Journal of the Royal Society of Medicine Volume 83 January 1990
25
When is treatment for cancer economically justified? Discussion paper
B Jennett MD FRCS Department of Neurosurgery, Institute of Neurological Sciences, Glasgow G51 4TF M Buxton BA Health Economics Research Group, Brunel University, Uxbridge UB8 3PH Keywords: cancer treatment; surgery, economics; outcome; patient choice
Public fear of cancer, together with clinical uncertainty about the effectiveness of various methods of management, conspire to leave clinicians wide discretion when deciding what to do for individual patients. It is easy to yield to the temptation to offer treatment for the sake of action rather than because there is a real expectation of bringing benefit by improving outcome. Treatment that is not effective cannot be cost-effective, however inexpensive it is. The effectiveness or value of any therapy depends on the probability that it will enhance the duration or quality of a patient's life -by making it less restricted and more bearable. But treatments impose burdens on the patients treated, who suffer discomfort, pain and sometimes mutilation, as well as the risks of other adverse effects. And they utilize health care resources that are denied to other patients. Were it not for these disadvantages, clinicians could justifiably pursue any treatment that offered even a slight probability of enhancing the quality or duration of life. In practice every treatment imposes some individual burdens and social costs. Management therefore involves balancing probable benefits against probable burdens - individual and social. These principles apply to many aspects of high technology medicine'. The management of cancer is further complicated by the number of disciplines that can be involved. For most cancer patients surgery is the only major therapeutic event, but the treatment of cancer is only part of the work of surgeons. Decisions by surgeons about how cancer should be treated affect competing patients with a variety of conditions. By contrast all the patients in departments of radiation and medical oncology have cancer, and resources are allocated between different cancer sufferers. Most patients in the UK with cancer who are likely to benefit from radiation do receive appropriate treatment by US standards, as do most patients who are likely to benefit from curative or palliative chemotherapy, according to an American report2. The five times greater provision and use of medical oncologists in the US results in much more frequent use there of chemotherapy for patients with advanced and incurable cancer who are unlikely to benefit. But surgery is also sometimes used inappropriately for treating cancer. This occurs even in Britain where a recent review of perioperative deaths showed that a substantial proportion followed operations on patients with advanced cancer, many of whom were also
elderly3. What drives the over-use of expensive and personally burdensome treatment for cancer is a hope for unrealistic outcomes. Whether such pleas are translated into treatment that is neither kind nor
cost-effective depends on whether doctors conspire with their patients to deny the reality of the situation. The emergence of the hospice movement in Britain ought to make it easier to consider the option of limiting active treatment. Because diagnosis and prognosis are now more often discussed truthfully with patients in the early stages of malignant disease, and doctors are more explicit about the intentions of treatment, it should become less difficult to act appropriately. When cure is a realistic objective it may be justified to impose considerable personal burdens on the patient, and to commit valuable health service resources. A reasonable prospect of substantial and sustained symptom relief(active palliation) may also justify some personal burdens and significant expenditure. When not even this much benefit is likely patients should not be exposed to serious hazards and discomfort, nor should scarce resources be used. Meeting the costs of treating the relatively limited range of currently curable cancers is not the issue. The problem is the unjustified use of resources for expensive therapies that offer little benefit. Therapy for the individual patient From the perspective of one clinician and his patient, the choice of appropriate treatment can focus on the benefits and burdens for that individual. What matters is how he values his possible gains relative to his possible losses, when comparing the alternative outcomes and the likelihood that each may be the result. Sometimes a choice has to be made between the duration and quality of life: the patient then needs information about the options and help in considering their relative values. Sometimes the doctor has to decide on a patient's behalf- in the light of that patient's known attitudes and wishes, perhaps with the help of his relatives and friends. Probabilities of different outcomes For many clinical interventions, the relative likelihoods of different outcomes are poorly defined. Even when data on perioperative mortality or long term outcomes are available, the probabilities for a particular patient can be difficult to estimate. He may not conform closely with any group in the published data. However, the track record of the local surgical or oncology unit may be less good than those of the centres for which results are available. But even if the true probabilities were known, how consistently would they be perceived? Experimental 0141-0768/90/ studies show that perception of risk is dependent on 010025-04/$02.00/0 how it is presented. An irrational cognitive illusion © 1990 can lead both doctors and patients to prefer a prospect The Royal of 90% chance of survival to a 10% risk of death4. Society of 'Rational' differences will occur between risk-takers Medicine
26
Journal of the Royal Society of Medicine Volume 83 January 1990
and those who are risk-averse. This applies both to patients and their doctors, and can crucially affect the choice between options when similar average expected values apply. Part of the growing concern with informed consent for treatment is that due weight must be given to the patient's attitude to the probabilities of benefit and burden5. Value of different health outcomes It is difficult to ascertain, let alone to measure, how different people value one health state relative to another. But that is no reason for ignoring the importance of patients' subjective valuations. When eliciting values for health states the selection and 'framing' of the information provided is important6. Individuals who have directly experienced a health state may value it differently than those who have not themselves experienced it7. But values also differ widely among those who have considerable observational experience through their professional dealings with similar patients. In this assumed context of an individual decision, what matters are the relative values, (more correctly the expected utilities), of the various management options that are available for that patient. The inevitable disutility of potentially curative treatment (eg mutilation) may be regarded by some patients as outweighing the probable benefits. For carcinoma of the larynx, laryngectomy offers the best prospect of long survival, but it involves the permanent loss of normal speech. In one study 20% of healthy volunteers indicated a preference for radiotherapy with preservation of speech, despite a significantly lower 3-year survival rate8. Surgeons need to be sensitive to such negative aspects oftheir operations, as well as to the obvious risks of perioperative death or complications. The disutility of surgery is relatively easy to identify and for patients to consider when choosing between alternative treatments. The choice may lie between more or less radical surgery (eg for breast cancer), or between surgery versus either radiotherapy or chemotherapy. Toxicity during treatment is commonly held to be the disadvantage of these alternatives to surgery. But the inconvenience and expense of attending hospital over weeks or months for radiation or drug treatment leads some patients to opt for the short, sharp shock of surgery. The successes claimed for the small proportion of cancers that can now quite often be 'cured' by chemotherapy can also be associated with long-term problems for some patients. Lymphoma has a much better prognosis than most other forms of cancer; yet more than half the successfully treated patients in one series still had many psychological problems more than a year later9"0. The success story of acute lymphoblastic leukaemia of childhood is an important force in driving oncologists to seek new drugs for treating other cancers. When this causes serious toxicity to patients without obvious benefit their defence is often to refer to the early days of the development of leukaemia treatment. Yet the cure rate for childhood leukaemia is still only 50%, and in one report almost half of those claimed to be cured still
had residual physical and psychological morbidity six or more years later". When estimating the cost per year of good quality of life gained by treating this disease account should be taken of these data.
The individual and other patients The individualistic perspective assumed above is not, however, realistic. Decisions to provide treatment to one patient almost always involve denying equivalent resources to other patients. There is always a range of alternative potentially beneficial uses for resources, not all of which can be accommodated within a fixed budget. Most clinicians implicitly take this into account when allocating their time between patients. Ending a consultation with a patient does not usually signify that there would be no additional benefit from further discussions. Rather it acknowledges that other patients are waiting, and that the extra time that could be spent on one patient would yield greater value if devoted to another. To rely on simple equity alone might require that each patient receive the same amount of resources (or time). Few clinicians would argue that this should override all considerations of 'need'. One patient may be clearly reassured by a short consultation, while another needs longer. Common sense dictates that more time be given to the latter patient. But if a patient is known to be unreassurable, time would be better spent on another patient who is more likely to benefit. When allocating resources in their departments, radiotherapists may choose to use short courses of radiotherapy for palliation in order to have more time available for longer and potentially curative regimens for other patients. However, some radiotherapists continue to use longer, fractionated courses of palliative treatment that they 'believe' to be 'better', although studies show little or no added benefit. As well as using more machine and professional time such regimens cause inconvenience to patients and their relatives and may incur transport costs. Radiotherapists should be required to state explicitly why they consider the benefits of their preferred method to be 'worth' the marginal extra costs. The same goes for surgeons who insist on more radical or elaborate procedures than their colleagues.
Planning allocation of health resources Issues ofchoice become more acute when planning the allocation of resources for the future, but the underlying arguments are the same. A truly egalitarian approach might allow the provision of a particular form of care only if it could be made available to all patients who might benefit. This would ignore the crucial role of clinical selection - of deciding which patients are most likely to benefit. Realistically equity should be pursued only in relation to treatment that is appropriate - defined as being necessary and effective. Necessity is, however, an elusive concept, easy to accept in principle but difficult to define in practice. Within a fixed budget, health care resources should be allocated so as to achieve the greatest aggregate ofwellbeing for patients. There is, however, room for considerable controversy about how the most good is likely to be done. One concept for allocating resources is the cost of gaining ('or buying') a 'quality adjusted life-year' (QALY). This provides a measure of how much benefit has to be foregone in one application of scarce resources to achieve a similar unit of benefit from an alternative health care activity. The cost element of this indicator measures the scarce resources that are involved; the QALY is the measure of the benefit achieved. This concept seems to be gaining cautious
acCeptance'U.
Journal of the Royal Society of Medicine Volume 83 January 1990 Table 1. Comparative costs per Quality-Adjusted-Life Year (QALY) gained for various health care procedures
Extra costs per QALY gained (£C at 1983/84 prices
Procedure
Pacemaker implantation for atrioventricular heart block Hip replacement Coronary artery bypass grafting (CABG) for main vessel disease Kidney transplantation Breast cancer screening estimated to be in the range Heart transplantation CABG for moderate angina with one vessel disease Hospital haemodialysis
700 750 1040
3000 3-5000 5000 12 000 14 000
From: A Williams'2 *By working group in DHSS Report"6 Table 2. Comparative cost-utility results for selected programmes
Adjusted cost/QALY gained in US dollars (at 1983 prices)
Programme Antepartum anti-D (rhesus incompatibility) CABG for main vessel disease Neonatal intensive care, 1000-1499 g Breast cancer screening estimated to be in this range T4 (thyroid) screening Treatment of severe hypertension (diastolic > 105 mmHg) in males age 40 Treatment of mild hypertension (diastolic 95-104 mmHg) in males age 40 Neonatal intensive care 500-999 g CABG for one vessel/moderate angina Continuous ambulatory peritoneal dialysis Hospital haemodialysis From: GW Torrance15 *By working group in DHSS
1220 4200 4500 4500-6300
6300 9400 19 100 31 800 36 300
47 100 54 000
Report'6
But some ethicists still argue for either a completely equitable or a completely random basis for resource allocation13; other critics are concerned about the inadequacy of the available data for QALY calcula-
tions'4. It is true that good costing data are available for only a limited number of interventions, and many important procedures and treatments are missing from the often quoted comparisons of Williams'2 and of Torrance15 (Tables 1 and 2). These data are quoted in the recent DHSS report on screening for breast cancer16, indicating the use of QALYs in real world decisions. There are few examples of QALYs for cancer therapy, reflecting the paucity of good economic studies in this field. In terms of calculating the unit of benefit, more fundamental doubts are reasonably expressed about
the existing empirical bases for measuring the relative values of health outcomes. That individual patients and society in general are not indifferent to quality of life, and are willing to make trade-offs between quality and quantity, is widely accepted. However, the strength and pattern of preferences of individuals vary widely and are difficult to elicit. The appropriateness of using as indicators of collective social preference the values derived from the small number of available studies is questionable. To establish a broader and more robust set of relative values of health states requires more data derived from a wider range of people. They should be of a range of ages, of social and ethnic backgrounds, and of experience of disease and disability'5. There is a need also to confirm findings that the relative values ascribed to various health states are influenced by the label of 'cancer', which can itself impose a 'disutility' regardless of what kind of treatment is used'7. The recent tendency for more open discussion of prognosis and management options with patients who have cancer may be changing the social attitude to this disease. Years of concealing the diagnosis of cancer from patients has meant that most people have known only of those who died of their disease, those who lived for years having mostly been left to assume that they never had cancer. Because the fortunate survivors have remained 'unlabelled', the impression has been reinforced that cancer is indeed a dread disease that is seldom cured. The concept of cost per QALY takes a utilitarian approach and assumes that the objectives of health interventions are both to extend the duration and to improve the quality of life. Would we wish instead to argue for action for action's sake? Accepting the objectives of utilitarianism, the QALY approach has the potential to handle explicitly the issue of the relative values of different health states, and of variations in mortality and morbidity consequent on interventions. It allows for preferences that indicate that the value of a year of life to an individual or to society is dependent on the quality of life. Not only is higher quality of life preferred to lower quality, but a shorter life of higher quality may be preferred to a longer life of lower quality. The QALY concept is underpinned by a clear concept of need, defined in terms of the potential of health resources to bring benefit. The 'cost per QALY' is not itself a measure of the value of life, but of the value of the resources that have to be expended on a particular intervention in order to achieve a year of full-quality life or its equivalent. Within a fixed budget, devoting health service resources to a use that produces QALYs more expensively than an alternative use inevitably reduces the total number of QALYs that can be 'bought'. The unenviable comparisons that have to be made between the list of potentially beneficial uses for limited health resources can then be made on a more rational basis. To suggest, at a philosophical level, that such comparisons are impossible is to deny the feasibility of planning 'health care'. Indeed it refuses to accept the moral duty to make the best use of scarce resources. Conclusions Whilst clinicians are gradually coming to accept the principle of opportunity costs, putting this into practice in daily dealings with patients can
27
28
Journal of the Royal Society of Medicine Volume 83 January 1990
be difficult. When treating cancer, clinicians are apt to regard any call for restraint that has an economic aspect as conflicting with their wish to do everything possible for the patient before them. On closer analysis this often proves to be a false antithesis, and what is economically sound coincides with what is ethically most appropriate18. Such analysis implies careful consideration of how much benefit this particular patient is likely to derive from a costly intervention under consideration. Clinical guidelines can be helpful in this situation. They act as an aide memoire to the clinician about the various factors that he should take into account before reaching a decision. They also make available the considered opinions of experts who have deliberated well away from the pressures associated with deciding about an individual patient. Guidelines can be helpful in giving permission to a clinician who wishes to avoid indulging in the use of expensive and unproven drugs, but who feels under pressure to do so. They may also restrain the enthusiast who is anxious to try anything new, regardless of expense. There are currently wide variations in how even common cancers are managed. It seems improbable that these each reflect equally appropriate care for all patients, let alone that each makes good sense economically. The recent call by some oncologists and others19'20 for expert groups to draw up consensus guidelines about the management of various cancers is therefore welcome. References 1 Jennett B. High technology medicine: benefits and burdens, 2nd edn. Oxford: Oxford University Press, 1986 2 Aaron HJ, Schwartz WB. The painful prescription: rationing hospital care. Washington DC: The Brookings Institutions, 1984 3 Anon. Accounting for perioperative deaths. Lancet 1987;ii:1369-71 4 McNeil BJ, Pauker SG, Sox HC, Tversky A. On the elicitation of preferences for alternative therapies. N Engl J Med 1982;306:1259-62
5 O'Brien BJ. What are my chances doctor? - A review of clinical risks. London: Office of Health Economics, 1986 6 Llewellyn-Thomas H, Sutherland HJ, Tibshirani R, et al. Describing health states: methodological issues in obtaining values for health states. Med Care 1984; 22:543-52 7 Sackett DL, Torrance GW. The utility of different health states as perceived by the general public. J Chronic Dis 1978;31:697-704 8 McNeil BJ, Weichselbaum R, Pauker SG. Speech and survival: tradeoffs between quality and quantity of life: laryngeal cancer. N Engl J Med 1981;305:982-7 9 Devlen J, Maguire P, Phillips P, et al Psychological problems associated with diagnosis and treatment of lymphomas. I: Retrospective study. Br Med J 1987; 295:953-4 10 Devlen J, Maguire P, Phillips P, et aL Psychological problems associated with diagnosis and treatment of lymphomas. II: Prospective study. Br Med J 1987; 295:955-7 11 Wheeler K, Leiper AD, Jannoun L, et al. Medical cost of curing childhood acute lymphoblastic leukaemia. Br Med J 1988;296:162-6 12 Williams A. Economics of coronary artery bypass grafting. Br Med J 1985;291:326-9 13 Harris J. QALYifying the value of life. J Med Ethics 1987;13:117-23 14 Smith A. Qualms about QALYs. Lancet 1987;i:1134-6,72 15 Torrance GW. Measurement of health state utilities for economic appraisal. J Health Econ 1986;5:1-30 16 DHSS. Breast Cancer Screening (Forrest Report). London: HMSO, 1988 17 Buxton MJ, Ashby J. The time trade-off approach to health state valuation. In: Teeling-Smith G, ed. Measuring Health: a practical approach. Chichester: John Wiley, 1988 18 Jennett B. Are ethics and economics incompatible in health care. Proc R Coll Phys Edin 1987;17:190-5 19 Timothy A, Brewin AR, Chamberlain J, et aL Cost versus benefit in non-surgical management of patients with cancer. Br Med J 1988;297:471-2 20 Smith A. Consensus on overtreating cancer. Br Med J
1988;297:438 (Accepted 26 June 1989)
25
When is treatment for cancer economically justified? Discussion paper
B Jennett MD FRCS Department of Neurosurgery, Institute of Neurological Sciences, Glasgow G51 4TF M Buxton BA Health Economics Research Group, Brunel University, Uxbridge UB8 3PH Keywords: cancer treatment; surgery, economics; outcome; patient choice
Public fear of cancer, together with clinical uncertainty about the effectiveness of various methods of management, conspire to leave clinicians wide discretion when deciding what to do for individual patients. It is easy to yield to the temptation to offer treatment for the sake of action rather than because there is a real expectation of bringing benefit by improving outcome. Treatment that is not effective cannot be cost-effective, however inexpensive it is. The effectiveness or value of any therapy depends on the probability that it will enhance the duration or quality of a patient's life -by making it less restricted and more bearable. But treatments impose burdens on the patients treated, who suffer discomfort, pain and sometimes mutilation, as well as the risks of other adverse effects. And they utilize health care resources that are denied to other patients. Were it not for these disadvantages, clinicians could justifiably pursue any treatment that offered even a slight probability of enhancing the quality or duration of life. In practice every treatment imposes some individual burdens and social costs. Management therefore involves balancing probable benefits against probable burdens - individual and social. These principles apply to many aspects of high technology medicine'. The management of cancer is further complicated by the number of disciplines that can be involved. For most cancer patients surgery is the only major therapeutic event, but the treatment of cancer is only part of the work of surgeons. Decisions by surgeons about how cancer should be treated affect competing patients with a variety of conditions. By contrast all the patients in departments of radiation and medical oncology have cancer, and resources are allocated between different cancer sufferers. Most patients in the UK with cancer who are likely to benefit from radiation do receive appropriate treatment by US standards, as do most patients who are likely to benefit from curative or palliative chemotherapy, according to an American report2. The five times greater provision and use of medical oncologists in the US results in much more frequent use there of chemotherapy for patients with advanced and incurable cancer who are unlikely to benefit. But surgery is also sometimes used inappropriately for treating cancer. This occurs even in Britain where a recent review of perioperative deaths showed that a substantial proportion followed operations on patients with advanced cancer, many of whom were also
elderly3. What drives the over-use of expensive and personally burdensome treatment for cancer is a hope for unrealistic outcomes. Whether such pleas are translated into treatment that is neither kind nor
cost-effective depends on whether doctors conspire with their patients to deny the reality of the situation. The emergence of the hospice movement in Britain ought to make it easier to consider the option of limiting active treatment. Because diagnosis and prognosis are now more often discussed truthfully with patients in the early stages of malignant disease, and doctors are more explicit about the intentions of treatment, it should become less difficult to act appropriately. When cure is a realistic objective it may be justified to impose considerable personal burdens on the patient, and to commit valuable health service resources. A reasonable prospect of substantial and sustained symptom relief(active palliation) may also justify some personal burdens and significant expenditure. When not even this much benefit is likely patients should not be exposed to serious hazards and discomfort, nor should scarce resources be used. Meeting the costs of treating the relatively limited range of currently curable cancers is not the issue. The problem is the unjustified use of resources for expensive therapies that offer little benefit. Therapy for the individual patient From the perspective of one clinician and his patient, the choice of appropriate treatment can focus on the benefits and burdens for that individual. What matters is how he values his possible gains relative to his possible losses, when comparing the alternative outcomes and the likelihood that each may be the result. Sometimes a choice has to be made between the duration and quality of life: the patient then needs information about the options and help in considering their relative values. Sometimes the doctor has to decide on a patient's behalf- in the light of that patient's known attitudes and wishes, perhaps with the help of his relatives and friends. Probabilities of different outcomes For many clinical interventions, the relative likelihoods of different outcomes are poorly defined. Even when data on perioperative mortality or long term outcomes are available, the probabilities for a particular patient can be difficult to estimate. He may not conform closely with any group in the published data. However, the track record of the local surgical or oncology unit may be less good than those of the centres for which results are available. But even if the true probabilities were known, how consistently would they be perceived? Experimental 0141-0768/90/ studies show that perception of risk is dependent on 010025-04/$02.00/0 how it is presented. An irrational cognitive illusion © 1990 can lead both doctors and patients to prefer a prospect The Royal of 90% chance of survival to a 10% risk of death4. Society of 'Rational' differences will occur between risk-takers Medicine
26
Journal of the Royal Society of Medicine Volume 83 January 1990
and those who are risk-averse. This applies both to patients and their doctors, and can crucially affect the choice between options when similar average expected values apply. Part of the growing concern with informed consent for treatment is that due weight must be given to the patient's attitude to the probabilities of benefit and burden5. Value of different health outcomes It is difficult to ascertain, let alone to measure, how different people value one health state relative to another. But that is no reason for ignoring the importance of patients' subjective valuations. When eliciting values for health states the selection and 'framing' of the information provided is important6. Individuals who have directly experienced a health state may value it differently than those who have not themselves experienced it7. But values also differ widely among those who have considerable observational experience through their professional dealings with similar patients. In this assumed context of an individual decision, what matters are the relative values, (more correctly the expected utilities), of the various management options that are available for that patient. The inevitable disutility of potentially curative treatment (eg mutilation) may be regarded by some patients as outweighing the probable benefits. For carcinoma of the larynx, laryngectomy offers the best prospect of long survival, but it involves the permanent loss of normal speech. In one study 20% of healthy volunteers indicated a preference for radiotherapy with preservation of speech, despite a significantly lower 3-year survival rate8. Surgeons need to be sensitive to such negative aspects oftheir operations, as well as to the obvious risks of perioperative death or complications. The disutility of surgery is relatively easy to identify and for patients to consider when choosing between alternative treatments. The choice may lie between more or less radical surgery (eg for breast cancer), or between surgery versus either radiotherapy or chemotherapy. Toxicity during treatment is commonly held to be the disadvantage of these alternatives to surgery. But the inconvenience and expense of attending hospital over weeks or months for radiation or drug treatment leads some patients to opt for the short, sharp shock of surgery. The successes claimed for the small proportion of cancers that can now quite often be 'cured' by chemotherapy can also be associated with long-term problems for some patients. Lymphoma has a much better prognosis than most other forms of cancer; yet more than half the successfully treated patients in one series still had many psychological problems more than a year later9"0. The success story of acute lymphoblastic leukaemia of childhood is an important force in driving oncologists to seek new drugs for treating other cancers. When this causes serious toxicity to patients without obvious benefit their defence is often to refer to the early days of the development of leukaemia treatment. Yet the cure rate for childhood leukaemia is still only 50%, and in one report almost half of those claimed to be cured still
had residual physical and psychological morbidity six or more years later". When estimating the cost per year of good quality of life gained by treating this disease account should be taken of these data.
The individual and other patients The individualistic perspective assumed above is not, however, realistic. Decisions to provide treatment to one patient almost always involve denying equivalent resources to other patients. There is always a range of alternative potentially beneficial uses for resources, not all of which can be accommodated within a fixed budget. Most clinicians implicitly take this into account when allocating their time between patients. Ending a consultation with a patient does not usually signify that there would be no additional benefit from further discussions. Rather it acknowledges that other patients are waiting, and that the extra time that could be spent on one patient would yield greater value if devoted to another. To rely on simple equity alone might require that each patient receive the same amount of resources (or time). Few clinicians would argue that this should override all considerations of 'need'. One patient may be clearly reassured by a short consultation, while another needs longer. Common sense dictates that more time be given to the latter patient. But if a patient is known to be unreassurable, time would be better spent on another patient who is more likely to benefit. When allocating resources in their departments, radiotherapists may choose to use short courses of radiotherapy for palliation in order to have more time available for longer and potentially curative regimens for other patients. However, some radiotherapists continue to use longer, fractionated courses of palliative treatment that they 'believe' to be 'better', although studies show little or no added benefit. As well as using more machine and professional time such regimens cause inconvenience to patients and their relatives and may incur transport costs. Radiotherapists should be required to state explicitly why they consider the benefits of their preferred method to be 'worth' the marginal extra costs. The same goes for surgeons who insist on more radical or elaborate procedures than their colleagues.
Planning allocation of health resources Issues ofchoice become more acute when planning the allocation of resources for the future, but the underlying arguments are the same. A truly egalitarian approach might allow the provision of a particular form of care only if it could be made available to all patients who might benefit. This would ignore the crucial role of clinical selection - of deciding which patients are most likely to benefit. Realistically equity should be pursued only in relation to treatment that is appropriate - defined as being necessary and effective. Necessity is, however, an elusive concept, easy to accept in principle but difficult to define in practice. Within a fixed budget, health care resources should be allocated so as to achieve the greatest aggregate ofwellbeing for patients. There is, however, room for considerable controversy about how the most good is likely to be done. One concept for allocating resources is the cost of gaining ('or buying') a 'quality adjusted life-year' (QALY). This provides a measure of how much benefit has to be foregone in one application of scarce resources to achieve a similar unit of benefit from an alternative health care activity. The cost element of this indicator measures the scarce resources that are involved; the QALY is the measure of the benefit achieved. This concept seems to be gaining cautious
acCeptance'U.
Journal of the Royal Society of Medicine Volume 83 January 1990 Table 1. Comparative costs per Quality-Adjusted-Life Year (QALY) gained for various health care procedures
Extra costs per QALY gained (£C at 1983/84 prices
Procedure
Pacemaker implantation for atrioventricular heart block Hip replacement Coronary artery bypass grafting (CABG) for main vessel disease Kidney transplantation Breast cancer screening estimated to be in the range Heart transplantation CABG for moderate angina with one vessel disease Hospital haemodialysis
700 750 1040
3000 3-5000 5000 12 000 14 000
From: A Williams'2 *By working group in DHSS Report"6 Table 2. Comparative cost-utility results for selected programmes
Adjusted cost/QALY gained in US dollars (at 1983 prices)
Programme Antepartum anti-D (rhesus incompatibility) CABG for main vessel disease Neonatal intensive care, 1000-1499 g Breast cancer screening estimated to be in this range T4 (thyroid) screening Treatment of severe hypertension (diastolic > 105 mmHg) in males age 40 Treatment of mild hypertension (diastolic 95-104 mmHg) in males age 40 Neonatal intensive care 500-999 g CABG for one vessel/moderate angina Continuous ambulatory peritoneal dialysis Hospital haemodialysis From: GW Torrance15 *By working group in DHSS
1220 4200 4500 4500-6300
6300 9400 19 100 31 800 36 300
47 100 54 000
Report'6
But some ethicists still argue for either a completely equitable or a completely random basis for resource allocation13; other critics are concerned about the inadequacy of the available data for QALY calcula-
tions'4. It is true that good costing data are available for only a limited number of interventions, and many important procedures and treatments are missing from the often quoted comparisons of Williams'2 and of Torrance15 (Tables 1 and 2). These data are quoted in the recent DHSS report on screening for breast cancer16, indicating the use of QALYs in real world decisions. There are few examples of QALYs for cancer therapy, reflecting the paucity of good economic studies in this field. In terms of calculating the unit of benefit, more fundamental doubts are reasonably expressed about
the existing empirical bases for measuring the relative values of health outcomes. That individual patients and society in general are not indifferent to quality of life, and are willing to make trade-offs between quality and quantity, is widely accepted. However, the strength and pattern of preferences of individuals vary widely and are difficult to elicit. The appropriateness of using as indicators of collective social preference the values derived from the small number of available studies is questionable. To establish a broader and more robust set of relative values of health states requires more data derived from a wider range of people. They should be of a range of ages, of social and ethnic backgrounds, and of experience of disease and disability'5. There is a need also to confirm findings that the relative values ascribed to various health states are influenced by the label of 'cancer', which can itself impose a 'disutility' regardless of what kind of treatment is used'7. The recent tendency for more open discussion of prognosis and management options with patients who have cancer may be changing the social attitude to this disease. Years of concealing the diagnosis of cancer from patients has meant that most people have known only of those who died of their disease, those who lived for years having mostly been left to assume that they never had cancer. Because the fortunate survivors have remained 'unlabelled', the impression has been reinforced that cancer is indeed a dread disease that is seldom cured. The concept of cost per QALY takes a utilitarian approach and assumes that the objectives of health interventions are both to extend the duration and to improve the quality of life. Would we wish instead to argue for action for action's sake? Accepting the objectives of utilitarianism, the QALY approach has the potential to handle explicitly the issue of the relative values of different health states, and of variations in mortality and morbidity consequent on interventions. It allows for preferences that indicate that the value of a year of life to an individual or to society is dependent on the quality of life. Not only is higher quality of life preferred to lower quality, but a shorter life of higher quality may be preferred to a longer life of lower quality. The QALY concept is underpinned by a clear concept of need, defined in terms of the potential of health resources to bring benefit. The 'cost per QALY' is not itself a measure of the value of life, but of the value of the resources that have to be expended on a particular intervention in order to achieve a year of full-quality life or its equivalent. Within a fixed budget, devoting health service resources to a use that produces QALYs more expensively than an alternative use inevitably reduces the total number of QALYs that can be 'bought'. The unenviable comparisons that have to be made between the list of potentially beneficial uses for limited health resources can then be made on a more rational basis. To suggest, at a philosophical level, that such comparisons are impossible is to deny the feasibility of planning 'health care'. Indeed it refuses to accept the moral duty to make the best use of scarce resources. Conclusions Whilst clinicians are gradually coming to accept the principle of opportunity costs, putting this into practice in daily dealings with patients can
27
28
Journal of the Royal Society of Medicine Volume 83 January 1990
be difficult. When treating cancer, clinicians are apt to regard any call for restraint that has an economic aspect as conflicting with their wish to do everything possible for the patient before them. On closer analysis this often proves to be a false antithesis, and what is economically sound coincides with what is ethically most appropriate18. Such analysis implies careful consideration of how much benefit this particular patient is likely to derive from a costly intervention under consideration. Clinical guidelines can be helpful in this situation. They act as an aide memoire to the clinician about the various factors that he should take into account before reaching a decision. They also make available the considered opinions of experts who have deliberated well away from the pressures associated with deciding about an individual patient. Guidelines can be helpful in giving permission to a clinician who wishes to avoid indulging in the use of expensive and unproven drugs, but who feels under pressure to do so. They may also restrain the enthusiast who is anxious to try anything new, regardless of expense. There are currently wide variations in how even common cancers are managed. It seems improbable that these each reflect equally appropriate care for all patients, let alone that each makes good sense economically. The recent call by some oncologists and others19'20 for expert groups to draw up consensus guidelines about the management of various cancers is therefore welcome. References 1 Jennett B. High technology medicine: benefits and burdens, 2nd edn. Oxford: Oxford University Press, 1986 2 Aaron HJ, Schwartz WB. The painful prescription: rationing hospital care. Washington DC: The Brookings Institutions, 1984 3 Anon. Accounting for perioperative deaths. Lancet 1987;ii:1369-71 4 McNeil BJ, Pauker SG, Sox HC, Tversky A. On the elicitation of preferences for alternative therapies. N Engl J Med 1982;306:1259-62
5 O'Brien BJ. What are my chances doctor? - A review of clinical risks. London: Office of Health Economics, 1986 6 Llewellyn-Thomas H, Sutherland HJ, Tibshirani R, et al. Describing health states: methodological issues in obtaining values for health states. Med Care 1984; 22:543-52 7 Sackett DL, Torrance GW. The utility of different health states as perceived by the general public. J Chronic Dis 1978;31:697-704 8 McNeil BJ, Weichselbaum R, Pauker SG. Speech and survival: tradeoffs between quality and quantity of life: laryngeal cancer. N Engl J Med 1981;305:982-7 9 Devlen J, Maguire P, Phillips P, et al Psychological problems associated with diagnosis and treatment of lymphomas. I: Retrospective study. Br Med J 1987; 295:953-4 10 Devlen J, Maguire P, Phillips P, et aL Psychological problems associated with diagnosis and treatment of lymphomas. II: Prospective study. Br Med J 1987; 295:955-7 11 Wheeler K, Leiper AD, Jannoun L, et al. Medical cost of curing childhood acute lymphoblastic leukaemia. Br Med J 1988;296:162-6 12 Williams A. Economics of coronary artery bypass grafting. Br Med J 1985;291:326-9 13 Harris J. QALYifying the value of life. J Med Ethics 1987;13:117-23 14 Smith A. Qualms about QALYs. Lancet 1987;i:1134-6,72 15 Torrance GW. Measurement of health state utilities for economic appraisal. J Health Econ 1986;5:1-30 16 DHSS. Breast Cancer Screening (Forrest Report). London: HMSO, 1988 17 Buxton MJ, Ashby J. The time trade-off approach to health state valuation. In: Teeling-Smith G, ed. Measuring Health: a practical approach. Chichester: John Wiley, 1988 18 Jennett B. Are ethics and economics incompatible in health care. Proc R Coll Phys Edin 1987;17:190-5 19 Timothy A, Brewin AR, Chamberlain J, et aL Cost versus benefit in non-surgical management of patients with cancer. Br Med J 1988;297:471-2 20 Smith A. Consensus on overtreating cancer. Br Med J
1988;297:438 (Accepted 26 June 1989)
