1279April 29, 1993

Article

To the Editor:

The study by Mitchell and Sunshine (Nov. 19 issue)1 contains serious technical flaws. Fundamental principles of experimental design dictate that a study must control adequately for confounding factors. The study by Mitchell and Sunshine, however, relies on a comparison between two unmatched and probably vastly different populations of patients. Its findings are based on medical data comparing the intensity of treatment in Florida with that in the United States as a whole. It attributes observed differences entirely to the higher percentage of facilities owned by referring physicians in Florida. The study fails to eliminate numerous alternative hypotheses, such as that Florida has more practitioners of radiation oncology, a different mix of cancer cases, different demographics, differences in wealth that affect the demand for medical care, or regional variations in practice patterns arising from custom, training, or both.

Even if self-referral often affects the cost and quality of care adversely, it can at times provide incentives for meeting patients' legitimate needs. Thus, we believe that decisions to prohibit self-referral should be made in response to credible evidence. We respect the view that self-referral creates conflicts of interest that inevitably undermine the physician-patient relationship and that therefore this practice should be banned altogether. But proponents of this position cannot claim that flawed studies provide supportive scientific evidence.

Despite its shortcomings, the study by Mitchell and Sunshine serves as a lightning rod for a useful professional dialogue about self-referral. The Journal's readership would have been better served, though, had it appeared in an opinion format such as a Sounding Board article.

Stan N. Finkelstein, M.D.
Massachusetts Institute of Technology, Cambridge, MA 02139

Kevin Neels, Ph.D.
Charles River Associates, Boston, MA 02116

1 References

1. 1

   Mitchell JM, Sunshine JH. Consequences of physicians' ownership of health care facilities -- joint ventures in radiation therapy. N Engl J Med 1992;327:1497-1501
   Full Text | Web of Science | Medline

To the Editor:

Mitchell and Sunshine attribute the increase in radiotherapy in Florida to the avarice of Florida physicians who wish to fatten their pockets by referring more patients for radiation to facilities they own. I would suggest another explanation: that Florida physicians treat their patients who have cancer more aggressively (from what I know of the demographics of Florida, who would they treat aggressively if not Medicare patients?). Furthermore, patients with cancer in Florida may be a lot healthier than their U.S. counterparts and may demand aggressive therapy. After all, they spent all those bucks to enjoy their golden years in the Florida sunshine, and they aren't willing to give that up.

This variation in practice patterns has perplexed researchers in health services for years. For example, Wennberg, who has done much of this work, found enormous differences in practice patterns between physicians in Boston and those in New Haven.1 Yet he never accused the Boston surgeons of greed, even though they performed more than twice as many endarterectomies as their New Haven counterparts.

Florida physicians are probably just as ethical as their U.S. counterparts; they just treat their patients who have cancer more aggressively, and their use of joint-venture facilities is good business. After all, medical practice in the United States is a business, and much of what physicians do can be suspect. But before practicing physicians are found guilty of unethical practice, better studies than that of Mitchell and Sunshine are needed.

Herman Kattlove, M.D., F.A.C.P.
2653 Elm, Long Beach, CA 90806

1 References

1. 1

   Wennberg JE, Freeman JL, Culp WJ. Are hospital services rationed in New Haven or over-utilised in Boston? Lancet 1987;1:1185-1189
   CrossRef | Web of Science | Medline

To the Editor:

Inasmuch as the Florida Health Care Cost Containment Board survey analyzed by Mitchell and Sunshine included data on cost and service use, a minimally critical reading of their article would lead one to ask why no results based on these data were presented. The answer is simple. The data do not support the interests of radiation oncologists who sponsored the work of Mitchell and Sunshine.

We analyzed the original data of the cost-containment board and found that fees at joint-venture centers were on average 9 percent lower than those at free-standing centers, whereas fees at centers owned wholly by radiation oncologists, radiologists, or both were 5 percent higher. Furthermore, the latter centers performed 11 percent more procedures per patient and had 8 percent higher net revenues per patient than joint-venture centers.

What about the data that Mitchell and Sunshine do present? Their most striking finding appears to be that “the frequency and costs of radiation-therapy treatments at free-standing centers were 40 to 60 percent higher in Florida” than the U.S. averages. However, because this comparison is based on use per capita rather than per patient, it merely reflects the greater prevalence of free-standing facilities in Florida. Would showing that per capita use of services of the University of Miami Hospital was higher among Florida Medicare beneficiaries than among those in other states demonstrate overuse?

The attempt by Mitchell and Sunshine to imply differences in quality on the basis of a reported 18 percent difference in time spent with patients by radiation physicists is also highly suspect. The underlying data on employee counts do not support the level of precision implied in the presentation of this result. Even if the difference is real, does the lower head count show poorer quality or greater efficiency?

Lastly, the difference in the mix of payers between joint-venture facilities and radiation oncologist-owned centers may merely reflect the better ability of the joint-venture centers to compete for managed-care business, because they have lower fees.

I share the widespread concern about physician self-referral. However, the truth is that events in Florida and elsewhere represent a turf battle over the profitable radiation-therapy market in which radiation oncologists have wrapped themselves in the anti-self-referral flag while they batter their market competitors. The piece by Mitchell and Sunshine appears to be more a part of this effort than of a search for the information policy makers need to help them address self-referral without promoting the anti-competitive interests of radiation oncologists or others.

Kevin F. O'Grady, M.D., M.S.P.H.
Center for Consumer Healthcare Information, Irvine, CA 92713

To the Editor:

The data presented by Mitchell and Sunshine about physicians' joint ventures in radiation oncology come as no surprise. Recent studies evaluating physician self-referral in diagnostic radiology and physical therapy, among other areas, have shown that such ventures result in higher costs and increased use1,2.

Aside from fiscal considerations, the study by Mitchell and Sunshine raises serious ethical issues. It suggests that in some cases physicians have placed their financial needs ahead of the health and well-being of their patients. For a long time the American Medical Association (AMA) believed a self-referral to be ethical as long as the patient was informed about the referring physician's financial interest in the facility. It is doubtful, however, that patients with cancer could independently evaluate the quality of radiation treatment in their community and that they would override the recommendations of a trusted referring physician.

Another ethical issue concerns the relationship between the employed radiation oncologist and the referring physician in a joint venture in radiation oncology. It is understood that physicians must be free of undue influence on their medical judgment. However, ownership of a high-cost facility in a time of decreasing reimbursement may increasingly predispose investing physicians to influence treatment decisions.

Furthermore, there is the unaddressed issue of the creation of classes of physicians, with one group or specialty potentially exploiting another in an employer-employee relationship. If medicine is to remain a profession rather than a business, there must be collegial and professional relationships among physicians. Influence exerted for economic reasons, whether to please a referring physician-owner or by such an owner, is not in patients' interest. The creation of a caste system among physicians and its effect on decision making and patient care need to be more fully addressed.

Alan A. Lewin, M.D.
James G. Schwade, M.D.
Sylvester Comprehensive Cancer Center, Miami, FL 33136

2 References

1. 1

   Hillman BJ, Joseph CA, Mabry MR, Sunshine JH, Kennedy SD, Noether M. Frequency and costs of diagnostic imaging in office practice -- a comparison of self-referring and radiologist-referring physicians. N Engl J Med 1990;323:1604-1608
   Full Text | Web of Science | Medline

2. 2

   Mitchell JM, Scott E. Physician ownership of physical therapy services: effects on charges, utilization, profits, and service characteristics. JAMA 1992;268:2055-2059
   CrossRef | Web of Science | Medline

Author/Editor Response

The authors reply:

To the Editor: We thank the writers for their interest and apologize for any perceived insults. None were intended. As we will show, none of these letters effectively challenge our principal findings -- that physicians' joint ventures negatively affected access, increased the use of services and costs by approximately 40 to 60 percent, and judging from limited indicators, did not improve quality.

The objections of Finkelstein and Neels are unsound. We concluded that joint ventures result in increased use of services and costs after carefully ruling out the main competing explanations (including demographic differences, a point Finkelstein and Neels apparently missed). Our findings accord with the extensive literature on physicians' financial self-interest. In the space the Journal permits, we can only note the following points. First, differences in the mix of cancer cases are an unlikely explanation, because overall cancer rates differ little between Florida and elsewhere. Moreover, radiation therapy is relevant (for cure or palliation) to all the most common cancers. In fact, the cancer mix among the elderly (our subject population) is similar in Florida and the entire United States. Second, Florida has fewer radiation oncologists than the U.S. average -- 4.1 per 1000 cancer deaths as compared with 4.3. Finally, differences in the use of Medicare services cannot be attributed to income, because four fifths of Medicare enrollees have supplemental insurance and thus face essentially no out-of-pocket costs.

If Kattlove is correct in equating higher use of services with more aggressive treatment, then our finding that cancer survival is not improved implies that more aggressive treatment does not yield better outcomes. Furthermore, his argument ignores an extensive literature that demonstrates that aggressive care is in large part attributable to financial self-interest. A better differentiation of use of services and aggressiveness seems necesssary.

In response to Dr. O'Grady: Our article notes that service use and costs can both increase when more patients receive services and when there are more services (or higher charges) per patient. The literature suggests that the former is the dominant factor, whereas O'Grady chooses to ignore it. Moreover, his analysis uses the data collected from radiation-therapy facilities as part of a Florida state survey. Since one of us was the principal author of the Florida report, we are aware of the problems that plague the radiation-therapy data. Miscounts of patients and services and possible confusion of global and professional charges by survey respondents make his figures unreliable.

Jean M. Mitchell, Ph.D.
Georgetown University, Washington, DC 20007

Jonathan H. Sunshine, Ph.D.
American College of Radiology, Reston, VA 22091

To the Editor:

As one who views the world through the eyes of a payer, I believe that the study by Swedlow et al. (Nov. 19 issue)1 is valuable as far as it goes. However, I do not believe that the authors have proved their point conclusively. It is possible that the early intense use of physical-therapy services and diagnostic methods in workers' compensation cases may actually reduce the duration of “lost time” and “modified work.” Such an outcome would result in a reduction in the overall cost of a work-related illness or injury to the employer. This, in and of itself, could represent a higher level of quality in medical care.

Although the medical cost per claim in workers' compensation cases is substantial, the continuation of wages (at some reduced level), administrative costs, and the cost to the employer of replacing the absent or less than fully productive worker may be higher than the cost of medical services. Therefore, for employers, the practice pattern that returns the injured worker to the usual level of function in the shortest time commensurate with recovery is the desired level of service.

I think it is unfortunate that the investigators looked at the cost of medical services without assessing the effect of the practice pattern on the duration of lost time, modified work, or both. Since the authors appear to have access to the additional data, they could have enhanced their research by extending their analysis to the effectiveness of the practice patterns, as indicated by the duration of lost time and modified work with the associated cost. A more valuable measure for assessing the effect of services and cost is perhaps the point at which the injured worker returns to full participation in productive work. The cost of medical practice patterns can then be properly interpreted as it relates to the total cost of an episode of work-related illness or injury.

W.W. Doane, M.D.
General Electric Plastics, Pittsfield, MA 01201

1 References

1. 1

   Swedlow A, Johnson G, Smithline N, Milstein A. Increased costs and rates of use in the California workers' compensation system as a result of self-referral by physicians. N Engl J Med 1992;327:1502-1506
   Full Text | Web of Science | Medline

To the Editor:

Swedlow et al. discuss physicians' self-referral, raising several important questions about the appropriateness of magnetic resonance imaging (MRI) scans in California. The authors do not indicate how many doctors were involved in the referral of these patients. It would be important to know the number and distribution of the doctors ordering “inappropriate” scans (i.e., were 10 percent of the doctors responsible for 75 percent of the inappropriate scans, or were such scans evenly distributed among all the referring physicians?). The authors found that one third of all referrals by the treating doctors were inappropriate in their assessment of the need for an MRI (173 of 492). This raises serious questions about either the judgment of the doctors seeing patients in California or the reasonableness of the precertification review used by the authors. The authors defined inappropriateness on the basis of a precertification review developed at the request of an insurance company. The insurance company has a financial interest in discouraging physicians from ordering scans. Another potentially reasonable conclusion from these data is that the screen applied by the insurance company was excessively restrictive. The authors indicate that the “reviewers” had conversations with the physicians requesting MRI. They do not indicate whether these reviewers were trained medical personnel or merely insurance-company adjusters. The authors imply that the absence of an accepted appeal within six months validates the finding that the scan was inappropriate. This reasoning is specious. Most of the injured workers need scans for the diagnosis and treatment of pain symptoms, not life-threatening illness. It is possible that many of the physicians did not want to spend the time on the telephone fighting with an adjuster and found it easier to declare the patient's condition unlikely to change on the basis of information available in the absence of the needed diagnostic tests.

What was the funding source for this study? The authors are listed as employees of William Mercer, Inc. That is an employee-benefits consulting firm that derives substantial income from the insurance industry in California, which has been a leading supporter of legislation to restrict ownership of imaging centers by physicians. This potential conflict of interest should have been disclosed in the article.

The authors indicate that greater availability of MRI scanners in California leads to greater use of services. This may indicate better care rather than inappropriate use. The data presented in this paper do not help answer the question.

Stanley Cohen, M.D.
UCLA School of Medicine, Los Angeles, CA 90024

To the Editor:

It would seem that “professional values” are now the “family values” of some medical arbiters. The quest to eliminate all self-referral may blind some to the occasional benefits. The report of increased use of physical therapy by some California physicians1 is a case in point. Physical therapy is much more strongly favored by some providers in the rehabilitation of occupational injuries2,3. As a non-self-referring physician in occupational medicine, I can sympathize with self-referrers who are able to provide therapy at a cost 10 percent below that of physicians who refer to independent facilities. Indeed, where I practice, that cost differential is much greater -- at times, nearly 100 percent more. Perhaps the California cohort may represent those who believe in physical therapy much more strongly than other occupational health care providers. Their self-referral may provide the more extensive use of physical therapy at a lower total cost than would be the case with free-standing referral and may have forced down the independent competition's charges.

A.J. Parmet, M.D., M.P.H.
St. Luke's Occupational Medicine, Kansas City, MO 64111

3 References

1. 1

   Swedlow A, Johnson G, Smithline N, Milstein A. Increased costs and rates of use in the California workers' compensation system as a result of self-referral by physicians. N Engl J Med 1992;327:1502-1506
   Full Text | Web of Science | Medline

2. 2

   Lee CK. Office management of low back pain. Orthop Clin North Am 1988;19:797-804
   Web of Science | Medline

3. 3

   Haldeman S. Spinal manipulative therapy in sports medicine. Clin Sports Med 1986;5:277-293
   Web of Science | Medline

Author/Editor Response

The authors reply:

To the Editor: Dr. Doane suggests that returning the injured worker to the usual level of function in the shortest time is highly desirable. We agree. Although it was not previously available, we will have access to this information for future studies.

Dr. Cohen raises several important issues. As we disclosed in our initial submission, the study was funded in part by the Industrial Indemnity Company, a California-based insurer of workers' compensation. Dr. Cohen hypothesizes that a small percentage of physicians may have been responsible for the majority of inappropriate scans. This was not the case. There was a total of 173 inappropriate scans ordered by 80 physicians. Seventy-seven percent of the inappropriate scans were ordered by 50 percent of the physicians.

Dr. Cohen suggests that the reason for the high rate of inappropriate MRI scans (38 percent) is that the screens used were “excessively restrictive.” His statement that “most of the injured workers need scans for the diagnosis and treatment of pain symptoms, not life-threatening illness” underscores the difference in philosophy between Dr. Cohen and the expert panel that designed the guidelines. The guidelines approved the use of scans in situations in which a positive finding would produce a change in treatment likely to benefit the patient. For back patients, that means surgery. As pointed out by Deyo et al.1 and others,2,3 scans are most appropriate for back patients who do not respond to six weeks of conservative therapy and have sciatic pain and corresponding neurologic deficits. Conversely, because disk herniation is so common in asymptomatic persons,4,5 a positive scan in the absence of confirmatory clinical findings may be misleading. We believe that the guidelines reflect the current literature and practice.

Now that health care costs consume 14 percent of the gross national product, we believe it is no longer reasonable, from the viewpoint of the patient, the physician, or the nation, to provide costly services that improve diagnostic specificity slightly but do not alter treatment to the patient's benefit.

Neil Smithline, M.D.
Current Health Concepts, Inc., Sausalito, CA 94965

Arnold Milstein, M.D.
Gregory Johnson, Ph.D.
Alex Swedlow, M.H.S.A.
William M. Mercer, Inc., San Francisco, CA 94111

5 References

1. 1

   Deyo RA, Loeser JD, Bigos SJ. Herniated lumbar intervertebral disk. Ann Intern Med 1990;112:598-603
   Web of Science | Medline

2. 2

   Wiesel SW, Feffer HL, Rothman RH. Industrial low-back pain: a prospective evaluation of a standardized diagnostic and treatment protocol. Spine 1984;9:199-203
   CrossRef | Web of Science | Medline

3. 3

   Quebec Task Force on Spinal Disorders. Scientific approach to the assessment and management of activity-related spinal disorders: a monograph for clinicians. Spine 1987;12:Suppl:S1-S59
   CrossRef | Web of Science | Medline

4. 4

   Hitselberger WE, Witten RM. Abnormal myelograms in asymptomatic patients. J Neurosurg 1968;28:204-206
   CrossRef | Web of Science | Medline

5. 5

   Wiesel SW, Tsourmas N, Feffer HL, Citrin CM, Patronas N. A study of computer-assisted tomography. I. The incidence of positive CAT scans in an asymptomatic group of patients. Spine 1984;9:549-551
   CrossRef | Web of Science | Medline

To the Editor:

As the debate rages about physician self-referral and joint ownership,1-3 a more common and egregious example of a conflict of interest is being ignored and even encouraged. Large multispecialty groups make up the backbone of modern American medicine, form the basis of the prepaid health system, and now even support most U.S. medical school faculties. These groups help practitioners cope with the financial and logistical strains of modern practice. One way they help is by providing a constant and predictable flow of patients through a pattern of internal referrals. Most of these large practices have implicit, if not explicit, requirements that their partners make internal referrals whenever possible. The unstated message is that the optimal physician referrals for all their patients are other physicians in the group. This is not usually the case.

Referral to self-owned facilities often represents an obvious ethical breach in physicians' duties. Yet it merely illustrates a much larger problem that the medical profession and society continue to ignore. Restriction of referral patterns by group practices and peer pressure is a more subtle and more pervasive ethical problem stemming from physicians' self-interest.

Kenneth V. Iserson, M.D., M.B.A.
University of Arizona Health Sciences Center, Tucson, AZ 85724

3 References

1. 1

   Relman AS. “Self-referral” -- what's at stake? N Engl J Med 1992;327:1522-1524
   Full Text | Web of Science | Medline

2. 2

   Mitchell JM, Sunshine JH. Consequences of physicians' ownership of health care facilities -- joint ventures in radiation therapy. N Engl J Med 1992;327:1497-1501
   Full Text | Web of Science | Medline

3. 3

   Swedlow A, Johnson G, Smithline N, Milstein A. Increased costs and rates of use in the California workers' compensation system as a result of self-referral by physicians. N Engl J Med 1992;327:1502-1506
   Full Text | Web of Science | Medline

To the Editor:

The November 19, 1992, issue of the Journal contained two articles, plus Dr. Relman's excellent editorial, on the self-serving practices of physicians who refer patients to facilities in which they have a financial interest. As the articles point out, such referrals increase physicians' incomes without a concomitant improvement in medical care. However, the articles fail to address the most burning issue of self-referral -- the practice of continually making appointments for return visits to one's own office. To all who are concerned with the declining ethics of medicine and the increasing pecuniary interests of private practitioners, such self-referrals must be especially galling.

Dr. Relman has editorialized for years about physicians' runaway incomes and the tendency of physicians to be self-serving. Now he must lead us to address this biggest of all boondoggles, self-referral to ourselves. Repeat visits are bankrupting our medical system.

Gerald S. Besses, M.D.
108 Lynch Creek Way, Petaluma, CA 94954

To the Editor:

Dr. Relman argues forcefully for the pre-1980 ethical code of the AMA. Most would agree with this simple, clear guideline. There is a flip side, however. A physician should not suffer personal financial harm because of professional decisions he or she makes. Perhaps to the code should be added, “In the practice of medicine a physician should not enter into agreements with managed care systems that put the physician at risk because of group utilization.” The concept of income at risk is certainly as antiethical to professionalism as that of self-referral.

Kurt H. Neumann, M.D.
Rose Cancer Center, Royal Oak, MI 48073

To the Editor:

Self-referral evolved as the result of two factors: the changes taking place in medicine and the inequities in the reimbursement system. The rewards of medicine are many, including service to humanity, intellectual stimulation, professional respect, and financial remuneration. Because the first three rewards are diminishing, it is inevitable that the concern for compensation will become more important.

Second, as this becomes paramount, physicians recognize the terrible inequity in the current reimbursement system. The essence of medicine, physicians' cognitive services, remains poorly compensated, while procedures and technical services are proportionally very highly and sometimes obscenely reimbursed.

A look at reimbursement schedules can easily explain why self-referral became a problem. Physicians' visits and consultations are reimbursed for a minute fraction of the radiologic and laboratory fees. Primary care physicians who receive little for their services see the riches others get from their work. No wonder they want a piece of that.

What we need is a major overhaul of the reimbursement system. The recent changes don't even scratch the surface. Reimbursement for laboratory services, home care, physical therapy, and radiology, to name several areas, is excessive in comparison to cognitive services. Balance in the reimbursement of these services is badly needed.

The new millennium calls for a new code of ethics, one that addresses the humane, scientific, professional, and business aspects of modern medicine.

Basel Yanes, M.D.
David L. Rike Cancer Center, Dayton, OH 45409

To the Editor:

On page 1500 of the November 19 issue, Mitchell and Sunshine mention “the AMA's repudiation of its strong stance in June 1992. . . .” Similarly, on page 1502 of the same issue, Swedlow et al. state that “in June 1992, however, the AMA's House of Delegates adopted a new policy that allows doctors to make such referrals. . . .” In the next sentence, Swedlow et al. characterize the House of Delegates as causing a “reversal on the part of the AMA. . . .”

Unfortunately, these statements create the false impression that the action taken by the House of Delegates in June 1992 had the effect of overriding the guidelines of the Council on Ethical and Judicial Affairs that were issued in December 1991. As Dr. Relman correctly observes in his editorial, “The vote could not change the council's report, which remains part of the AMA's code of ethics. . . .”

It is critical for physicians and others to understand that the House of Delegates cannot reverse or repudiate the ethical guidelines of the Council on Ethical and Judicial Affairs. This is important for two reasons. First, on specific issues such as self-referral, physicians need to know that the AMA's ethical guidelines are found in the statements of the Council on Ethical and Judicial Affairs. Second, the legitimacy of the council's guidelines will be undermined if people believe that the guidelines are decided by popular vote, rather than by an independent tribunal that grounds its statements in fundamental ethical principles.

At its December 1992 meeting, the AMA's House of Delegates reaffirmed the council's authority to issue ethical guidelines; the House of Delegates also resolved its disagreement with the council on self-referral by formally expressing its concurrence with the council's self-referral guidelines.

Oscar W. Clark, M.D.
David Orentlicher, M.D., J.D.
Council on Ethical and Judicial Affairs, American Medical Association, Chicago, IL 60610

Author/Editor Response

Dr. Relman replies:

To the Editor: Dr. Iserson evidently considers referral of patients to one's colleagues in a prepaid multispecialty group practice to be just as unethical as referral of patients to an outside facility in which the referring physician has a financial interest. I do not, and I doubt that many others do, either. When patients join a prepaid arrangement such as a health maintenance organization, it is clearly understood that their medical care will be provided by the physicians in the group, who are practicing together as an integrated team.

Dr. Besses raises an important issue -- “self-referral to ourselves.” It certainly contributes to the rising cost of medical care, but it doesn't violate the basic ethical principle that physicians should be paid only for services they personally render or supervise. To eliminate this form of conflict of interest, the fee-for-service system would have to be changed, whereas elimination of the conflict of interest represented by self-referral to outside facilities doesn't require very drastic measures or any basic change in the present fee-for-service system.

I agree with Dr. Neumann. Physicians should not be at personal financial risk for the professional decisions they make. Professionalism is best served when physicians' decisions are economically neutral for the physician -- although prudent medical judgment should certainly consider costs.

I also agree with Dr. Yanes' explanation of the forces that have been encouraging self-referral. He is right in urging “a major overhaul of the reimbursement system” and “a new code of ethics.”

Finally, I am grateful to Dr. Clark and Dr. Orentlicher for their clarification of the AMA's position on this issue. My editorial commends the AMA for its renewed commitment to professional values and expresses the hope that it will take equally strong stands against the participation of practitioners in all other forms of joint ventures and business deals involving the drugs, equipment, and clinical facilities they recommend to their patients.

Arnold S. Relman, M.D.
Harvard Medical School, Boston, MA 02115