- Email: [email protected]
The QALY: A guide for the public health physician
Public Health (1993), 107,327-336
O The Society of Public Health, 1993
The QALY: A Guide for the Public Health Physician S. Petrou, BSc PhD 1 and A. Renton, MBBS MFPHM 2
7Research Fellow in Health Economics, 2Senior Lecturer in Public Health Medicine, Academic Department of Public Health, St Mary's Hospital Medical School, Norfolk Place, London W2 1PG
This paper considers the background to the rational allocation of health resources at DHA level within the context of the new purchaser/provider regime. A popular approach to this problem has been that of cost utility analysis, using quality adjusted life years (QALYs) as the measure of the benefit derived from a health intervention. This paper seeks to offer the practising public health physician an appraisal of the practical value and pitfalls of a QALY-based approach. To this end, the theoretical background to QALYs and their practical use are described and some of the practical and ethical problems and pitfalls are discussed. It it concluded that the QALY concept provides a useful tool in aiding decision making in the allocation of health care resources, but that it is too early for a universal application of the approach.
Introduction T h e National H e a l t h Service ( N H S ) review has defined the role of District H e a l t h Authorities ( D H A s ) as purchasers of health care. 1 U n d e r the new a r r a n g e m e n t s , D H A s are funded on the basis of the size and characteristics of their resident populations. Within these financial constraints they are charged with purchasing services to m e e t the health care needs of their residents. In response to these new d e m a n d s , a variety of agencies and professional groups have set out to elaborate the concepts and methodologies required for rational allocation of resources. Currently, health care resources in the N H S are allocated by a mixture of formal and informal devices. 2 It has b e e n argued, 3 however, that the new regime provides us with a unique o p p o r t u n i t y b o t h to place resource allocation on a rational footing, and to m a k e explicit choices and priorities which until recently have b e e n m a d e on an ad hoc or implicit basis. Cost utility analysis represents a standard economic technique for maximising allocative efficiency within health care provision. It allows all health interventions to b e c o m p a r e d in terms of their costs and the health i m p r o v e m e n t s they procure. T w o types of information are required to construct a cost utility analysis: the health gain accrued through interventions and the costs of those interventions. District-specific information of each of these types will be h a r d to c o m e by, although m u c h can be achieved through an integrated purchasing intelligence function. 4 H o w e v e r , with regard to the health gain derived f r o m interventions, there is a particular p r o b l e m : that of quantification. W h e r e the health o u t c o m e s of interventions have b e e n assessed they are often m e a s u r e d in qualitative terms, or as quantitative measures derived f r o m the specific clinical features of the health problem. While such disease-specific
Correspondence to: Dr Stavros Petrou, c/o Academic Department of Public Health, St Mary's Hospital Medical School, Imperial College of Science, Technology and Medicine, Norfolk Place, London W2 1PG.
328
S. Petrou and A. Renton
outcomes may be extremely useful in making comparisons between different interventions for the same disease, they will be of little value in the construction of a general health gain function for the District as a whole. There is, therefore, a requirement that we measure the health gain derived from an intervention in units which are independent of the particular disease or intervention considered. It is to provide this unit of health gain that the QALY has been developed. QALYs Quality adjusted life years or QALYs measure the health gains derived from health interventions in a single unit which combines increased survival and enhancement of health-related quality of life. They have been calculated for interventions for a number of illnesses, usually together with costs, to inform resource allocation decisions within specialities. Stason and Weinstein, 5 for example, used QALY data to determine how resources could be used most efficiently within programmes to treat hypertension. More recently, attempts have been made to use QALY data to compare the relative cost-utility of interventions between diseases. 6-7
Calculating QALYs Two quantities must be known in order to calculate the gain in QALYs which an individual with a specific health problem can expect to derive from a health intervention: the likely effect of the intervention on his/her survival, and its effect on his/her health-related quality of life.
Life years gained An estimate of the average number of life-years gained by individuals receiving an intervention would ideally be achieved through clinical trials of the intervention, incorporating long-term follow up of patients and controls. Such studies may however be extremely expensive and difficult to carry out. Consequently, many trials identify normalisation of clinical pathological or pathological indicators of disease rather than death as study end-points. Estimates of increased survival are therefore more generally available for interventions for diseases which are more rapidly fatal.
Quality o f life gained A number of approaches have been used to estimate the health-related quality of life of individuals with different diseases and the gains derived from interventions. The task is made complex by the need to make value judgements about the relative importance of the different dimensions of 'quality' and to devise scales to measure them, and by the problem of securing representative samples, comprising appropriate numbers and combinations of subjects. Many approaches have been adopted, and these fall broadly into four categories.
Decision analysis Groups of experts in the management of an illness (usually doctors) are asked to attribute health-related quality of life values to the different health states derived
The Q A L Y
329
from different interventions in a disease. 8-11 While this approach is quick and relatively inexpensive, professional and patient valuations have been shown to be at variance in some cases.12
Direct empirical measurement These approaches use various devices to measure preferences for different health states. These may be measured in general population samples or patients with specific health problems, and can be used in the aggregate to generate QALYs. Measurement techniques used include rating scales, the standard gamble approach, the time trade-off approach, the equivalence technique and ratio scaling. We will deal briefly with each of these. Rating scales are a relatively straightforward method of measuring health-related quality of life. Typically, individuals are offered several case descriptions covering a wide range of health states and are asked to assign values to each on the same scale whose end-points are usually defined as death on the one hand and perfect health on the other. The ratings correspond to the individual's expressed preference for the alternative health states. The standard gamble approach 13-16 uses hypothetical lotteries as a means of measuring people's preferences. These lotteries involve a choice between two alternatives: the certainty of one outcome and a gamble with two possible outcomes. For example an individual might be asked to choose between the certainty of surviving for a fixed period in a particular state of ill health and the gamble between surviving for the same fixed period without disability on the one hand and immediate death on the other. The probability of surviving without disability, as opposed to dying, is then varied until the person shows no preference between the certain option and the gamble. This probability then defines the preference of an individual for the disabled state on a scale between nought and one, whose end-points are death and perfect health. The time trade-off approach 17-21 involves asking subjects to consider the relative amounts of time they would be willing to trade in order to survive in various health states. The choice may lie between continuing in a present defined state of ill health or moving to a shorter but healthier life. The duration of survival in the healthier state is varied until the subject is indifferent between the two alternatives, at which point his/her utility for the health state can be calculated. The equivalence technique. Preferences for alternative health states are measured by asking subjects to relate groups of individuals in these states. Subjects may be asked how many sick people in one health state are equivalent in total health status to a given number of perfectly healthy people. In this way, all health states can be related to each other in terms of preference. Ratio scaling, 12 which is very similar to the equivalence method, provides a ratio comparison of health states. Here subjects are asked to provide a value that represents the ratio of the desirability of each health state to an arbitrarily chosen one. The reference state is assigned a random value from which the preference values of the other health states can be calculated. In this way the preference values for all the health states can be represented on a scale.
330
S. Petrou and A. Renton
These measurement techniques are often used in combination and are the most reliable methods of measuring health-related quality of life associated with different health states. However, they can be time-consuming and people often have difficulty understanding the concepts underlying them.
Health status indices These indices or scales are essentially weighting schemes which assign weights to each definable health status, based on the premise that health-related quality of life is a complex, multidimensional phenomenon. Researchers themselves precisely predetermine what they consider to be the relative importance of each constituent component of disability and discomfort and so construct overall indicators of health-related quality of life. A number of these health status indices have been developed. 22-27 However, they tend to be disease-specific and generate intermediate output effectiveness data which cannot easily be converted into QALYs. In addition, they often lack sensitivity in picking up small improvements in the health-related quality of life associated with some treatments. Moreover, as Culyer2s has argued, they tend to incorporate value judgements which appear not to be identically shared by individuals in the community. Existing values available from the medical literature A fourth approach to measuring health-related quality of life is to use existing utility values available in the literature. This approach is relatively quick, straightforward and inexpensive though it is not as reliable as the measurement approach since it is difficult to ensure that the health states, subjects and measurement instruments used are representative. A number of studies in the medical literature report utility values. The study by Kaplan et al. ,29 for example, reports utility values for 36 different health states, ranging from no symptoms or problems to a loss of consciousness such as seizures, fainting or comas.
Calculating Health Gain in QALYs QALYs are units of health gain which combine the survival period and health-related quality of life during the period survived. Quality of life is usually measured on a scale from nought (death) to one (full health). Once the change in both life expectancy (e) and health-related quality of life (q) derived from an intervention have been estimated, the QALY gain is calculated simply as: QALY gain = e × q
Discounting future benefits In performing cost-utility analyses, economists generally operate on the assumption that society and individuals value current utility higher than future utility, and find current costs more painful to bear than future costs. Future benefits and costs are therefore 'discounted', usually at a fixed rate per annum. In the light of this, there has recently been some debate over whether and at what rate QALYs, as measures of utility in health gain, should be discounted. The issue is particularly pertinent since
The Q A L Y
331
QALYs entail measurement of survival which must of necessity pertain to the future. Whichever discount rate is selected, a value judgement is in fact being made on one's preference for current over future consumption. As the discount rate is increased, progressively less weight is given to future benefits. Thus a high discount rate will favour an intervention which substantially enhances health-related quality of life but has little effect on survival, over one which enhances survival but has little effect on health-related quality of life, even if the Q A L Y gains for the two interventions are identical. Drummond et al. 3° argue that failing to discount future health effects, whilst at the same time discounting future health costs, could lead to inconsistencies and impossible conclusions in resource allocation. The main argument against discounting future health effects is that it is not clear that people value future health any less than their present health. Indeed, concern for fitness and general health, and the willingness of individuals to spend large sums of money on health care, are taken as evidence that human beings are naturally concerned about their future health. Cullis and West 31 argue that, as a result, a zero discount rate or even a negative rate might be appropriate. This line of thinking has recently found support in the Department of Health which has proposed that a zero discount rate should be applied to future health benefits. 32 We have no firm opinion about which discount rate should be applied, but believe that the recent debate on the subject makes it imperative that sensitivity analyses are performed on the rates selected to discount future costs and benefits of health interventions. This will be particularly important when cost per Q A L Y league tables are presented to health authorities and other bodies with exact cost utility estimates for interventions.
Cost p e r Q A L Y
Often the Q A L Y health gain derived from interventions is combined with the cost of delivering those interventions in a cost utility measure. Because D H A s will already be providing many of the services, this is usually expressed as the 'marginal cost per QALY'. This is the marginal cost of increasing provision of an intervention to a level that secures one extra QALY.
Using QALYs A QALY-based system would work by allocating health care resources according to the relative positions of interventions in a cost per Q A L Y league table. Such a league table would list all interventions according to their relative cost effectiveness, with the intervention providing the greatest benefits (in terms of QALYs) per unit of resource placed at the top of the table. Theoretically, the most cost-effective intervention would be supplied first. The authority or governing body would then proceed to work down the league table and place decreasing priority on the less cost-effective interventions, until the health budget is exhausted. Those interventions at the bottom of the league table would fare badly, but would not necessarily be overlooked. Stevenson 2 argues that the gaps would be filled by private insurance, voluntary agencies and special government funds. An example of a cost per Q A L Y league table is provided by Culyer 33 (Table I).
332 Table I
S. Petrou and A . Renton Cost per QALY league table for selected health care interventions (1983-84 prices)
Intervention
Present value of extra cost per QALY gained (£)
Benign intracranial tumours Subarachnoid haemorrhage Pacemaker implantation for heart block Hip replacement CABG for severe angina LMD GP control of total serum cholesterol CABG for severe angina with 2VD Kidney transplantation (cadaver) Breast cancer screening Heart transplantation Metastatic tumours in central nervous system CABG for mild angina 2VD Hospital haemodialysis Malignant brain turnout
240 310 700 750 1040 1700 2280 3000 3500 5000 11000 12600 14000 69000
Source: Culyer33 Problems with a QALY-based Approach to Resource Allocation
The many problems encountered by practising public health physicians in using a cost utility approach based on QALYs to inform the development of targets and priorities for purchasing may be summarised under three main headings. Inadequate information There is a well-recognised paucity of adequate data on the effects of health interventions on patient survival and health-related quality of life for many diseases. In many cases this information will also prove technically difficult and expensive to generate in the future. This means that cost per Q A L Y estimates may be available for only a small proportion of the range of service options competing for priority. Ethical problems Quite apart from the information problems, the use of cost utility analysis in general, and a QALY-based system in particular, have raised substantial ethical objections. Whether doctors can ever be justified in considering financial costs in making clinical decisions has been questioned. More specific concerns have not been restricted to the utility side of the equation but, as Drummond et al. 3° have suggested, also involve important questions as to whether and which indirect costs should be included in the analysis. On the utility side Harris, 34 for example, raises two ethical concerns. First, QALY-based systems will tend to discriminate against elderly people and those with shorter life expectancies because greater Q A L Y benefits can be obtained by treating younger people and people with longer life expectancies. Second, QALYs tend to be inflexible in tending to place low values on health states with a high degree of discomfort and distress, and do not cater for the tendency of human beings both to cope with their compromised health states and to place greater value on them as time passes.
The QAL Y
333
Methodological problems Cost per QALY league tables which have been constructed to compare the relative cost effectiveness of various medical interventions have based their results on studies which were performed at different times and in different places. Gudex ~5 converted information from the British and overseas medical literature, which covered a 17-year time-span (1968-85) into one league table. This hides considerable dangers because the benefits of medical interventions are likely to improve over time. Similarly, the relative costs of interventions are likely to vary substantially over time as a result of both technical improvements and economies of scale. Benefits and costs may also be expected to vary between nations and between regions within nations, implying that the results of studies not performed in the same location may be affected by extraneous factors. A Check-list for Public Health Physicians The above considerations suggest a series of questions which need to be addressed before QALYs are used to establish priorities in resource allocation. 1. Is a cost utility type analysis appropriate? 2. Do I have information on enough interventions to make the approach viable? 3. Were QALYs and costs for each intervention measured in comparable places at comparable times? 4. Has the health-related quality of life been measured appropriately? 5. Are there factors specific to my district which may make the published estimates inapplicable? 6. Are there demographic or other features of the population concerned which require special consideration or make ethical and equity issues especially sharp? Conclusion Whatever shape tomorrow's health service takes, the requirement to make equitable and efficient use of resources through a rational and explicit resource allocation policy will remain. The QALY concept provides a standard unit for measuring health gain across diseases and specialties which enables a systematic approach t o be taken to maximising efficiency while making more explicit the choices and priorities which this entails. Even if all the problems of validity and data availability can be overcome, decision makers will still face a number of other problems when they consider the final cost utility estimates. If cost per QALY estimates are to be used as a guide to resource allocation, then a continuous updating will be necessary to allow for technological advances. Some sort of check in the system may be necessary to prevent discrimination against new and innovative interventions and programmes which are likely to improve their relative cost effectiveness at a later stage of development. Costing information must be adjusted to allow for local variation and economies of scale. The practical and ethical appropriateness of a cost utility approach to particular decisions should always be considered. For the present, the use of QALYs in cost utility analysis should be viewed with a degree of caution, and we have attempted to provide a structure within which that caution might be exercised. We believe that in its current state of development the
334
S. Petrou and A. Renton
Q A L Y concept provides a useful tool in aiding decision making in the allocation of health care resources, and highlights the need for the measurement in commensurable units of the outcomes of interventions for different diseases. H o w e v e r , it is too early for a universal application of the approach. Q A L Y s measured in a variety of different ways, in different places and at different times, aggregated into league tables, are of questionable value. Different methods of measuring health-related quality of life yield different results. If the Q A L Y concept is to be applied on a wide scale, it will be necessary to reach a consensus on which measurement technique or combination is the most accurate and most appropriate for the exercise, and which costs are to be included in calculations. D r u m m o n d 36 considered that a highly summarized presentation of data suggesting quick and easy solutions is dangerous. Decision makers will need to b e c o m e m o r e involved in the data collection process, so that they can become more aware of the vulnerability of the final analyses. Cost per Q A L Y figures are best estimates of averages. Their presentation, like that of any statistical estimate, should include some information about their precision and validity, and the variance of the underlying distribution. If public health physicians believe that cost utility analyses are inappropriate to particular situations, then other economic techniques such as cost-effectiveness analysis, cost benefit analysis, option appraisal and p r o g r a m m e budgeting may be useful in addressing the key issue of confronting resource scarcity in an explicit m a n n e r by comparing the costs and benefits at the margin of alternative health care programmes.
Acknowledgements We are grateful to Dr Andrew Stevens and Professor David Miller for their comments on this paper.
References 1. Government White Paper (1989). Working for Patients. London: HMSO. 2. Stevenson, R. (1991). The Oregon formula: health economists' dream or Stalinist nightmare? Archives of Disease in Childhood, 66,990-993. 3. Dixon, J. & Welch, H. G. (1991). Priority setting: lessons from Oregon. Lancet, 337, 891-894. 4. NHS Management Executive (1991). Purchasing Intelligence. London: NHS Mangement Executive. 5. Stason, W. B. & Weinstein, M. C. (1977). Allocation of resources to manage hypertension. New England Journal of Medicine, 296, 732-739. 6. Williams, A. (1985). Economics of coronary artery bypass grafting. British Medical Journal, 291,326-329. 7. Torrance, G. W. & Zipursky, A. (1984). Cost-effectiveness of ante-partum prevention of Rh immunization. Clinics in Perinatology, 11,267-281. 8. Pauker, S. G. (1976). Coronary artery surgery: the use of decision analysis. Annals of Internal Medicine, 85, 8-18. 9. Pauker, S. P. & Pauker S. G. (1977). Prenatal diagnosis: a directive approach to genetic counselling using decision analysis. Yale Journal of Biology and Medicine, 50, 275-289. 10. Weinstein, M. C. (1980). Estrogen use in postmenopausal women--costs, risks and benefits. New England Journal of Medicine, 303,308-316.
The Q A L Y
335
11. Weinstein, M. C. (1986). Risky choices in medical decision making: a survey. The Geneva Papers on Risk and Insurance, 11,197-216. 12. Rosser, R. M. & Kind, P. (1978). A scale of valuations of states of illness: is there a social consensus? International Journal of Epidemiology, 7,347-358. 13. McNeil, B. J., Weichselbaum, R. & Pauker, S. G. (1978). Fallacy of the five-year survival in lung cancer. New England Journal of Medicine, 299, 1397-1401. 14. Eraker, S. A. & Sox, H. C. (1981). Assessment of patients' preferences for therapeutic outcomes. Medical Decision Making, 1, 29-39. 15. Breyer, F. & Fuchs, V. R. (1982). Risk Attitudes in Health: An Exploratory Study, Working Paper No. 875. Cambridge, MA: NBER. 16. Elstein, A. S., Holzman, G. B., Ravitch, M. M., Metheny, W. A., Holmes, M. M., Hoppe, R. B., Rothert, M. L. & Rovner, D. R. (1986). Comparison of physicians decisions regarding estrogen replacement therapy for menopausal women and decisions derived from a decision analytic model. The American Journal of Medicine, 80,246-258. 17. Pliskin, J. S., Shepard, D. S. & Weinstein, M. C. (1980). Utility functions for life years and health status. Operations Research, 28,206-224. 18. McNeil, B. J., Weichselbaum, R. & Pauker, S. G. (1981). Tradeoffs between quality and quantity of life in laryngeal cancer. New England Journal of Medicine, 305 (17), 982-987. 19. Torrance, G. W., Boyle, M. H. & Horwood, S. P. (1982). Application of multi-attribute utility theory to measure social preferences for health states. Operations Research, 30 (6), 1043-1069. 20. Weinstein, M. C. & Stason, W. B. (1982). Cost-effectiveness of coronary artery bypass surgery. Circulation, 66 (III), 56-66. 21. Buxton, M., Ashby, J. & O'Hanlon, M. (1986). Valuation of Health States using the Time Trade-off Approach: Report of a Pilot Study Relating to Health States One Year after Treatment for Breast Cancer, Discussion Paper No. 2. Brunel University: HERG. 22. Karnofsky, D. A., Abelmann, W. H., Craver, L. F. & Burchenal, J. H. (1948). The use of the nitrogen mustards in the palliative treatment of carcinoma, with particular reference to bronchogenic carcinoma. Cancer, 634-656. 23. Grogono, A. W. & Woodgate, D. J. (1971). Index for measuring health. Lancet, 2, 1024-1026. 24. Izsak, F. C. & Medalie, J. H. (1971). Comprehensive follow-up of carcinoma patients. Journal of Chronic Diseases, 24, 179-191. 25. Bergner, M., Bobbitt, R. A., Kressel, S., Pollard, W. E., Gilson, B. S. & Morris, J. R. (1976). The Sickness Impact Profile: conceptual formulation and methodology for the development of a health status measure. International Journal of Health Services, 6, 393-415. 26. Katz, S. & Akpom, C. A. (1976). A measure of primary sociobiological functions. International Journal of Health Services, 6, 493-507. 27. Spitzer, W. O., Dobson, A. J., Hall, J., Chesterman, E., Levi, J., Shepherd, R., Battista, R. N. & Catchlove, B. R. (1981). Measuring the quality of life of cancer patients: a concise QL-Index for use by physicians. Journal of Chronic Diseases, 34, 585-597. 28. Culyer, A. J. (1978). Measuring Health: Lessons for Ontario. Toronto, Canada: Ontario Economic Council. 29. Kaplan, R. M., Bush, J. W. & Berry, C. C. (1976). Health status. Types of validity and the index of well-being. Health Services Research, 11 (4), 478-507. 30. Drummond, M. F., Stoddard, G. L. & Torrance, G. W. (1987). Methods for the Economic Evaluation of Health Care Programmes. Oxford: Oxford University Press, 81-82. 31. Cullis, J. G. & West, P. A. (1979). The Economics of Health: An Introduction. London: Martin Robertson, 191-192. 32. Parsonage, M. & Neuburger, H. (1992). Discounting and health benefits. Health Economics, 1, 71-76. 33. Culyer, A. J. (1991). The promise of a reformed NHS: an economist's angle. British Medical Journal, 302, 1253-1256.
336
S. Petrou and A. Renton
34. Harris, J. (1991). Unprincipled QALYs: a response to Cubbon. Journal of Medical Ethics, 17,185-188. 35. Gudex, C. (1986). QALY's and Their Use by the Health Service, Discussion Paper 20. York: Centre for Health Economics, University of York. 36. Drummond, M. F. (1989). Output measurement for resource allocation decisions in health care. Oxford Review of Economic Policy, 5 (1), 59-74.
O The Society of Public Health, 1993
The QALY: A Guide for the Public Health Physician S. Petrou, BSc PhD 1 and A. Renton, MBBS MFPHM 2
7Research Fellow in Health Economics, 2Senior Lecturer in Public Health Medicine, Academic Department of Public Health, St Mary's Hospital Medical School, Norfolk Place, London W2 1PG
This paper considers the background to the rational allocation of health resources at DHA level within the context of the new purchaser/provider regime. A popular approach to this problem has been that of cost utility analysis, using quality adjusted life years (QALYs) as the measure of the benefit derived from a health intervention. This paper seeks to offer the practising public health physician an appraisal of the practical value and pitfalls of a QALY-based approach. To this end, the theoretical background to QALYs and their practical use are described and some of the practical and ethical problems and pitfalls are discussed. It it concluded that the QALY concept provides a useful tool in aiding decision making in the allocation of health care resources, but that it is too early for a universal application of the approach.
Introduction T h e National H e a l t h Service ( N H S ) review has defined the role of District H e a l t h Authorities ( D H A s ) as purchasers of health care. 1 U n d e r the new a r r a n g e m e n t s , D H A s are funded on the basis of the size and characteristics of their resident populations. Within these financial constraints they are charged with purchasing services to m e e t the health care needs of their residents. In response to these new d e m a n d s , a variety of agencies and professional groups have set out to elaborate the concepts and methodologies required for rational allocation of resources. Currently, health care resources in the N H S are allocated by a mixture of formal and informal devices. 2 It has b e e n argued, 3 however, that the new regime provides us with a unique o p p o r t u n i t y b o t h to place resource allocation on a rational footing, and to m a k e explicit choices and priorities which until recently have b e e n m a d e on an ad hoc or implicit basis. Cost utility analysis represents a standard economic technique for maximising allocative efficiency within health care provision. It allows all health interventions to b e c o m p a r e d in terms of their costs and the health i m p r o v e m e n t s they procure. T w o types of information are required to construct a cost utility analysis: the health gain accrued through interventions and the costs of those interventions. District-specific information of each of these types will be h a r d to c o m e by, although m u c h can be achieved through an integrated purchasing intelligence function. 4 H o w e v e r , with regard to the health gain derived f r o m interventions, there is a particular p r o b l e m : that of quantification. W h e r e the health o u t c o m e s of interventions have b e e n assessed they are often m e a s u r e d in qualitative terms, or as quantitative measures derived f r o m the specific clinical features of the health problem. While such disease-specific
Correspondence to: Dr Stavros Petrou, c/o Academic Department of Public Health, St Mary's Hospital Medical School, Imperial College of Science, Technology and Medicine, Norfolk Place, London W2 1PG.
328
S. Petrou and A. Renton
outcomes may be extremely useful in making comparisons between different interventions for the same disease, they will be of little value in the construction of a general health gain function for the District as a whole. There is, therefore, a requirement that we measure the health gain derived from an intervention in units which are independent of the particular disease or intervention considered. It is to provide this unit of health gain that the QALY has been developed. QALYs Quality adjusted life years or QALYs measure the health gains derived from health interventions in a single unit which combines increased survival and enhancement of health-related quality of life. They have been calculated for interventions for a number of illnesses, usually together with costs, to inform resource allocation decisions within specialities. Stason and Weinstein, 5 for example, used QALY data to determine how resources could be used most efficiently within programmes to treat hypertension. More recently, attempts have been made to use QALY data to compare the relative cost-utility of interventions between diseases. 6-7
Calculating QALYs Two quantities must be known in order to calculate the gain in QALYs which an individual with a specific health problem can expect to derive from a health intervention: the likely effect of the intervention on his/her survival, and its effect on his/her health-related quality of life.
Life years gained An estimate of the average number of life-years gained by individuals receiving an intervention would ideally be achieved through clinical trials of the intervention, incorporating long-term follow up of patients and controls. Such studies may however be extremely expensive and difficult to carry out. Consequently, many trials identify normalisation of clinical pathological or pathological indicators of disease rather than death as study end-points. Estimates of increased survival are therefore more generally available for interventions for diseases which are more rapidly fatal.
Quality o f life gained A number of approaches have been used to estimate the health-related quality of life of individuals with different diseases and the gains derived from interventions. The task is made complex by the need to make value judgements about the relative importance of the different dimensions of 'quality' and to devise scales to measure them, and by the problem of securing representative samples, comprising appropriate numbers and combinations of subjects. Many approaches have been adopted, and these fall broadly into four categories.
Decision analysis Groups of experts in the management of an illness (usually doctors) are asked to attribute health-related quality of life values to the different health states derived
The Q A L Y
329
from different interventions in a disease. 8-11 While this approach is quick and relatively inexpensive, professional and patient valuations have been shown to be at variance in some cases.12
Direct empirical measurement These approaches use various devices to measure preferences for different health states. These may be measured in general population samples or patients with specific health problems, and can be used in the aggregate to generate QALYs. Measurement techniques used include rating scales, the standard gamble approach, the time trade-off approach, the equivalence technique and ratio scaling. We will deal briefly with each of these. Rating scales are a relatively straightforward method of measuring health-related quality of life. Typically, individuals are offered several case descriptions covering a wide range of health states and are asked to assign values to each on the same scale whose end-points are usually defined as death on the one hand and perfect health on the other. The ratings correspond to the individual's expressed preference for the alternative health states. The standard gamble approach 13-16 uses hypothetical lotteries as a means of measuring people's preferences. These lotteries involve a choice between two alternatives: the certainty of one outcome and a gamble with two possible outcomes. For example an individual might be asked to choose between the certainty of surviving for a fixed period in a particular state of ill health and the gamble between surviving for the same fixed period without disability on the one hand and immediate death on the other. The probability of surviving without disability, as opposed to dying, is then varied until the person shows no preference between the certain option and the gamble. This probability then defines the preference of an individual for the disabled state on a scale between nought and one, whose end-points are death and perfect health. The time trade-off approach 17-21 involves asking subjects to consider the relative amounts of time they would be willing to trade in order to survive in various health states. The choice may lie between continuing in a present defined state of ill health or moving to a shorter but healthier life. The duration of survival in the healthier state is varied until the subject is indifferent between the two alternatives, at which point his/her utility for the health state can be calculated. The equivalence technique. Preferences for alternative health states are measured by asking subjects to relate groups of individuals in these states. Subjects may be asked how many sick people in one health state are equivalent in total health status to a given number of perfectly healthy people. In this way, all health states can be related to each other in terms of preference. Ratio scaling, 12 which is very similar to the equivalence method, provides a ratio comparison of health states. Here subjects are asked to provide a value that represents the ratio of the desirability of each health state to an arbitrarily chosen one. The reference state is assigned a random value from which the preference values of the other health states can be calculated. In this way the preference values for all the health states can be represented on a scale.
330
S. Petrou and A. Renton
These measurement techniques are often used in combination and are the most reliable methods of measuring health-related quality of life associated with different health states. However, they can be time-consuming and people often have difficulty understanding the concepts underlying them.
Health status indices These indices or scales are essentially weighting schemes which assign weights to each definable health status, based on the premise that health-related quality of life is a complex, multidimensional phenomenon. Researchers themselves precisely predetermine what they consider to be the relative importance of each constituent component of disability and discomfort and so construct overall indicators of health-related quality of life. A number of these health status indices have been developed. 22-27 However, they tend to be disease-specific and generate intermediate output effectiveness data which cannot easily be converted into QALYs. In addition, they often lack sensitivity in picking up small improvements in the health-related quality of life associated with some treatments. Moreover, as Culyer2s has argued, they tend to incorporate value judgements which appear not to be identically shared by individuals in the community. Existing values available from the medical literature A fourth approach to measuring health-related quality of life is to use existing utility values available in the literature. This approach is relatively quick, straightforward and inexpensive though it is not as reliable as the measurement approach since it is difficult to ensure that the health states, subjects and measurement instruments used are representative. A number of studies in the medical literature report utility values. The study by Kaplan et al. ,29 for example, reports utility values for 36 different health states, ranging from no symptoms or problems to a loss of consciousness such as seizures, fainting or comas.
Calculating Health Gain in QALYs QALYs are units of health gain which combine the survival period and health-related quality of life during the period survived. Quality of life is usually measured on a scale from nought (death) to one (full health). Once the change in both life expectancy (e) and health-related quality of life (q) derived from an intervention have been estimated, the QALY gain is calculated simply as: QALY gain = e × q
Discounting future benefits In performing cost-utility analyses, economists generally operate on the assumption that society and individuals value current utility higher than future utility, and find current costs more painful to bear than future costs. Future benefits and costs are therefore 'discounted', usually at a fixed rate per annum. In the light of this, there has recently been some debate over whether and at what rate QALYs, as measures of utility in health gain, should be discounted. The issue is particularly pertinent since
The Q A L Y
331
QALYs entail measurement of survival which must of necessity pertain to the future. Whichever discount rate is selected, a value judgement is in fact being made on one's preference for current over future consumption. As the discount rate is increased, progressively less weight is given to future benefits. Thus a high discount rate will favour an intervention which substantially enhances health-related quality of life but has little effect on survival, over one which enhances survival but has little effect on health-related quality of life, even if the Q A L Y gains for the two interventions are identical. Drummond et al. 3° argue that failing to discount future health effects, whilst at the same time discounting future health costs, could lead to inconsistencies and impossible conclusions in resource allocation. The main argument against discounting future health effects is that it is not clear that people value future health any less than their present health. Indeed, concern for fitness and general health, and the willingness of individuals to spend large sums of money on health care, are taken as evidence that human beings are naturally concerned about their future health. Cullis and West 31 argue that, as a result, a zero discount rate or even a negative rate might be appropriate. This line of thinking has recently found support in the Department of Health which has proposed that a zero discount rate should be applied to future health benefits. 32 We have no firm opinion about which discount rate should be applied, but believe that the recent debate on the subject makes it imperative that sensitivity analyses are performed on the rates selected to discount future costs and benefits of health interventions. This will be particularly important when cost per Q A L Y league tables are presented to health authorities and other bodies with exact cost utility estimates for interventions.
Cost p e r Q A L Y
Often the Q A L Y health gain derived from interventions is combined with the cost of delivering those interventions in a cost utility measure. Because D H A s will already be providing many of the services, this is usually expressed as the 'marginal cost per QALY'. This is the marginal cost of increasing provision of an intervention to a level that secures one extra QALY.
Using QALYs A QALY-based system would work by allocating health care resources according to the relative positions of interventions in a cost per Q A L Y league table. Such a league table would list all interventions according to their relative cost effectiveness, with the intervention providing the greatest benefits (in terms of QALYs) per unit of resource placed at the top of the table. Theoretically, the most cost-effective intervention would be supplied first. The authority or governing body would then proceed to work down the league table and place decreasing priority on the less cost-effective interventions, until the health budget is exhausted. Those interventions at the bottom of the league table would fare badly, but would not necessarily be overlooked. Stevenson 2 argues that the gaps would be filled by private insurance, voluntary agencies and special government funds. An example of a cost per Q A L Y league table is provided by Culyer 33 (Table I).
332 Table I
S. Petrou and A . Renton Cost per QALY league table for selected health care interventions (1983-84 prices)
Intervention
Present value of extra cost per QALY gained (£)
Benign intracranial tumours Subarachnoid haemorrhage Pacemaker implantation for heart block Hip replacement CABG for severe angina LMD GP control of total serum cholesterol CABG for severe angina with 2VD Kidney transplantation (cadaver) Breast cancer screening Heart transplantation Metastatic tumours in central nervous system CABG for mild angina 2VD Hospital haemodialysis Malignant brain turnout
240 310 700 750 1040 1700 2280 3000 3500 5000 11000 12600 14000 69000
Source: Culyer33 Problems with a QALY-based Approach to Resource Allocation
The many problems encountered by practising public health physicians in using a cost utility approach based on QALYs to inform the development of targets and priorities for purchasing may be summarised under three main headings. Inadequate information There is a well-recognised paucity of adequate data on the effects of health interventions on patient survival and health-related quality of life for many diseases. In many cases this information will also prove technically difficult and expensive to generate in the future. This means that cost per Q A L Y estimates may be available for only a small proportion of the range of service options competing for priority. Ethical problems Quite apart from the information problems, the use of cost utility analysis in general, and a QALY-based system in particular, have raised substantial ethical objections. Whether doctors can ever be justified in considering financial costs in making clinical decisions has been questioned. More specific concerns have not been restricted to the utility side of the equation but, as Drummond et al. 3° have suggested, also involve important questions as to whether and which indirect costs should be included in the analysis. On the utility side Harris, 34 for example, raises two ethical concerns. First, QALY-based systems will tend to discriminate against elderly people and those with shorter life expectancies because greater Q A L Y benefits can be obtained by treating younger people and people with longer life expectancies. Second, QALYs tend to be inflexible in tending to place low values on health states with a high degree of discomfort and distress, and do not cater for the tendency of human beings both to cope with their compromised health states and to place greater value on them as time passes.
The QAL Y
333
Methodological problems Cost per QALY league tables which have been constructed to compare the relative cost effectiveness of various medical interventions have based their results on studies which were performed at different times and in different places. Gudex ~5 converted information from the British and overseas medical literature, which covered a 17-year time-span (1968-85) into one league table. This hides considerable dangers because the benefits of medical interventions are likely to improve over time. Similarly, the relative costs of interventions are likely to vary substantially over time as a result of both technical improvements and economies of scale. Benefits and costs may also be expected to vary between nations and between regions within nations, implying that the results of studies not performed in the same location may be affected by extraneous factors. A Check-list for Public Health Physicians The above considerations suggest a series of questions which need to be addressed before QALYs are used to establish priorities in resource allocation. 1. Is a cost utility type analysis appropriate? 2. Do I have information on enough interventions to make the approach viable? 3. Were QALYs and costs for each intervention measured in comparable places at comparable times? 4. Has the health-related quality of life been measured appropriately? 5. Are there factors specific to my district which may make the published estimates inapplicable? 6. Are there demographic or other features of the population concerned which require special consideration or make ethical and equity issues especially sharp? Conclusion Whatever shape tomorrow's health service takes, the requirement to make equitable and efficient use of resources through a rational and explicit resource allocation policy will remain. The QALY concept provides a standard unit for measuring health gain across diseases and specialties which enables a systematic approach t o be taken to maximising efficiency while making more explicit the choices and priorities which this entails. Even if all the problems of validity and data availability can be overcome, decision makers will still face a number of other problems when they consider the final cost utility estimates. If cost per QALY estimates are to be used as a guide to resource allocation, then a continuous updating will be necessary to allow for technological advances. Some sort of check in the system may be necessary to prevent discrimination against new and innovative interventions and programmes which are likely to improve their relative cost effectiveness at a later stage of development. Costing information must be adjusted to allow for local variation and economies of scale. The practical and ethical appropriateness of a cost utility approach to particular decisions should always be considered. For the present, the use of QALYs in cost utility analysis should be viewed with a degree of caution, and we have attempted to provide a structure within which that caution might be exercised. We believe that in its current state of development the
334
S. Petrou and A. Renton
Q A L Y concept provides a useful tool in aiding decision making in the allocation of health care resources, and highlights the need for the measurement in commensurable units of the outcomes of interventions for different diseases. H o w e v e r , it is too early for a universal application of the approach. Q A L Y s measured in a variety of different ways, in different places and at different times, aggregated into league tables, are of questionable value. Different methods of measuring health-related quality of life yield different results. If the Q A L Y concept is to be applied on a wide scale, it will be necessary to reach a consensus on which measurement technique or combination is the most accurate and most appropriate for the exercise, and which costs are to be included in calculations. D r u m m o n d 36 considered that a highly summarized presentation of data suggesting quick and easy solutions is dangerous. Decision makers will need to b e c o m e m o r e involved in the data collection process, so that they can become more aware of the vulnerability of the final analyses. Cost per Q A L Y figures are best estimates of averages. Their presentation, like that of any statistical estimate, should include some information about their precision and validity, and the variance of the underlying distribution. If public health physicians believe that cost utility analyses are inappropriate to particular situations, then other economic techniques such as cost-effectiveness analysis, cost benefit analysis, option appraisal and p r o g r a m m e budgeting may be useful in addressing the key issue of confronting resource scarcity in an explicit m a n n e r by comparing the costs and benefits at the margin of alternative health care programmes.
Acknowledgements We are grateful to Dr Andrew Stevens and Professor David Miller for their comments on this paper.
References 1. Government White Paper (1989). Working for Patients. London: HMSO. 2. Stevenson, R. (1991). The Oregon formula: health economists' dream or Stalinist nightmare? Archives of Disease in Childhood, 66,990-993. 3. Dixon, J. & Welch, H. G. (1991). Priority setting: lessons from Oregon. Lancet, 337, 891-894. 4. NHS Management Executive (1991). Purchasing Intelligence. London: NHS Mangement Executive. 5. Stason, W. B. & Weinstein, M. C. (1977). Allocation of resources to manage hypertension. New England Journal of Medicine, 296, 732-739. 6. Williams, A. (1985). Economics of coronary artery bypass grafting. British Medical Journal, 291,326-329. 7. Torrance, G. W. & Zipursky, A. (1984). Cost-effectiveness of ante-partum prevention of Rh immunization. Clinics in Perinatology, 11,267-281. 8. Pauker, S. G. (1976). Coronary artery surgery: the use of decision analysis. Annals of Internal Medicine, 85, 8-18. 9. Pauker, S. P. & Pauker S. G. (1977). Prenatal diagnosis: a directive approach to genetic counselling using decision analysis. Yale Journal of Biology and Medicine, 50, 275-289. 10. Weinstein, M. C. (1980). Estrogen use in postmenopausal women--costs, risks and benefits. New England Journal of Medicine, 303,308-316.
The Q A L Y
335
11. Weinstein, M. C. (1986). Risky choices in medical decision making: a survey. The Geneva Papers on Risk and Insurance, 11,197-216. 12. Rosser, R. M. & Kind, P. (1978). A scale of valuations of states of illness: is there a social consensus? International Journal of Epidemiology, 7,347-358. 13. McNeil, B. J., Weichselbaum, R. & Pauker, S. G. (1978). Fallacy of the five-year survival in lung cancer. New England Journal of Medicine, 299, 1397-1401. 14. Eraker, S. A. & Sox, H. C. (1981). Assessment of patients' preferences for therapeutic outcomes. Medical Decision Making, 1, 29-39. 15. Breyer, F. & Fuchs, V. R. (1982). Risk Attitudes in Health: An Exploratory Study, Working Paper No. 875. Cambridge, MA: NBER. 16. Elstein, A. S., Holzman, G. B., Ravitch, M. M., Metheny, W. A., Holmes, M. M., Hoppe, R. B., Rothert, M. L. & Rovner, D. R. (1986). Comparison of physicians decisions regarding estrogen replacement therapy for menopausal women and decisions derived from a decision analytic model. The American Journal of Medicine, 80,246-258. 17. Pliskin, J. S., Shepard, D. S. & Weinstein, M. C. (1980). Utility functions for life years and health status. Operations Research, 28,206-224. 18. McNeil, B. J., Weichselbaum, R. & Pauker, S. G. (1981). Tradeoffs between quality and quantity of life in laryngeal cancer. New England Journal of Medicine, 305 (17), 982-987. 19. Torrance, G. W., Boyle, M. H. & Horwood, S. P. (1982). Application of multi-attribute utility theory to measure social preferences for health states. Operations Research, 30 (6), 1043-1069. 20. Weinstein, M. C. & Stason, W. B. (1982). Cost-effectiveness of coronary artery bypass surgery. Circulation, 66 (III), 56-66. 21. Buxton, M., Ashby, J. & O'Hanlon, M. (1986). Valuation of Health States using the Time Trade-off Approach: Report of a Pilot Study Relating to Health States One Year after Treatment for Breast Cancer, Discussion Paper No. 2. Brunel University: HERG. 22. Karnofsky, D. A., Abelmann, W. H., Craver, L. F. & Burchenal, J. H. (1948). The use of the nitrogen mustards in the palliative treatment of carcinoma, with particular reference to bronchogenic carcinoma. Cancer, 634-656. 23. Grogono, A. W. & Woodgate, D. J. (1971). Index for measuring health. Lancet, 2, 1024-1026. 24. Izsak, F. C. & Medalie, J. H. (1971). Comprehensive follow-up of carcinoma patients. Journal of Chronic Diseases, 24, 179-191. 25. Bergner, M., Bobbitt, R. A., Kressel, S., Pollard, W. E., Gilson, B. S. & Morris, J. R. (1976). The Sickness Impact Profile: conceptual formulation and methodology for the development of a health status measure. International Journal of Health Services, 6, 393-415. 26. Katz, S. & Akpom, C. A. (1976). A measure of primary sociobiological functions. International Journal of Health Services, 6, 493-507. 27. Spitzer, W. O., Dobson, A. J., Hall, J., Chesterman, E., Levi, J., Shepherd, R., Battista, R. N. & Catchlove, B. R. (1981). Measuring the quality of life of cancer patients: a concise QL-Index for use by physicians. Journal of Chronic Diseases, 34, 585-597. 28. Culyer, A. J. (1978). Measuring Health: Lessons for Ontario. Toronto, Canada: Ontario Economic Council. 29. Kaplan, R. M., Bush, J. W. & Berry, C. C. (1976). Health status. Types of validity and the index of well-being. Health Services Research, 11 (4), 478-507. 30. Drummond, M. F., Stoddard, G. L. & Torrance, G. W. (1987). Methods for the Economic Evaluation of Health Care Programmes. Oxford: Oxford University Press, 81-82. 31. Cullis, J. G. & West, P. A. (1979). The Economics of Health: An Introduction. London: Martin Robertson, 191-192. 32. Parsonage, M. & Neuburger, H. (1992). Discounting and health benefits. Health Economics, 1, 71-76. 33. Culyer, A. J. (1991). The promise of a reformed NHS: an economist's angle. British Medical Journal, 302, 1253-1256.
336
S. Petrou and A. Renton
34. Harris, J. (1991). Unprincipled QALYs: a response to Cubbon. Journal of Medical Ethics, 17,185-188. 35. Gudex, C. (1986). QALY's and Their Use by the Health Service, Discussion Paper 20. York: Centre for Health Economics, University of York. 36. Drummond, M. F. (1989). Output measurement for resource allocation decisions in health care. Oxford Review of Economic Policy, 5 (1), 59-74.