I would like to thank the commentators for their thoughtful and thought-provoking responses. I commend the different perspectives on the relationship between coverage and user charges that they offer. In particular, the issue of moral hazard (raised by Hurley) deserves specially close attention – one of the key roles of value-based user charges is precisely to address this phenomenon. McIntyre and Meheus emphasize the crucial issue of practicality in low-income settings and the importance of considering how the staged transition to universal health coverage is managed. I have raised this issue briefly in the initial paper, but shall expand on it further below. It is also interesting to note that both responses mention the importance of strategic purchasing of health services. I completely agree with this observation, but would argue that user-charge policy (or its mirror image – where to direct limited public funds) is not an alternative to strategic purchasing. Rather, value-based user charges should be a pillar of strategic purchasing, answering the fundamental question of what services should be purchased from public funds and at what price.
As I see it, the big themes that underlie the commentaries are (a) the informational limitations and (b) the transaction costs associated with the real-world application of cost-effectiveness principles to user-charge policy. I hope that we agree that – in principle – the limited resources available for publicly funded health service should be used to best effect. Although the definition of ‘best effect’ is open to policy debate, in most health systems, this means applying a criterion such as ‘expected cost per QALY’ to decide whether or not to fund a treatment. This will ensure that the payer secures maximum benefit (in this case, increments to health) for a fixed available budget. Of course, it is possible to adjust the precise metric, perhaps to incorporate equity considerations or other policy objectives, but the principle remains unchanged.
The ultimate logic of these optimization principles is that the chosen cost-effectiveness criterion should be applied to each patient. That is – rather than follow some generic guidelines – a clinician should assess the expected costs and health benefits specific to the current patient. That assessment should then determine whether or not a treatment under consideration can be offered if financed from public funds. Although perhaps a little fanciful, it is worth noting that the notion of clinical triage, developed to assess priorities among wartime casualties, may come close to an operational example of patient-level cost-effectiveness analysis, by seeking to focus limited clinical resources on patients for whom the greatest health gains could be secured.
In the context of more conventional health services, patient-level CEA is ruled out for numerous practical and political reasons, most notably that the information requirements are infeasible, the burden on clinicians would be intolerable and the political consequences of variations in treatment are enormous. In the same way, it is difficult to see any practical way in which clinical triage could become a basis for setting value-based user charges. It would, for example, be extraordinarily difficult for a clinician to explain to an older person that a charge must be applied because the health benefits of treatment are expected to be small relative to those secured for a younger person who can be offered the treatment free of charge. For this reason, general rules of entitlement and the associated charges have to be based on average cost-effectiveness calculations for chosen groups of patients.
However, the types of patient groupings, and the charges to be applied, have varied enormously between countries and have not historically been subject to careful scrutiny. Thus, it is common to find crude access charges for ambulatory care applied uniformly, regardless of the type of patient or the care being sought. It is these blanket user charges that are so pernicious. They are likely to deter certain patients, regardless of the cost-effectiveness of potential treatment. Pursuing this argument, the emphasis in South Africa on reducing financial barriers to primary health-care services is likely to improve the coverage of cost-effective primary-care treatments, but may also stimulate the use of certain primary-care services that are not cost-effective in the South African setting. This may be appropriate if the net benefits of the policy are positive, and there is no possibility of a more refined system of financial subsidies and revenue collection. However, it does appear to embody the judgement that public funds are best directed at primary care in its entirety, in preference to secondary care. Yet there are also likely to be certain secondary care services that offer very good levels of cost-effectiveness. If information systems permit, I would therefore argue that some of the public subsidy should be directed at those services. One would, in my view, advocate a blanket targeting of primary care with the available public funds only if the current state of administrative arrangements precludes reliable recording of specific diagnoses and services provided, or if the capture and audit of such data impose unacceptably high costs on the health system or if relevant cost-effectiveness data are not available.
As soon as information systems and revenue collection mechanisms permit, I would argue that lower income countries should move away from the systems of blanket coverage of classes of health services (or blanket exclusions) towards more nuanced systems. High-value interventions should be offered at zero price (or even in some cases given an additional subsidy so that the user can be incentivized to seek treatment), whereas partial or full user charges are attached to less-valued interventions. As explained in my initial paper, charges can in principle be varied according to ‘ability to pay’ through the use of fee exemptions, although of course accommodating such equity considerations may reduce the capacity to maximize other objectives such as aggregate health gain.
A key issue raised in the commentaries is the extent to which more refined charging regimes are practical and can be implemented uniformly. This will vary between countries. The South African example suggests that even relatively advanced health systems may struggle in the short term to secure advances in this respect. However, this should not blind us to the observation that the public subsidy of health services could secure higher levels of health gains if the necessary administrative arrangements can be put in place.
In their pursuit of higher levels of coverage, lower income countries suffer from a severe difficulty that did not arise for higher income countries when they secured universal coverage. Their citizens and clinicians can observe many advanced technologies in use in high-income settings that cannot feasibly be provided from public funds in low-income settings. The pressure to adopt these technologies is nevertheless fierce, even though their adoption may pre-empt the use of more prosaic technologies that are less fashionable but are more cost-effective. Policymakers need strong cost-effectiveness evidence to pick out the technologies they can afford to fund from public sources. For interventions with intermediate levels of cost-effectiveness, value-based user charges offer the potential for a more subtle and effective use of public funds than blanket inclusion or exclusion, and should be part of the armoury of all strategic purchasers of health services.
