Evidence on the economic value of end-of-life and palliative care interventions: a narrative review of reviews

Background As the demand for palliative care increases, more information is needed on how efficient different types of palliative care models are for providing care to dying patients and their caregivers. Evidence on the economic value of treatments and interventions is key to informing resource allocation and ultimately improving the quality and efficiency of healthcare delivery. We assessed the available evidence on the economic value of palliative and end-of-life care interventions across various settings. Methods Reviews published between 2000 and 2019 were included. We included reviews that focused on cost-effectiveness, intervention costs and/or healthcare resource use. Two reviewers extracted data independently and in duplicate from the included studies. Data on the key characteristics of the studies were extracted, including the aim of the study, design, population, type of intervention and comparator, (cost-) effectiveness resource use, main findings and conclusions. Results A total of 43 reviews were included in the analysis. Overall, most evidence on cost-effectiveness relates to home-based interventions and suggests that they offer substantial savings to the health system, including a decrease in total healthcare costs, resource use and improvement in patient and caregivers’ outcomes. The evidence of interventions delivered across other settings was generally inconsistent. Conclusions Some palliative care models may contribute to dual improvement in quality of care via lower rates of aggressive medicalization in the last phase of life accompanied by a reduction in costs. Hospital-based palliative care interventions may improve patient outcomes, healthcare utilization and costs. There is a need for greater consistency in reporting outcome measures, the informal costs of caring, and costs associated with hospice.


Background
In the context of increasing pressures on health budgets, evidence on the economic value of treatments and interventions is key to informing resource allocation and improving the quality and efficiency of healthcare delivery. As an important share of healthcare expenditures occurs in the last months of life [1][2][3], a good understanding of the costs and benefits of care delivered during this period is particularly important. Whether dying patients are receiving "appropriate" care has been the focus of recent debates [4,5]. Notably, there are concerns that many patients are admitted to the hospital and undergo invasive procedures at the end-of-life without evidence of clinical benefits and improved quality of life (QoL); more than 80% of decedents are hospitalized at least once in the last 180 days of life in several countries [6]. In addition, research shows that such treatments are often counter to patient and/or caregiver preferences [7,8]. Palliative and end-of-life care services and interventions have the potential to improve the quality, appropriateness and efficiency of care provided at the end-of-life from both the perspective of patients and their families and the health system as a whole [9]. Palliative care has been associated with improved patient outcomes, such as pain and symptom management [10], improved communication [11], higher satisfaction with care [12], improved QoL [13], reduced healthcare costs [14,15] and an increased likelihood of dying in one's preferred place [16]. Additionally, evidence suggests that such models of care could reduce healthcare resource use, including emergency hospital admissions, length of hospital stays (LOS) and ICU admissions [17][18][19].
As the demand for palliative care services increases, more information is needed on how effective and efficient different types of palliative care models are for providing care to patients and their caregivers. In particular, high-quality economic evidence regarding end-of-life and palliative care interventions is crucial to adequately support and develop new models of care that have the potential to improve the experience of patients and their relatives, avoid unnecessary treatments, and potentially reduce healthcare costs. Prior research has shown that the quality of economic evidence in the area is mixed [20]; it is often limited to assessments of reduced healthcare utilization, without proper measurement of intervention costs and/or valuation of benefits [21]. While several reviews on the topic of palliative care have been published in recent years, few have assessed the effectiveness and cost-effectiveness of palliative care interventions. Existing reviews have mainly focused on particular populations (e.g., cancer) [22], specific interventions (e.g., advance care planning, ACP) [23], or settings (e.g., home-setting) [24]. A recent review of international evidence provided important insights into a wide range of palliative care models. It found that irrespective of setting or patient characteristics, palliative care seem to be beneficial to the patients and may reduce total healthcare costs [25]. However, a more detailed assessment of effectiveness and cost-effectiveness has not been identified.
The aim of this systematic review is to provide a comprehensive overview of the available evidence published between 2000 and 2019 on the economic value of palliative and end-of-life care interventions across various settings (i.e., community, home, hospital, etc.). Due to growing evidence from multiple systematic reviews, a review of reviews is considered the most appropriate approach for bringing together the available evidence together on different interventions [26,27].
We focused our search on published reviews that contained at least some economic evidence, including measures of cost-effectiveness, intervention costs and/or impact on healthcare use. Such knowledge might be valuable to support health policy makers in making resource allocation decisions and commissioning palliative care services.

Methods
This systematic review was performed according to PRIS MA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [28]. This review is guided by Smith et al [26] methodology for conducting systematic reviews of systematic reviews. This review is registered in PROSPERO (registration number: CRD42018110910).

Search strategies
We performed systematic searches using Ovid in the following databases: Medline, Embase, PubMed, Cinahl, Psychinfo, Scisearch and Cochrane. We searched reviews published between January 2000 and September 24, 2019, using both MESH and keywords in the fields of palliative and end-of-life care and health economics ( Table 1). Reference lists of included reviews were screened for relevant studies. We also performed Google Scholar searches using key search terms ('systematic review', 'palliative care', 'cost-effectiveness').

Study selection
We included systematic reviews focusing on quantitative assessments of intervention effectiveness that contained economic evidence on cost-effectiveness, intervention costs and/or resource use. We included interventions  [25], palliative and end-of-life care models/interventions were defined as "any structured care model involving multiple components, including who delivers (e.g., professionals, caregivers) the intervention (specialist or generalist palliative care), where (e.g., hospital), to whom (care recipients), when (i.e., timing and duration), how (e.g., face to face) and for what purpose (i.e., expected outcomes)." Details on inclusion and exclusion criteria are reported in Table 2.

Study selection
Two reviewers (BO, XL) screened independently and in duplicate all titles and abstracts using the inclusion and exclusion criteria (Table 1). A third reviewer (JM) provided arbitration in the event of disagreement. The same reviewers assessed the full texts of potentially relevant for eligibility. Any disagreement was resolved by discussion and by consulting the third reviewer.

Assessment of quality of included studies
We assessed the methodological quality of included studies using the AMSTAR. We assessed the methodological quality of included studies using the AMSTAR 2 (A Measurement Tool to Assess Reviews) which is a critical appraisal tool for systematic reviews that include randomized or non-randomised studies of healthcare interventions, or both [29]. The reviews were rated as low, medium quality or high quality.

Data extraction
Two authors (XL, BO) extracted data using a predefined extraction form, and any discrepancies were resolved by a third reviewer. Data on the key characteristics of the studies were extracted, including information about the aim of the review, design, population, type of intervention and comparator, information on (cost) effectiveness and resource use, and main findings and conclusions.

Data synthesis and analysis
The reviews were too heterogeneous in terms of disease, setting and outcomes measured to perform a metaanalysis; a narrative synthesis was considered more appropriate. We organized the evidence according to the main outcomes reported, i.e.: (1) cost-effectiveness, (2) costs, and (3) resource use. In each section, we discuss evidence by intervention setting and/or type.

Results
The initial searches identified 435reviews. Finally, 43 review articles were included in the analysis. Figure 1 provides details on searches and study selection. Over half of the reviews (Table 3) were published between 2011 and 2018 and included predominantly cancer patients. A wide range of palliative and end-of-life care models were described and comparator models were often described as usual care, which varied across studies, or was not specified. Most of the reviews (n = 21) involved interventions delivered in multiple settings [20-23, 25, 30-50]. Nine reviews examined interventions in the hospital setting [20,[50][51][52][53][54][55][56][57][58]. Of these, 3 focused on ICU [51][52][53], and 1 on surgical patients [57]. Three reviews involved Reviews not focusing on palliative care. Reviews focusing on a single procedure (i.e., a focus on treatment rather than palliative care) such as chemotherapy, radiotherapy, surgery and other curative strategies. Reviews focusing on palliative care, but not on interventions.
Reviews focusing on treatment rather than palliative care.
Economic outcomes: Information on costeffectiveness and costs Reviews including any information regarding costs, whether cost, cost for caregivers cost of intervention, hospitalization, or general costs at end-of-life.
Reviews not including quantitative cost elements.
Other outcomes Resource use -Language Written in English. Not written in English.
Most of the included reviews were rated as having a moderate quality (Table 3).

Evidence synthesis Cost-effectiveness
In this section, we focus on studies that measured both the costs and intervention outcomes. Ideally, these components would be combined in an incremental costeffectiveness framework, following best-practice economic evaluation guidelines [64]. Such evidence is rare [20,21]; therefore, the evidence shown below (Table 4) also includes studies that measured both components independently. Overall, only 10 reviews included evidence on cost-effectiveness of the interventions, while the remaining 33 reviews included details on intervention costs and/or resource use.

Home-based interventions
Overall, most cost-effectiveness evidence relates to home-based interventions. One review [32] showed that home-based services may reduce resource use and costs and improve pain management and increase death outside the hospital. Furthermore, home-based teams may generate substantial savings for the health system and improve patient outcomes. A review [47] identified 2 studies [65,66] providing evidence of the cost-effectiveness of home-based teams for specific patient groups (i.e., cancer AIDS). A review [59] identified 6 studies of homebased palliative care, but only 2 of them provided clear evidence of cost-effectiveness. Finally, various models showed potential benefits for patients and caregivers and reduction in total healthcare costs; most such evidence pertains to home-based interventions [25].

Other settings
The remaining evidence relates to the cost-effectiveness of palliative care across multiple settings [20,21,25,37,44,54,61]. A review on [61] palliative care interventions outside the home, identified two studies that included an economic evaluation [67,68], but only one [68] performed a full cost-effectiveness analysis of the effect of psychosocial support on survival in women with breast cancer. No significant difference in healthcare resource use between the intervention and control groups was            [20] identified one costeffectiveness study on short-term palliative care in multiple sclerosis patients; it suggested potential costeffectiveness when caregiver burden was used as the main outcome measure [69]. Salamanca et al. (2018) [37] provided inconclusive evidence on the costeffectiveness of clinical nurse specialists (CNSs): of 13 economic analyses, 7 reported improvement in health economic outcomes, mainly in cardiac patients. CNSs were cost-effective in several countries. In Austria home-based CNSs were both cost-effective and less expensive than standard care for HF patients [70]. In the United Kingdom, CNSs for heart disease patients generated an incremental cost of £13,158 per qualityadjusted life year (QALY) gained compared to the control group [71]. In England CNS intervention improved QoL and reduced readmissions and costs in HF patients [72]. Similarly, a study in the Netherlands CNSs improved health outcomes of patients with severe HF, with a slight increase in costs [73]. Douglas et al. [44]  Two other reviews reported some evidence on the benefits of hospital-based palliative care consultations on patient outcomes, healthcare utilization and cost [54,75]. Finally, a review by Dixon et al. (2015 [21] found no evidence on cost-effectiveness of ACP.

Cost analyses
The following studies reported on the cost and resource use implications of palliative care with various degrees of comprehensiveness. Some were limited to intervention cost while others compared total healthcare costs between an intervention and a control group. Five reviews reported the cost of home-based palliative interventions [31,76] [24,59,60]. Among those, 1 [76] found significant improvement in healthcare use or costs in 16 of 21 (76%) studies that reported those outcomes; the majority of evidence was obtained from quasiexperimental designs. Similarly, Gomes et al. (2013) [59] showed lower costs for home-based group compared with usual care (18 to 35%). Institutional and noninstitutional costs, medications costs and in 1 study, informal care costs, were reported. Other reviews found relatively low-quality and inconsistent evidence that home-based interventions might reduce healthcare costs [24,60,77], and the costs included varied across studies.
The evidence regarding the impact of ACP on costs is limited. One review [21] showed savings ranging from USD 64,827 to USD 56,700 in total healthcare costs over 6 months. Klinger et al. (2016) [23] reported net cost savings ranging from USD 1041 to USD 64,830 per patient (relative cost reduction 68 to 5%), with the greatest reductions found for sicker patients. Martin et al. (2016) [62] found limited evidence of impact on costs in 1 study [80] suggesting a significant decrease in both hospital costs and total healthcare costs [81]. Another study [63] identified cost savings for patients with severe dementia who received ACP in the community [82] and for care home residents [80]. Another study [83] on ACP for home health staff reported showed significant reductions in hospitalisation rate, days and healthcare cost.

Impact on resource use
Five studies [24, 30-32, 36, 38, 60, 76] examined the impact of home-based interventions on resource use. Bainbridge et al. (2016) [76] showed a significant reduction in healthcare use in the home care group. Other reviews found inconclusive evidence regarding the impact of home-based interventions on resource use, including general healthcare utilization [32], LOS and hospitalizations [24,31,38,60]. Candy et al. (2011) [32] showed that patients receiving hospice care at home had lower healthcare utilization. The strength of this evidence, however, was limited because few of the findings originated from RCTs. Davis et al. (2015) [31] showed that early integration of outpatient and home palliative care may reduce LOS and hospitalizations. In another study [60], it was unclear whether home-based care impacts hospital admissions. Admission to hospital while receiving home-based care varied among trials. One study [38] found that palliative interventions (65%, 11 of 17) reduced hospital use and that home interventions were more effective than other interventions in cancer, COPD and dementia patients.
Evidence regarding the impact of hospital-based palliative care interventions was scarce. One review [58] suggested that hospital-based palliative care may have a positive impact on resource use (i.e., reduced hospitalizations). Two reviews showed a reduction in ICU LOS for intervention patients receiving inpatient palliative care consultations compared with control patients [50,51]. Two other reviews found that communication interventions may decrease healthcare resource use in multiple dimensions, including duration of mechanical ventilation [53] and hospital and ICU LOS [41,52,53]. Waller et al. (2017) [54] found limited evidence that hospital-based interventions improved processes and outcomes, with only 1 study reporting longer hospice stays.
Nine reviews reported inconsistent findings on resource use across multiple settings. Three reviews found limited evidence that palliative care interventions, including case management [20], decreased resource use, [20,43,45] and improved hospice use. The evidence on effects of specialized palliative care is sparse, with 3 reviews reporting fewer hospitalizations [22,33] and higher hospice use [42] in the intervention group. Some evidence showed that CNS might reduce hospitalizations, rehospitalizations/admissions, and LOS [37,44,78]. Outpatient palliative care decreased healthcare utilization, including ambulatory care, physician visits, hospitalizations, LOS and readmissions [49]. Datla et al. (2019) [79] reported reduction in health care resource use as a result of multi-disciplinary palliative interventions in patients with symptomatic heart failure.
ACP and palliative care interventions consistently demonstrated a trend toward decreased ICU admissions and reduced ICU and hospital LOS [35,83]. The mean relative risk reduction in the percentage of admissions with ACP and palliative care consultations was 37%. These findings were supported by two other reviews showing decreased hospitalization rates [62] among patients receiving ACP, including dementia patients [63].

Discussion
The strongest evidence of cost-effectiveness relates to home-based interventions, suggesting high potential efficiency gains for the health system through a decrease in total healthcare costs and resource use and improvements in patient and caregiver outcomes. The evidence of interventions in other settings was inconsistent. There was some evidence suggesting potential benefits of CNS interventions for patients with heart disease. Overall, the majority of studies focused on cancer patients. One explanation is that disease trajectories for cancer are less heterogeneous than for other diseases [80,81] and that the end-of-life phase is easier to predict in this patient group [84].
Our findings reveal a potential for reduction in direct healthcare costs and resource use mainly in home-based programs. Inpatient palliative care consultations were consistently associated with fewer hospitalizations, readmissions and reduced costs. We found some evidence of the effectiveness of ACP for reducing resource use. Such interventions have been considered important, especially for dementia patients [85]; however, further research is needed regarding the effectiveness and cost-effectiveness of ACP interventions.
Overall, most studies overlooked important costs, including out-of-pocket, hospice and informal care costs. It is important that future studies systematically report these to get a comprehensive understanding of the health economic implications of palliative and end-oflife care, despite the complexity of their measurement and valuation [86]. Another issue is the lack of consistency in outcomes measurement that hinders comparability across studies. While quality-adjusted life years (QALYs) are widely used in cost-effectiveness, they are limited in their ability to capture the benefits of endof-life and palliative care. Tools that have been developed to measure wellbeing in the broader sense should be more widely used [87,88]. More generally, future economic studies in the area should more systematically follow existing economic evaluation guidelines and adopt a societal perspective.

Strengths and limitations
An important strength of this study is the focus on the economic value of a wide range of interventions. The effectiveness of palliative care interventions has been studied previously, but the focus on cost-effectiveness, resource use and costs has been scarce [25,89]. By conducting this review of reviews we believe that we have identified the main sources of (high-quality) evidence. This work has several limitations. One key challenge was related to the wide range of settings and intervention types included as well as the diversity of reporting of findings across reviews (e.g., outcome measurement, costing approaches, etc.). While a broad inclusion was a deliberate feature of our study, it made the synthesis of evidence difficult. Most approaches focused on the hospital setting. Hospital costs are known to escalate in the last phase of life, with questionable benefit to patients [90,91]. Furthermore, the evidence is subject to a range of methodological limitations as discussed above. Also, as evidence often comes from retrospective data important information (e.g., patient preferences) were not be available [92].

Conclusions
Home-based palliative care may contribute to a dual improvement in quality of care by reducing aggressive medicalization end-of-life and concomitantly reducing costs. Hospital-based palliative care interventions may improve patient outcomes, healthcare utilization and costs. Evidence regarding other approaches is less conclusive. This study provides a foundation for discussions between policy makers and clinical services managers regarding resource allocation and the commissioning of palliative care services. There is a need for greater consistency in costs and outcome measures reporting, including breadth of capture (e.g., informal care costs, hospice costs, better-suited measure of patient outcomes, etc.). In addition, as RCTs are rarely feasible in the area, a particular focus on the quality of observational, and quasi-experimental evidence, is warranted. On the clinical side, further examination of who can deliver interventions is important for clinical practicecan generalists (non palliative care specialists) deliver these interventions as well as specialists? What are the enablers and barriers to successful interventions? Also, information on timing in relation to death would be useful to understand more fully the impact and resource implications of palliative care interventions. Research that addresses these practical issues as well as examining effectiveness and cost would help policy makers and clinical teams know where to invest resources. Better patient care can be provided at lower costs for people in the last phase of life with intentional palliative care interventions.
Abbreviations ACP: Advance care planning; CNS: Clinical nurse specialist interventions; ICU: Intensive care unit; LOS: Length of stay; QOL: Quality of life