2.6 The economic challenges of palliative medicine

The compelling reasons for palliative care continue to be better symptom management, better advanced care planning and medically appropriate goal setting, and transitions to hospice care (Morrison and Meier 2004). Since the first edition of this book, over 80% of hospitals in the United States now have palliative care programmes (Center to Advance Palliative Care, 2014), the United Kingdom has strongly endorsed palliative care in the End of Life Care Strategy as a way to allow more patients to die at home with better quality of care (Department of Health, 2008), and the American Society of Clinical Oncology has endorsed concurrent palliative care for all seriously ill cancer outpatients (Smith et al., 2012a). Other new-found compelling reasons include better survival with hospice care, better survival with concurrent palliative care, and lower cost to hospitals and government and insurance funders. In every study to date, hospice and palliative care have been associated with equal or better survival and lower cost. A compilation of the available studies, in Fig. 2.6.1, shows that the group that receives palliative care has lower in-hospital or total costs.

In this chapter, we will define the various types of cost and clinical studies and discuss the available data about the economic challenges of palliative care, how to apply the available data, how to collect and present some useful and useable data, and new directions for research. We have summarized the key learning points in Box 2.6.1.

The United States is an example of uncontrolled health spending. Currently the United States spends 18% of the gross domestic product (GDP) on health care, which translates into $8000 per person per year; Canada and most other countries spend $4000–5000 per person per year and have equivalent health outcomes (Organisation for Economic Co-operation and Development, 2012). Insurance premiums for a family of four have risen from $6000 in 2000 to over $15,000 in 2011, with co-payments rising from $1000 to over $4000 (Claxton et al., 2010). There are over a million medical bankruptcies each year, with over half amongst insured people (Himmelstein et al., 2009). Almost one-quarter of Medicare spending (the single payer system for those over age 65) is in the last year of life and over 9% in the last month of life (Riley and Lubitz, 2010). There is great variation in end-of-life care in the United States, such that over half of patients die in the ICU at some hospitals, and the use of hospice care varies from 20% to 40%, when it should be over 60% (The Dartmouth Institute for Health Policy and Clinical Practice, 2014).

Other countries have different issues. In the United Kingdom, and much of the world, the dramatic rise of deaths in the hospital has driven up health-care costs. Before the Second World War this rarely happened but at present 58% of people die in hospital, which has increased the need for beds, with attendant costs (Gardiner et al., 2013). The barriers to better palliative care are familiar: lack of discussions about palliative care, avoidance of discussions to maintain false hope, and lack of mechanisms for a smooth transition (Gott et al., 2011). Over half of all complaints about medical care in the United Kingdom relate to poor end-of-life care, which the National Health Service (NHS) is addressing with the End of Life Care Programme. In the Middle East and countries like Portugal there has been an even more rapid shift, with over half of all patients now dying in hospital within the past generation, which is simply unsustainable financially. In most cultures, when people are given a choice, they prefer to die at home. Despite the large percentage of deaths in UK hospitals, current UK data show an increase in deaths at home and decline in deaths in hospital, which suggests that such programmes can be successful (Gomes et al., 2012).

The traditional ways of balancing cost and health outcomes are reviewed briefly in Table 2.6.1. In decades past, clinical outcomes alone were sufficient. But with palliative drugs now costing over $100,000 a year it is critical that we ascertain if the benefit is affordable. Some examples include palliative sipuleucel-T for metastatic castrate-resistant prostate cancer (survival benefit 4 months, cost US$90,000), and palliative treprostinil (Remodulin^®, US$120,000/year) and epoprostenol (Flolan^® US$100,000/year) for pulmonary hypertension, with better exercise tolerance, but no clear cut improvement in overall survival (Macchia et al., 2011). It is clear that the problems of balancing cost and effectiveness will persist.

When a drug is much better than any other treatment, for instance, imatinib mesylate (Gleevec^®) in the treatment of chronic myelogenous leukaemia or gastrointestinal stromal tumour, medical care decisions are clear-cut. It is a lifesaving drug but the cost ranges from $32,000 to $98,000 per year depending on dose. With such blockbuster drugs, the only real issue is how to make room in the budget; another issue is that patients who bear some or all of the financial cost will stop taking expensive drugs in an economic downturn (Kelley and Venook, 2010). The second and third drugs on the market for the same indication were priced even higher and to date there has been no price competition (Tuma, 2007), so alternative strategies on cost control will continue to be needed. The NHS in England is trying several, including 2 free months of the drug (paid for by the manufacturer), and the NHS only pays if the drug is successful (Rawlins et al., 2010; Yong et al., 2010).

Very few, if any, new drugs cost less than the treatment they replace. Knowing that a drug or treatment is inexpensive does not help decision-makers without some indication that a treatment actually works. For example, a widely used anti-nausea remedy lorazepam (Ativan^®), diphenhydramine (Benadryl^®) and haloperidol gel (‘ABH gel’) costs pennies for each dose, but is not absorbed, so cannot be effective (Smith et al., 2012b).

Cost minimization combines clinical results with cost. Two equally effective strategies are compared, and the one that has the least total cost is preferred. An example is drugs which are designed simply to replace expensive drugs with equivalent ones. In palliative care, Bruera and colleagues showed that methadone was equal to sustained-release morphine in the relief of pain (Bruera et al., 2004); the choice should depend on which is less expensive. This approach is not without flaws. First, it requires that there be a direct comparison of the drugs or technologies. Second, it may put too much emphasis on immediate treatments costs alone, and not enough on long-term costs; for instance, if ABH gel is less expensive than other more effective anti-nausea drugs, one hospitalization for refractory nausea would outweigh hundreds of pounds of savings on the cheaper but less effective remedy (Smith et al., 2012b).

Palliative and supportive care, including home and inpatient hospice care, is a good example of cost minimization. If we assume that survival is equal and the care is at least as good and costs are lower, then we should choose hospice and palliative care. Palliative care consultation improves appropriate hospice enrolment. Morrison et al. compared 1427 patients who did not have a palliative care consult with 296 patients who did, matched in every other respect, and found that if the palliative care team were involved, 30% of hospice eligible patients were discharged to hospice compared with just 1% of patients not seen by palliative care (Morrison et al., 2011). Patients enrolled in palliative care programmes at Sutter Health enrolled in hospice 47% of the time, compared to 20% of similar patients (Meyers et al., 2004, 2011). Appropriate transition to hospice not only improves care (Peppercorn et al., 2011), but is also associated with longer survival (Connor et al., 2007) and lower costs (Pyenson et al., 2004). The most recent data show that hospice saves US Medicare about US$2309 in the last year of life compared to conventional care, and that the savings increase the longer hospice is used (Taylor et al., 2007).

This method compares two known strategies in both effectiveness (years of additional life gained) and cost, then assigns a cost per additional year of life (LY) saved by the best (‘dominant’) strategy. For instance, adjuvant chemotherapy for a 45-year-old woman with early-stage breast cancer would be expensive, but would add 5.1 months of additional life at a cost of about $15,000 per year (Hillner and Smith, 2007). Expensive therapies can be ‘cost-effective’ too if they provide substantial benefit. For instance, trastuzumab (Herceptin^®) for adjuvant treatment of breast cancer patients is expensive (over $50,000 in all countries) but the 50% relative reduction in recurrence risk and 1.5 additional years for the average patient makes the cost-effectiveness ratio of $14,000 to $40,000/LY acceptable (Kievit et al., 2005).

There are several assumptions about this method that make it difficult to apply in the real world. First, it implies a limit to resources such that there must be some ‘cap’ on the cost-effectiveness ratio. The World Health Organization has defined ‘reasonable’ cost-effectiveness as interventions with a cost-effectiveness ratio that is less than the per capita GDP (Lillemoe et al., 1993; World Health Organization, 2001). Second, it assumes that a single week of added life for 52 persons equals 1 full year of life for another person. Third, it assumes that all people are somehow in the same health system and share a single set of values. Fourth, it requires someone to fund these difficult studies to generate cost-effectiveness data, when many patients are only interested in cure, doctors have a strong interest in benefit, and drug companies are interested in profit, and not usually social justice. Currently, the United Kingdom, Australia, Canada, and most countries use explicit cost-effectiveness to determine insurance coverage, but the United States does not.

Another issue is the question of ‘whose costs’ are being saved. In the United States and other countries with large hospital infrastructures and large third-party funder markets (e.g. Germany), one must keep in mind that money is changing hands several times. One can reduce hospital costs in ways that do not affect the amount paid to the hospital by the funders. In instances where funders do cut costs, typically this means lower revenue to providers (hospitals and doctors).

Palliative care brings an additional challenge in that very few palliative treatments make people live longer so that additional LYs are not saved. Nearly all the available studies show equal survival, with only one showing improved survival of 2.7 months with concurrent oncology and palliative care (Temel et al., 2010), with about $2000 less overall cost (Greer et al., 2012).

In cost-utility analysis the quality of life in a given state of disease or treatment (‘health state’), is converted to a utility ratio where 1 equals perfect health and 0 equals death. For example, a patient whose pain has been relieved may have a utility range of 0.90 of 1.00, whereas a patient with continued pain may have a utility range of 0.50. The utility ratio converts ‘cost-effectiveness’ into cost per quality-adjusted life year (QALY). For example, an intervention that costs $100,000 per year of life gained would be converted to $50,000 per QALY if the treatment doubled the patient’s utility score from 0.5 to 1.0. This intervention, however welcomed, would not save money but would cost an additional $50,000 to save an additional QALY with the intervention.

There are several difficulties with the cost-utility method. One is assigning proper utility ratios to the various health states. Controversy exists as to the most appropriate source of values: patients, healthcare workers, or the general population (society) due to differences in perspective. Patients and clinicians place greater emphasis on the issues that matter most to them, such as life, morbidity, or expense. For example, patients are more likely than clinicians to accept toxic treatments in exchange for minor clinical benefits, and a 1% chance of cure or 10% chance of symptom relief may be sufficient for them to choose treatment over supportive care (reviewed by Russell et al., 1996). The most objective source is probably lay people (jurors) whose money is being used (Russell et al., 1996; Smith et al., 2005); the difficulty lies in educating them properly in all aspects of a patient’s experience, for example, emotional turmoil, treatment toxicity, inconvenience, and other physical and psychosocial stressors.

A second problem is the magnitude of palliative care interventions, which is usually not enough to change overall (‘global’) quality of life or utility values. A recent exception was the randomized trial of palliative care alongside usual oncology care versus usual oncology care alone which showed a substantial difference in global quality of life (Zimmerman et al., 2012). Whether the magnitude is enough to change utility scores has not been studied.

Several recent cost-effectiveness or cost-utility studies have produced results that will be of use to decision-makers. Single-fraction radiation for painful bone metastases instead of the usual six treatments was as effective and cost substantially less (van den Hout et al., 2006). Shorter-fraction regimens for palliative treatment of lung cancer, 2 × 8 Gy versus 3 × 10 Gy, showed the opposite effect; the cost-utility ratio for the 10 × 3 Gy schedule versus the 2 × 8 Gy schedule was acceptable at $40,900/QALY. The longer regimen had an acceptable cost-effectiveness ratio because people lived longer with the 10 × 3 Gy treatment, offsetting the higher treatment cost (Mason et al., 2011). In countries with nationalized healthcare systems, the QALY is more important and more frequently used (e.g. in the United Kingdom with its NHS) than in other countries such as the United States where there is neither a single, nationalized healthcare system providing care, nor a single, nationalized payer.

This method compares two interventions then assigns a monetary value to the added clinical benefit of the best strategy, based on the overall economic productivity of an individual. Therefore, a treatment that costs $50,000 to prolong life 1 year would be acceptable if the person could produce $50,000 of economic worth in that year, but not if the person only produced $10,000 that year. This kind of analysis, while useful in business decisions, is rarely used in clinical studies because of the difficulties in assigning a monetary value to a human life, especially if that value is tied to an individual’s socioeconomic status. Since palliative care patients rarely produce income, even if their symptoms are well controlled, palliative care interventions would be at a distinct disadvantage.

This concept has been widely used as an economic argument for palliative care. If a patient can be moved from a US$5000/day ICU bed to a US$1500/day palliative care bed, with equal or better care that meets the patient’s goals, a US$3500 cost will have been avoided each day for each patient. Taken in aggregate, these costs can be substantial and more than offset palliative care programme costs. For instance, palliative care programmes have reported only a small profit, but annual savings of over US$1 million due to ‘cost avoidance’ defined as costs saved by transferring a patient from a high-cost venue to a low-cost one, such as from ICU to a regular hospital bed (Brumley et al., 2007; Gade et al., 2008; Smith and Cassel, 2009; Morrison et al., 2011). Similarly, not doing 50 computed tomography (CT) scans each year by careful matching of the goals of care with the diagnostic tests could save up to US$2500 × 50 = US$125,000. In a cost-based system such as national health insurance or diagnosis-related group (DRG) payment system, this would be a desired outcome; however, in a profit-based system this could represent substantial lost profit.

Cost avoidance is relatively easy to measure from usual data bases in US hospitals if there are comparable patients to match before the intervention, because billing systems record what was actually done. A simple historical analysis would show that more patients receive comfort care once a palliative care (PC) programme is established than under standard care. However it is much more difficult to quantify patient-level (or day-level) hospital cost savings in other countries, such as England, where fewer hospitals have invested in systems to do patient-level (or episode-level) cost accounting and cost allocation. In such countries or scenarios, the simplest way to collect data about cost savings is a notebook of ‘avoided tests’ after palliative care consult, for example, for the first 50 PC consult patients, 50 planned CTs or similar procedures were cancelled. Or to use hospital data on the frequency, duration, and intensity of hospitalizations per patient, comparing PC and non-PC patients.

Large insurers have noted the benefits of enhanced palliative care. In a series of landmark observational trials, researchers at Aetna introduced an expanded access ‘Compassionate Care Programme’. They identified cancer patients being treated for terminal conditions through their billing patterns, offered expanded access hospice without having to give up conventional treatment, and offered 24-hour daily help with symptoms by a nurse call-in line. Patients were also assigned a care manager. Hospice use doubled, hospice days doubled, hospital and ICU admissions were dramatically reduced (Spettell et al., 2009), and costs were reduced 22% in the last 40 days of life (Krakauer et al., 2011). Whether the change in resources and costs is due to the PC consult by itself, or the underlying reasons for the PC consult (change in medical goals, change in family or patient perspective, change in physician perspective) cannot be answered from the available data. Since these studies have now been replicated in numerous settings, it seems unlikely that the reduction in resources and costs is unrelated to PC consultation; but, one cannot ascribe all the change to the consultation itself. In the case of PC, the retrospective correlational studies correlate well with the available randomized clinical trials that show equal survival with substantial cost savings (Smith and Cassel, 2009).

There are a few important other considerations in the economics of PC. The first is perspective: whose money is being spent, on whom? In general, health economists take the societal viewpoint; that is, society has limited resources and wants to spend it to gain the most health for the most people. This runs directly counter to the individual viewpoint that only one person’s health—‘mine’—matters. The societal impact would also take into account work and time lost, and all the additional costs of care that are absorbed by families. Since studies often cannot take all these into account, a common compromise is to take the view of a health system or funder, for example, the NHS in the United Kingdom or Canada or a large insurer in the United States. As an example, a palliative care programme that accepts patients from the hospital will lower the hospital costs; if however, that palliative care programme transfers these patients to hospice, and the patients continue to use more care (IVs, pain medicines, diagnostic tests) then hospice costs increase even though the total costs decrease (Passik et al., 2004). Palliative care programmes for inpatients have consistently shown cost savings of 25–60% (Smith et al., 2003; Naik, 2004; Morrison et al., 2008, 2011).

A second concern is opportunity cost, simply defined as the additional revenue that could be gained if money had been used in a different way (this will only be relevant in health systems where third-party revenue is at stake). For instance, an ICU bed that is filled by someone who is not getting better may force the hospital into diversion. In a natural experiment, Oregon Health Center found that every hour of Emergency Room diversion cost the hospital system over $1100 per hour in profit, the opportunity cost of diversion. Transferring patients from the ICU to a more appropriate venue may give cost avoidance of $2500/day, and open that bed for an additional patient (McConnell et al., 2006). The opportunity cost of filled ICU beds is a sound financial concept that is understandable to most chief executives; if a palliative care programme assists in the transfer of 200 patients 2 days earlier, then the medical staff will have 400 more ICU bed days available.

A third concern is mortality in the hospital. If a hospital has a robust palliative care programme, will patients come there, die, and worsen the hospital mortality statistics? The two reports available strongly suggest that mortality is not increased in programmes that have the best palliative care systems (Elsayem et al., 2004; Cassel et al., 2010a).

A fourth issue is who funds palliative care specialists. In the United States, the Veterans Administration healthcare system and the Kaiser Permanente system, both of which have annual healthcare expenditure budgets and do not rely primarily on revenue from third parties, have invested millions of dollars in funding specialists, teams and palliative care units. In other US hospitals, a mixture of direct clinical revenue for specialists and a subsidy from the hospital together fund the teams (see Morrison et al., 2008). In terms of hospice provision, the US Medicare Hospice Benefit provides a reliable funding stream for patients in hospice and hospice touches over 40% of deaths in the United States (National Hospice and Palliative Care Organization, 2013). Other countries have struggled to provide an adequate funding stream for hospice—for example, in the United Kingdom, a large portion of hospice expenses are paid for through charity, and hospice touches far fewer of the deaths.

We have organized the available randomized clinical trial data into one table of important clinical effectiveness and cost data (Table 2.6.2). No trial has shown worse care or an increase in costs, or any sort of harm to patients.

The effectiveness data for palliative care is now sufficient for the American Society of Clinical Oncology to recommend concurrent palliative care alongside usual oncology care for all seriously cancer patients (Smith et al., 2012a). The most recent randomized clinical trial from Canada confirms and extends these findings (Zimmerman et al., 2012). The only negative randomized clinical trials used a model of consultant care outside the usual care team, such that only about 20% of recommendations were followed (Pantilat et al., 2010); we would not reject a trial of a new drug if compliance were only 20%.

Hospice was designed to improve care, not save money, but a good side effect has been cost control. The original Medicare hospice benefit was developed with the intent to improve care, not save money, but was required to be ‘revenue neutral’ (or, not cost Medicare any additional money). As noted above, hospice care saves money compared to usual care at the end of life, and the longer hospice is used, the more money is saved (Taylor et al., 007).

The available studies of palliative care strongly suggested substantial cost savings as shown in Fig. 2.6.1. Originally, Smith et al. reported over 60% hospital cost savings when patients died under the care of the palliative care team, matched to other patients who died in the hospital (Smith et al., 2003; Naik, 2004). The same group subsequently reported substantial reductions in all symptoms measured, if present, within 48 hours suggesting that the care was equal or better than usual care (Khatcheressian et al., 2005). Elsayem and colleagues at M.D. Anderson Cancer Center reported good symptom control accompanied by a dramatic decrease in costs (Elsayem et al., 2004). More recently, Morrison and colleagues showed that in New York State, hospitals with palliative care programmes increased appropriate referrals to hospice by tenfold, and saved over $5000 per admission (Naik, 2004; Morrison et al., 2008).

There are now several randomized trials that show substantial improvements in care associated with equal or reduced costs, as shown in Table 2.6.3. The data from the two randomized trials done within the Kaiser Permanente health maintenance organization were so compelling, with better care and $4500–7000 savings per person, that this is now standard care at all their hospitals (Brumley et al., 2007; Gade et al., 2008). The study of non-small cell lung cancer patients randomized at presentation to usual oncology care versus usual oncology care plus concurrent care demonstrated substantially improved quality of life, quality of care, better survival (Temel et al.,2010), and a cost saving of over $2000 per person (Greer et al., 2012). Patients with multiple sclerosis who received palliative care, cost the UK NHS £1200 less than their peers who did not receive it, even incorporating the cost of the care (Higginson et al., 2009) and had better clinical outcomes during and after the trial for up to 6 weeks (Currow et al., 2011).

A key question before collecting data is to decide beforehand on its use. Is the interested party interested in passing accreditation, improving market share, improving patient satisfaction, reducing health system costs, or increasing profits? The goal will dictate what data to collect and how to present it. In collecting and presenting the data, we have found that the simplest approach is the most effective. Most administrators want to know that the programme is successful, defined as providing good care, keeping busy, and not losing too much money.

We advise programmes to collect useful data that are being requested, and no more. Some common metrics are listed in Table 2.6.3. To collect more data than is actually used, takes time, effort, and money that would be better spent doing clinical care or research.

Presentation of the data should be as simple as possible. We have found that a simple ‘improved care at no extra cost’ approach is well received, as shown in Table 2.6.4. Again, most administrators will not be expecting palliative care to make a large profit. The goal of this report is fairly and accurately to reflect the activity and impact of the programme.

Palliative care is now well established as a necessary part of modern clinical care that can improve quality of life and quality of care and can save money. Additional research is needed to confirm results in different diseases, especially non-cancer (Farquhar et al., 2011; Higginson et al., 2011). As a field, we must determine a simple, easy to understand metric of the net benefits and burdens of palliative care in order for funders to justify new and continuing investments (Currow et al., 2011).

Palliative Care Leadership Center Curriculum (PCLC)—Center to Advance Palliative Care. The PCLC curriculum addresses issues relevant to each stage of palliative care programme development: structural, organizational, and financial. See ‘Module 3: Making the Financial Case’, available at <http://www.capc.org/palliative-care-leadership-initiative/curriculum>. This curriculum shows a programme how to build a financial case for palliative care, alongside the clinical case. There are core, custom, and advanced modules available.

Inter-Institutional Collaborating Network On End Of Life Care (IICN) website: ‘free access to over 4,000 pages of high-quality education materials about end-of-life care, palliative medicine, and hospice care, including the full text of several books’. This includes a very helpful guide to starting PC programmes: TriCentral Palliative Care Programme Toolkit, available at <http://www.growthhouse.org/palliative/>.