Aligning policy objectives and payment design in palliative care

The World Health Organisation explicitly includes palliative care in its goal for universal health care, and a recent resolution of the World Health Assembly called on member states to ensure good access to palliative care [36]. Most jurisdictional policies express the same aspiration but, unfortunately, the reality of palliative care provision appears to fall short of that aspiration [37‐39], although probably mainly because of the much higher political profile accorded to acute care, it may also be due to ‘conventional health economic analyses … undervaluing the benefits derived by society from the provision of palliative care’ [40]. These aspirational goals need to be achieved within the context of both constrained public funding and ensuring value for money.

Funding policy design needs to be informed by two critical considerations to ensure value for money: the risk of supply- and demand-side moral hazard and the risk of provider adverse selection [41].

Supply-side moral hazard involves providers delivering more services than consumers require. It is particularly relevant in fee-for-service funding environments, where providers are reimbursed for each additional service [42, 43], and is especially a risk in environments, such as the United States, with a large corporate for-profit sector [44]. Contemporary payment system design in health care balances payer and provider risk: fee-for-service payment involves the greatest payer risk and the least provider risk and, for this reason, payers are replacing fee-for-service reimbursement by funding strategies which ‘bundle’ payments either over time – known as ‘capitation’ approaches – or across an episode of care using some form of classification of the episode [45].

Adverse selection occurs when a payer (such as an insurance company) or a provider has a disproportionate share of ‘bad risks’ –patients who have greater than expected health care costs. Adverse selection is mitigated by use of risk adjustment in either episode payment or capitation formulae [46]. The corollary of adverse selection is ‘cream skimming’ which involves providers’ selecting patients who are expected to have lower needs than the reimbursement formula predicts [47]. This is a risk either because the provider has access to additional information about the patient not captured in the risk-adjustment formula [48] and/or because of behavioural responses by consumers [49]. Again, the risk can be mitigated, but not eliminated, by good risk adjustment and limiting the ability of palliative care providers to select patients, for example, by implementing geographic assignment of patients.

In budget-constrained systems, administrative rules might be used to limit access to palliative care, such as limiting access to those with an anticipated life expectancy of six months or less [50]. Crude rules such as this take no account of different illness trajectories and so may discriminate against those with severe, on-going illness [51], create ethical issues for the clinicians involved as gatekeepers [52], rely on potentially inaccurate estimates of life expectancy [53, 54] and undermine a more holistic vision for the role of palliative care [55]. They are especially cruel if accompanied by limitations on the extent to which patients receiving palliative care can also receive intensive disease–directed treatments such as blood transfusions, which may be aimed at improving quality of life [56]. Rules put in place in the United States may have been developed because of the absence of universal coverage for substitute services such as residential (nursing home) care. However, the significant increase in spending on hospice care in the United States has been driven by an increase in the proportion of decedents who have received hospice services rather than by spending per patient which has been constant [57], and so can be interpreted as an objective of policy not as a cost ‘blow out’.

In countries with universal health care, there seems to be no justification to limit access to palliative care services to people who expect to die within a defined period and so the risk of demand-side moral hazard (e.g. substitution from other forms of care) is lower. It is unlikely that people would seek palliative care unless the prospect of death were high and so demand for these services is naturally capped to those who expect to die in the near future. If funding for palliative care services were uncapped, supply-side moral hazard could be mitigated using performance metrics e.g. that 90 per cent of clients managed by a palliative care service died within x months of admission to care. Components of funding could still be capped to encourage desirable policy directions e.g. hospital funding could have tighter caps than out-of-hospital funding if the policy directions were to encourage dying outside hospital.

Palliative care as a distinct and recognised service is a relatively recent newcomer to health care – in the United Kingdom it is often dated to the pioneering work of Dame Cicely Saunders in the 1960s and 1970s, with its emergence in the United States dating from the same period [58‐60]. With advances in medical technologies, and ever increasing opportunities for intervention, the challenge of developing and meeting the appropriate goals of care [61] becomes greater and the place of care becomes even more important. The result appears to be that the supply of palliative care lags behind need and demand [37‐39].

The importance of ensuring appropriate funding design is increased as countries seek to expand palliative care, to meet the universality aspiration. Inappropriate funding design may mean that additional palliative care investment is not used as efficiently as possible that is, funding increases do not generate a commensurate increase in the extent to which palliative care needs are met.

A second aspect of allocative efficiency in palliative care is ensuring the appropriate balance of investments in supporting home vs hospital-based palliative care programs. There is an imbalance in actual compared to preferred location of death in most countries, with most people preferring a death at home [62, 63] but with most deaths occurring in institutional settings [64].

Importantly, better palliative care or other supports can influence the likelihood of death in the preferred setting [26, 27, 65], and the length of time at home even if death occurs in hospital [66]. As indicated above, in addition to reflecting people’s priorities better, increasing the proportion of deaths occurring at home could also improve technical efficiency [15, 16].

Funding policies can inadvertently preference one type of care over another. In Belgium, for example, basic palliative care is funded through uncapped social insurance funding but specialist, hospital-based palliative care is block funded leading to overwork issues [34]. This could potentially lead to problems of gaining access to palliative care.

In Victoria, Australia, hospital-based palliative care is funded under that state’s activity based funding arrangements, so each additional hospital admission attracts additional funding (although overall hospital funding, across all types of patients, is capped). In contrast, community or home based palliative care is funded on a population basis, so additional patients do not attract additional funding. The different funding mechanisms thus create a greater incentive for admissions to hospitals than for community-based treatment.

There are two main theoretical justifications for personal out-of-pocket costs in health care. The first is demand-side moral hazard: that unless there is a personal contribution, patients will consume too much of the ‘free’ service. Co-payments reduce the demand for non-institutional services (such as general practitioners and pharmaceuticals) but not for hospitals, and have a higher impact on low income people [67]. Aside from the problematic equity issues, such co-payments may thus perversely encourage hospital, rather than community-based palliative care options [14], even more so if co-payment policies lead to higher out-of-pockets for home-based care. In the context of freely available substitutes, as is the case in most countries with universal health care, the risk of demand-side moral hazard is low. Together these factors would argue against significant use of patient co-payments.

The second justification for co-payments is normative: that the alternative to co-payments is additional collective funding - either through taxation or social insurance – and that taxes and social insurance contributions need to be minimised because of the ‘dead weight’ burden of taxation [68]. The level of taxation is a social choice and perceptions of the appropriate level of taxation will differ across countries and the perceived purposes of taxation. There is a strong ethical argument for public financing of health care [69] ,with end-of-life care potentially having an even stronger ethical justification for public funding.

Activity-based funding involves paying for health services according to a patient classification system which groups patients into clusters of patients expected to have a similar resource profile and which themselves are clinically meaningful [72] .Although initially developed to describe acute inpatient care, classification systems have now been developed for a range of types of health care [73].

A number of classification systems to describe palliative care patients either have been or are being developed, the most comprehensive being the Australian National Sub-acute and Non-acute Patient classification which has a palliative care sub-classification [74]. NHS England has proposed using a classification analogous to Australian National Sub-acute and Non-acute Patient classification [75].

The current version of the Australian classification (version 4, available at https://​www.​ihpa.​gov.​au/​what-we-do/​subacute-and-non-acute-care accessed 18 February 2018) divides palliative care patients into admitted and non-admitted branches, then into adult and paediatric branches. Patients are then split by ‘phase of care’: stable, unstable, deteriorating and terminal [76]. Further splits are made according to other measures such as ability to perform activities of daily living.

The classification has 20 admitted patient groups (14 adult, 6 paediatric) and 14 non-admitted groups (9 adult, 5 paediatric; with the unit of grouping being the phase of care [77]), although prices have only been assigned to some of the groups of the classification. The current relative prices (in A$) for each of the adult groups which have been assigned prices are shown in Table 1.

All activity classifications – and the Australian palliative care classification is no exception - assume homogeneous quality, an assumption belied in practice. Although it has been shown that it is possible to incorporate quality of care measures into an efficiency measure [78, 79], generally quality variation in practice has been addressed in payment policies through supplements or ‘pay for performance’ incentives, as discussed below.

An alternative activity-based funding approach is to pay for each death managed. This has the advantage of simplicity and would create a strong incentive on the palliative care organisation to manage care in the least cost setting. Such a model may enhance continuity of care and recognise that patients’ needs evolve over their end-of-life path.

Unfortunately, there is no published evidence of such a system in practice, nor has a risk-adjustment model been developed to recognise the different costs involved in managing different types of patients (e.g. paediatric patients compared to adult, patients with different levels of carer support, and patients with different family circumstances such as experiencing domestic violence or family estrangement, or with different levels of acceptance of the prognosis). However, it is possible that such a model could be developed for larger organisations that could manage the risk of patient variation. Performance benchmarks would also need to be developed to mitigate the risk of under-servicing by providers.

Payments based on the population covered provide incentives to keep populations healthy and manage care in the most appropriate setting [45]. Well-designed capitation payments could be used to address issues of geographic equity [48, 80]. Capitation (or per member per month) payments are common in a number of countries and generally cover all health care needs. However, in the United States, where capitation payments are a common payment approach, Medicare population-based funding arrangements ‘carve out’ hospice programs which are reimbursed on the typical mix of fee-for-service and per diem payments [81].

Specialist palliative care organisations could be paid on the basis of the population served and the expected number of deaths in that population. However, again there is no published risk-adjustment formula for such a payment model. As with a ‘per death’ activity-based model, a capitation based model involves a risk of underservicing, as there is no incentive on the organisation to maximise referrals or reach out to populations known to have a low take-up of palliative care [82]. Depending on how capitation rates are determined, it may enshrine historic inequitable service access.

Further, earlier involvement of palliative care teams leads to better outcomes [83‐88], leading to heterogeneity in palliative care activity as a result of the variability in length of life after initial involvement. In turn this creates heterogeneity in costs for capitated organisations and, because risk adjustment requires developing homogeneous categories, makes it harder to develop an adequate formula to set a fair capitation payment for providers.

Finally, in the absence of good measurement of the extent to which a population’s palliative care needs are met, if the palliative care experience is similar to primary care, capitation funding is likely to lead to lower levels of productivity and efficiency than activity-based funding [43, 89‐91].

Many countries and health systems are adopting add-on or bonus payments to reward better performance on policy-relevant metrics [45, 92]. Often referred to as ‘pay-for-performance’, some are more accurately described as paying providers for following preferred processes (such as referrals to specialists) or paying for additional data collection. The evidence about whether pay-for-performance works in practice is quite mixed [93‐97], although generally out-of-hospital programs which involve larger rewards or focused on process measures seem to be more likely to have positive impacts [98].

Poorly designed and conceptualised pay-for-performance can have quite negative impacts on care. The English experience with the Liverpool Care Path is an unfortunate one: what was proposed initially as a way of improving end of life care, became an impersonal, politicised and discredited approach which was abandoned [99‐104]. If pay-for-performance were to be incorporated into palliative care funding arrangements, it should adopt ethical design principles [105] and be rigorously evaluated [106].

Regardless of whether they are used in pay-for-performance arrangements, there are a number of palliative care measures which should be used in benchmarking or as contract performance criteria, including the Palliative Care Outcome Scale [107], and other proposed measures of palliative care impact which better account for the value of palliative care to families and carers [108, 109]. National benchmarking organisations such as the United States National Palliative Care Registry (see https://​registry.​capc.​org 18 February 2018) or the Australian Palliative Care Outcomes Collaboration (see http://​www.​pcoc.​org.​au/ accessed 18 February 2018) already collect and report on many such indicators. Appropriate measures might include measures of timeliness of responding to initial referrals and to requests for assistance (both in normal business hours and out-of-hours), emergency department visits in the last six months of life, whether expectations of place of care and place of death have been realised; and patient reported outcome measures [108, 110‐115].

Performance supplements should not only provide incentives for appropriate use of services, or meeting expectations of place of death, but should also address the all-important question of style of care provision. Patients consistently report that how care is provided to them is critical [116], and integrated care - care which ensures continuity - should be measured and valued in performance metrics [117‐120].

Over time, these measures could be used in the design of pay-for-performance supplements.

Early referral to palliative care is beneficial [83], in these circumstances palliative care should not be an alternative to curative treatment, but at least in the early stages, should be a complement. This will change over the course of the illness trajectory as the goals of treatment and care change [61].

Integration of services with a curative intent and provision of palliative care services involves change to contemporary service provider structures and processes [39, 121], made more difficult because of the multiple care settings involved in palliative care [66] cultural differences between different types of care and health system fragmentation [38]. Payment for palliative care should recognise the complementary nature of palliative care services and services with a curative intent, especially at the early stages of the disease trajectory.

Early involvement of palliative care in the disease trajectory will often mean that discussions about treatment choices and planning about palliative care will take place simultaneously with treatments with a curative intent. If there are separate funding streams for curatively-oriented and palliative care services, consideration will need to be given to how they mesh together.

There are two basic options. First, the two different funding systems could be designed and implemented in parallel, complementary funding existing side-by-side. In community or hospital-based ambulatory care, separate activity-based payments could be made for curative intent and palliative care services, even for overlapping payment periods (e.g. if a 30-day window is used as the payment episode. This approach has the advantage of simplicity.

The second, alternative approach, is for integrated funding -with some form of bundling of the palliative and curatively-oriented components of care. This would provide stronger incentives for care continuity between palliative and curative providers, albeit at the possible expense of administrative complexity. Integrated funding would probably stimulate discussions and negotiations about how curative-intent and palliative care teams will work together and how leadership will shift as the goals of care evolve [61]. In addition to the benefit that improved clarity of roles will bring to patient care, such discussion and negotiation will serve the pragmatic purpose of resolving how funding will be divided and also who will be accountable for which performance metrics (and how joint accountability will be handled). Bundled payment of this kind is a desirable direction for funding design in the medium to longer term.

An important interaction is between palliative care payment policy and payment arrangements for nursing homes and other residential aged care facilities [14]. A high proportion of older people die in aged care facilities [25, 64, 122], and the converse is also true, that the mortality rate in nursing homes is high, a natural consequence of the age and frailty of nursing home residents. Importantly, the risk of death in a nursing home can be predicted reasonably well [123]. Death in a nursing home is ‘normal’, and addressing the palliative care needs of residents, such as improved emotional and spiritual care, help with personal cleanliness, and treatment for pain [124], should be core business of the nursing home.

Depending on overall service system design, the additional services for dying patients might be incorporated in routine nursing home funding formulae, either as part of the needs-based funding or as a ‘death supplement’; provided by additional specialist palliative care services by community palliative care agencies and funded through that mechanism; or a combination of both approaches.

For inpatient episodes, standard care-bundling rules that are already used for activity-based funding would apply.

An important contemporary care issue is the provision of clinically futile care, or more accurately, non-beneficial care: care which has no reasonable expectation of leading to survival outside an acute setting [125‐127]. Greater clarity about the goals of care [61], involvement of palliative care services in treatment decisions at an early stage, and inclusion of measures of inappropriate care at the end-of life [114, 128], may all help to reduce the incidence of futile care. Although current economic incentives may play a role in perversely encouraging futile care [129], a host of other factors are also involved [130, 131]. In terms of palliative care payment design, the objective should be to encourage a ‘goals of care’ approach, although this may be better facilitated by organisational processes rather than financial incentives.