Introduction
The escalating costs of healthcare in the United States demand innovative solutions that not only enhance patient care but also offer substantial cost savings. Palliative care, particularly through hospital consultation programs, has emerged as a vital approach to achieving this dual goal. While often associated with improved quality of life and patient-centered care at the end-of-life, the economic benefits of palliative care consultation programs are increasingly recognized. These programs, designed to support patients and families facing serious illnesses, are demonstrating significant potential in reducing healthcare expenditures while aligning care with patient values and preferences.
A pivotal report from the Institute of Medicine, “Dying in America,” underscored the critical need for integrating palliative care and advance care planning (ACP) into the US healthcare framework. This integration is not merely about compassionate care; it’s also about responsible resource management. Studies have consistently shown that palliative services can lead to more appropriate care, enhanced symptom management, and, crucially, reduced costs. However, awareness and access to these beneficial services remain a challenge, often leaving patients and their families unprepared to navigate complex end-of-life decisions.
Advance Care Planning (ACP) plays a crucial role within the broader palliative care model. By facilitating structured conversations, ACP empowers patients to articulate their values and preferences for future medical care. This proactive approach ensures that patient wishes are documented, understood, and respected as their health conditions evolve. While ACP itself can be delivered in various settings, it is often a key component of comprehensive palliative care programs, including hospital-based consultation services. The presence of ACP has been linked to improved care quality and decreased stress for caregivers, alongside its potential impact on healthcare utilization and costs. With the increasing engagement of physicians, partly driven by CMS billing options for ACP introduced in 2016, the integration of ACP within palliative care is poised to expand further.
The core components of ACP, such as completing Healthcare Power of Attorney (HCPOA) and Practitioner Orders for Life-Sustaining Treatment (POLST) documents, are instrumental in clarifying patient preferences regarding resuscitation and other critical medical decisions across different care settings. While the ACP discussion itself is not a medical order, it serves as a crucial guide for families and healthcare providers, ensuring informed choices that align with patient values as illness progresses. A frequently cited aspiration of ACP and palliative care is to enable more patients to experience death outside of the hospital environment, reflecting a preference for home or hospice care.
Inpatient palliative care consultation has a well-established track record of delivering both cost and utilization benefits. Furthermore, the growing field of outpatient palliative care is also demonstrating promising results in cost reduction. While the impact of hospice services on costs has shown varied outcomes, the overall evidence points towards the economic advantages of early and integrated palliative care approaches. However, the specific financial impact of outpatient ACP, particularly from a system-wide perspective encompassing both inpatient and outpatient costs, remains an area requiring further investigation. Existing studies often differ in patient populations, disease categories, and the types of costs analyzed, highlighting the need for more comprehensive research.
This study delves into the association between outpatient ACP and advance directive documentation rates, healthcare utilization, and costs. By examining a cohort of patients who died and comparing them with matched controls within an Accountable Care Organization (ACO) framework, this research aims to contribute to the growing body of evidence supporting the cost savings associated with palliative care consultation programs. The ACO under study has made substantial investments in expanding ACP access and improving facilitator qualifications, providing a robust setting to evaluate the real-world impact of these programs.
FIG. 1.
Cost comparison. Costs were compared with a difference-in-difference method, illustrating the cost benefits of advance care planning.
Methods
This research employed a case-control study design, comparing patients who underwent ACP (cases) with matched controls who did not. Data was drawn from three primary sources: (1) ACO data, (2) an ACP facilitator logging database (IDEA), and (3) EPIC Electronic Medical Record (EMR) data. The study received approval from the regional institutional review board and CMS for the use of ACO data for research purposes.
The ACO data spanned from January 2013 to April 2016, covering a large multisite healthcare system with 11 hospitals and 75 outpatient practices serving a population of nearly 3 million individuals. During the study period, the system maintained a network of over 150 active ACP facilitators. The study population included Medicare beneficiaries attributed to the ACO, determined by CMS attribution algorithms.
The IDEA ACP facilitator logging database, a HIPAA-compliant web-based system, was utilized to track ACP discussions conducted by trained facilitators. All facilitators underwent comprehensive training, including didactic sessions and standardized patient simulations, to ensure consistent and high-quality ACP discussions.
The EPIC EMR served as the repository for ACP documentation. ACP discussion records, HCPOA, and POLST forms were scanned into the EMR under the advanced directives tab. The EMR also provided patient problem lists, comorbidity data for matching, and information on inpatient palliative care consultations.
Cases with ACP were identified by cross-referencing ACO data with the ACP logging database (Fig. 1). A 1:1 matching algorithm (Appendix Fig. 1) was employed, using variables such as ACP date (snapshot index date for controls), gender, race, age, Deyo-Charlson Comorbidity Index (CCI), and an internally developed utilization risk score. Further details on the matching approach are available in Appendix Figure 2.
Outcome variables included the presence of HCPOA and POLST forms, healthcare costs, and utilization metrics. The focus on HCPOA and POLST forms was to assess the documentation of patient surrogate decision-makers and end-of-life preferences. The study acknowledged that advance directives could originate from sources outside the ACP process, categorizing patients into those with no ADs, ADs from external sources, ADs from within the system but outside ACP, and ADs completed as part of the system’s ACP process.
Chart reviews were conducted to verify AD types and relevant preferences, with adjudication by co-authors (W.B., L.F.) for unclear cases. Exclusions were made for cases with no ACP discussion documentation, ACP discussions during hospice admission, or improper group assignment based on log data and ACP discussion dates. Corresponding matches from excluded charts were also removed.
Healthcare cost and utilization data encompassed total healthcare expenditure, ED visits, hospital admissions, 30-day readmissions, hospital days, ICU use, hospice use, and skilled nursing facility (SNF) and home health agency (HHA) utilization. Data was sourced from the ACO dataset, including all Medicare claims within and outside the system, covering all age groups insured by Medicare. Total expenditure included all Medicare payments.
Statistical Analysis
Demographic summaries were presented using means and standard deviations for continuous variables and frequencies and percentages for categorical variables. Unpaired t-tests and Chi-squared tests were used for comparisons of continuous and categorical variables, respectively. Logit models, adjusted for age, gender, race, CMS risk score, and comorbidities (CHF, COPD, ESRD, cancer), were used to analyze HCPOA and POLST presence.
Difference-in-difference analysis, using generalized linear models with log link and gamma distribution (expenditure) or Poisson distribution (utilizations), adjusted for the same covariates, was employed for utilization and cost comparisons. This method was chosen to account for baseline differences in illness severity between the ACP and control groups. Data for utilization and cost were collected for the 12-month periods before ACP/match date and before death. Multiple imputation was used to address missing data for patients without a full 12-month period between ACP/match date and death, with ten imputed values per missing observation. All costs were converted to 2016 US dollars using the Medical Component of the Consumer Price Index. Sensitivity analyses were conducted to validate the results, including analyses using only complete data and six-month data periods. A statistical significance level of 0.05 was used for all tests, and Stata version 12 was used for all analyses.
Return on Investment (ROI) was calculated by considering ACP program costs (start-up, administration, education, facilitator time) and the observed Medicare expenditure benefits. Facilitator time was estimated at one hour per ACP discussion, with salary costs based on a weighted average of interprofessional facilitators.
Results
The study cohort comprised 325 ACP cases and 325 matched controls, with a mean age of 81 years and balanced gender distribution (51.1% female, 48.9% male). Table 1 details demographics and comorbidities, revealing slight differences in CCI, predicted utilization risk, and proportions of CHF and COPD between the groups. Rural-urban code distributions also showed statistical differences.
Demographics of ACP Cases and Matched Controls
| Continuous variables | ACP (N = 325) | Match (N = 325) | |
|---|---|---|---|
| age | Mean (SD) | 81.13 (8.54) | 81.18 (8.19) |
| CMS risk score | Mean (SD) | 1.91 (1.16) | 1.68 (1.12) |
| Predicted utilization rank | Mean (SD) | 0.69 (0.25) | 0.66 (0.25) |
| CCI | Mean (SD) | 9.36 (2.95) | 9.13 (2.68) |
| Categorical variables | N (%) | N (%) | |
|---|---|---|---|
| Gender | Female | 166 (51.1) | 166 (51.1) |
| Male | 159 (48.9) | 159 (48.9) | |
| Race | White or Caucasian | 319 (98.2) | 319 (98.2) |
| Black | 6 (1.8) | 6 (1.8) | |
| CHF | No | 157 (48.3) | 182 (56.0) |
| Yes | 168 (51.7) | 143 (44.0) | |
| COPD | No | 125 (38.5) | 152 (46.8) |
| Yes | 200 (61.5) | 173 (53.2) | |
| Cancer | No | 318 (97.8) | 315 (96.9) |
| Yes | 7 (2.2) | 10 (3.1) | |
| ESRD | No | 185 (56.9) | 201 (61.8) |
| Yes | 140 (43.1) | 124 (38.2) | |
| Medicare beneficiary entitlement reason | Aged without ESRD | 307 (94.5) | 314 (96.6) |
| Aged with ESRD | 13 (4.0) | 8 (2.5) | |
| Disabled without ESRD | 4 (1.2) | 3 (0.9) | |
| ESRD only | 1 (0.3) | 0 (0.0) | |
| USDA rural-urban continuum codes | Metro: 1M+ population | 15 (4.6) | 11 (3.4) |
| Metro: 250K-1M population | 144 (44.3) | 157 (48.3) | |
| Metro: <250K population | 92 (28.3) | 54 (16.6) | |
| Nonmetro: urban 20K+, adjacent metro | 59 (18.2) | 77 (23.7) | |
| Nonmetro: urban 20K+, not adjacent metro | 0 (0.0) | 1 (0.3) | |
| Nonmetro: urban 2.5K-19.9K, adjacent metro | 15 (4.6) | 18 (5.5) | |
| Nonmetro: urban 2.5K-19.9K, not adjacent metro | 0 (0.0) | 3 (0.9) | |
| Nonmetro: rural, adjacent metro | 0 (0.0) | 4 (1.2) |
In the 12 months preceding death, HCPOA completion was significantly higher in the ACP group (98.5%) compared to controls (74.8%) (OR: 21.6, 95% CI 8.6–54.1, p < 0.001). POLST completion was also higher in the ACP group (52.9% vs 44.6%), although the difference was less pronounced (OR 1.40, 95% CI 1.02–1.90, p = 0.034). Adjusted results mirrored these findings (HCPOA OR 22.6, POLST OR 1.39). Appendix Fig. 3 visually represents these documentation rates.
Appendix Fig. 3.
Documentation of Health Care Power of Attorney (HCPOA) and Practitioner Orders for Life Sustaining Treatment (POLST) in the twelve months prior to death.
Total costs, both unadjusted and adjusted, are shown in Figure 2. Table 2 provides detailed utilization and cost data. Adjusted analyses revealed that ACP cases experienced fewer inpatient admissions (-0.37, 95% CI -0.66 to -0.08) and inpatient days (-3.66, 95% CI -6.23 to -1.09). No significant differences were observed in hospice use, hospice days, SNF use, HHA use, 30-day readmissions, or ED visits. Overall adjusted costs were significantly lower in the ACP group by $9,500 (95% CI -$16,207 to -$2,793).
FIG. 2.
Cost comparison between ACP and control groups, demonstrating significant cost savings associated with ACP.
Utilization and Costs Using a Difference in Difference Method
| ACP (N = 325) | Match (N = 325) | Unadjusted | Adjusted | |
|---|---|---|---|---|
| Pre-12 months before ACP date A | Last 12 months before death B | Pre-12 months before match date C | Last 12 months before death D | |
| Total expenditure | $12,252 | $31,961 | $9936 | $38,544 |
| 95% CI | ||||
| Utilization | ||||
| Inpatient admissions | 0.47 | 1.42 | 0.28 | 1.55 |
| 95% CI | ||||
| Inpatient days | 3.07 | 9.29 | 1.90 | 11.47 |
| 95% CI | ||||
| ICU days (inpatient) | 0.06 | 1.87 | 0.01 | 1.98 |
| 95% CI | ||||
| Hospice use | 0.04 | 1.20 | 0.12 | 1.25 |
| 95% CI | ||||
| Hospice days | 0.90 | 23.85 | 3.33 | 26.81 |
| 95% CI | ||||
| SNF use | 0.18 | 0.80 | 0.27 | 1.01 |
| 95% CI | ||||
| HHA use | 0.32 | 0.60 | 0.21 | 0.56 |
| 95% CI | ||||
| 30-day readmissions | 0.11 | 0.23 | 0.03 | 0.23 |
| 95% CI | ||||
| ED visits | 0.10 | 1.62 | 0.09 | 1.55 |
| 95% CI |
Sensitivity analyses, using complete data and six-month data periods, corroborated the primary findings, demonstrating consistent cost savings in the ACP group (Appendix Table 1, Appendix Table 2). Specifically, adjusted annual costs in the complete data sensitivity analysis were $10,433 lower in the ACP group (95% CI −$18,467 −$2,398).
Inpatient palliative care consultation rates did not differ significantly between groups (ACP 30% vs controls 25%, p = 0.16). However, patients receiving inpatient palliative care consults incurred higher costs ($10,394, 95% CI $5,575–$15,213). Even after adjusting for inpatient palliative care consults, ACP remained associated with lower costs ($-8,963, 95% CI -$15,411 to -$2,515).
The ACP program demonstrated a 104% ROI. With an intervention cost of $1,515,170, the program generated $3,087,500 in Medicare expenditure savings, resulting in a net saving of $1,572,330 (Table 3).
Return on Investment of the Advance Care Planning Program
| 2013 | 2014 | 2015 | 2016 | Total | ||
|---|---|---|---|---|---|---|
| Cost of ACP | ||||||
| Education program cost ($) | A | $43,980 | $42,932 | $56,791 | $18,237 | $161,941 |
| Number of ACPs done (person) | B | 4672 | 3985 | 4448 | 1489 | 14,594 |
| Weighted average cost per ACP ($) | C | $29.54 | ||||
| Execution cost ($) | D = B × C | $431,107 | ||||
| Maintenance cost of program ($) | E | $120,054 | $159,047 | $155,015 | $49,282 | $483,398 |
| Start-up cost ($) | F | $438,724 | ||||
| Total ACP intervention cost ($) | G = A+D+E+F | $1,515,170 | ||||
| Benefit of ACP | ||||||
| Reduced expenditure of last 12 months of life per patient ($) | H | $9,500 | ||||
| Number of patients died (person) | I | 325 | ||||
| Total benefit of ACP intervention ($) | J = H × I | $3,087,500 | ||||
| ROI | ||||||
| Total cost ($) | K = G | $1,515,170 | ||||
| Net benefit (total benefit-total cost) ($) | L = J−K | $1,572,330 | ||||
| ROI | M = (L/K) × 100 | 104% |
Discussion
This study provides compelling evidence that outpatient ACP is associated with increased documentation of advance directives and significant cost savings, primarily driven by reduced inpatient healthcare utilization. The increased completion of HCPOA and POLST documents suggests that ACP facilitates meaningful conversations between patients, families, and healthcare providers, aligning care with patient values and preferences. While cost savings are a significant outcome, the primary value of ACP lies in empowering patients and families, reducing anxiety, and promoting shared decision-making. The ethical underpinnings of ACP, emphasizing respect for autonomy, beneficence, and justice, are paramount, and the observed cost savings should be considered a positive, albeit secondary, benefit.
The findings on ACP and documentation rates are consistent with prior research, including a systematic review by Brinkman-Stoppelenburg et al., which highlighted the positive impact of ACP on advance directive completion and adherence to end-of-life wishes. Other studies, such as the prospective randomized trial by Detering et al. and the retrospective cohort study by Chen et al., have also reported significant increases in advance directive completion following ACP interventions. While Bischoff et al. found ACP associated with reduced in-hospital deaths and increased hospice use (contrasting slightly with our hospice findings), the overall trend supports the positive influence of ACP on end-of-life care preferences and documentation.
The observed cost reduction of $9,500 per patient in the ACP group, primarily due to decreased inpatient utilization, adds to the growing body of evidence supporting the economic value of palliative care. While randomized trials specifically focusing on ACP cost impacts are limited, studies on related interventions, such as ACP-like discussions in nursing homes (Molloy et al.) and care coordination with AD assistance for high-risk patients (Engelhardt et al.), have also indicated potential cost reductions. The SUPPORT trial, while not primarily focused on cost savings, also showed reduced Medicare costs in the intervention group in the last six months of life in secondary analyses (Hamlet et al.).
Inpatient palliative care consultation is already recognized for its cost-effectiveness. While our study did not find a statistically significant difference in inpatient palliative care consultation rates between ACP and control groups, ACP may still indirectly contribute to cost savings by influencing care pathways and reducing the need for resource-intensive inpatient care. The emergence of outpatient palliative care services further strengthens the potential for cost-effective palliative care delivery across different settings.
Limitations of this study include its retrospective, correlative design, which cannot establish causality. Unmeasured variables, such as socioeconomic status and family support, could influence both ACP access and healthcare utilization outcomes. The ACP cohort was inherently sicker than the control group at baseline, although the difference-in-difference method was used to mitigate this bias. Geographic variations and the underrepresentation of minorities and cancer patients in the dataset also limit the generalizability of the findings. Despite these limitations, the study’s strengths lie in its use of real-world ACO data, a robust matching methodology, and consistent findings across sensitivity analyses.
Considering the ROI of ACP programs, it is crucial to acknowledge the substantial noneconomic benefits for patients and families, alongside the measurable cost savings for healthcare systems and payers. The 104% ROI calculated in this study suggests a strong financial argument for investing in ACP programs. Given the large Medicare beneficiary population and the potential for even modest per-patient savings to aggregate into significant system-wide reductions, the economic benefits of widespread ACP adoption are substantial.
In conclusion, this study reinforces the value of ACP as a patient-centered service that not only improves documentation of patient preferences and end-of-life wishes but also yields significant cost savings through reduced inpatient healthcare utilization. These findings advocate for the broader integration of palliative care consultation programs, including ACP, within healthcare systems to enhance both patient care and healthcare resource management.
APPENDIX FIG. 1.
Conceptual map of the matching process used in the study to ensure comparable groups.
Acknowledgments
The authors express gratitude to Mark Hohulin and Bryan Kaiser for their initial data acquisition support, Jinma Ren for early design contributions, and Karen Renken, Jessica Fish, Kyle Mou, and Rebecca Ebert-Allen for their assistance in data acquisition.
Author Disclosure Statement
The authors declare no competing financial interests.
