Theory of Competing Risks
One of the presumptions upon which we currently allocate health resources is that the greatest cost effectiveness will be achieved if money is spent on the most common causes of death and morbidity,e.g., cancer, stroke and heart disease. This presumption may be incorrect. Although intuition suggests that life expectancy would be benefited greatly by the discovery of a cancer cure, close scrutiny suggests otherwise.
Assume the following to be true:
- That all individuals die of either cancer, stroke or heart disease, and that the frequency as a cause of death for each disease is the same;
- That the cumulative death rate with age for each of the diseases is the same;
- That the acquisition of one disease, and its subsequent cure, does not protect the individual from the other two diseases.
Under these assumptions, it is obvious that the discovery of a cure for cancer would not affect life expectancy at birth. It would only increase by 50% the number of individuals dying from heart disease and stroke at the same age that they would have died had no cure been discovered for cancer. The cure for cancer only changes the cause of death. The foregoing assumptions are not precise (e.g., some deaths due to malignancy occur in childhood), and a cancer mortality rate of zero would therefore actually increase life expectancy by about 2.3 years.6
Our theory of competing risks states: “In a developed society where the exponentials describing the cumulative rates with age for the more common causes of death have similar intercepts, the elimination of one cause of death does not significantly affect life expectancy.”
Expressed simply, if you are 80 years old and your doctor cures you of one ailment, you will likely die of another quite soon.
The percentages of death from all causes naturally add up to 100. As medical science eliminates one cause of death after another, the percentage must still add up to 100. That is what confuses people. We all must die of something. There is only one way to decrease our percentage of death from heart disease, and that is to increase our percentage of death from other things. What would you like to have increased?”
Competing risks explain some of the difficulties encountered with computerized, prospective, randomized studies which attempt to identify a subset of patients whose lives are prolonged significantly by coronary artery bypass grafting. Competing risks may be implicated in the recent observation of an increased incidence of cancer of the colon in the Framingham study.8 In view of the decrease in deaths due to stroke,9 it should be so.
As noticed by Fries, “the medical and social task of eliminating premature death is largely accomplished.”4
Quality of Life
Developing countries can anticipate increases in life expectancy with control of infectious disease, elimination of malnutrition, immunization, and introduction of sanitary sewage and water systems. Medical care systems in industrialized western societies, it would appear, are largely constrained to improving the quality of life. Some benefits may be objective, such as the relief of arthritic pain by medication; other benefits may be related to what Fuchs has described as the “caring function”—reassuring the patient by providing compassionate understanding.10 This caring function is a worthy activity of the physician.
The medical “house call” is a paradigm of the caring function. A physician would not ordinarily seek from a professional colleague a house call for a member of his own family. He is aware that, without a hemacytometer, a radiographic machine, or plaster of Paris, a doctor’s function is limited to giving reassurance. Yet, a Gallup poll would likely show that what the American mother wants most is “a good old-fashioned doctor who makes house calls.” Few events are as threatening to the maternal instinct as a bedridden child with a fever of 105°F. What the mother wants is a physician to come to the secure environment of her home and provide reassurance. What she needs is to meet the physician in the emergency room of a community hospital where something effective can be done about the child’s illness.
The demand for medical care can be divided into two components: 1) Medical wants for services and resources which cannot be shown by objective analysis to be beneficial, and 2) Medical needs consisting of services and resources which favorably affect mortality and the quality of life.
It is relatively easy to distinguish some radical wants (house calls, vitamins, antiseptic mouthwash, face lifts) from some medical needs (water and sewage systems, immunizations, antibiotics, insulin, appendectomies). Other health services are difficult to classify. A “worried well” individual may visit a physician who does nothing which objectively improves the patient’s health; but, if reassurance alone improves the quality of life by eliminating worry, then the consultation was indeed a health “need.” If the physician prescribes a hematinic, or vitamin injection (motivated by his previous observation that the patient will feel better sooner if something positive is done), he plants a seed which grows into a major problem. Because the “worried well” individual attributes his continued good health to the unnecessary therapy, a pattern of demand for health care (want) is established. The popularity of the tribal medicine man arises from an analogous sequence, i.e., most “worried well” warriors who consult him survive the battle, therefore creating a future demand for his services. It is probable that Chairman Mao Tse-Tung recognized that his nation possessed a highly cost-effective means of satisfying medical “wants” when he was faced with the problem of providing medical care for 900 million people with 4,000 conventionally trained physicians. He decreed that “western medicine” and traditional “eastern medicine” (herbs, roots, teas and acupuncture) must coexist in the same facility. Today, traditional herbalists carry a large fraction of the patient load in modern Chinese hospitals where heart surgery and transplantation are performed.
It is appropriate to ask the question: can the consumer be educated to accept satisfaction of wants from a nonphysician? The barefoot doctor in China, the feldsher in Russia, and the American military corpsman represent a very cost-effective means of providing the caring function. In addition to satisfying wants, these physician-extenders may supply medical needs which have great import on population structure, i.e., health education, immunization, and population control. Their effectiveness seems to be determined, not by what they can do but, by the attitudes of those whom they serve. The military corpsman was accepted by 12 million Americans in uniform because the soldier was educated to do so. He would unlikely be accepted in the same role by the civilian public. This suggests that it may be more cost-effective to alter by education the attitudes (that is, the “wants”) of the recipients of medical care than it is to attempt to satisfy those wants by manipulating the health care system.
Health, Health Care Delivery, and Poverty
Two decades ago, there was in the United States, a perceived crisis in the delivery of health care. Society’s response was based on the tacit assumption that a close relationship exists between health on one hand, and health care delivery on the other. From the accumulating mass of data suggesting that this assumption is invalid, I shall cite only a single study.
The investigation by Kohn and White et al.,11,12 under the auspices of the World Health Organization, made comparisons of medical care in 12 areas of seven countries in the western hemisphere and in western and eastern Europe. The study evaluated the health of the populace, the availability and use of health resources, the system of medical care (“socialized,” fee-for-service, etc.), incidence of sick days, rates of acute and chronic illness, incidence of physician contacts, availability of hospital beds, expenditure per capita for health care, hospitalization rate, physician-to-population ratio, the use of prescription and nonprescription drugs, and other factors. This investigation involving a sample of 48,000 people representing 15 million individuals led to, among others, two observations: 1) “in general, the populations do not differ materially in health status as judged by ... traditional indices ... ,” and 2) “People’s perception of their health is clearly a major determinant in the use of services ...”
Health had little relation to health care delivery; the use of health resources was not related to illness. Contrary to conventional wisdom, the organization of the health care system (“access block”) did not determine use.
Lack of a close correlation between health care delivery and health is a signal observation in assigning priorities related to health care. The euphemism “health maintenance organization” applied to prepaid health care plans rekindled interest in the concept of preventing disease rather than treating it. Although public health measures, particularly sanitary water supply and sewage disposal, have profoundly beneficial effects on developing nations, many of the types of preventative medicine practiced in industrialized western societies (annual physical examination, blood tests, screening radiographs, etc.) cannot be shown by scientific study to favorably affect health. As Lewis has said, “Preventative medicine contains more advocacy than reality and suffers from overpromotion in the face of underachievement.”13
Poverty stands in stark contrast to those determinants which might be expected to, but actually do not, affect health. Both life expectancy and standard indices of health show close correlation to gross national product per capita (Fig. 3). As wealth increases above the minimum essential for subsistence, societies appear to assign priority to those resources that favorably affect health, i.e., water supply, sewage treatment, nutrition, housing, and immunization.
The concept of poverty as it affects health extends beyond a consideration of material wealth, or lack of it. People who have been reared in a population that has been poor for generations, acquire a set of culturally and ethnically determined values which impair health, even in the presence of adequate medical services.15 This “culture of poverty” causes the ghetto-dwelling child to suffer the deformity of rickets even though he lives in the shadow of a modern medical center.
Cost of Medical Care
Two examples from my own hospital environment serve to illustrate the common wisdom that effective cost containment is not feasible unless the provider personally shares in the costs which he is responsible for generating.
In almost identical adjacent outpatient suites where the surgical residents and the surgical faculty see their patients in an academic medical center, the overhead costs per patient visit were—resident suite: $25.67; faculty suite: $13.56. The difference was largely explained by the differing efficiencies resulting from the fact that residents had no responsibility for cost-sharing, but the faculty were responsible for 100% of the overhead costs of seeing their private patients.
An experiment was performed to determine the effect of cost-sharing on the ordering of radiographic and laboratory studies on hospitalized patients by a group of six surgical and medical residents. Each resident was offered $100 for each of two weeks to participate in an experiment on “laboratory use.” During the “control” week, they were asked to record every study they ordered in the treatment of patients for whom they were responsible. At the end of this week, the total cost of all radiographic procedures and laboratory tests ordered was computed. Each resident was then told that he had a “bank account” in the amount of 10% of the cost of all tests ordered during the “control” week. During the succeeding “incentive” week, he was told he would receive the agreed-upon $100, but would receive a bonus computed in the following manner: 10% of the cost of every test ordered during the “incentive” week would be subtracted from his “bank account” and, at the end of the week, he would be paid a bonus consisting of all money left in his account. Without adversely affecting patient care, the residents were able, by more prudent and cost-conscious behavior, to reduce laboratory expenditures during the second week by 29.5%. The projected annual saving for the participation of all resident staff in the hospital was $6.6 million.
On reflection, it occurred to me that my efforts were a model of what is wrong with the current national effort at cost containment,i.e., we frequently err in seeking answers to inappropriate questions (e.g., how can we shorten hospital stay, decrease hospital admissions, reduce use of laboratory procedures, develop an effective second opinion program?). A question more pertinent addressing the cost containment issue is: what is responsible for the increase in hospital costs beyond that caused by inflation? Table 1 indicates that hotel-type expenses at the UCLA Medical Center have increased at the rate of general inflation.16 In contrast, costs related to the application of new technology to patient care (renal dialysis, heart-lung machines, radioimmunoassay, new drugs, monitoring, etc.) show cost increases ranging from two to four times that reflected by the consumer price index. Our cost problem clearly stems from the changing nature of medical care in which better clinical results are sought by applying expensive new technology to the care of the sick.
Figure 4 shows the increasing costs of a hospital bed over 20 years stated in constant dollars (inflation adjusted). The rate of increase is, therefore, a measure of the cost of the application of new technology. If it were possible to institute immediately an ideal cost-containment program of 20% and to maintain it, the cost of a hospital bed in 1987 would be 194% of the 1967 cost, rather than 212%. Therefore, the current national cost-containment effort, even if perfectly conceived and executed, will have minimal effect on hospital costs, since it addresses inappropriate questions. In contrast, limiting the application of new technology to patient care could have a very beneficial effect on costs.
The physician is not in a position to limit the application of technology, inasmuch as it is his Hippocratic responsibility to use all available resources, irrespective of the relationship of marginal costs to marginal benefits, in caring for his patient. A number of developed nations have dealt with this ethical/social problem by limiting the resources available to the physician, so that he is forced to select application of expensive technology to circumstances where the greatest marginal benefit exists. For example, queues for herniorrhaphy in some countries are responsible for large numbers of patients dying with a condition which was a minor inconvenience and no threat to life, therefore reserving resources for diseases which have a greater impact on mortality and quality of life. In other countries, limiting facilities for cardiac surgery results in the application of coronary bypass to improving the quality of life, rather than applying it in circumstances where it merely shifts the cause of death from myocardial infarction to a competing risk.
The infinite value we place on human life makes it repugnant to physicians to accept any limit on the effort to preserve it. Nevertheless, the progressively larger fraction of total personal spending for medical care (now 9.7%)17 will soon lead the consumer to attach relative values to, and to select among, medical care and other desired commodities. Each Monday, he now works until noon to pay his medical costs, which is half again more time than he has available to work for his recreational spending. When he perceives the difficulty we are having in demonstrating the relationship between health and health care delivery, he may elect nonmedical options such as home ownership, recreation, and children’s education.
It is difficult to know at what level of spending there will be a populist revolt against investment in health care. Figure 5 makes the assumption that the present 9% fraction of gross national product consumed by medical care will increase at the historical rate, and that the revolt will occur at 12.5%. So long as we continue to develop and apply new technology at the current rate, an immediate and ideally effective 20% cost containment will delay the revolt only slightly.
Again it appears that our present cost containment efforts are addressing inappropriate questions. A more appropriate question to ask might be: how do we ration, in a manner which is ethical and acceptable to the consumer, the application of new technology to patient care?
The Physician’s Self-Perception and the Political Process
It is difficult for the physician to perceive realistically the impact of his personal contributions to the health of the populace at large. The internist who revives the patient from diabetic coma, and the surgeon who excises the burst appendix, have immediate evidence of their personal value to the health of at least one individual. They appreciate less well the fraction of their time that is spent satisfying medical wants, and the role of diabetes and appendicitis on health viewed on a global scale. The statistical impact of the average medical practice is considerably diluted by the fact that 80% of disease is functional or self-limited, 10% is incurable, and only the remaining 10% is susceptible to therapy.
It is essential that we physicians understand the relationship of medical practice to health if we are to participate effectively in decisions relating to the medical care system. The direction in which a free society moves is usually determined by a dynamic balance between the counterposed forces of public desire, scientific advance, political expediency, state of the economy, and other factors. Unfortunately, policy-making in the health field in the United States has, for the past 20 years, been determined to a large extent by slogans: “the doctor shortage,” “crisis in health care delivery,” “maldistribution,” “health maintenance,” “access block,” “continuing medical education,” “relicensure and recertification,” “cost containment,” and “primary care.”
The perceived inadequacy of the number of physicians in the 1960s shares common characteristics with other contemporary crises such as the “missile gap” and “population explosion” in the United States. Each crisis was the source of profound public concern and ultimately proved to be nonexistent. For example, under legislative stimulus, a 50% increase in the number of students admitted to medical school occurred. By 1978, before most of these physicians had the opportunity to complete medical school and postgraduate training, an oversupply of physicians was perceived. Continuing medical education was mandated by a score of state legislatures without any evidence that it affects the quality of care delivered. Recertification and relicensure were accepted in principle before the development of examination procedures that met the tests of validity and reliability.
Each crusade received attention in the media, stimulated legislation, cost much money, and had a life expectancy of about eight years. None had a demonstrable effect on health.
A recent Secretary of Health, Education and Welfare recalled, nostalgically, 1931, when 94% of physicians were engaged in primary care.18 I, too, recall house calls made that year by our family doctor to attend me for purulent otitis media accompanied by fever of 105°. On his daily house calls, he provided analgesic ear drops and aspirin for me, and reassurance to my parents. On this, and four subsequent occasions, a surgical myringotomy was required. It was only with the development of the sulfonamides some years later that he was able to affect, beneficially, the course of my repeated infections. If the 94% of resources allocated in 1931 to primary care had been diverted to research in the development of penicillin and other effective remedies, one wonders if I would not have been healthier sooner.
Primary care, more than any medical specialty, is directed at the satisfaction of wants. I do not impugn the value of primary care, but cite it because it is the currently fashionable shibboleth and illustrates our penchant for mandating expensive national programs without knowledge of their effect on health.
For two decades, we have been witnessing expensive attempts to find answers to inappropriate questions. As a practicing surgeon, I am not qualified to suggest the solutions to global problems in the health field. As a scientist, however, I can identify a problem that may be solved by the application of the scientific method. The first step in the solution to any complex problem is to identify, in precise terms, the critical questions. These might include: what fraction of the gross national product can we afford, and are we willing, to spend on medical services? Of this fraction, how much can we afford to divert to medical wants which have no demonstrable effect on health? Could the consumer be induced to accept satisfaction of his medical wants from a nonphysician? Would greater cost-effectiveness be achieved by changing the attitudes of the medical consumer than by manipulating the health system? Assuming that medical care in an industrialized society does not significantly affect longevity, how much does it affect the quality of life? Could the health of those isolated by poverty and its cultural characteristics be improved more by money spent on education, housing and integration than by the same amount spent on medical care?
It is interesting that the questions address a world which is, as Ramo has observed, a “paradox of technological progressivism and social primitivism.”19
The scientific method can and should be applied in seeking answers to the critical questions, once they are identified. It is too expensive to address a problem of this dimension with intuition, prejudice, political expediency, nostalgia, self-interest, and slogans. The methods of operations research are appropriate to the task (operations research is a scientific method which provides a quantitative basis for decision making).
- Formulating the problem (what are the critical questions?).
- Constructing a model of the system (determining what factors affect health, and what do not; separating wants from needs; attaching values to parameters; establishing goals.)
- Deriving alternative solutions from the model (testing the outcome of alternate strategies for the investment of human and material resources).
- Field testing the solutions (a system which doesn’t survive locally is unlikely to succeed nationwide).
- Imposing constraints on the system (establishing maxima for investment of resources; setting minima on the services and facilities available to geographic, economic and social groups).
- Implementing the best solution.
You may, with reason, wonder why I choose this occasion, before a most distinguished medical audience, to speak of the limitations, rather than of the glories, of our contributions to the commonweal. I do so because, as English historian A. L. Rowse has observed, the great tragedies in history share one thing in common: the people involved at the time did not perceive what was happening around them. Although the relationship between medicine and society does not yet qualify as a tragedy in the historical sense, it is true that there seem to be very few who appreciate what is happening around us.
We physicians have a special responsibility to be aware of social and scientific trends, and to exercise leadership in providing longevity and improved quality of life to our fellows. Our Hippocratic responsibility is not limited to the provision of medical care, but to encouraging whatever steps are necessary to improve the public health. Given the constraints imposed by biology and the limitations of medicine, the requisite action to improve health will, with increasing frequency, be found to lie in the public, rather than in the medical, domain.
It has been politically popular in the recent past to declare that resources should be diverted from research to the application of what we already know. All the evidence I have cited is contrary to this view. Communication in the field of medicine is so effective that new discoveries are known and applied in the developed nations within weeks or months of their announcement. It is the lack of knowledge which keeps us from dealing effectively with chronic illness and debility which impair the quality of life, and which forces us to accept the concept of maximum life span. New knowledge is what is required to improve health in the developed nations. The search for this knowledge should be funded by the termination of expensive and poorly conceived programs which do not beneficially affect health.
Comprehending the limits of medicine will enable us to separate the achievable from the impossible. It will force us to identify those questions, the answers to which will affect health. It will eliminate expensive governmental programs that do not affect health, and will focus attention on our need to develop new knowledge. It will fix the fraction of the working man’s effort directed to the costs of health care. Most importantly, it will eliminate slogans as a basis for health programs and substitute rational planning based on the scientific method.
1. Comfort A. Biology of Senescence, 3rd Ed., New York, Elsevier Press. 1979; pp. 81–86.
2. Hayflick L. The cellular basis for biologic aging. In Finch CE, Hayflick L (eds.) The Biology of Aging. New York, Van Nostrond Reinhold Co. 1977.
3. Hayflick L. The cell biology of human aging. Sci Am
4. Fries JF. Aging, natural death and the compression of morbidity. N Eng J Med
5. Fox RC. The medicalization and demedicalization of American society. Proc Am Acad Arts Sci
6. Keyfitz N. How much would a cure for cancer affect life expectancy? Surgical Rounds July 1978; 24–29.
7. Morrison H. Letter to the editor. Time, Nov. 21, pp. 6, 8, 1955.
8. Williams RR, Sorlie PD, Feinlieb M. Cancer incidence by levels of cholesterol. JAMA
9. Walker, WJ. Changing United States life style and declining vascular mortality: cause or coincidence? N Eng J Med 297:163-5, 1977;297:163–165.
10. Fuchs VR. Who shall live? In Health, Economics and Social Choice. New York, Basic Books. 1974.
11. Kohn R, White KL. (eds.). Health Care: An International Study. Report of the World Health Organization collaborative study of medical care utilization. London, Oxford University Press 1976. p. 392.
12. White KL. International comparisons of medical care. Sci Am
13. Lewis C. DHEW Publication No. (PHS) 78-1232, National-Center for Health Statistics (Med. World News quote), p. 100.
14. Mahler H. People. Sci Am
15. Lewis G. The culture of poverty. Sci Am
16. Lamson BG. Inflation in the Health Care Industry. MEDINFO
17. U.S. Bureau of Economic Analysis. Statistical Abstract of the U.S., 1978 Government Printing Office. 1978; unpublished data, U.S. Bureau of Economic Analysis. Quoted in Socio-economic Fact Book for Surgery. 1980, Haug, J.N., and Kuntzman, J., American College of Surgeons, Chicago, Illinois, 1980.
18. Califano JA Jr. Memorandum #78–56, American Assoc. Med. Colleges, Washington, D.C., 1978.
19. Ramo S. Cure for Chaos. New York, David McKay Co. 1969.
© 2004 Lippincott Williams & Wilkins, Inc.
20. Krishna R. Economic development in India. Sci Am