Economics of Palliative Care: Measuring the Full Value of an Intervention

Introduction

The importance of developing and strengthening the evidence base on palliative care is well established. Fiscal pressures and rising healthcare costs among seriously ill populations who often receive inappropriate care make health economics a key part of this research agenda.

Randomized trials reporting high-level evidence of survival effects have provided new insight into the value of palliative care interventions.^1,2 These studies also have important implications for the economics of palliative care as currently practiced and understood.

The purpose of this article is to highlight the most important of these implications, and to give food for thought to investigators planning studies to strengthen the economic evidence based on palliative care. There are fundamentally two challenges in economic evaluation of healthcare: the measurement of treatment effect on costs and measurement of effects on outcomes, and we discuss issues in both domains in turn.

Evaluating Intervention Effect on Costs

There are two established research designs for assessing association between an intervention and costs. There is the “forward-counting” approach, a conventional research design comparing outcomes of interest between groups from the point of intervention to the end of the study period or death. ³ And there is the “backward-counting” approach, which identifies a population of decedents and compares utilization between treatment and comparison groups for a specified period before death. ⁴

Treatment effect estimates derived from these two approaches are often discussed as if interchangeable, but there are important differences. An excellent illustration of these differences was provided recently, when both approaches were reported side by side for a randomized trial of palliative care for metastatic lung cancer from point of diagnosis. ⁵ The authors report that for the last 30 days of life, early palliative care was cheaper: intervention patients had $2,527 lower mean costs. But for the whole study period, intervention patients were on average $11,260 more expensive.

The apparent inconsistency is easily explained: patients who received early palliative care lived longer, and the costs of additional care eclipsed the savings accrued through lower intensity at end of life. Simply put, we have known for a long time that dying is an expensive business, but these data demonstrate that living is not cheap either.

These results provide a valuable new perspective for the economics of palliative care as currently practiced and understood.

First, it is clear that treatment effects derived using “backward-counting” and “forward-counting” approaches are not interchangeable. Specifically, “forward-counting” methods capture overall resource usage for the selected perspective and study period by definition, but “backward-counting” methods do so only on an assumption of treatment futility. If there is a difference in survival/mortality between treatment and comparison groups in either direction, then data counting backwards from death for a specified period no longer provide a valid comparison: any reported “cost savings” are neither reliable nor useful.

“Backward-counting” approaches retain other applications in describing end-of-life (EOL) experience, and in particular in capturing that experience at a population level when enrollment in a prospective study is not feasible. ⁶ In this framework, utilization may be a useful proxy for quality of life and death, but treatment effect on costs in last days of life is emphatically not the same thing as effect on total cost of care. Future economic studies in palliative care would be strengthened considerably by recognizing the distinction.

Second, growth of palliative care programs in the United States has been based partly on the incentive of cost savings for hospitals. Although evidence on the capacity of palliative care to reduce the cost of inpatient admissions is now clear, the same cannot be said for whole trajectories of care. There may be lower scope to reduce futile care or change treatment choices in an ongoing community care setting than in a resource-intensive inpatient admission. Moreover, the issue of perspective (whose costs are included) is critical: a purely hospital perspective risks missing transfer of costs from inpatient to outpatient services or informal networks, as well as the scope for hospital admission avoidance to reduce healthcare costs. And the growing evidence of survival effects for palliative care interventions suggests that, even where cost savings are realized, this may not result in lower healthcare expenditure overall, but rather in people living longer and spending more.

None of this is to say that the results published by Greer ⁵ undermine the economic case for palliative care. Rather, to date that case has been examined by evaluations asking only some of the relevant questions. Taken with the clinical results from the same study, ² the cost data published by Greer ⁵ show that palliative care improved both quality of life and survival, and for only a modest increase in costs. This is a proposition that demonstrates the value of the intervention. Renewed consideration must be given to the “effectiveness” side of the “cost-effectiveness” equation, so that palliative care is rewarded for its overall value and not reliant on demonstrating cost savings for continued development.

Measuring and Valuing Outcomes

The quality-adjusted life year (QALY) has become the gold standard measure of outcomes for economic analysis. ⁷ The method combines information about the impact of an intervention on both quantity and quality of life into a single metric and the superiority of one intervention over another is expressed as a gain in QALYs. The appeal in using the QALY framework is that it facilitates the comparison of health benefits across a range of interventions. Where policy makers are primarily interested in maximizing health outcomes using available resources, there priority is given to funding interventions that have the lowest cost per QALY.

In practice, QALY analyses of palliative care have been rare. This reflects both practical constraints and an ongoing debate about the appropriateness of the QALY framework in an EOL context.^7,8 However, even in the context of these issues, it is notable that only a fraction of more than 50 economic evaluations of EOL interventions have incorporated outcomes in any way. ⁹

Ongoing neglect of outcome measurement in economic studies is likely to have negative implications for the field of palliative care. Evidence of extended life expectancy has attracted considerable attention, but a sole focus on improved survival is unlikely to capture fully the relevant impact of an EOL intervention. For example, the use of life years gained as an outcome measure can be problematic in the context of palliative care given that the primary objective, more often than not, is improving quality of life. Where an intervention improves quality of life, an evaluation that measures only life expectancy will systematically underestimate its effectiveness.

Data on quality of life (utility), along with overall survival, are essential if conclusions are to be drawn about their effectiveness and value for money. The absence of such evidence makes it difficult to ensure patients are receiving the best possible care and limits scope for improving the quality of services. And in the context of limited resources, failure to generate evidence about the cost-effectiveness of palliative care could result in inefficient or inequitable funding decisions.

Although outcome measurement has a significant role to play in improving efficiency and access to palliative care services, there are some practical challenges to consider during study design. Gatekeeping by healthcare professionals or patients' family and friends may hinder recruitment to studies. In addition, patients may lack capacity to consent or participate, or be unable or unwilling to participate. In such cases, researchers may rely on information reported by proxies, but this approach may produce unreliable results given the potential for response bias on the part of caregivers. This was demonstrated in a recent study that found caregivers had a higher willingness to pay for life-extending treatments than patients did. ¹⁰ Although researchers may consider accepting proxy reports for observable or straightforward outcomes, differences in priorities and perceptions of quality of life between patients and caregivers emphasize the need to elicit preferences and outcomes directly from patients whenever possible.

This lack of clarity over ideal methodological or philosophical frameworks by no means precludes researchers from undertaking evaluations that include some element of outcome measurement. Rather than waiting for the perfect tool, the science would be better served by approaches that incorporate appropriate, if noncomprehensive, methods. A good starting point would be identifying the objective of the intervention, for example, addressing specific symptoms, reducing caregiver burden, or improving quality of life. The next step would be to choose outcome measures that reflect these goals—with recent publications providing guidance on the use of quality of life and other outcome measures in palliative care populations. ¹¹ Finally, the measures should be repeated as often as possible to capture the full impact of the intervention.

Summary

Current economic evidence on palliative and EOL care has been generated by studies with limited methodological scope. These limitations are understandable in the context of both palliative care's relative newness and the formidable challenges in EOL research, and the evidence provides a valuable foundation for future work.

In considering this future work and in the context of recent randomized trials reporting survival effects, it is timely to expand the methodological scope of studies. In particular, cost evaluations will ideally count forward from the point of intervention and using the broadest possible perspective. Where “backward-counting” methods continue to be used in evaluating intervention impact on cost of care, an evidence-based assumption of treatment futility ought to be declared. For studies to measure fully the value of interventions, it is essential that evaluations also assess outcomes, covering not only survival effects but also quality of life. This outcome measurement need not incorporate the theoretically optimal QALY approach to make a worthwhile contribution.