Advance Care Planning (ACP) in Primary Care Settings for People with Mild Cognitive Impairment or Alzheimer’s Disease/Alzheimer’s Disease and Related Dementias (AD/ADRD)

Opening Remarks and Workshop Goals

Theresa Kim, NIA, NIH

NIA, which is one of 27 NIH institutes and centers, has four extramural divisions, including the Division of Behavioral and Social Research (BSR). BSR focuses on social, behavioral, and economic research; research infrastructure; and research training on processes of aging at the individual and societal levels, including life course health and aging research and research on Alzheimer’s disease and Alzheimer’s disease-related dementias (AD/ADRD). Within BSR, the Population and Social Processes Branch focuses on demographic, institutional, and other population-level factors (such as health systems, Medicare, Medicaid, and nationally representative longitudinal studies), and the Individual and Behavioral Processes Branch focuses on interventions and psychosocial, behavioral, and interpersonal processes at the individual and dyadic levels.

The 2019 National Advisory Council on Aging recommended developing strategies to improve interactions of older adults with the health system. ACP cannot only benefit patients—it must also benefit clinicians and care partners by helping them honor patients’ care preferences. However, ACP is not commonly practiced in primary care settings, despite Centers for Medicare & Medicaid Services (CMS) reimbursing ACP discussions. In fact, fewer than 7.5% of fee-for-service Medicare beneficiaries discussed ACP in outpatient settings, suggesting that nonfinancial barriers may be what impedes ACP discussions with people living with dementia (PLWD). Thus, this workshop aimed to address the following questions:

• What makes ACP an appropriate component of AD/ADRD care?
• What are the existing and potential positive and negative aspects of ACP conversations in AD/ADRD care?
• Outside of AD/ADRD care, what are ACP successes and complications?
• What are the consequences of offering ACP to patients with AD/ADRD in primary care settings in the presence of care partners?
• If ACP is more appropriate in primary care and at earlier stages of AD/ADRD, what modifications, if any, should occur?

Keynote: What Is ACP for PLWD and Their Caregivers?

Lauren Nicholas, University of Colorado Anschutz Medical Campus

Older adults frequently experience cognitive impairment and physical comorbidities late in life. Health and Retirement Study (HRS) data show that at least two-thirds of Medicare beneficiaries experience some degree of cognitive impairment prior to death. Thus, surrogates must frequently make decisions regarding invasive medical treatments that have limited medical benefits for hospitalized older adults, often under stressful circumstances and with little information about prognosis or patient preferences. Individuals often express preferences to avoid life-extending treatment (LET), such as feeding tubes, mechanical ventilation, and extended hospitalizations. Yet, despite these general preferences and the limited benefits of LET, the default is to provide all possible care to sustain life. ACP has the potential to improve surrogate decision-making, especially when developed prior to the onset of cognitive impairment, but its utility has been questioned because preferences can change, and advance directives may not always apply to future decisions. ACP may also reduce provider communication with patient surrogates, and patient surrogates can ignore these directives.

Dr. Nicholas has studied patient EOL preferences and how ACP and advance directives affect surrogate and provider decision-making. A series of studies on patient EOL care preferences found that individuals with mild cognitive impairment (MCI) were more likely to express a preference for LET following a heart attack than were cognitively normal older adults, but their EOL care preferences for severe dementia did not differ from cognitively normal older adults.

Using a large online sample, Dr. Nicholas probed how potential surrogates would make decisions regarding LET that would not restore health. Surrogates were more likely to choose LET for people who were bedridden but cognitively normal than they were for PLWD who were no longer able to communicate and not bedridden. Advance directives that only requested comfort care were more likely to be honored than directives requesting LET, and potential surrogates’ personal preferences regarding EOL care influenced their decisions for others.

Dr. Nicholas has also examined how advance directives affect care decisions in practice using EOL hospitalization data for 3,252 HRS participants who died between 2002 and2015 and whose data were linked to Medicare claims. She assessed the impacts of advance directives, multiple decision-makers versus a single decision-maker, and patient dementia on EOL hospitalizations. For cognitively normal decedents, having an advance directive or a single decision-maker was associated with receiving less aggressive EOL care. However, for patients with severe dementia, the number of decision-makers and having an advance directive had no impact on care decisions. Thus, in hypothetical and real-world studies, surrogates sometimes disregard patients’ preferences, particularly for PLWD, and they are generally less likely to recommend LET for PLWD.

Discussion

Dr. Nicholas confirmed that individuals with MCI in HRS were identified by cognitive testing. Dr. Nicholas shared that she was unable to accurately determine when in the course of the disease these individuals created advance directives. Dr. Nicholas added that studying ways to include individuals with cognitive impairment in care decisions is an important avenue for future research.

ACP in Primary Care Settings for People with Cognitive Impairment of ADRD: Lessons Learned from Two Primary Care-Based Trials

Jennifer L. Wolff, Johns Hopkins University

Dr. Wolff recently led two sister randomized trials to test interventions for improving the quality of communication about EOL and ACP processes in primary care settings. Individuals randomized to the intervention arms received a letter from the clinic, which introduced an initiative to improve communication and normalize ACP, as well as a patient-family checklist to stimulate ACP discussions and align perspectives on the family’s role in care decisions. If patients desired, they also received help registering themselves and family members in a patient portal, and they were provided access to a facilitator trained in leading ACP discussions. Clinical staff in the intervention arms received education and resources on ADRD through Respecting Choices, an evidence-based program for non-clinicians to facilitate ACP conversations with patients and decision-makers.

The SHARE Efficacy Trial tested the intervention among 273 dyads of highly selected patients, aged 80 and older with cognitive impairment, and their care partners. Extensive oversight and fidelity monitoring were used. The SHARE trial found that longer conversations, greater facilitator experience, and more engaged care partners were associated with greater fidelity to the Respecting Choices guidelines. In addition, care partners of those with more severe cognitive impairment tended to be more engaged in the ACP process than care partners of individuals with milder impairment.

The Sharing Access to Health Records, Agenda Setting, and RespectING (SHARING) Choices pragmatic trial randomized 51 primary care clinics to test the same intervention without monitoring its delivery in a nonselected group of patients. The most difficult aspects of the intervention to implement were family engagement and discussions regarding ACP and dementia.

Overall, the two studies demonstrated that primary care settings must adapt ACP discussions to a patient’s degree of cognitive impairment, and these ACP discussions require organizational commitment. Effective ACP may require the use of skilled facilitators, adaptations to settings and populations, and relatively long conversations.

Guide to Identify Barriers and Enablers to Implementing ACP Video Intervention in Nursing Homes

Latarsha Chisholm, University of Central Florida

Although nursing homes (NHs) are required to have discussions regarding ACP and EOL care, their rates of engagement in ACP discussions vary. Video-based interventions are potentially low-cost ways to promote ACP discussions in NHs, but little is known regarding the best practices for their implementation. To determine the barriers and enablers of video-based ACP interventions in NHs and to identify potential strategies for mitigating those barriers, Dr. Chisholm conducted 12 semi-structured interviews with ACP champions at 6 Florida-based NHs between September 2023 and May 2024. The ACP champions watched the “Goals of Care” video and then identified the following potential enablers of its use: physical infrastructure, information technology, a focus on learning, and a mission to provide person-centered care. ACP champions also identified potential barriers to the video’s use: an inability to share the video with incoming patients during the care plan meeting, a lack of access to easy-to-understand health educational materials, a lack of patient and caregiver readiness, and cultural and family backgrounds.

To address these barriers, researchers implementing video-based interventions should promote adaptability, develop easy-to-understand educational materials for NH patients, and obtain and use patient and family feedback. Holding educational meetings and developing and distributing educational materials could help address concerns about lack of health education. Finally, researchers implementing video-based interventions may need to assess patient and caregiver readiness for ACP; forge collaborations with patients, their care partners, and their family members; identify and prepare potential ACP champions in NHs; and conduct qualitative research with key stakeholders prior to developing an intervention that promotes ACP.

Dr. Chisholm next plans to conduct online surveys with the NH care plan team. To date, the results underscore that NHs vary in their abilities to implement ACP interventions and in the populations they serve. Qualitative research with key stakeholders prior to intervening may help mitigate barriers to implementation.

ACP by Physicians: A Behavioral Decision Science Odyssey

Amber E. Barnato, Dartmouth College

Dr. Barnato has conducted a series of studies that demonstrate how patient characteristics, including cognitive status, and the broader health care ecosystem affect physicians’ initiation of ACP discussions. Two pragmatic trials in a California-based Medicaid population examined how adding a $50 patient incentive to complete an advance care directive to an existing $100 physician incentive affected ACP. Patient incentives boosted ACP only when they were delivered by physicians (not when delivered by mail), and physicians were more likely to hand out ACP pamphlets in the patient incentive condition. During post-trial debriefs, physicians reported greater comfort broaching ACP when they had an incentive to offer patients for completing a directive. Another important factor affecting physician ACP initiation was the frequency of ACP at hospitals and the broader care community. Intensivists’ responses to a hypothetical scenario regarding a decision to intubate a patient with advanced cancer who did not have an advance directive varied by how common ACPs were in the broader oncology care community. Intensivists at a low-intensity care hospital where ACP was commonplace inferred that lack of an advance directive indicated that dying was unexpected and, thus, were more likely to intubate the patient. In contrast, intensivists at a hospital where ACP was uncommon did not infer that a lack of ACP meant that a patient’s death was unexpected and, thus, were less likely to intubate the patient. In another set of studies, hospitalists were more likely to recommend immediate ACP for PLWD than they were for other patients with identical mortality risks.
Dr. Barnato completed a series of studies with a large acute care staffing group for 250 hospitals across 41 states that implemented an ACP quality improvement intervention. The intervention included education, a $25 financial incentive, and a primer for mortality risk reflection upon hospital admission. This intervention increased ACP billing rates from 2% to 10%. Dr. Barnato also tested a videogame-based intervention to promote ACP, but the intervention trial occurred during a spike in COVID-19 cases and did not affect ACP.

ACP for Seriously Ill Primary Care Patients and Physician Orders for Life-Sustaining Treatment in the Nursing Home

Neil S. Wenger, University of California, Los Angeles

Current methods for identifying seriously ill patients have limited accuracy. For example, algorithms often focus on high-utilizing patients, many of whom will have regression to the mean level of health. Palliative care registries often contain very sick patients who join these registries past the ideal time for shared decision-making and ACP, and problem lists and billing codes are not specific to a serious illness. To address these challenges, Dr. Wenger tested a method of identifying seriously ill patients with electronic health records (EHRs) in a pragmatic trial involving three large University of California health systems. The method searched for six serious illnesses in EHRs: advanced cancer, advanced chronic obstructive pulmonary disorder, dialysis-dependent chronic renal failure, advanced heart failure, end-stage liver disease, and amyotrophic lateral sclerosis. Seriously ill primary care patients identified with this strategy had a 50% overlap with a utilization-based, high-risk primary care population.

These seriously ill patients were also evaluated based on the following criteria for high-priority ACP: poor short-term survival prognosis (i.e., 1–2 years), development of incapacity, worsening of functional status, or high burden of disease. White individuals, English speakers, and those age 85 and older were more likely to have advance directives or portable medical orders (POLSTs). Seriously ill patients who received information on an advance directive or a POLST prior to a primary care appointment were more likely to create an advance directive or a POLST within the next 2 years compared to seriously ill patients who received the information without an upcoming primary care appointment.

In a separate line of research, Dr. Wenger examined POLST use from 2011 to 2016 in all Californian NH residents by using Medicare data in conjunction with the Minimum Data Set, which contains federally mandated cognitive assessments for NH residents. He found that overall POLST utilization increased from 2011 to 2016. Surprisingly, residents with cognitive impairment were less likely to have a POLST than other residents but, as expected, those with a POLST tended to have orders for less aggressive care. Racial and ethnic minorities were more likely than Whites to have a POLST, indicating a preference for more aggressive care. These differences have important implications for EOL care because having a POLST is strongly associated with rehospitalization, intensive care unit stays, and time spent hospitalized.

Discussion

Impacts of Cultural Barriers and Time Constraints on Interventions to Promote ACP in Nursing Homes

Two major barriers to ACP discussions in NHs are time constraints and cultural barriers. Dr. Chisholm emphasized that several NH staff she interviewed preferred that patients and their families receive ACP-related educational materials and videos prior to admission and the initial care plan discussion. She also emphasized that NHs varied in their barriers and enablers of ACP, so researchers need to design ACP interventions that can be implemented with flexibility while maintaining fidelity.

Racial and Ethnic Disparities in ACP and Honoring of Advance Directives

In Dr. Barnato’s case study that involved physicians at a high-intensity hospital and a low-intensity hospital, physicians at the high-intensity hospital assumed that admission to their hospital implied a desire for more intensive care. These physicians were also more likely to make inferences about ACP preferences based on the racial and ethnic backgrounds of their patients than physicians at the low-intensity hospital who provided care for a broader, more diverse range of patients. While financial incentives, feedback, audits, and other types of interventions can promote care discussions, addressing implicit racial or ethnic biases requires a different approach, such as introducing new heuristics and challenging faulty assumptions.

Although NH residents in California with cognitive impairment were less likely to have a POLST than other residents, residents who were members of racial and ethnic minorities were more likely to have a POLST than White residents, perhaps because the minorities were more likely to receive diagnoses of cognitive impairment later in the disease course. Additional research on individual NHs and care discussions is needed to understand the factors influencing POLST use among NH residents. Further research is also needed to understand care preferences of residents who lack a POLST.

Target Outcomes for ACP Interventions

A major debate in ACP research concerns overall goals and outcome measures. Dr. Wolff contended that, rather than ensuring ACP documentation, the overarching goals for the field should be improving EOL care so that it is more aligned with patient preferences and reducing the strain on surrogate decision-makers. Meeting those goals will require having ongoing conversations with patients, tailoring interventions to populations that need additional support, and improving the efficiency and effectiveness of patient outreach. An additional challenge is updating advance directives as patients’ preferences change. Although advance directives typically ask about a patient’s values, goals, and preferences regarding various health states in order to reduce the need for updates, updates are required when decision surrogates change.

Gaps and Opportunities in Primary Care

Annette M. Totten and David Dorr, Oregon Health & Science University

Dr. Totten recently worked on two trials to promote ACP that underscore the challenges of improving ACP in primary care. The Meta-LARC trial was a Patient-Centered Outcomes Research Institute–funded cluster randomized trial that compared team-based to individual clinician-based ACP in 42 primary care practices. These practices were in five states and two Canadian provinces and were part of primary care practice-based research networks (PBRNs) engaged in research and quality improvement. Although the Meta-LARC trial did not include PLWD, a subsequent NIA IMPACT Collaboratory pilot tested an adaptation of its intervention for people with cognitive impairment and PLWD in four primary care practices in two of the U.S. PBRNs.

The NIA IMPACT Collaboratory pilot intervention consisted of six 1-hour video sessions that included training, discussion of cases, and implementation support. Technical assistance was provided between sessions. The primary care practices found the pilot difficult to implement because of competing demands and pressures as well as staffing constraints stemming from the COVID-19 pandemic. Although the practices could use EHRs to identify patients with cognitive impairment or serious illness, ACP conversations were not recorded in EHRs, which complicated the collection of outcome data. An important lesson from these trials is that primary care practices differ from one another, not just by type of practices (e.g., rural, urban) but also within these broad categories. Therefore, high levels of intervention fidelity are not realistic; instead, interventions must be adaptable to be successful. Notably, leadership and experience implementing trials that required role flexibility were the most important determinants of intervention fidelity, not practice type. Another lesson from these trials is that primary care has a broad scope and is generally under-resourced and understaffed. The fallout from the COVID-19 pandemic exacerbated these challenges. Thus, interventions that ask primary care staff to assume additional responsibilities without also providing assistance and direct benefits are unlikely to succeed.

A further challenge is that busy primary care providers are not primarily focused on providing care for PLWD, who typically represent only a minority of their patients. Finally, underdiagnosis and misdiagnosis of dementia present challenges to implementing ACP interventions for PLWD. Identifying PLWD using EHRs is a complex task. Diagnosis codes for dementia miss more than half of PLWD, and approximately half of the coded diagnoses are misdiagnoses. Screening tools can be used to identify PLWD, but these require 3–10 minutes to administer, and they are not in widespread use.

To address challenges related to staffing, the broad focus of primary care, technical difficulties, and difficulties identifying PLWD, researchers need to:

• Attach ACP to the concerns of the primary care practices, such as longitudinal care experiences and whole person care
• Integrate ACP into existing workflows
• Encourage flexible and creative use of nonclinical staff for initial ACP discussions
• Integrate interventions with older EHR technologies until better technology is in widespread use

Interventions should also have a plan to address challenges in identifying individuals who could benefit from ACP, such as initiating conversations with a broad range of patients who may benefit from ACP and then adapting those conversations for the patients with cognitive impairment or dementia.

One potential solution for identifying PLWD is to use an algorithm based on EHR codes that identify or are strongly correlated with dementia, followed by an in-clinic screening. Primary care settings currently collect much of the information that is used by established EHR algorithms to identify PLWD and those with serious illness. Validating these identifications through manual chart review, observation, and patient self-report is essential because EHR-based algorithms miss diagnoses and misidentify some patients. For patients with MCI or AD/ADRD, researchers may also consider developing interventions that integrate ACP into existing primary care workflows for screening, diagnosis, and referral. Finally, interventions should incorporate implementation science frameworks to assess primary care practices’ perceptions of feasibility, measurement of ACP conversations, cost, practitioner acceptability, practitioner alignment with patients’ goals, and impacts for patients.

Discussion

A formal diagnosis of dementia requires extensive neuropsychological testing and, potentially, imaging; however, in many care systems, including primary care, these tests are rarely performed. Instead, providers often code dementia based on patients’ or care partners’ concerns about memory loss or other signs and symptoms. This approach misses many diagnoses because people are often unaware of cognitive loss and, thus, fail to report cognitive impairment. An additional problem is that not all primary care practices use EHRs that have a bioinformatics infrastructure. The four primary care practices in the NIA IMPACT Collaboratory pilot used EHR to identify people who could benefit from ACP, but they were unable to use it to document whether ACP discussions occurred (in order to track compliance with the intervention) or whether advance directives were signed. A related concern is ensuring that advance directives are accessible when patients are hospitalized. Providing technical assistance to primary care practices and patients may help overcome some of these issues; other issues require changes beyond the walls of the primary care practice.

Policy Tools for ACP

Norma B. Coe, University of Pennsylvania

EOL care is intensive, expensive, and highly dependent on where someone lives. Large geographic differences in EOL care, including the use of hospice and transfers from NHs to hospitals for PLWD, underscore that this geographic variation is not driven by patient preferences. The variation in EOL care poses concerns regarding its equity, quality (EOL care is often inconsistent with patient preferences and does not improve overall outcomes), and expense (25% of health care spending is for 6% of the patients who die each year). Many policies have been enacted or are being considered to improve EOL care and to better align it with patient preferences. Ensuring goal-concordant care (GCC) requires planning, communication with surrogates, facilitation of care requests, and transfer of patient wishes across different health care systems. Notably, PLWD face unique challenges meeting these requirements because they are often eligible for insurance through multiple payers, receive services across multiple systems, and have impaired communication during the later stages of disease.

Although fee-for-service care encourages quantity and a high quality of care, it does not typically incentivize coordination of care. Fortunately, several recent changes in CMS reimbursements to better incentivize care coordination in order to reduce avoidable hospitalizations may help address these challenges. Several states have created ACP registries to facilitate access to advance directives across health care systems. However, these registries vary in effectiveness for reasons that remain unclear. In addition, enrollment in Medicare Advantage continues to grow and is associated with less intensive EOL care and lower rates of dying while hospitalized. Whether these trends indicate that Medicare Advantage is better at matching care to patient preferences remains an open question. Overall, the percentage of Medicare recipients dying in hospitals declined from 2000 to 2015; however, not all measures of EOL quality improved during this period. For example, the proportion of decedents who received invasive ventilation during the last 100 days of life rose during this period. Further improvements in EOL care will likely require a combination of policies and attention to coordination of care across payers and health care systems.

Medicare Reimbursement for ACP

Mireille Jacobson, University of Southern California

Since 2016, traditional Medicare has reimbursed qualified health care professionals for in-person ACP conversations, but this change has not substantially increased the rate of billing for ACP conversations. From 2016 to 2018, for example, ACP billing rates for beneficiaries increased from 1% to 3%, and data from CMS suggest that approximately 25% of beneficiaries have engaged in ACP conversations with their health care provider. The overall ACP billing rate remains low despite ACP billing increasing as the number of serious illness diagnoses increases. ACP billing is twice as common among those diagnosed with AD/ADRD. The overall low rates may stem, in part, from Medicare providing coverage with no cost-sharing only for ACP conversations that occur during annual wellness, not during other visits, and from the lack of reimbursement for ACP conversations with social workers, registered nurses, or chaplains. Another contributor may be that Medicare Advantage plans, which cover approximately half of Medicare beneficiaries, do not necessarily cover ACP.

In addition, providers may fail to bill for ACP conversations because of the time costs associated with documenting these conversations. Medicare reimburses $86 for the first 30 minutes of a visit, and at least 16 of those 30 minutes must be spent discussing ACP; however, documentation of an ACP conversation is burdensome. In one survey, only 16% of generalists and 4% of specialists perceived $86 as sufficient reimbursement for using the ACP code. Qualitative work has found that many providers prefer to use less-burdensome evaluation and management billing codes for ACP conversations. Thus, billing codes may not accurately capture when ACP discussions are occurring. Finally, although low levels of awareness of reimbursement have been hypothesized to contribute to the low billing rates, the data largely do not support this explanation. For example, in the first quarter of 2016, 35% of primary care providers and 24% of specialists in a large academic medical center were aware of ACP reimbursement billing codes. Moreover, many national and local organizations, associations, and agencies have developed educational materials to raise awareness about reimbursement for ACP.

Several policy changes, financial incentives, and provider prompts could help increase engagement in ACP conversations. For example, financial incentives and prompting providers to answer a question about their patients’ risks of mortality have been shown to increase initiation of ACP conversations with patients who have serious illnesses. However, even with these prompts and incentives, the overall rate of ACP reimbursement billing remains low.

System-level changes to Medicare reimbursements could have a greater impact on ACP billing rates. Such changes could include:

• Eliminating co-pays for conversations occurring outside of annual wellness visits
• Providing reimbursement for non-physicians who engage in ACP conversation
• Removing the 16-minute requirement for ACP-related conversations or allowing that time to occur across multiple visits

Ethics and Equity in ACP for PLWD

Daniel Dohan, University of California, San Francisco

As dementia progresses, the decisions that PLWD, their care partners, and others face regarding care shift dramatically, from decisions on whether to undergo various treatments to whether to receive LET. These decisions occur within a broader social context of persistent structural inequities and a social welfare system poorly designed to provide the collaborative, coordinated care PLWD need. Furthermore, the preferences and needs of PLWD and their care partners are not always aligned.

To understand how individuals experience the care they or their partners receive for dementia, Dr. Dohan conducted the Diverse Cultures, Ethnography, and Decision-making in Dementia—Implications for Medical Culture (DISCERN-MC) study, a multi-site, community-based ethnographic study over the course of 3 years in Maine and California. Dr. Dohan interviewed PLWD and their care partners about their experiences with ACP and examined how social structures, local institutions, and practices shaped their care experiences. Several themes emerged from these interviews. Individuals struggled with the lack of coordination of care, particularly when a longtime provider retired or when a patient had to navigate changing care systems. Luck played a substantial role in whether PLWD ended up in a health care system that provided proper care and ACP. Care partners struggled with deciding whether their promises to keep their partners living with dementia home were no longer reasonable and with their own feelings of sadness and loss of an imagined future with their partners.

These experiences raise broader questions concerning how ACP in primary care settings contends with a culture of individualism riven with inequalities, how to incentivize ACP in such settings, and how interventions can leverage health professionals’ good intentions and care partners’ strengths to improve the lives of those affected by dementia.

Discussion

Urban Versus Rural Disparities in ACP

Research has not found clear disparities between rural and urban care settings in ACP. For example, the Dartmouth Atlas of Health Care shows large geographic variations in EOL care but no pattern in rural versus urban differences. In addition, the use of billing codes for ACP tends to be higher among smaller practices compared to larger practices, perhaps because the value of the reimbursement is relatively larger for smaller practices. However, this difference in billing may not actually reflect differences in whether ACP conversations are occurring.

Impact of Educational Campaigns to Increase Awareness of ACP

Several recent awareness campaigns aimed at patients have emphasized the importance of ACP. Although the presenters were unsure of whether these campaigns have been effective, HRS participants increasingly report having advance care directives in place, which may reflect improvements in overall awareness of ACP.

Unbilled and Outside of Health Care Settings ACP Conversations

Not all ACP conversations are billed, in part because of physicians’ reluctance to have them and in part because not all staff who engage in them are qualified to bill for the conversations. However, it is not understood how frequently uncoded ACP conversations occur. In addition, some advance directives are created outside of the health care system, such as in a legal context. Notably, legal assistance in ACP is outside the means of many Americans, so enabling these conversations in the context of Medicare can address economic inequities in ACP.

Data Needed to Understand How System-Level Changes in Health Care and Policies Affect ACP

To better understand how policies and health care system–level changes affect ACP, researchers need better access to large EHR datasets and to datasets that link claims and EHR data with more qualitative data on individual preferences and concerns. Greater access to EHR data could also improve researchers’ understanding of the gap between billing and practices’ ACP conversations. The ability to link EHR data across multiple systems could improve researchers’ understanding of how people are navigating care across multiple health care systems and transitions from one system to another. These questions are particularly timely given the recent widespread closures of primary care clinics and the loss of health care providers stemming from the COVID-19 pandemic that affected ACP in terms of creating advance directives and having those directives honored. Finally, additional mixed methods research could improve the ability to connect EHR and claims data to outcomes that individuals care about.

Translation of Research on ACP for Cancer Patients to PLWD

Research on ACPs for patients with cancer typically follows a diagnosis of metastatic cancer and concerns plans for EOL. In contrast, research on ACP for PLWD needs to occur early in the disease course and concern plans for preferred living situations and the loss of one’s decision-making capacity and their ability to drive.

Gaps in ACP Research & Recommendations

Kara Dassel, University of Utah
Susan Hickman, Indiana University-Purdue University
Rebecca Sudore, University of California, San Francisco
Angelo Volandes, Massachusetts General, Brigham

Definitions in Advance Care Planning

ACP’s definition has evolved since the Patient Self-Determination Act of 1991, which based ACP’s definition on legal advance directive forms and emphasized treatment decisions about medical procedures (e.g., cardiopulmonary resuscitation) in advance of serious illness. However, the older ACP definition applied only to individuals who could make decisions independently (i.e., retain capacity) and a focus on medical procedures and only on advance directive forms have not been shown to be beneficial for patients or caregivers. Two interdisciplinary Delphi studies, one international and one in the U.S., considered ACP and its definition in the context of AD/ADRD.

During the U.S.-based Delphi study in 2017, panelists considered whether relying on one-time advance directive forms was the most effective ACP method and instead suggested that people living with dementia or MCI and their surrogates should be prepared for communication and decision-making throughout the disease course. Ideally, ACP should begin prior to a dementia or cognitive impairment diagnosis, enabling the patient and their family members or trusted persons (i.e., potential future surrogates) to be maximally involved in care planning. Moreover, early conversations could help prepare surrogates and families for their future roles as decision-makers and encourage collective decision-making. The consensus definition for ACP was “ACP is a process that supports adults at any age or stage of health in understanding and sharing their personal values, life goals, and preferences regarding current or future medical care and prepares them and their surrogate decision-makers for communication and medical decision-making.” This panel marked the first time ACP was formally defined by consensus.

In 2023, a review of the ACP landscape found that ACP had further evolved to a focus on preparing people for communication and decision-making throughout the life course. This new ACP framework focuses on current and future health care decisions (i.e., anticipatory decision-making) grounded in the individual’s views about quality of life.

The 2023 international Delphi study included a literature review on ACP and dementia, and it convened 107 panelists, from 33 countries, who had expertise in ACP research, practice, and policy. The Delphi panel’s abbreviated consensus definition of ACP for PLWD was “[a] process of communication about future care and treatment preferences, values, and goals with the person with dementia, family, and the health care team, preferably with ongoing conversations and documentation.” This definition was based on a relationship-centered framework that focused on involvement and communication between family, patients, and health care providers across the disease course, regardless of the capacity of the PLWD. Key components of the panel-determined ACP definition included involving PLWD throughout the ACP process, adapting conversations to an individual’s capacity, and maintaining maximal patient involvement for as long as possible. ACP then continues when PLWD become unable to make their own decisions.

In 2024, Dr. Dassel and her colleagues sought to define “concordance” in ACP, ultimately concluding that ACP care dyad concordance occurs “when care recipients and care partners both understand the care recipient’s EOL values and EOL preferences informed by those values, and the care partner expresses a willingness to accomplish the care recipient’s wishes to the best of their ability.”

ACP Outcomes

ACP can inform several different outcomes for patients, surrogates, communities, clinicians, health systems, and policies. However, the longstanding focus on the completion of an advance directive document fails to capture the range of other potential ACP successes and known patient behaviors, such as discussing medical preferences with family, friends, and clinicians. A Delphi panel convened in 2018 sought to create an organizing framework for addressing measurement complexity in ACP outcomes. Panelists identified continuously measurable outcomes in four categories:

• Action (e.g., documentation, communication)
• Process (e.g., behavior changes, patient perceptions)
• Quality of care (e.g., meeting care goals, patient satisfaction, patient-provider trust)
• Health outcomes (e.g., physical and mental health symptoms)

In 2021, a scoping review of ACP and its outcomes stemming from all ACP intervention types (written, multimedia, facilitated discussion, and clinician training) found that patients and surrogates, particularly those with experience making medical decisions related to serious illness, want ACP. Patients want to talk to clinicians about ACP in order to prepare surrogates and decrease surrogates’ decision-making burden. ACP outcomes were mostly positive in terms of self-efficacy, congruence between patients, surrogates, and clinicians, and communication satisfaction. ACP also decreased surrogate anxiety, depression, post-traumatic stress disorder, complicated grief, and caregiver burden. One study also found that ACP reduced clinician distress. However, the review found several studies reporting mixed (below 50% positive) outcomes in patient health status, health care utilization, and GCC. Notably, patient health status was not expected to change in response to ACP, health care utilization was not considered a patient-centered outcome, and GCC was non-validated, difficult to measure, and may be better studied by asking patients or care partners about real-time care alignment instead of waiting until after patients have died.

POLST and ACP

A review of studies conducted between 1996 and 2023 that compared outcomes between patients with and without POLST forms found that using a POLST as an ACP intervention resulted in overall positive outcomes, particularly related to quality of care and action. However, health care utilization—the most widely examined outcome in the reviewed studies—displayed mixed outcomes. None of the studies reviewed focused on process outcomes, and only one study focused on health outcomes (e.g., quality of life). This review also revealed four gaps in ACP outcomes measures: (a) a lack of standardized GCC measures, (b) few validated outcomes focused on care partners, (c) insufficient patient-focused outcome measures, and (d) incompatible outcome measures for differing care settings and workflows.

Acknowledging Complexity in ACP

ACP innovation and implementation are complex and evolving processes. Recognizing ACP’s complexity may allow for more accurate ACP evaluation. Some studies examine ACP disjointedly, as a series of unconnected steps, rather than conceptualizing ACP as a whole process comprising many interconnected elements and stakeholders. The Consolidated Framework for Implementation Research 2.0 captures this interconnectedness by recognizing five multi-level domains, including the outer setting (i.e., day-to-day environment) and the inner setting (i.e., within a health care environment). Many environmental conditions can affect the outer setting, including critical incidents (e.g., pandemics), community conditions, finances, policies, and laws, whereas ACP’s inner setting depends on structural characteristics (e.g., physical space, staffing, time), health practice culture and willingness to integrate ACP, and relational communication. At the individual level, an ACP implementation team must be multidisciplinary, consisting of leaders, implementers, innovators, and recipients.

Policy and Legal Barriers to ACP

Poor form readability, additional legal hurdles (e.g., required notary or witnesses, social isolation of individuals), and inaccessible technology requirements for advance directive implementation are barriers that affect accessibility of ACP, particularly for individuals with cognitive impairment. Policies allowing verbal advance directives in discussions with clinicians, video advance directives, virtual or asynchronous notaries and witnesses, and e-signatures could reduce some of the legal barriers. Another legal barrier to ACP is the loss of religious, cultural, or social contexts when complex legal language is translated into advance directives.

PREPARE Program

Rebecca Sudore, University of California, San Francisco

The PREPARETM program, created by Dr. Sudore and her colleagues, is an online, secure ACP resource focused on people in the community and primary care for patients, including those with MCI or early ADRD. PREPARE was co-developed with patients, caregivers, and communities to address culture, language, and literacy barriers in ACP. Patients and care partners can use PREPARE to identify what quality of life is important to them; learn how to talk to family, friends, and medical providers; understand how to make informed medical decisions; receive the care that is best suited to the patient’s needs; and prepare care partners. PREPARE tools include “how to” scenario videos, easy-to-read advance directives, and step-by-step guides to complete advance directives, with closed-captioning and multi-language options available. The PREPARE program uses a stepwise question and answer format, with tailored answer automation that will pre-populate a personalized advance directive form for all U.S. states. People may also skip the stepwise approach and download an advance directive or fill one out online. The PREPARE advance directives align with updated definitions of ACP and include information that has been shown to be important to patients and caregivers, such as what quality of life means to someone, what brings them joy, what would be hard on their quality of life, and what their personal values, hopes, and wishes are for their medical care, in addition to treatments that may or may not be important to them during serious illness or at the end of life.

During PREPARE’s initial rollout at the University of California, San Francisco, in 2019, ACP documentation for patients increased from 18% to 69% for all patients and to 91% for seriously ill patients. Several randomized control trials (RCTs) found that PREPARE’s multi-language resources successfully primed English and Spanish-speaking patients for ACP, with 98% of patients in the intervention arms engaging in some form of ACP. In one RCT, PREPARE was found to increase ACP documentation from 8.5% to 43%, and it increased patient empowerment, clinician responsiveness, and real-time GCC compared to control groups. In one pilot study that included participants with MCI and early ADRD, as well as limited health literacy, English- and Spanish-speaking participants rated PREPARE as easy to use and feasible, and 80% of participants increased their ACP engagement levels with the help of caregivers. Moreover, by addressing limited health and digital literacy, cultural and language diversity, and visual, hearing, and cognitive impairment, PREPARE has been shown in several trials to decrease health disparities.

Surrogates often want to speak to their loved ones about their medical preferences or may be in a situation where they need to make medical decisions for someone else who has lost their decision-making capacity, such as in ADRD. To address caregivers’ needs, PREPARE developed a care partner-facing component through a co-development process with surrogate decision-makers (PREPARE for Their Care). The program includes video modules that show caregivers how to have conversations, for example, to resolve family conflict. In a caregiver study, Dr. Sudore and her colleagues found that 13% of surrogates were not formally asked to play a decision-making role. In addition, surrogate confidence and readiness were lower among historically marginalized populations, Spanish-speaking populations, and populations with limited health literacy. Moreover, 46% of patients had not discussed their medical wishes prior to appointing their surrogate and 51% had no formal ACP documentation. PREPARE’s care partner component includes resources on how to help other people with their medical planning and how to make medical decisions for other people. In a care partner-focused pilot study, PREPARE resources increased surrogate readiness to assist others with ACP and to engage in their own ACP.

Life-Planning in Early Alzheimer’s and Other Dementias ACP Intervention

Kara Dassel, University of Utah

Although ACP can reduce unnecessary medical interventions, lower hospital readmission rates, lessen transitions between health care settings, lower EOL health care costs for PLWD, decrease physical pain, and lower emotional distress for care partners, PLWD are less likely than those without dementia to complete an advance directive, appoint a surrogate decision-maker, or engage in ACP conversations with family. The Life-Planning in Early Alzheimer’s and other Dementias (LEAD) Guide focuses on ACP communication, documentation, and continuation for PLWD. The Guide has three main parts that discuss:

• EOL documents (e.g., living wills, medical durable powers of attorney [MDPOAs])
• Medical care values, such as quality of life versus length of life, burden, and autonomous versus shared decision-making
• Medical care preferences for the patient’s present and future self, such as location of care, life-prolonging measures, and location of death

The LEAD Guide includes free-response fields that allow the patient to fill in supplementary information (e.g., “What does quality of life mean to you?") to aid future decision-makers.

Dr. Dassel and her colleagues created a pilot project using the LEAD Guide to facilitate ACP in community-based dementia dyads between persons at risk of developing dementia, persons concerned about dementia, or patients in preclinical or early-stage dementia and their respective care partners. This pilot aimed to:

• Adapt the LEAD Guide to a web-based format
• Evaluate the LEAD intervention’s acceptability, usability, and feasibility
• Explore the LEAD Guide’s initial efficacy by examining decision-making self-efficacy, patient/partner relationship quality and concordance, documentation completion, and advance directive sharing with health care providers

Patients were primarily White, college-educated females either at risk for dementia or diagnosed with MCI or early-stage dementia. Patients and care partners completed the LEAD Guide separately, with care partners attempting to accurately reflect their respective patient’s wishes and values. Patients and their care partners then discussed each other’s answers and clarified any differing responses. Participants were then encouraged to fill out an advance directive to be shared with health care providers. Preliminary results showed improved decision-making self-efficacy for patients and care partners, ACP concordance, advance directive completion and sharing, and reduced relationship strain.

Based on data from this pilot study, NIA funded a clinical trial that will explore whether ACP concordance increases surrogate decision-making confidence and improves relationship quality and subjective well-being for patients and care partners. Patients and their care partners will fill out the LEAD Guide online, enabling researchers to track participants’ progress. After the Guide has been completed by both parties, the website will identify differing answers and prompt conversation to clarify areas of misunderstanding. Additionally, the LEAD Guide will soon be translated into Spanish and culturally adapted for the Hispanic AD/ADRD community.

Dr. Dassel and her colleagues also examined ACP for individuals with one of three diseases—AD/ADRD, pancreatic cancer, or congestive heart failure—which have different death trajectories (i.e., gradual, intermittent, and acute). Compared to those with pancreatic cancer or congestive heart failure, individuals with dementia had unique care preferences and were less likely to want to live at home or live as long as possible. These findings underscore that, although many ACP tools are disease-agnostic, dementia’s unique needs necessitate specific ACP tools that are community-based, self-administered, valid, and reliable. These tools should promote active engagement in meaningful conversations among care teams, increase ACP understanding among care partners, and enable anticipatory decision-making.

ACP Decision-Making in Different Settings

Angelo Volandes¹ Massachusetts General, Brigham

Hilo Medical Center Trials

An NIA-funded trial at the Hilo Medical Center in rural Hilo, Hawaii, tested the impact of a nurse-led, video-aided ACP discussion on ACP document completion for individuals with advanced illnesses who were admitted to the hospital overnight. This trial was designed to be reflective of the clinical environment, to ensure elements of both flexibility and specificity, and it allowed patients to opt out. Following the ACP discussion’s implementation, ACP document completion rates increased from 3.2% to nearly 40%. However, this study had several limitations, including the use of a single paid nurse and no involvement of PLWD or their care partners.

A second iteration of this study expanded to two hospitals and included any patient over age 65 or any PLWD. The study examined the impact of social worker-led, video-aided ACP discussions on ACP documentation in EHR using natural language processing. The intervention was delivered as part of routine care, and participating social workers could approach any qualifying patient without consulting with a hospitalist. This trial was more pragmatic than the first trial, allowing for more flexibility in areas such as eligibility, recruitment, and setting. Of the 11,000 study participants, nearly half were from non-White populations and 12% were PLWD. Results showed a significant 12% increase in EHR documentation of ACP for participants who received the trial intervention, including an almost 15% increase in EHR ACP documentation among non-Hispanic White participants and a 10% increase among non-White participants. Notably, the intervention also increased EHR ACP documentation among PLWD by 14%. This study had a few limitations, which included involving only two hospitals in urban settings, using paid research staff, and not examining care delivery.

A third ACP study at the Hilo Medical Center would be maximally pragmatic by using a parallel cluster randomized trial design to evaluate whether an intervention is scalable to hospital networks. In this study, the ACP intervention will be delivered by non-research (i.e., clinical) staff; include content in different languages, as well as ADRD-specific content; and examine patient and care partner reported outcomes, changes in patient care preferences, and longitudinal outcomes, particularly at EOL.

¹Dr. Volandes has a financial interest in ACP Decisions, a nonprofit organization developing ACP video decision support tools. Dr. Volandes’ interests were reviewed and are managed by Massachusetts General Hospital and Mass General Brigham, in accordance with their conflict-of-interest policies.

Honoring and Preserving the Voices of PLWD

Because advance directives do not always successfully honor and preserve autonomy for PLWD, NIA funded two studies to examine the impact of ACP video declarations (ViDecs) versus advance directives in patients with cancer and end-stage renal disease. ViDecs were created by the patient in collaboration with their care partner and a clinician. After watching an informational video on ACP (e.g., videos found on the PREPARE website), a patient recorded a short, 3-minute video of themselves discussing their hopes, fears, values, decisions, and any other important aspects of care. ViDecs were then shared with the patient’s family, friends, physicians, and specialists and uploaded to the EHR. The majority of patients reported that ViDecs were helpful, easy to create, and easy to navigate. By visually capturing a patient’s medical desires, ViDecs provided care partners with a tangible, specific conceptualization of patient autonomy.

Limitations of ViDecs include digital literacy—especially for patients in rural or historically underserved communities—and their legal recognition as valid care directives. Additionally, although patients can revise (i.e., re-record) a ViDec more easily than a written advance directive, developing or changing patient care preferences remains a potential limitation in ADRD ACP. Particularly for PLWD, an ongoing debate exists about whether to honor a patient’s care decisions from “then” (i.e., prior to substantial capacity loss) or “now” (i.e., during or after substantial capacity loss). Notably, ViDecs allow PLWD to specify whether care partners should honor “then” or “now” decisions. Future ViDecs studies should include patient and clinician reported outcomes, as well as longitudinal outcomes, compared to usual care.

POLSTs: Portable Medical Orders

Susan Hickman, Indiana University-Purdue University

Advance directives are legal documents that apply to all competent adults who independently create them, but they are not recognized by emergency medical services (EMS). Conversely, POLSTs are medical orders relevant to seriously ill patients nearing EOL that are completed by a patient or surrogate in collaboration with a health care professional. These can be honored by EMS in emergency situations. National POLST provides best practice guidance and education for programs based on POLSTs throughout the United States.

In 2005, the National Institute of Nursing Research (NINR) funded a study examining POLST implementation in NHs. The study concluded that, compared to individuals with comfort care orders, individuals with a POLST who requested full treatment were significantly more likely to receive interventions (e.g., hospitalization), and individuals without a POLST received significantly more life-sustaining interventions.

Because nearly 70% of PLWD will spend time in an NH near EOL, POLSTs should accurately reflect patients’ and surrogates’ care preferences. A 2016 study funded by NINR and the Retirement Research Foundation, examined concordance between documentation and current preferences of retirement facility residents without capacity and their surrogates. Facilities that used POLSTs or had them available had 3 times the odds of accurately reflecting patient and surrogate preferences for life-sustaining treatment in comparison to settings that did not use POLSTs at all. However, nearly half of patients without POLSTs reported specific preferences for interventions, such as hospitalization or intubation, that were not reflected in their medical records, whereas surrogates were more likely to want intervention limitations. Dr. Hickman and colleagues asked participating residents and surrogates to comment on the discordance in POLST and non-POLST care preference reflections. Participants with a POLST cited problematic NH practices (e.g., receiving the form at check-in), missing information, changes in preference, or deferred decisions as the main reasons for discordance. Participants without a POLST cited absence of a POLST-related discussion with facility staff, inadequate information about care options, prior preferences missing from medical records, or family involvement.

When asked about recall following a POLST’s completion, only 50% of participants remembered seeing, completing, or discussing the POLST form. Participants were more likely to remember the form based on age, cognition, health, general education level, and POLST knowledge. Recall was not associated with the length of time since obtaining a POLST, which indicated that completion date is not indicative of a POLST’s concordance with current preferences.

Finally, Dr. Hickman and colleagues found that POLSTs and intervention preferences differ based on who is making decisions. For example, surrogates making decisions on behalf of PLWD are likely to make different intervention decisions than non-cognitively impaired NH residents making decisions for themselves. Moreover, decisions to focus on comfort rather than cure were more stable.

NIA is funding a pragmatic embedded RCT , led by Dr. Hickman and colleagues, that examines ACP training for NH staff. The trial includes 64 intervention facilities, with one to two staff ACP specialists per NH. Staff ACP specialists completed online training to facilitate ACP conversations about prolonging life, maintaining function, and implementing comfort care, with a focus on PWLD. Preliminary trial EHR data indicate that 40% of residents had decisions made by a surrogate. Two other NIA-funded studies currently incorporating POLSTs are the Indiana Palliative Excellence in Alzheimer’s Care Efforts study led by Dr. Greg Sachs and the POLST Facilitation in Community Dwelling Older Adults With and Without Dementia study led by Dr. Alexia Torke.

ACP Synergy

ACP strategies should be constructed collaboratively by patients, caregivers, and health providers to fit a diverse audience. Most advance directives are written above a 12th grade reading level even though the average reading level in the U.S. is 8th grade, the average reading level for older adults is 5th grade, and nearly 61% of Americans speak a language other than English at home. Online advance directives are similarly inaccessible because, although 73% of older adults use the internet, only 60% can fill out an online form. Older individuals might experience incompatibilities with Western views of autonomy and decision-making. For example, nearly 20% of people do not want to make their own medical decisions, even prior to being diagnosed with cognitive impairment, and nearly 20% report not having a trusted decision-maker. Some individuals also experience racism within the health system or mistrust of the health system, and many individuals distrust legal forms and documents.
Different ACP strategies, such as different patient facing interventions (e.g., web- and video-based tools, online forms, written forms), surrogate interventions, clinician interventions (e.g., communication education), and health system interventions (e.g., EHR innovations), can work synergistically with one another to assist patients, caregivers, and clinicians within different contexts and throughout the disease course.

ACP Recommendations

Kara Dassel, University of Utah
Susan Hickman, Indiana University-Purdue University
Rebecca Sudore, University of California, San Francisco
Angelo Volandes, Massachusetts General, Brigham

Old definitions of ACP do not consider the unique needs of PLWD across the disease course. A new ACP framework in AD/ADRD should:

• Be implemented prior to diagnosis of AD/ADRD
• Incorporate loved ones and caregivers from the outset
• Include the patient in care decisions for as long as their cognitive state allows
• Prioritize current and future decision-making

This new ACP framework foregrounds consistent communication between a patient's entire care team (i.e., patient, health care providers, home caregivers, loved ones) in different settings throughout the life course, acknowledging that ACP outcomes and definitions may differ between populations and settings.
Advance directives can be inaccessible to patients because they are written in complex language or require technological capabilities exceeding those possessed by the patient. Thus, care providers should use multiple methods to ensure that ACP information is understood, that communication is facilitated, and that advance directives are not only completed but also understood.

Outdated ACP frameworks emphasize advance directives as outcomes of successful ACP, but successful ACP can also be demonstrated by person-focused outcomes, such as patient satisfaction and surrogate preparedness. This new framework suggests a revised approach to ACP outcome measurement based on action, process, quality of care, and patient health outcomes. Additional outcome follow-up should occur at longer periods over the life course (e.g., 10 years post-intervention), and researchers should prioritize creating standardized, validated ACP measures that incorporate real-time GCC, EHR integration, and caregiver and dyadic outcomes.

ACP research can be constrained by funding budgets. Research funding should prioritize cultural adaptation of existing tools. Moreover, larger grant mechanisms, such as U54 Cooperative Agreements, would allow for holistic, system-based, synergistic ACP research. Policy can also facilitate ACP development and improvement by creating guides (e.g., a CMS ACP guide model) and decreasing legal barriers to ACP (e.g., legally recognizing ACP methods such as ViDecs). Policy could also incorporate billing and quality incentives for health systems to encourage ACP implementation.

Discussion

Comprehensive ACP Discussions

ACP prepares people to make informed medical decisions and reduces surrogate distress and burden. It also provides people with time to learn about treatment options and alternatives. Care preference changes are sometimes made based on patients not fully understanding their treatment options, and individuals who are fully informed about ACP and related health information are less likely to change their care preferences. Successful ACP extends beyond making decisions about hospice versus aggressive care, and care teams should consistently facilitate informed, comprehensive discussions about patients’ goals for medical care over the life course. Comprehensive ACP discussions should include disease-modifying treatments as they emerge, but health providers should avoid potentially perpetuating health inequities for individuals who cannot reasonably access certain interventions.

Identifying MCI and ADRD

MCI identification can be skewed by cultural and health disparities, thus affecting an MCI or AD/ADRD diagnosis. Individuals in historically underserved communities might report subjective cognitive complaints, but they are often left without a diagnosis despite being at higher risk for dementia than the general population. Thus, broadening AD/ADRD and MCI definitions is important in community-based studies.

Early ACP

Although the AD/ADRD decline trajectory may be long, ACP for PLWD should start early, ideally prior to AD/ADRD diagnosis. Early ACP conversations, held while an individual still retains decision-making capacity, can illuminate important baseline knowledge, identify important values and goals, and inform care partner selection. Care partner support should also be incorporated as early as possible into ACP across the AD/ADRD trajectory. Individuals with MCI and early ADRD may also continue to maintain decision-making capacity and should be included in ACP for as long as possible.

Recommendations and Opportunities

The panel presented the following recommendations for ACP and ACP research, emphasizing that ACP definitions and outcomes continue to evolve and may differ between settings and populations:

• Start ACP early and normalize ACP along the care continuum (i.e., separately from seeking cognitive impairment diagnoses)
• Include caregivers early in the ACP process
• Include shared and supportive decision-making as ADRD progresses
• Develop standardized, validated measures for real-time patient-reported outcome-based, goal-concordant care
• Leverage artificial intelligence and machine learning to capture and analyze EHR data
• Create standardized caregiver and dyadic outcomes
• Create a systems-based, pragmatic, and synergistic approach to ACP
• Leverage policies, such as the CMS Guide Model, and reduce legal barriers to increase ACP access
• Fund evidenced-based ACP programs to determine: (a) ACP’s use in different settings and life stages and (b) program synergies to inform holistic approaches to ACP
• Develop synergistic and system-based programs to create meaningful, real-life change by: (a) examining ACP interventions across care settings and transitions; (b) including caregivers early in ACP and along the care continuum; (c) integrating and collecting EHR data across care settings; (d) coordinating programs with the GUIDE model and other Integrated Healthcare Association age-friendly health systems; (e) leveraging large grant mechanisms, such as U54, to promote holistic, system-based, synergistic work
• Promote patient follow-up for 6–10 years over the life course
• Fund cultural adaptation of existing ACP tools
• Incentivize health systems use by providers (e.g., via billing and quality incentives)

Improving Community-Based Serious Illness Communication Among PLWD and Their Family Care Partners

Suzanne S. Sullivan, SUNY Upstate Medical University; University at Buffalo, State University of New York

Caregiving and ACP

According to the Alzheimer’s Association, 6.9 million individuals are currently living with dementia in the U.S.; this number is expected to increase nearly twofold by 2060. In 2023 alone, 11 million family care partners provided care to PLWD at an estimated unpaid labor cost of $346.6 billion. Dementia care partners are mostly women, 30% of whom live with PLWD with an annual household income less than $50,000. Eighty-five percent of family care partners assist care recipients with mobility, health, and medical needs. While many family care partners acknowledge positive caregiving aspects, family care partners also face caregiving-related emotional, financial, and physical burdens.

Dr. Sullivan and her colleagues conducted an initial study to examine the relationship between caregiver burdens and care transitions during the last year of life for PLWD. Leveraging the National Health and Aging Trends Study (NHATS) and National Study of Caregiving (NSOC), researchers found that PLWD were more likely to experience overnight hospitalizations when care partners reported facing physical, financial, or emotional difficulties while caring for a loved one. PLWD were nearly 6% more likely to transition to hospice care when care partners reported having emotional difficulty while caregiving.

Because hospice care improves quality of life, pain, symptoms, care satisfaction, and care quality in the last month of life, researchers examined PLWD–care partner dyads to understand which demographic features predict community-based hospice care use. No reliance on food stamps, better physical or cognitive health, and regular physician access were qualities that indicated a lower burden of illness and treatment for PLWD and predicted that they would not likely need hospice care. During participant interviews, researchers also found that PLWD in hospice care experienced feelings of uncertainty or fear related to losing hospice care, even after ACP.

The primary care Coordinating Transitions Intervention was developed to better identify conversation opportunities in ACP. Researchers collaborated with a Regional Health Information Exchange Organization (RHIO) to transfer information about PLWDs’ hospitalizations, chronic conditions, and emergency department visits to the primary care setting. After reviewing a patient’s information, a nurse in the primary care setting would call the patient for further assessment using an electronically adapted version of the Patient Centered Assessment Method (ePCAM). The nurse would enter the patient’s information into the EHR. Researchers then used the COMPLEXedex clinical algorithm to classify participating patients based on health risk indicators within the EHR (e.g., inpatient, outpatient, and emergency department visits) in order to identify vulnerable populations. The Coordinating Transitions Intervention resulted in 25% fewer inpatient visits, 35% fewer emergency department visits, and 27% more outpatient visits. This intervention could improve ACP by expanding the nurse outreach phone call to evaluate PLWD and family care partners’ support needs and to schedule an in-person ACP visit.

COMPLEXedex

To build on the Coordinating Transitions Intervention and to better understand the burden of treatment and illness in chronic disease management, Dr. Sullivan and her colleagues expanded the COMPLEXedex algorithm to include social determinants of health (COMPLEXedex-SDH). Developing data definitions to identify vulnerable populations is difficult because most health systems data are biased toward patients who have access to the health care system (i.e., patients outside of vulnerable populations). However, using disease complexity factors, community-based service use, health care utilization patterns, and social determinants of health, researchers added the following new data terms and their definitions to the COMPLEXedex-SDH:

• Cross-cohort needs: a person with a major chronic illness necessitating care from two or more community-based organizations
• High-need: individual with economic disadvantage or cross-cohort needs
• Suboptimal utilization: individual with utilization patterns that indicate higher risk
• High-need with suboptimal utilization: individual with high hospital encounters and outpatient underutilization

In an Agency for Healthcare Research and Quality–funded trial led by a multidisciplinary team of experts in nursing, social work, industrial systems, implementation science, and computer science, RHIO is being used to share daily patient care and health information with community-based providers in primary care, social services, and behavioral health clinics to facilitate personalized cross-sector transitional care management. The trial is now in its fourth year, focusing on persons with high medical and social needs, including PLWD and family care partners.

Dr. Sullivan and her colleagues are conducting a study to identify whether certain care-related data patterns are specific to PLWD. For example, length between medical visits decreases and medical visit frequency increases as PLWD approach death. Researchers plan to use preliminary data from cluster analysis to characterize individuals who may benefit from a specific intervention. For example, one cluster was made for individuals with the following data patterns who should be highly prioritized for ACP and community-based support:

• Short intervals between health care encounters (i.e., fewer than 20 days between visits)
• Nearly 16 outpatient visits per year
• Two or more inpatient or emergency department visits
• Housing needs
• A significant burden of illness and caregiver burden

ACP and Nursing

Dr. Sullivan and her colleagues interviewed home health nurses to develop two programs:

• The Home Healthcare Communication Openings Framework, which is intended to help homecare nurses, PLWD, and families discuss serious illness care goals and ACP.
• The TILE-12 Index, which calculates risk factors to develop a score corresponding to the probability of an individual transitioning to a higher care level within 12 months. Risk factors include age, sex, marital status, race and ethnicity, insurance status, daily activities, social life, and disease conditions.

The TILE-12 Index was formed in response to finding that relationships between nurses, patients, and families are central to engaging in serious care-related communication. However, to prevent transitions to hospice care severing nurse-patient relationships, some home nurses choose to provide at-home EOL care without proper training and send patients to the hospital only when it is unsafe to keep the patient at home.

Creating Change: Opportunities to Translate Work into Practice

Aretha Delight Davis² ACP Decisions Foundation

Successful ACP recognizes individuals’ care values, goals, and preferences. ACP can empower PLWD at all ages and stages of health across the lifespan by improving current and future decision-making in a relationship-centered way that involves loved ones and health care teams. ACP should be implemented early and focus on communication, not just documentation, and draw upon patient life experiences and values to make ideal care choices.

Although providers report that ACP is a high priority, it requires training, staffing, and patient and family education materials. Systematic ACP clinical routines and policies and ACP-specific billing language would make ACP facilitation easier for clinicians. Additionally, many ACP providers recognize that individuals trained in navigating ACP conversations may be useful to have in clinics. Dr. Davis discussed several studies that examined the impact of ACP provided by different navigators: social workers, laypeople, nurses, community health workers, and clinical health practitioners.

Advance Care Planning Navigators

One study used trained social workers to facilitate and document care information conversations between NH residents and their proxies. Care and health record information was shared with participating patients’ health teams, and participating residents could provide feedback on information accuracy to health teams. Residents with dementia who received the intervention were more likely to document certain treatment and nutrition preferences, whereas non-intervention residents were more likely to receive discordant care.
A similar care conversation-centered study examined the impact of layperson-led ACP conversation interventions in oncology centers. The intervention was based in conversation rather than documentation and 40% of participating layperson navigators were African American. Researchers supported navigators by introducing them to participating patients and making space for conversations and questions. Post-intervention, participants’ hospitalization, intensive care unit visits, emergency department visits, and chemotherapy utilization decreased.

In a study comparing ACP intervention to usual care, participants with multiple comorbidities and cognitive or physical impairments received nurse-led ACP and readiness assessments to determine care goals, values, and phases (i.e., pre-contemplative, contemplative, or action phases). Nurses populated participants’ assessment information into EHRs, which informed patients’ follow-up visits with their primary care providers. As a result of the nurses’ actions, primary care providers only needed to add ACP or readiness information to the EHR for less than 20% of patients, which allowed the providers more time to discuss other topics during the health visits. In addition, significantly more intervention participants (42%) had ACP documentation compared to just 4% of participants receiving usual care.

Another study focused on care conversations led by community health workers in older adult patients, 24% of whom were PLWD. Navigators took part in ACP training simulations and had access to clinical support tools that suggested prompts about patients’ histories, values, and health preferences. Navigators documented care conversations in EHRs, identified proxies, and made referrals, when necessary. After care conversations with navigators, each participant received a letter intended for their proxy decision-maker that detailed the participant’s care values and preferences. Nearly half of participants who received the care conversation intervention had ACP discussions, and 34% were significantly less likely to be admitted to the hospital compared to participants who did not interact with the navigators.

Group visits can also be beneficial intervention styles for ACP by enabling scalability and efficient information sharing. One small study examined the effectiveness of two 2-hour ACP group visits to older adults in a senior clinic—the first session focused on sharing stories and discussing values, and the second session focused on common medical decisions and health care discussions. Six months after the interventions, researchers found that EHR advance directives increased by 26% and surrogate appointments increased by 30%. Another group visit study involved only one 1-hour session that was facilitated by clinicians, physicians, or nurse practitioners who provided a lecture and a video-led group discussion. Participants accessed the session via web platforms that were available in multiple languages and accommodated differing literacy needs. Following the session, advance directive comprehension and completion rates increased for both English- and Spanish-speaking participants and both participants and facilitators reported enjoying the session.

²Dr. Davis is the chief executive officer of the ACP Decisions Foundation and receives a salary from the foundation.

Technology and ACP

Patient portals can aid in ACP understanding and output. In one study evaluating a novel patient portal-based ACP tool, researchers created a webpage containing ACP educational resources, which included links to the PREPARE and NIA websites. Participants, aged 17–98, used the portal to view, complete, and electronically sign an MDPOA. An ACP support team could answer participants’ questions via direct message in the portal, and an ACP coordinator monitored the portal and verified participants’ capacity, when necessary. Eighty nine percent of participants completed an MDPOA, almost all of whom had not completed an MDPOA prior to the study, and participants rated the portal’s usability at 89% . The study had several strengths, including internal funding, operational and clinical support, repetitive communication with clinical teams and staff, and allowances for proxy portal access.

Another patient- and clinician-focused study examined the effects of an ACP video decision aid intervention on ACP documentation. Clinicians participated in an intensive, 4-hour long communication training focused on ACP in inpatient settings, and patients received ACP video decision aids 2 weeks prior to scheduled health appointments. Six months after the intervention, ACP increased for all patients by nearly tripling among Black/African American patients, doubling among Hispanic patients, and nearly doubling among non-Hispanic White patients. Future work with ACP should consider how to expand research synergy beyond disciplinary silos to best serve patients and clinicians.

Discussion

Medical Information Integration and Exchange

Medical information exchange can be difficult for certain states. For example, many of Indiana’s medical records are written on paper, which makes translation into compatible online formats difficult. Western New York is developing an ACP exchange model using RHIO, and California is working toward POLST integration and exchange into clinical protocols.

Early ACP

Conversations about EOL decisions and ACP often occur too late in the care cycle of any disease. When care providers suggest pursuing palliative and hospice-focused care for PLWD and critically ill patients, surrogates and families are sometimes relieved and express their harbored uncertainty over knowing when to switch to an EOL care model. Conversely, some care partners might disagree with experts’ suggestions by not yet wanting to pursue EOL care. Medical providers should not wait to facilitate ACP when a patient is in an emergent or life-threatening condition that necessitates major life-extending treatments (e.g., respiratory intubation, feeding tube).

Importantly, medical providers are not the only parties responsible for ACP. The Home Healthcare Communication Openings Framework illuminates instances in which patients hold an opinion or value that indirectly informs ACP. For example, a provider or care partner can respond to a patient saying, “I don’t want to go back to the hospital anymore,” by opening a conversation about other care options. EOL conversations should be fostered beyond medical settings by taking place in homes and in communities.
Health Care Team Communication

Communication between health providers can be challenging. RHIO seeks to provide a way for health care teams to more easily and effectively share information and facilitate interdisciplinary communication. For example, RHIO can connect home health teams and social service agencies to primary care and hospitals.

ACP Data

Inadequate data are a continuous challenge in ACP research. Present data offer information about patients and caregivers separately, but data on dyads are lacking. Research on ACP use by EMS and emergency responders (e.g., paramedics) is also lacking.