Click here for the PDF version

Contents

All systems of health care are under pressure financially, fuelled by rising public expectations for good health, an ageing population, and the advent of expensive but effective health technologies [1]. The particular point on the cost-effectiveness continuum at which the pressure is felt most keenly – the point where demand just outstrips the level of provision – differs between countries, and between health-care systems within countries. Cochlear implantation (CI) rests close to that point in the publicly funded parts of the health-care systems in the United Kingdom (UK) and United States (US), both of which consume a relatively small percentage (~6%) of their country's Gross Domestic Product (GDP). Access to CI is restricted in these systems, though by different mechanisms – under-reimbursement in the US Medicare system and occasional non-reimbursement in the UK National Health Service (NHS) [2]. In comparison, access to CI is easier in Germany and the privately funded parts of the US health-care system, which consume greater percentages (10% and 14%, respectively) of their country's GDP.

In the UK, the National Institute for Clinical Excellence (NICE) takes account of cost effectiveness in recommending treatments for adoption by the NHS [3]. NICE's judgements are highly educative. They challenge the popular conception that all treatments in health care are effective and should be provided to deserving patients regardless of cost. They establish the principle that economic evaluation should contribute to setting priorities. These points are relevant to CI. Total health-care costs (at net present values) have been estimated to be €60,000 for post-lingually deafened adults accumulated over 30 years of expected use, €74,000 for children accumulated over 15 years, and 96,000 for children accumulated over an expected lifetime [4]. Our aim in this article is to assess whether these high costs justify any restriction on access to CI in the UK, given current estimates of the cost-effectiveness of CI.

The economic approach to prioritising the allocation of resources in health care is not to everybody's taste because of its emphasis on the average rather than the individual. Nonetheless, it is hard to argue with the following two propositions: (i) health-care interventions that are beneficial to patients and which save more money than they cost should be adopted. (ii) Among interventions that do not save money, resources should be allocated preferentially to interventions that generate the greatest increase in quality of life at the lowest cost. The propositions distinguish two approaches to the economic analysis of programmes of health care. Cost–benefit analysis values the consequences of, as well as the inputs to, programmes in monetary terms. The greater the extent to which the balance sheet shows a cost saving, the stronger the support for a programme should be. The fact that all inputs and consequences are expressed in a common currency means that comparisons can be made not only within the health-care domain but also between domains (health care, education, and civil engineering, for example). For this reason, the approach can be powerful. However, it has the limitation that the most important outcome of health care is (hopefully) better health, independent of the costs or savings that result from better health. Yet, it is difficult to put a monetary value on "better health" itself – much medico-legal argument surrounds the issue of placing a monetary value on an additional 10 dB of hearing sensitivity.

This problem is avoided in the second style of analysis, cost–utility analysis, which relates the net cost of a programme of health care to the net gain in health-related quality of life (HRQL) that results from the programme. Researchers are encouraged to take account both of direct costs (and savings) incurred in the health-care system and indirect costs (and savings) incurred in other domains. Cost–utility analysis then determines the cost of gaining a time-integrated unit of HRQL, usually referred to as a quality-adjusted life year (QALY). The potential for making comparisons between domains is lost – we no longer have a disciplined basis for deciding whether to re-allocate resources between road building and health care, for example – but the capacity to make comparisons within the health-care domain is enhanced. If the aim of commissioning health care is to maximise the gain in HRQL for each Euro, Dollar, or Pound that is spent, then resources should be allocated to treatments that gain a QALY at higher cost only after meeting the demand for all treatments that gain a QALY at lower cost. As a benchmark, scrutiny of the judgements of NICE shows that the treatments that have been recommended gain a QALY for less than about 50,000 (£30,000) [3].

The cost–utility approach has limitations, however. Although rigorous psychophysical procedures have been developed for measuring the utility of states of health [5], the procedures invoke the concept of early death and so may raise unacceptable levels of anxiety in patients, or the parents of patients, who are considering a surgical procedure. For example, one of the preferred procedures, the Time Trade-off Technique, estimates the utility of an informant's health state by requiring the informant to indicate the proportion of their expected remaining years of life they would be prepared to give up in exchange for perfect health. To avoid making patients anxious, systems based on questionnaires have been developed in order to map informants onto states of health whose utilities have been valued independently by members of the public who have been asked to consider what it would be like to experience that state of health. One such system, the Mark III Health Utilities Index (HUI-III) [6], is widely used in assessing the utility of states involving profound hearing impairment. It is favoured because it measures function in vision, hearing, and speaking as well as the conventional dimensions of mobility, dexterity, emotion, cognition, and pain. It returns a single value on a scale where zero corresponds to the state of death, and unity to the state of perfect health. However, while many members of the public have experienced temporary states of health involving pain and immobility, many fewer have experienced chronic sensory disabilities. If so, they might misjudge the loss of utility associated with hearing impairment, with the result that utilities estimated by the HUI-III would lack credibility.

Four issues are relevant: first, is the philosophy of HUI-III appropriate for valuing states involving hearing impairment? Second, does the HUI-III produce plausible utilities when completed by adults, or the parents of children, who receive CI? Third, are the medical costs of paediatric CI offset by indirect cost-savings in the domain of education? Fourth, taking account of direct health-care costs and indirect cost savings, does cost–utility analysis indicate that CI represents acceptable value for money for the NHS?

Hitherto in this article, we have asserted that the HUI-III measures HRQL. That choice of terminology implies that bilateral profound deafness is associated with a loss of HRQL and that CI is associated with a gain in HRQL. Two dilemmas are created by this orientation. The first is that hearing-impaired people rarely regard hearing impairment as a manifestation of "illness". For example, where profoundly hearing-impaired adults who are about to undergo CI are asked to complete the EuroQol questionnaire (a self-report measure of health status, which emphasises health), they report a status that does not differ from population norms [7]. Moreover, their score does not change significantly following CI (unless their tinnitus has been attenuated or exacerbated), despite a large improvement in the ability to hear and to understand speech. Second, by the same logic, it seems odd to assert that congenital deafness is a manifestation of reduced HRQL. The two dilemmas can be resolved by recognising that the HUI-III measures the "utility" of states of being, and that the word "utility" in this context has the meaning of "preference". Thus, one state has a higher utility than another if, on average, it is preferred to the other. Expressed in that way, it becomes more logical to assert that the state of being able to hear has higher utility than the state of being hearing-impaired because, on average, members of the public prefer to hear than to have impaired hearing. Evidence of this preference is provided by the fact that people seek interventions, including hearing aids and cochlear implants, for themselves and for their children when they are hearing-impaired. Thus, from a societal perspective, CI is associated with a gain in utility because, on average, it places users of implants in a state that society judges to be more preferable. In terms of values prevalent in society, therefore, the HUI-III has the potential to measure the extent to which the goal of CI has been achieved.

Two comparisons have been made between utilities estimated with the HUI-III and utilities estimated directly with patients or their parents. Both comparisons indicate that values obtained from the HUI-III are plausible.

We have worked with 311 post-lingually deafened adult patients who completed the HUI-III before and after undergoing CI. Their average utility prior to CI was 0.43, their average utility 9 months after CI was 0.63, yielding a gain in utility of +0.20 (95% confidence interval: 0.18 to 0.22). This increase is similar to the average increase estimated by averaging data from several studies in which patients had valued their quality of life, with and without a cochlear implant, on a visual-analogue scale extending from zero (death) to unity (perfect health). The mean value without an implant was 0.54, and with an implant was 0.80, yielding a gain of +0.26 [8].

We have also measured the average difference in utility when the HUI-III is completed by the parents of children with CI, and children without CI. We controlled for the number of significant additional disabilities experienced by the child, the age of onset of impaired hearing, the socioeconomic status of the child's family, the average pre-operative hearing level of the child, and the current age of the child. The average difference was +0.19. For children with congenital hearing impairment, the value ranged from +0.29 for children implanted before the age of 4 years to +0.11 for children implanted after the age of 6 years. These values embrace estimates by the parents of children with implants who made retrospective judgements of their child's utility before and after implantation, yielding a gain of +0.22 with the Time Trade-off Technique and of +0.29 using a visual-analogue scale [9].

Gains in utility measured with the HUI-III are credible, therefore, insofar as they are similar to those measured directly using psychophysical methods with relevant informants.

There are two main reasons for establishing whether paediatric CI is associated with savings in indirect costs. First, such savings are an informative outcome in their own right, because they indicate that CI has an impact downstream of improving hearing. Second, such savings can be offset against the medical costs of CI in calculations of cost-effectiveness. To date, research has focussed on the issue of whether savings occur in the domain of education. In the US, age-matched groups of children were identified with and without implants. Prior to implantation, the implanted group displayed the same profile of educational placements as the un-implanted group. Two or more years after implantation, the profile of the implanted group had shifted in the direction of mainstreaming. Based on the profiles observed, the costs of different types of change in educational placement were measured, and were shown, with one exception, to more than offset the health-care costs of providing CI [10]. In Germany, total educational and medical costs to the age of 16 years were estimated for three groups of children with implants and for a control group with acoustic hearing aids. The groups of implanted children were distinguished by their age at implantation: I: 0–1.9 years, II: 2–3.9 years, and III: 4–6.9 years. Educational costs were similar for Groups II (DM231,000) and III (DM257,000) and the control group (DM277,000), but were lower for Group I (DM159,000). Net costs (educational and medical) were lower for Group I (DM271,000) than the control group (DM313,000) [11]. In the UK, we have valued the special educational resources allocated to approximately 400 children with implants and 1800 children without implants. The data were analysed with control exercised over the number of additional disabilities experienced by each child, the age of onset of hearing impairment, the socioeconomic status of the child's family, the average pre-operative hearing level, and the child's current age. Possession of an implant was associated with an average cost saving of €2384 per year.

All three sets of results indicate that paediatric CI is associated with cost-savings in education. The German data indicate that the savings are greater the younger the age of the child at the time of implantation.

Fig. 1 plots cost–utility ratios, which we have estimated for the provision of unilateral CI to children and to post-lingually deafened adults, and the provision of bilateral CI to adults. The unilateral ratios were estimated empirically, with the HUI-III being completed by patients or their parents. Total health-care costs were projected over patients' expected remaining lifetimes taking into account costs incurred in remedying device failures, managing other medical and surgical complications, and providing CI to a small percentage of patients who elect to become non-users of their devices. In the case of children, indirect cost savings in education were also taken into account. The bilateral ratios were estimated in a scenario analysis in which clinicians and researchers with knowledge of cochlear implantation estimated the gain in utility that would result from receiving a second implant in either the same surgical session (simultaneous bilateral) or an additional session (additional bilateral) [12]. The first five unilateral ratios are competitive within the British health-care system, in that each ratio falls significantly below the value of €50,000/QALY inferred from the judgements of NICE to represent the upper limit of acceptability. The sixth ratio, which does not differ significantly from the limit, arises when CI is provided in the less physiologically responsive ear of adults who already benefit slightly from acoustic hearing aids. The two bilateral ratios fall outside the limit. They may fall within acceptable limits in the insurance-based parts of the health-care systems of countries such as the US and Germany that devote a greater percentage of their national wealth to health care than the UK does at present. The ordering of the ratios is credible in that it implies that the highest priority should be given to implanting young children, followed by adults and older children. Then, only after needs for unilateral implantation have been met, should bilateral implantation be contemplated.

The results of analyses of the cost–utility of CI are credible when viewed from a societal perspective. They generate an ordering of priorities for the provision of CI between groups of patients that is compatible with intuitive logic. The results can therefore guide, and justify, clinical priorities. Cost–utility ratios for CI are sufficiently favourable to challenge the practice of restricting the access of suitable candidates to unilateral implantation in the UK.