1. Equity and political economy

Tamire Woldemariam (Reference Tamire Woldemariam2015) claims that the Report’s conception of fairness in service provision unduly prioritizes ‘horizontal equity’ (providing the same quality services to everyone with the same need) over ‘vertical equity’ (which he understands as the provision of care suited to differential needs, e.g. antimalarials for a malaria sufferer and cardiac surgery for a heart condition).

In reply: the Report differs from Tamire Woldemariam’s interpretation of vertical equity in the use of health care resources, on which both the malaria sufferer and the person with an operable heart condition must be treated. Instead, the Report follows Morris et al. (Reference Morris, Sutton and Gravelle2005: 1251) in understanding vertical equity in health service provision as demanding that “individuals with different levels of need consume appropriately different amounts of health care”. The question is then what level of health care is appropriate for different needs. The Report denies that under circumstances of severe resource scarcity, it is always appropriate to treat both the malaria sufferer and the person with an operable heart condition. Instead, it holds that coverage should be based on (among other factors) both the health gain per unit of expenditure and the level of advantage from which this gain takes place, with greater importance being assigned to gains that come to the worst off. On these grounds, the Report holds, it may be fair to offer coverage for malaria while failing to offer coverage for a heart condition, if treatment for the heart condition (a) is much less cost-effective than other as-yet-uncovered treatments; and (b) benefits people who are typically among the better-off (say, because they are already long-lived and are among a relatively well-off urban class).

Tamire Woldemariam (Reference Tamire Woldemariam2015) also suggests that the Report add to the criteria proposed for deciding on coverage a criterion of ‘achieving high total return’. As the Report’s three core criteria already aim at gains in aggregate health, improvements in the situation of the worst off, and a reduction in health-related poverty, we assume that he has in mind still broader effects, such as improved economic growth or strengthening the social fabric (e.g. by engaging a broad segment of society, as in the recruitment of health-promotion volunteers in Ethiopia to which he refers). These broader effects are certainly among the permissible aims of government policy. It is, however, a deeply controversial issue how far, if at all, such additional benefits should count in priority setting within the health sector (Brock, Reference Brock2003). The Report therefore merely raises them as factors that are worth considering (see Box 3.3).

Littlejohns and Chalkidou (Reference Littlejohns and Chalkidou2015) worry that the Report’s frequent use of the word ‘should’ ‘makes it seem all too easy’ to implement its recommendations. Apart from its recognition of resource constraints, they suggest, the Report is blind to the political constraints faced by decision-makers.

Their challenge offers an opportunity to clarify the Report’s intended role. In ordinary discourse, sometimes ‘should’ implies ‘can’. For example, if your doctor says, ‘you should walk more’, she must assume that you can walk more. Nonetheless, ‘should’ may also be used in an explanation of what it would take for policies to conform to a standard, without implying that meeting this standard is currently feasible for the policy-maker in question. This is the way it is used in the Report. Take, for example, its claim that “the health system should first expand coverage for [services] in the highest priority class” (WHO, 2014: 22). Suppose that a policy-maker aims to move towards Universal Health Coverage (UHC) in circumstances in which many rural poor still lack access to skilled birth attendance. Also assume that in order to somewhat improve their access to very cost-effective skilled birth attendance at a later point in time, this policy-maker must first secure the support of an urban elite, which demands an expansion of coverage for only moderately cost-effective tertiary care in the cities. By compelling the policy-maker to devote to an expansion of coverage for low-priority services resources which she could otherwise have used to expand a high-priority service, this interest group forces her to depart from a fair path to UHC. There is then a distance between what should, in fairness, be done first and what a pragmatic policy-maker will have to do. We agree with Tamire Woldemariam (Reference Tamire Woldemariam2015) that in such circumstances it may be necessary to make ‘wise choices’ that are not wholly fair. Nevertheless, it is important that the resulting unfairness is recognized and minimized. Therefore, even when the fairest path is rendered unfeasible by political constraints, the Report’s recommendations can play a twofold role. First, to evaluate the policies demanded by the interest group against a standard of fairness. Second, to help a policy-maker select the least unfair among the policies that remain feasible once the constraints imposed by this interest group have been taken into account. The Report therefore does not assume away political constraints. Rather, it aims to offer a critical perspective on their effect and to guide choice given their operation.

Weale (Reference Weale2015) raises a question about the Report’s recommendation that countries “include disadvantaged groups from the outset and make sure that these groups are not left behind” (WHO, 2014: 38). He notes that this recommendation is sensible only if a good way to improve the lot of the worst off in the long run is to choose policies that benefit them at each stage on the path to UHC. However, he argues, though this empirical assumption may be true, it also might be false. For example, it might be necessary to develop competence in running an insurance scheme by starting it among formal sector workers. Or the most likely route to UHC might be to first expand coverage for well-off groups, in the expectation that this will eventually generate a demand for an expanding circle of coverage.

We believe that, at every step on the way to UHC, policy-makers should give particular weight to the interests of the worst off. However, we acknowledge (as does the Report, see WHO, 2014: 37ni) that if it were established that attempting to include the worst off from the start would severely impede progress towards UHC, then those especially concerned with the worst off should accept policies that, temporarily, leave them behind.Footnote ¹ The Report’s recommendation should therefore be taken to apply only when including the worst off from the start is an effective way of improving their lot in the long run.

2. Starting points and rankings

Littlejohns and Chalkidou (Reference Littlejohns and Chalkidou2015) and Weale (Reference Weale2015) point out that, while the Report focuses on expansion of coverage, its criteria have implications for disinvestment as well. For example, the criteria might require dropping coverage for some low-priority services (e.g. dialysis) in order to expand coverage for high-priority services (e.g. secondary prevention of diabetes and cardiovascular disease). Disinvestment is, of course, politically difficult (as one government official remarked to some authors of the Report: “Don’t take sugarcane from the elephant’s mouth!”). Moreover, as Weale (Reference Weale2015) argues, there may be moral reasons that distinguish disinvestment from expansion. For example, people who have paid in to receive a particular package of coverage may have a claim to that coverage. The Report does not discuss how to balance such claims against the claims of others to high-priority treatments, but we agree that guidance on this difficult issue is required.

Weale (Reference Weale2015) also questions the Report’s proposed assignment of available interventions into priority tiers, partly on the grounds that it is too demanding on limited cognitive resources. In reply, we agree that constructing a well-informed, complete ranking of all possible interventions is unfeasible. However, the criteria recommended by the Report can provide useful guidance without such a ranking. Consider two common types of situation. In a first type, the decision maker is considering only a single intervention and lacks information about exactly which intervention(s) will lose out if the intervention in question is funded. In these cases, she can gather estimates of the proposed new intervention’s (i) cost-effectiveness; (ii) impacts on disadvantaged groups; and (iii) effects on the incidence of catastrophic health expenditure and apply a set of thresholds (in the manner illustrated in Box 1 of our Précis), with the thresholds determined by an estimate of the opportunity costs of funding that intervention (Claxton et al., Reference Claxton, Martin, Soares, Rice, Spackman and Hinde2013). If the intervention falls into the high-priority class, then that is a good reason to fund it. If the intervention falls in the low-priority class, then she has a good reason not to fund it, as it is likely that the resources are better used on other interventions, especially in a setting where many high-priority interventions are yet to be fully covered.

In a second type of decision scenario, the decision maker has to choose one intervention among a limited set of interventions competing for the same resources. If she can gather decent estimates of how these interventions score on the three core criteria, then she can attempt to rank them and fund the intervention that is highest ranked. By way of illustration, consider a simplified version of a choice faced by administrators during the expansion of the Thai Universal Coverage System (UCS) in the mid-2000s.Footnote ² The question was whether to (i) first expand coverage for secondary prevention for diabetes and cardiovascular disease; or instead (ii) first offer coverage for dialysis. Making this decision did not require a ranking of all available interventions in the health system – just of these two interventions (prevention was ‘on the menu’ because it was known to typically be cost-effective; dialysis due to advocacy on behalf of patient groups and nephrologists). Table 1 gives estimates for how these interventions performed on the Report’s three key criteria. In brief, secondary prevention had an overwhelming advantage in terms of cost-effectiveness, but dialysis helped those who were worse-off on one important indicator, namely their individual disease burden. Moreover, considerations of financial risk protection did not clearly favour one of the two policies. Coverage for dialysis offered a clear reduction in catastrophic health expenditure for an identified group of sufferers. However, secondary preventive services would, eventually, prevent many serious cases of illness, thereby also preventing some cases of catastrophic expenditure and income loss. It was therefore plausible that, in the long run, secondary preventive services would, on balance, offer more financial risk protection than dialysis.

Table 1

Relevant data for dialysis and secondary prevention

QALY=quality-adjusted life year; GDP=gross domestic product.

^a Teerawattanon et al. (Reference Teerawattananon, Mugford and Tangcharoensathien2007) and DCP2 http://dcp2.org/interventions/17/interventions-for-diabetes (accessed June 2014), using per capita income of USD 3000.

^b Authors’ own calculation from GBD Database (spreadsheet available upon request). These numbers are relative to the GBD global norm of 86 years.

^c This is based on an estimate by Phusit Prakongsai, assuming 31,000 cases annually, 75% of which fall under informal sector health insurance with 90% of these cases facing catastrophic health expenditure (>10% of monthly consumption expenditure) if not covered. See also Prakongsai et al. (Reference Prakongsai, Palmer, Uay-Trakul, Tangcharoensathien and Mills2009).

The Report recognizes that there is room for reasonable disagreement about such cases. It requires only that the three core values of cost-effectiveness, priority to the worst off and financial risk protection be considered, but prescribes no single way of balancing them (WHO, 2014: 20). Moreover, these criteria are not exhaustive – other considerations will be morally relevant. In the Thai context, for example, it was significant that two other public insurance schemes (those for civil servants and formal sector workers) already covered dialysis, so that the lack of coverage in the UCS became, to some, a symbol of second-class treatment of those in the informal sector (WHO, 2010). Nonetheless, the Report suggests a way of combining criteria that would yield a clear judgment in this case (WHO, 2014: 20–22; see also Box 1 in our Précis). Roughly, it proposes that highly cost-effective interventions should definitely have high priority and very cost-ineffective interventions should definitely have low priority, and that other criteria may determine the priority class of the intervention only in the (substantial) range in between these extremes. Now, the estimated range of cost-effectiveness of secondary prevention arguably places it within the high-priority category, whereas the estimated range of cost-effectiveness for dialysis places it firmly within the low-priority category. The Report’s suggested judgment is therefore that secondary prevention should be prioritized over dialysis.

3. Unacceptable trade-offs

The Report allows for many different ways of making trade-offs on the path to UHC. However, it also argues that particular ways of making trade-offs are generally unacceptable. Several respondents question this claim.

Littlejohns and Chalkidou (Reference Littlejohns and Chalkidou2015) ask to whom and why these trade-offs are said to be unacceptable. We take this question to be about the Report’s moral universalism. When the Report claims that particular ways of making trade-offs are generally unacceptable, it claims that they are generally incompatible with what the large and diverse group of authors collectively agreed were reasonable ways of weighing the three core values and therefore presumptively unjustifiable to those affected by or concerned with the decision. (It is allowed that other moral factors may give grounds for overriding this presumption.) However, one might ask, what justifies taking the promotion of aggregate health, improving the condition of those worst off and financial risk protection as values which should guide policy in all countries on the path to UHC and which should each have substantial independent weight in their policy decisions? In reply: these three aims have support from a multitude of moral perspectives; they are also widely endorsed as proper goals for health systems. Moreover, they play a central role in motivating the pursuit of UHC. In endorsing UHC as a goal (as in the 2005 58th World Health Assembly), the member states of the WHO therefore expressed support for these values. It is thus fitting that the Report clarifies these values and provides advice on how they should guide policy-makers on the path to UHC.

Rumbold and Wilson (Reference Rumbold and Wilson2015) discern a conflict between the Report’s tolerance for various ways of making trade-offs and its argument that particular trade-offs are nonetheless unacceptable. Before we remark on the detail of their argument, we note that the Report’s general strategy here is straightforward: the unacceptable trade-offs mark out a ‘no go area’; the Report is tolerant about where one goes within the remaining area. However, Rumbold and Wilson question the way these boundaries are drawn. One of their arguments concerns the second unacceptable trade-off (UT2): “to give high priority to very costly services whose coverage will provide substantial financial protection when the health benefits are very small compared to alternative, less costly services” (WHO, 2014: 39). Rumbold and Wilson (Reference Rumbold and Wilson2015) understand UT2 as implying that in almost all cases cost-effectiveness should trump financial risk protection. In their view, it therefore contradicts what they take to be the Report’s “attempts to stay neutral as to how the values of cost-effectiveness, benefit to the worse off and financial protection ought to be weighed against one another”.

We acknowledge that the wording of UT2 can suggest Rumbold and Wilson’s interpretation. We therefore welcome the chance to explain how regarding this trade-off as generally unacceptable is consistent with the Report’s aforementioned ecumenical attitude. The justification offered in the Report for UT2 is that prioritizing coverage for very cost-ineffective interventions that offer substantial financial risk protection over very cost-effective interventions is presumptively unjustifiable because coverage for the latter typically dominates the former both in terms of health gains and in terms of financial risk protection. As the Report states (WHO, 2014: 39), favouring costly services that offer low health gains:

There is, therefore, no conflict between UT2 and the Report’s tolerance for a variety of ways of trading off financial risk protection against health gains. As a further clarification, we note that the Report’s tolerance does not equate with an attempt to ‘stay neutral’ on how these values should be traded off. Rather, as our discussion of the case of dialysis vs secondary prevention shows, the Report proposes (though it does not prescribe) an approach to weighing them. Like many an opinionated liberal, the Report is therefore both partisan and tolerant.

Rumbold and Wilson also challenge UT1: “to expand coverage for low- or medium-priority services before there is near universal coverage for high-priority services” (WHO, 2014: 38). This, they write, proscribes some perfectly reasonable choices: “[Contra UT1,] it does not seem to be unacceptable for a policy-maker to seek to remove, say cultural or ethnic barriers to health care for all [existing low, medium and high-priority] services, before seeking to ensure maximal coverage for [high priority] ones” (Rumbold and Wilson, Reference Rumbold and Wilson2015) In reply: we agree that overcoming barriers posed by stigmatization or discrimination offers grounds for prioritizing effective and affordable access for a currently excluded group to already provided services. Ensuring such access may be an important step towards ensuring its members’ rights are met; it may also affirm their status as citizens. Cases such as this show why expanding low-priority services before high-priority services is only presumptively unacceptable.

Finally, Rumbold and Wilson (Reference Rumbold and Wilson2015) discern a conflict between UT1 and UT3 (“to expand coverage for well-off groups before doing so for worse-off groups when the costs and benefits are not vastly different”; WHO, 2014: 39). They ask: “What ought a policy-maker to do when faced with the question of whether to expand coverage for low- and medium-priority services for the worse-off or expand coverage for high-priority services for the already well-off? Both options here, on different UTs, look generally unacceptable”.

The following may be an example of what they have in mind. Let us revise our ‘dialysis vs secondary prevention’ case as follows. Suppose the decision-maker must choose between (i) adding high-priority secondary prevention to the already more generous packages for the formal sector, or instead (ii) adding the low-priority dialysis to the more limited UCS (for the informal sector). Option (ii) appears to be ruled out by the prohibition on expanding coverage for low-priority services before high-priority services (UT1). However, (i) is ruled out by the prohibition on expanding coverage for the better-off before expanding coverage for the worse-off (UT3) – or so Rumbold and Wilson seem to think.

In reply: this is not really a conflict between two of the Report’s ‘thou shalt not’s’. UT3 only proscribes expanding coverage for the better-off rather than the worse-off when the costs and benefits are not vastly different. However, of course, in this case, they are vastly different – it is precisely because they are so different that secondary prevention is a high-priority service and dialysis a low-priority service. Thus, contrary to Rumbold and Wilson, UT3 is mute in cases of this kind. Nonetheless, the example could be seen as a challenge to UT1. After all, the example involves a trade-off in which various answers may be reasonable. We therefore conclude that Rumbold and Wilson’s discussion suggests a second case in which UT1’s presumption favouring an expansion of high-priority services may be overridden.

4. The need for more information

We end on a topic of agreement. Littlejohns and Chalkidou (Reference Littlejohns and Chalkidou2015) argue that successful implementation will depend on more data, especially on what works (and what fails) in practice. We welcome their many constructive suggestions for future research and share their belief that developing a catalogue of implementation case studies should be a priority and that it would be especially valuable to see whether these studies confirm or challenge the general approach advocated in the Report. Fortunately, the WHO agrees. Under the leadership of Tessa Tan-Torres Edejer it has gathered nearly two dozen academics and policy-makers from low- and middle-income countries to draft 10 implementation case studies. These studies examine stylized versions of actual decisions taken on the route towards UHC and assess them in the light of the Report’s recommendations, and vice versa. (The WHO plans to make them available on the WHO-CHOICE website.)

In closing, Weale (Reference Weale2015) rightly remarks that understanding the circumstances in which UHC is pursued and knowing how to act effectively in these circumstances are prerequisites for progress. We thank our critics for their contributions to this understanding and know-how.