The limits to demand for health careCommentary: An open debate is not an admission of failureBMJ 2000; 321 doi: https://doi.org/10.1136/bmj.321.7252.40 (Published 01 July 2000) Cite this as: BMJ 2000;321:40
The limits to demand for health care
- Stephen Frankel, professor, epidemiology and public health medicine (, )
- Shah Ebrahim, professor, epidemiology of ageing,
- George Davey Smith, professor, clinical epidemiology
- Department of Social Medicine, University of Bristol, Bristol BS8 2PR
- 9 Countess Road, London NW5 2NS
- Correspondence to: S Frankel
- Accepted 8 June 2000
There are a number of historical, political, and economic explanations for the tradition of underinvestment in the NHS, but one insidious disincentive will provide a continuing brake on expansion. This is the expectation that in a publicly funded healthcare system any increase in expenditure will be followed only by further failure, although at an increased cost. Further, the imbalance between demand and supply is assumed to be deteriorating as the population ages, new technologies appear, and expectations rise.1 This orthodoxy has become so pronounced that questioning the linkage between particular aspects of failure in healthcare delivery and an overarching mismatch between supply and demand can be dismissed as an idealistic act of denial.
These problems are most commonly discussed in the idiom of rationing. In rationing the concern is to curtail access to health care through measures that may be, according to the perspective of the commentator, most likely to enhance effectiveness; be efficient, equitable, ethical, or beneficial to the independent sector; or be politically feasible. The rationing debate came to the fore during a period of recession, so that “choices” in health care involved choosing to offer less. The conventional, but implausible, insistence that the term rationing has a neutral meaning will be tested in the welcome new circumstance in which healthcare expenditure in the United Kingdom is set to expand substantially. But what are the grounds for the view that legitimate demand in most areas of provision must exceed realisable supply?
The rationing debate has been conducted almost exclusively through assertion and political analysis
Conventional assumptions of an imbalance between demand and supply are not supported by evidence
Pessimism about adverse future trends in demand arising from an ageing population, the costs of innovation, and rising public expectations are similarly unsupported by good evidence
Many perceived deficiencies in health care are attributable to issues other than overwhelming demand, such as the unwillingness of the public to accept the limits of effectiveness and the self interest of professionals
The proposition that the limits to demand lie within the capacity of a properly resourced NHS should be tested explicitly
Rationing the evidence
The rationing debate is profoundly unscientific. The problem it seeks to address is assumed rather than expressed in any refutable form, and the literature is dominated by assertion, political analysis, and ethical debate. In other areas of public health, assertions must be supported by data. For example, the case for a national concern with the health implications of obesity in childhood would not be accepted were it based on haphazard sightings of fat children; representative data describing body mass index would be required, drawn from samples adequate to support robust population estimates. In contrast the discussion of rationing is largely free of appropriate data. Any particular instance of failure is taken as evidence of a global shortfall in provision, which is the epidemiological equivalent of pointing at the occasional fat person as sufficient justification for an antiobesity programme.
In the clinical context the ideas of bias and representativeness are now accepted to the extent that selected observations are disparaged as anecdotal and thus not generalisable to the wider population. The epidemiological naivety that suffuses discussions of health policy can be illustrated in the context of current patterns of provision and in relation to trends in utilisation.
The longstanding failure to satisfy the demand for a number of effective and high volume procedures has been the benchmark of demand mismanagement in the NHS for many decades,2 but a more intimate examination may not support the common interpretation of a global mismatch between supply and demand. For example, in general surgery in England, about 110 000 people were waiting for day case treatment in December 1999, of whom some 21 000 had waited over six months. This is not, however, a general problem. In a few parts of the country there are serious shortfalls, but in most there are none (fig 1). If most centres satisfy demand successfully while a few do not, which group is representative: the 10% who account for 35% of the problem or the 10% who have no problem at all? This question is not asked, but the conventional, implicit interpretation is that centres that fail to satisfy demand are representative of a global pattern of failure. This is the equivalent of taking a few fat individuals as a proxy for a problem in the population.
When demand for health care is assumed to be increasing it is more common to express concern about this trend than to question it. In the United Kingdom high rates of bed occupancy lead to inevitable bed crises during periods of raised demand.3 These periodic problems are interpreted as expressions of inexorable increases in demand that arise from an ageing population. The reality is that emergency admission rates are certainly increasing in some areas4 while in others supposed increases are largely artefacts of re-counting those individuals who are moved between consultants within a hospital. 5 6 This is less an instance of a global and inexorable problem than an issue requiring more contextual inquiry. It is interesting to note that in the United Kingdom, population based measures of utilisation, such as the proportion of the population admitted to hospital each year, have remained remarkably constant for the past 15 years.7 The “increase” in total ordinary admissions in the Oxford area between 1988 and 1993 was −0.3%.6 In primary care, consultations increased by only 2.4%, from 33 961 to 34 785 per 10 000 person years at risk during the entire decade from 1981–2 to 1991-2, hardly an insurmountable challenge for the NHS.8
The epidemiology of indications
It is difficult to justify the conventionally unfavourable forecasts of the relation between demand and supply in terms of overall epidemiological trends, as these are broadly favourable. Among the main disease groups that underlie demand, the rates of coronary heart disease and stroke have been declining since the 1970s. Trends in cancer differ between younger and older people. Rates of cancer among men younger than 70 years have decreased but have increased among those who are older. For women an equivalent change occurs at age 50. As these cohort effects work through the population the overall outlook is therefore encouraging.9 This broad approach, however, is of little value since legitimate demand on the NHS reflects not so much incidence of disease as the incidence of indications for worthwhile forms of care.10
Examples of failure
In contrast to such a global approach, much can also be learnt from disaggregating areas of care when attempting to judge the relation between demand and supply. In most areas of provision demand is met—that is, all individuals at risk are offered care, for example in cases of childbirth, cancer, and major trauma. In others, such as counselling and cosmetic surgery, the responses of the NHS are unpredictable. The poles of the continuum from obligatory care to discretionary care are clear but the boundaries between them become increasingly arbitrary in the area where the marginal cases lie. Examples of the failure to meet demand are those conditions for which there are waiting lists, conditions for which relief is nominally an obligatory NHS service, although the obligation is often not fulfilled. Among the most notorious examples of these are two forms of elective surgery which are comparatively cheap and have remarkable benefits in transforming patients' lives: total hip replacement and cataract extraction.
The potential demand for total hip replacement cannot be estimated from conventional epidemiology because there is an unclear relation between pathology and the probability of benefiting from treatment. A study of the need for total hip replacement among the population in England was therefore conducted on a sample of 28 080 individuals.11 The overall requirement for primary replacement was 46 600 operations per year for patients who expressed a preference and were suitable for surgery. This compares with the actual level of provision in NHS and independent hospitals in England of some 43 500 primary total hip replacements, a trivial mismatch which would be annulled were each NHS orthopaedic hospital to perform a little over one more hip replacement each month. Therefore, the scale of potential demand does not explain the current disorder surrounding access to this surgery. It is noteworthy that in the United States levels of need for total hip replacement surgery are comparable to those in the United Kingdom, although they are not accompanied by underprovision.
Cataract surgery is an effective, safe, and comparatively cheap remedy for one of the commonest causes of incapacitating visual loss in older people. Lack of access to cataract surgery is expected in poorer countries, but it is unclear why such readily reversible morbidity burdens older people in wealthy countries like the United Kingdom. It is necessary to go beyond the conventional measures of pathology to incorporate those additional features that inform decisions to treat, such as the impact of the pathology on the individual's quality of life and fitness for surgery. A study based on a random population sample of 2783 individuals in England generated a prevalence estimate of those requiring cataract extraction of some 325 000 people.12 About 600 ophthalmic surgical teams serve this population. If they were each to offer two surgical lists a week and perform five cataract extractions on each list (an accepted level of activity13), the capacity of the service would be some 300 000 cataract extractions per year. Demand satisfaction was achieved once in Brighton.14 There seem to be no real barriers to this being achieved for the whole country were current uncertainties of access to be replaced by booking systems informed by agreed criteria for surgery.
Such treatments have been rationed by a willingness to wait or a willingness to pay, and past failures to satisfy demand have often been used as justification for yet more energetic rationing. Disaggregating areas of care in which demand has patently not been met suggests that even these examples of failure could be transferred, with little change in current treatment rates, from the discretionary areas of NHS care to the obligatory areas. The fundamental flaw belying the health economists' hegemony in this debate is the belief that a correct assumption at the macro level—that choices must be made in the use of scarce resources—can then transposed unquestioningly to the micro level, where it may or may not apply. If the assumptions of an inevitable mismatch between supply and demand have little epidemiological basis in the sentinel conditions considered here, it is worth questioning conventional assumptions in other areas of care.
Changing patterns of demand and supply
The question of whether we can catch up with today's problems may seem irrelevant when the current adverse balance is seen as deteriorating under the burden of the three commonly cited adverse pressures: increasing life expectancy, new technologies, and raised public expectations.1 What is the justification for portraying this triad in such pessimistic terms when each could otherwise be seen as a hallmark of desirable progress?
Doomsday scenarios and demographic time bombs are popular with policymakers; they do not seem to be inhibited by the fact that crises predicted earlier have not materialised. The number of older people is increasing and will continue to increase if life expectancy continues to rise; it is important to recognise that the relation between ageing and healthcare costs is, in important respects, one of association rather than causation. The costs of health care relate strongly to the process of dying. Time spent in hospital increases somewhat with increasing age, but even in elderly people the time spent in hospital is generally confined to the time before death.15 If people die later the costs of health care will fall later, but this is the cost of dying, not of ageing. Cross sectional data are particularly misleading in this regard since as people get older it becomes more likely that they are in their final years, a time during which healthcare costs will be concentrated. In the United States there is no evidence that older people make greater demands on acute care, and longitudinal data suggest that healthcare expenditure does not depend on age once remaining lifetime is controlled for.16 Furthermore, the period between the onset of illness and death may become shorter as prosperity rises, resulting in the compression of morbidity; indeed the proportion of those in the United Kingdom who required help with activities of daily living was halved between 1976 and 1991.17 Recent projections show that taking account of the increased fitness of successive cohorts of older people in the United Kingdom reduces the population burden of disability by about threefold, from 3.5 million to 1 million affected by 2051.18
Longevity is one of the prizes of prosperity, so it is contrary to portray it only as a burden on society. An ageing population will clearly have implications for health care, but the extent of these implications will only emerge empirically over time. The common view that a wealthy country such as the United Kingdom must necessarily have difficulty caring for people at the very age when they are most likely to need such care cannot be accepted by default.
The cost of innovation
In most fields innovation is welcomed, but in health care new technologies are commonly presented as a destabilising threat. The assumption that new technology must lead to substantial increases in cost that probably cannot be met is neither subject to sufficient questioning nor supported by empirical evidence. An analysis of new medical technology introduced during the 1970s would not support the assumption that new technologies increase costs,19 and it is unclear whether new technology was the cause or an effect of increased spending during that period.20
One problem with expensive new technologies is that the hopes for them are so often belied by the reality of modest clinical effectiveness. Some innovations will incur greater costs while others will result in savings: the issue can be judged only in relation to specific innovations and specific problems. A study of the changing patterns in diagnosing and treating peptic ulcers found that new technology had cut costs.21 The same may or may not be the case for other conditions. For example, at today's prices the costs of late mobilisation after myocardial infarction as practised in the 1950s may well prove comparable to those of modern intensive interventions and short hospital stays. Drugs such as statins, deemed too expensive for widespread use, reduce the number of coronary revascularisations, thus resulting in cost savings but in a different healthcare sector.22
The recent political acknowledgement that an open ended commitment was unsustainable for the NHS was prompted by the introduction of a number of expensive drugs which were either of marginal benefit, such as the influenza drug zanamivir (Relenza), or marginal to conventional clinical practice, such as sildenafil (Viagra). These discussions tend to present the rising cost of drugs as an inevitable constraint. This is misleading: prices are the outcome of complex considerations of commercial judgment and wider political interest in a key industry. For example, assumptions that statins must be rationed and made available only to treat those who have an arbitrarily high level of risk to reduce mortality from coronary heart disease23 are belied by the fact that in India statins cost about one seventh of the cost paid by the NHS.24 Equivalent pricing in the United Kingdom would make the cost effectiveness of these drugs comparable to that of other secondary preventive measures. At the very least, there is clearly some discretion in the costs of such innovations.
The assertion that the demands of a better informed population must increase more rapidly than the health system's capacity to satisfy them is the third support of pessimism. When so little is known of current expectations the empirical basis for such future projections must be weak indeed. When conditions are not life threatening or when treatments offer only equivocal benefits, the true expert, and the key influence on management, should be the individual who is ill. There can be a striking mismatch between what is offered and the treatments that individuals prefer. Non-compliance with treatment occurs consistently, but the interpretation of this is shifting. Instead of seeing the 30-50% of people who do not take prescribed medicines as failing to comply, their actions are increasingly being seen as rational decisions made by individuals who might understand their problems better than the practitioner.25 There is strong evidence that a better informed public is as likely to reject treatments as to demand more of them. 26 27
Finally, most people want to die at home,28 but there has been an increasing trend for people to die in institutions.29 The simplistic assumption that “increased public expectations” equates with overwhelming demand must be tempered by evidence that other public expectations can also increase, particularly the expectation of being protected from treatments for which, from the point of view of the individual, the risks are likely to outweigh the benefits. The balance between these conflicting implications cannot be predicted, although it may be open to influence if the public is guided towards safety rather than false expectations.
The social construction of pessimism
The scale of legitimate demand will therefore be moderated over time by this range of factors. If an informed prediction of the balance of the outcome is so uncertain, why does the pessimistic interpretation predominate? One answer is that healthcare experiences are constantly being examined informally,30 are given continuous media attention, and provide the seasoning to much political debate. This discourse reflects a number of moral and political undercurrents.
Modern health systems resolve or ameliorate many problems with illnesses that until recently were insoluble and they may provide reassurance to those who are well, but their predominant concern is with a tide of pain, fear, indignity, and death. A happy health service is therefore something of an oxymoron. Many of the problems that prompt debate about rationing have more to do with an unwillingness to acknowledge that some disease processes are inexorable rather than reflecting an inadequate health service. The child B case, in which a health authority denied further treatment to a child with non-Hodgkin's lymphoma, was a public example of this,31 but there are numerous such cases in which informed comment is restrained by sensitivity to the feelings of ill people and their families or to the bereaved. Such a denial of mortality and of the irreversibility of much morbidity is accentuated by prosperity, and this may underlie the otherwise odd association between the wealth of a country and its preoccupation with rationing.
There is also a strand of professional self interest in the common focus on failure. The NHS is one of the largest employers in the United Kingdom. Evidence of unsatisfied demand is not necessarily disinterested. Bidding for extra resources is a competitive process for which ordered coping is rarely rewarded. Surgical waiting lists still offer the potential for abuse that is inherent in an arrangement which offers financial incentives for selectively low productivity. The NHS still draws on an impressive fund of altruism, but it would be naive to assert that the interests of staff always coincide with those of the public. These undercurrents to the flow of information make a balanced assessment of the system's capacity to meet demand less likely. It can be difficult to understand why the NHS should be seen as drowning when the pool is so shallow (fig 2). In Ophelia's case the explanation is that
her garments, heavy with their drink,
Pull'd the poor wretch from her melodious lay
To muddy death.
For the NHS this social construction of pessimism adds unacknowledged weight to the demand side of the scales where the possible future influences of age, technology, and expectations are judged.
Finite demand, infinite supply?
Even in its underfunded state, the capacity and productivity of the NHS are prodigious. In England over 11 million hospital episodes are managed each year. Within this capacity there is considerable discretion. For example, up to one quarter of admissions cannot be justified on clinical grounds, implying flexibility for reducing any apparent mismatch between supply and demand. 6 32
The demand for health care must be finite: the population is finite and only a proportion of the population can benefit from and want treatment.33 Care is offered at all levels within society, from the most specialist provision to the most informal support within the neighbourhood or family. Slogans about healthcare capacity should have no place in serious analyses, but as there is a tendency to oversimplify these issues it may be wiser to encompass this informal sector, invert conventional assumptions, and think in terms of finite demand and infinite supply.34 The working hypothesis should be that selected services can be expanded in the expectation that legitimate demand will be satisfied. This is a proposition that can be tested empirically. The establishment of programmes to test whether demand will be satisfied has been inhibited by the wrong assumptions that are conveyed by conventional metaphors such as the “iceberg of morbidity.” In many areas of the NHS's failure to meet demand this metaphor may be strained by the fact that key NHS icebergs are carrying all, or almost all, of their bulk above the surface. Often we are dealing with an iceflow of morbidity that can be as readily melted as allowed to deepen; but in a world dominated by the language of rationing, where every solution has a problem, it can be difficult to attract attention to the fact that much of the failure to meet demand is unnecessary. Instead the rationing gaze wanders restlessly towards other deficiencies or takes refuge within the safety of “dilution,”35 although the fact that some aspects of care could be offered more agreeably is a platitude.
Misunderstanding, vested interests, and parsimony are greater problems than the potential level of demand. The conventional null, or nihilist, hypothesis that demand always exceeds supply within a public health system reflects neither hope nor experience. The proposed expansion in investment in the NHS, including the targeted use of resources to address the politically serious issue of waiting lists,36 provides an opportunity to establish whether, after over 50 years of equivocation, it is possible to counter the professional37 as well as the intellectual barriers to satisfying demand. This programme must be ordered to allow us to judge the merits of the alternative hypothesis: that the limits to demand for key categories of health care lie within the capacity of a properly resourced NHS.
The Department of Social Medicine of the University of Bristol is the main centre of the MRC Health Services Research Collaboration.
Competing interests None declared.
Commentary: An open debate is not an admission of failure
- Bill New, independent health policy analyst ()
- Department of Social Medicine, University of Bristol, Bristol BS8 2PR
- 9 Countess Road, London NW5 2NS
The paper by Frankel and colleagues makes the point that there is much evidence to suggest that the demand for healthcare can be managed successfully within the NHS and that, in this regard, many commentators have been overly pessimistic about the future. However, the question is whether one should go further and claim that a debate about rationing is unwarranted and contributes to pessimism and defeatism within the health service. I want to suggest that implicit in the authors' own case is an acknowledgement that aggregate demand will always exceed the ability of resources to satisfy it and that, therefore, some kind of rationing is inevitable—that is, benefits will have to be denied to people when scarce resources are allocated. Admitting to something that is inevitable is not pessimistic or an admission of failure but is the only politically sustainable position in an increasingly open and well informed democracy.
Demand is problematic
In the first place the concept of demand is rather more problematic than the authors allow. For goods and services rationed by price, demand is “satisfied” because the price rises until no additional people come to the market. Similarly, in a system that is not rationed by price, demand can be “satisfied” by a number of other techniques including, to use some of the authors' words, setting “agreed criteria,” estimating “requirements,” or establishing “legitimate demand.” There may, however, still be unsatisfied demands for which benefits could be provided.
Discussing legitimate demand introduces the need to make a judgment about costs and benefits—that is, a decision about when the provision of a wanted thing no longer warrants the cost to others. This is the language of rationing. Unless there is unanimity about what constitutes legitimate demand, then agreeing what this level of demand should be will involve denying some people things that they believe will benefit them. These denials should not be hidden behind technical language but defended openly. It seems to me that those who use the NHS will settle for nothing less.
Secondly, the authors criticise the rationing debate for being selective: “any particular instance of failure is taken as evidence of a global shortfall.” This is certainly not a valid inference, although I would point out again that rationing does not necessarily imply a failure or even a shortfall. But neither does the reverse apply: instances of (potential) success do not imply the possibility of a global avoidance of rationing. The examples of hip replacement and cataract surgery are used as a means of demonstrating “the flaws belying the health economists' hegemony,” but the argument falls foul of the authors' own criticisms.
To borrow their analogy, choosing these two cases is a bit like pointing out that there are a number of thin people (or people who could be made thin if we concentrated resources on them) and then using this fact as a means of showing that the whole population could be thin. This does not follow, and it is certainly no more scientific a means of using evidence than indicating that there has never been a time when all demands have been satisfied in the NHS. However, the authors go on to accept that “at the macro level, choices must be made in the use of scarce resources.” This is confusing: why should admitting that rationing occurs at the macro level, rather than at the micro level, be acceptable? Furthermore, surely it must be acknowledged that global rationing directly implies that micro rationing occurs somewhere in the NHS? In fact, on closer examination Frankel and colleagues seem to be arguing that rationing is currently done rather badly, not that it should not happen at all. In this case the need for a debate about rationing seems all the more important.
The third point relates to the authors' narrow understanding of what rationing means, as revealed by the phrase: “The demand for care must be finite: the population is finite.” This suggests that each individual is limited to a discrete package of demand and that the degree to which their needs should be satisfied is not problematic. In fact the concept of rationing by dilution is crucial. It refers to the possibility that the degree, intensity, or quality of an intervention can be varied endlessly so as to marginally change the overall level of health improvement gained by a patient. It is often a matter of probabilities: just how small of a risk of an adverse incident do we wish to achieve? Adding to these small but real benefits will inevitably bring health systems up against resource constraints even in countries that spend much more on health care than the United Kingdom. The fact that it is sensible to avoid trying to exhaust these benefits should not lead us to pretend that they do not exist. (I am unclear why this type of rationing is considered to refer only to the possibility of offering care “more agreeably.”)
A matter of belief?
Any discussion about rationing must be unscientific in one sense because it is impossible to test whether a society can ever have all its desires satisfied, in health or anything else. Nevertheless, many of those who argue that rationing is inevitable would support much of this paper; it produces an excellent catalogue of evidence opposing the view that the NHS is incapable of coping and it also complements less epidemiologically based work that makes the same case.1 But surely all experience of health care over time, not to mention the experience of economic exchange in general, is that new wants are continually developing and that human ingenuity will continue to strive to meet, but never exhaust, them. Perhaps this comes down to a matter of belief, but for those who support an open debate about rationing this is not an admission of failure but an attempt to sustain the NHS by being brutally honest about its inevitable limitations as well as its considerable capabilities.
Competing interests None declared.