Advance Care Planning, Palliative Care, and End-of-life Care Interventions for Racial and Ethnic Underrepresented Groups: A Systematic Review

Abstract

Context:

Persons from underrepresented racial and ethnic groups experience disparities in access to and quality of palliative and end-of-life care.

Objectives:

To summarize and evaluate existing palliative and end-of-life care interventions that aim to improve outcomes for racial and ethnic underrepresented populations in the United States.

Methods:

We conducted a systematic review of the literature in the English language from four databases through January 2020. Peer-reviewed studies that implemented interventions on palliative care, advance care planning, or end-of-life care were considered eligible. Data were extracted from 16 articles using pre-specified inclusion and exclusion criteria. Quality was appraised using the modified Downs and Black tool for assessing risk of bias in quantitative studies.

Results:

Five studies were randomized controlled trials, and the remainder were quasi-experiments. Six studies targeted Latino/Hispanic Americans, five African Americans, and five, Asian-American and/or Pacific Islanders. The two randomized control trials reviewed and rated “very high” quality, found educational interventions to have significant positive effects on advance care planning and advance directive completion and engagement for underrepresented racial or ethnic groups.

Conclusion:

The effectiveness of advance care planning, end-of-life, and palliative care interventions in improving outcomes for underrepresented racial and ethnic populations remains uncertain. RCTs and educational interventions indicate that interventions targeting URGs can have significant and positive effects on AD and/or ACP-related outcomes. More high-quality intervention studies that address racial and ethnic health disparities in palliative care are needed, particularly those that address systemic racism and other complex multilevel factors that influence disparities in health.

Introduction

In the United States (US), non-white racial groups have been historically marginalized and underrepresented in healthcare settings. Persons from racial and ethnic underrepresented groups (URGs) (e.g., African Americans, Latino/Hispanic Americans, Asian Pacific Islanders and Native Americans) in the US face significant disparities in access to health care, quality of care, and health outcomes, which extend to healthcare in serious illness and at the end-of-life.(1,2) Palliative and end-of-life care services are underutilized overall, and especially amongst URGs.(3,4) At the same time, the US is experiencing a rapid increase in the populations of URGs.(5) By 2050, 40% of adults age 65 and older will be from an URG.(6) Health care systems must rise to meet the needs of an increasingly diverse, and historically underserved, patient population. Compared to non-Hispanic whites, persons from racial URGs have been found to have lower patient and family satisfaction at the end-of-life, be at higher risk for not receiving goal-concordant care, are more likely to use aggressive treatments, and have higher healthcare expenditures.(3,4,7) Thus placing a greater focus on achieving equitable, highly accessible serious illness care must take on increasing prominence in both research and delivery than is currently present.

Significance of the problem

Although racial and ethnic disparities in health outcomes have long been recognized in the US, little progress has been made at reducing these disparities.(2) Members of URGs suffer disproportionately from preventable and treatable conditions when compared to non-Hispanic whites.(8) Part of these disparities are attributable to differences in access to high quality health care, regardless of patient insurance or ability to pay for care.(9) This healthcare gap is also linked to higher death rates for persons from URGs.(9)

Racial and ethnic disparities in health-related outcomes extend to serious illness and the end of life. Compared to non-Hispanic whites, persons from URGs are less likely to receive hospice care and more likely to receive aggressive treatments and interventions including mechanical ventilation and cardiopulmonary resuscitation (CPR), all of which are associated with lower quality of life and higher costs.(10) In addition, compared to non-Hispanic whites, persons from URGs experience more in-hospital deaths, ICU and hospital admissions, and longer lengths of stay.(10,11) They also experience lower quality of acute, chronic, cancer, and palliative pain care.(12)

Racial and ethnic disparities in palliative and end-of-life care have been partially attributed to differences in completion of advance directives (AD), engagement in advance care planning (ACP), perceptions of care quality, hospice enrollment, and end-of-life care service utilization.(4,7,13–15) Compared with white counterparts, African Americans have been found to be less likely to complete AD and have care preferences respected.(16) In addition, independent of diagnosis, setting of care, or geographic location, African American and Hispanic persons have been found to enroll in hospice at significantly lower rates than whites and are more likely to be hospitalized and use aggressive treatment interventions in the last 6 months of life.(4) These disparities persist, and do so despite the documentation that high-quality serious illness care is important and valued by persons from URGs.(17)

Researchers propose a socioecological framework to address inequities and the many interacting layers of systemic racism in palliative care.(18) A socioecological framework parses out the ways in which oppressive practices and structures create barriers and influence health at the individual, interpersonal, organizational, community, and policy levels. At the individual level, there may be barriers to equitable care related to the philosophy of care’s discordance with patient beliefs, patient education, or health literacy. For example, it has been documented that the underlying goals of hospice and palliative care may not align with the spiritual and religious beliefs of all groups.(13–15) At the interpersonal level, issues with provider bias and communication (less information provided or fewer clinical interactions) generate disparities in care. At the organizational level, fewer resources including sites of care and trained specialists create inequities. At the community level, a history of racial injustices and disparities in health care may cause apprehension for some groups regarding the use of hospice care and forgoing curative care at the end of life.(4,19) And finally, at the policy level, barriers to equitable care are created by the lack of available comprehensive health insurance.

Although the reasons for health disparities for URGs in the US in palliative care have been studied for over a decade, there had been a lag in the development of evidence-based interventions targeting the aforementioned outcomes, and to some extent a dearth still exists in fully understanding some of the underlying factors so that they can be addressed in targeted interventions to achieve equity. Belatedly, intervention research has commenced aiming to improve palliative and end-of-life care for persons from URGs.

The National Consensus Project (NCP) guidelines for Quality Palliative Care, first published in 2004 and now in its 4^th edition(20) has played a significant role in consolidating and guiding improvements in high quality palliative care practice. These guidelines may be useful as a framework for identifying disparities and targeting interventions that can address issues of quality and equity in URG. The NCP guidelines are assembled by the NCP Coalition, a not-for-profit organization consisting of 12 national organizations committed to improving access to and quality of palliative and hospice care. The NCP’s importance in guiding practice for quality in the field is widely recognized. The NCP identifies 8 domains for improving the quality of palliative care for patients and families including: structure and process of care, physical aspects of care, physiological and psychiatric aspects of care, social aspects of care, spiritual, religious, and existential aspects of care, cultural aspects of care, ethical and legal aspects of care, and care of the patient nearing the end of life.(20) The last domain exists to emphasize terminal care as an explicit facet of palliative care.

Objectives of the current review

The purpose of this review is to synthesize and evaluate existing literature on interventions that aim to improve palliative and end-of-life care outcomes (e.g., quality of life, health outcomes including pain and symptom management, and completion of and engagement in ACP or AD) for URGs in the US. More specifically, this review aims to answer the following questions: 1. What sub-groups and characteristics of URGs are being studied in intervention research that aim to address racial and ethnic disparities in serious illness? 2. What outcomes were studied and how do they align with the quality domains outlined by the NCP Guidelines? 3. What serious illness care intervention strategies are used to reduce disparities in outcomes for URGs? 4. How effective were these interventions and strategies reported in studies with and without intervention comparators? 5. What is the quality of intervention research that seeks to reduce disparities for URGs in serious illness care? 6. What levels of the socioecological model do the existing interventions target?

Methods

A systematic review was conducted in January 2020. The review protocol is registered with and published by PROSPERO (ID# CRD42020161496).

Data Sources

Following the PRISMA guidelines,(21) a systematic electronic literature search was conducted to include all English-language publications on the topic of serious illness care interventions for URGs until January 2020. The electronic search was conducted in the following databases: MEDLINE, PsychINFO, CINAHL, and EMBASE. We chose these databases because they include studies covering areas of medicine, nursing, social work, psychology, sociology, and health sciences. We worked with a research librarian and drew upon previous literature reviews to identify Boolean search terms. The main components of the search strategy consisted of terms related to URGs, serious and terminal illness care, and AD or ACP. An example of the search strategy has been included in Supplementary Material. We also included related terms generated by each unique database interface. As an example, our complete MEDLINE search strategies are provided in Supplementary Material. To identify any non-peer reviewed intervention studies, we searched theses and dissertations, performed hand-searches of any relevant review references, and in some cases first-authors were contacted to ensure most up-to-date versions of study publications. The studies identified by the initial search were uploaded to COVIDENCE systematic review software where duplicates among and within databases were removed. Two of the authors independently screened titles and abstracts to identify relevant studies for full-text review, and examined the full text of relevant studies. Authors discussed and reached consensus on disagreements arising during the screening process to arrive at the final selection of studies to be included for the review.

Inclusion and Exclusion Criteria

Articles were included in this systematic review if they met the criteria outlined below and in Table 1. As appropriate for areas with limited research,(22) this review was designed to be as comprehensive as possible with regard to the participants, interventions, and comparator groups eligible for review. Studies needed to be published in English, involve ACP, palliative care, end-of-life care, hospice or terminal care interventions, and target racial or ethnic URGs in the US.

Table 1:

Inclusion/ Exclusion Criteria

Inclusion criteria (organized using the PICOS framework)
1. Population: Studies on racial/ethnic underrepresented populations specifically 2. Intervention: Articles describe empirical study on the effect of advance care planning, palliative care, end-of-life care, or terminal care 3. Comparisons: To maximize inclusiveness, studies that involve any intervention comparators and those without a comparator are all considered 4. Outcomes: Study outcomes fit within the 8 domains of outcomes outlined by NCP for quality palliative care 5. Study Design: Experimental and quasi-experimental studies that evaluate an eligible intervention 6. Studies published in English up until January 2020
Exclusion criteria
1. Studies on psychiatric patients 2. Studies on hypothetical situations (e.g. exclusively vignette or simulation studies) 3. Qualitative, descriptive, observational, nonexperimental, or retrospective studies 4. Studies on populations outside of the U.S.

Eligible study designs included randomized controlled trials (RCTs), pre/post studies, and intervention studies without comparators. To narrow the search results to these study methodologies and identify appropriate filters for each database, we consulted an institutional librarian and the Cochrane guidelines.(23) Filters were selected and utilized to cast the widest net possible while narrowing search results to quantitative studies relevant for further review.

To identify study outcomes included in the review, we followed the National Consensus Project’s (NCP) eight domains for assessing quality care in interventions at the end of life.(20) These include structure and process; physical aspects; psychological and psychiatric aspects; social aspects; spiritual, religious, and existential aspects; cultural aspects; care provided to the patient at the end of life; and ethical and legal aspects of care.

Data Extraction and Quality Appraisal

Each study meeting eligibility criteria through the screening process was carefully reviewed and data extracted and verified by a second reviewer, based on the Cochrane checklist for data extraction,(23) including study methods/design, participants, setting, interventions, and outcomes (see Table 2). Both randomized and non-randomized studies were appraised using the modified Downs and Black checklist for quality assessment.(24,25) The Downs and Black checklist consists of 27 questions and has been designed and validated to evaluate articles in four domains: reporting, external validity, internal validity, selection bias, and power. For the purposes of our study, the study quality scores were used to objectively describe the studies. For contextualization of the scores, we followed previously published categorizations: >=20 “very high”; 15–19 “high”; 11–14 “moderate”; <=10 “low”.(26) Two reviewers independently completed the checklist for each included article, resolving conflicts by consensus.

Table 2.

Characteristics of included studies

Author, year	N, participants, setting	Duration/ Frequency	Design/ Intervention / Comparison	Outcomes	Results	Downs and Black Score
EDUCATIONAL INTERVENTIONS
Bonner et al., 2014	68 African American caregivers of relatives over the age 65 diagnosed with dementia in Chicago at 5 adult day care centers.	4 weekly group sessions over 4 weeks. First and last sessions 90 min, 2nd and 3rd sessions, 60 min.	Pre/posttest. ACT-Plan - Provide self-reflection, storytelling, education, and behavioral skills with educational content on ACP. Control (attention-control condition): interactive discussion on health promotion.	Knowledge, self-efficacy, and skills related to EOL decision-making.	Increased self-efficacy (p = 0.02) at posttest completed immediately after 4th session.	12 (moderate)
Braun et al, 2006	570 family caregivers of racially and ethnically diverse homebound or institutionalized older adults age 60 or over located in a community-based setting in Hawaii.	Educational booklets sent to participants by mail.	Single group pre/posttest repeated measures. Provide 5, 20-page booklets on ACP, funeral and memorial services, care for dying, what to do when a loved on dies, and bereavement support.	Familiarity with hospice and ADs; attitudes towards ACP; openness to hospice and home death for self and older adult; presence of AD, funeral/burial plan.	Significant increase in number of elders with living wills (p < .05)and in numbers of caregivers and elders with funeral or burial plans (p < .05) at 3-month follow-up.	12 (moderate)
Cruz-Oliver et al., 2016	145 family caregivers of Latino older adults in community-based setting in Missouri, Texas, and Puerto Rico.	1, 1-hour session.	Single group pre/posttest. Oral presentation, 15 minute video telenovela on EOL caregiving, and structured discussions.	Knowledge and attitudes regarding EOL, intention to complete ACP or use EOL services.	At 4–7 month posttest, significant improvements were found in caregiver stress self-awareness (p < .001) and willingness to accept professional help (p < .05).	14 (moderate)
Cruz-Oliver et al., 2017	142 health care providers of Latino patients in community-based settings in Florida, Missouri, and New York.	1, 1-hour session.	Pre/posttest. Oral presentation, 15-minute telenovela video on EOL caregiving, and structured discussion. Comparison group: same procedures with a generic EOL caregiving experience video.	Knowledge and attitudes regarding EOL, openness to discussing EOL issues with culturally diverse patients.	Immediately after the intervention, awareness of health literacy, cultural competence, cultural difference about EOL decisions, caregiver stress, and interventions for caregiver stress improved (p < 0.001), and were significantly higher in the intervention group (p < 0.05).	16 (high)
Heyman et al., 2010	84 Latino older adults receiving home care in New York (home care agency-based).	1, 45- to 60-minute session.	Posttest only control group. Intervention group 1 ‘Conversation A’: Bilingual AD/ACP education. Intervention group 2 ‘Conversation B’: AD/ACP education plus additional topics found important by focus group. Control: usual care.	Attitudes toward and comfort with EOL planning.	At 2 week posttest interview both intervention groups had higher ACP attitude scores than control (p < .001). Conversation A had higher ACP comfort score than control (p < .01).	15 (high)
Hinderer et al., 2019	72 Chinese Americans at a community center in the Washington, DC area.	1, 90-minute workshop.	Single group pre/posttest repeated measures. Provide culturally sensitive, bilingual, nurse-led ACP PowerPoint seminar and ACP tool.	Experiences with and attitudes towards ACP.	Attitudes towards AD higher immediately after intervention (p=0.014), and no significant change between immediately after and 1 month follow-up.	9 (low)
Lee et al, 2015	72 urban, community dwelling Chinese Americans in Washington, DC area.	1, 1-hour session.	Single group pre/posttest, repeated measures. Deliver culturally sensitive, bilingual, nurse-led ACP PowerPoint seminar and ACP tool.	AD knowledge, engagement, and completion.	AD knowledge increased significantly immediately after attending the seminar (p < .01). No significant change between immediately after and 1 month follow-up.	12 (moderate)
Lyon et al., 2019	233 dyads: adult with HIV and surrogate in 5 HIV clinics in Washington, DC area; 88% African American.	2, 1-hour sessions.	RCT. Provide two sessions on ACP completion to adolescents with HIV. Control: two sessions on developmental history and nutrition.	ACP and AD documentation in medical record.	At 3 months post intervention a significant Increase was found in AD completion (p < .0001).	20 (very high)
Song et al, 2009	58 dyads: African American dialysis patient near EOL and family/chosen surrogate decision maker in 6 dialysis clinics in Pennsylvania.	1, 1-hour session.	RCT. Illness education intervention delivered to dialysis patients and family caregivers. Control: usual care- one-time discussion and written materials on ACP with social worker.	Patient-surrogate congruence in EOL treatment preferences, patient and surrogate comfort in decision-making, and psychosocial-spiritual well-being.	Intervention dyads improved congruence in EOL treatment preferences (p = 0.04) at 1 week; no significant improvement from baseline at 3 month follow-up.	17 (high)
Song et al., 2010	19 dyads: African American dialysis patients in dialysis clinic nearing EOL and families/chosen surrogate decision maker in Pennsylvania.	1, 1-hour session.	RCT. Illness education intervention delivered to dialysis patients and family caregivers. Control: usual care: one-time discussion and written materials on ACP with social worker.	Patient-surrogate congruence in EOL treatment preferences, patient and surrogate comfort in decision-making, and psychosocial-spiritual well-being.	Intervention dyads improved congruence in EOL treatment preferences (p < .05) at 1 week.	18 (high)
Sudore et al., 2018	968 English or Spanish speaking primary care patients aged 55 or older with two or more chronic medical conditions at 4 safety-net, primary-care clinics in San Francisco.	5, 10-minute videos.	RCT. Interactive, online ACP program in English and Spanish plus easy- to-read AD. Control: easy-to-read AD only.	Primary: documentation of new ACP at 15 months. Secondary: engagement in ACP process.	Intervention increased documentation of AD at 15 months (p < .001), and increased patient-reported ACP engagement (p < .001), and behavior change (p < .001) at all time points (1 week, 3, 6, and 12 months)	23 (very high)
Sun et al., 2017	174 Chinese and Vietnamese American older adults. 2 Chinese Protestant and Vietnamese Catholic churches.	2, 2-hour group sessions held 4 weeks apart.	Single group pre/posttest. Culturally tailored intervention in Cantonese or Vietnamese including church leader endorsement of ACP, ACP education, and support completing ACP.	AD knowledge, beliefs, attitudes, intention to complete, and completion.	A significant increase in proportion with AD knowledge (p < .01), proportion with supportive beliefs and attitudes about AD (p ≤ .05), and intention to complete and discuss AD (p < .01) was found immediately following session 2.	11 (moderate)
Volandes et al., 2008	104 Spanish-speaking patients aged 40 and older in 2 urban primary care clinics in the Boston area.	1, 15-minute intervention.	Single group pre/posttest. Education about advanced dementia and video of person with advanced dementia in Spanish.	Preferences for EOL care.	Significant change in reported care preferences immediately following intervention (p < .001).	10 (low)
CLINICAL CARE INTERVENTIONS
Fernandes et al, 2010	45 dyads: patients and caregivers, plus 1 patient. Ethnically and racially diverse patients eligible for home based palliative care in a community health center in Hawaii.	Visits as needed over 12 months. Median number of visits: 3.5, median length of stay in program: 7 months.	Single group with historical control. Deliver home-based culturally sensitive palliative care program with interdisciplinary team.	Quality of life, AD completion, ED use, and hospitalizations.	By the end of the study a significant improvement was found in sense of well-being (p= 0.03) in quality of life scale. No change on other quality of life domains. Increase in EOL preferences discussion and documentation (p < 0.01). No change in ED visits. Decrease in hospitalizations (p = 0.002).	10 (low)
COMMUNITY SUPPORT INTERVENTIONS
Fischer et al, 2018	223 adults with advanced cancer who self-identified as Latino in Colorado. 3 urban comprehensive cancer centers and 7 rural community cancer clinics.	5 or more in-home peer navigation visits. No duration specified however each visit was a mean of 105 min/visit.	RCT. Culturally tailored patient navigator intervention (5 or more in-home peer navigator visits). Control & Intervention: culturally tailored bilingual educational packet on ACP, AD, pain management and hospice.	Primary: documentation of ACP, pain management, hospice use. Secondary: quality of life, length of hospice stay, intensive care at EOL.	At month 46, a significant increase was found in documentation of ACP among intervention participants (p < 0.001). No difference in pain management, quality of life, hospice use, length of stay, or intensive care at EOL between groups.	16 (high)
Hanson et al., 2014	130 community volunteers and 48 African American patients with advanced cancer or comparable serious illness in North Carolina. Community-based.	Training: 3 hours for volunteers, 6 hours for volunteer team leaders. Patient participation: variable based on needs.	Pre/ posttest. Provide training on practical, emotional and spiritual support for cancer to community volunteers. Teams of 3–10 volunteers provided support to patient participants based on volunteer time and patient needs. Comparison: patients who refused a support team and received usual care.	Patients: support needs, awareness of services, health-related quality of life. Volunteers: comfort with providing checklist of support activities.	2 months after enrollment, patients with support teams reported fewer needs for practical, emotional and spiritual support on structured checklist. They more often communicated with someone about their cancer care needs (p=0.04), and were more aware of Hospice (p=0.04). Quality of life scores were unchanged.	12 (moderate)

Abbreviations: ACP: advance care planning; AD: advance directive; EOL: end of life; RCT: randomized controlled trial

Data synthesis

Due to the heterogeneous nature of the studies and the lack of comparator group, a formal evaluation and meta-analysis examining the comparative effectiveness of the interventions was not possible. Therefore, a narrative data synthesis was utilized.

Results

The overall process of the review is summarized in Figure 1. Electronic database searches generated 1,597 records and hand searches generated seven (7) results. After removing duplicates, we screened 642 records for eligibility by reviewing title and abstract. Through this screening we excluded 607 records that failed to meet eligibility criteria. Following title and abstract screening, 35 articles remained for full-text review. After the final full-text review, 16 articles remained to be included in the systematic review.

Figure 1:

Flow diagram for the study selection process

Study Characteristics

Table 2 summarizes the study characteristics; the sample, study design, intervention, and outcomes of reviewed studies. All studies were conducted in the US, and despite a lack of date parameter utilized, all of the studies included in the final review were published from 2006–2020.

The 16 studies included for final review had heterogeneous designs: five (5) employed an RCT design,(27–31) three (3) used a pre/post-test two-group design,(32–34) one (1) study utilized posttest only two-group design,(35) and the remaining seven (7) studies employed a pre/posttest single group design.(36–42)

Participant Characteristics / Target Population

Included studies targeted a variety of URGs. Of the 16 intervention studies included in the review, six (6) included Latino/Hispanic Americans,(27,31,33,35,37,41) five (5) African Americans,(28–30,32,34) and five (5) Asian or Pacific Islander Americans.(36,38–40,42) Sample sizes in the studies varied from 30 to 986. The mean age of care-recipients in reviewed studies ranged from 51–71 years old. Most of the studies (9) did not target a specific disease population.(33,35–42) The remaining seven studies targeted advanced cancer (2),(31,34) dialysis (2),(28,29) dementia (1),(32) HIV (1),(30) and multi-morbidity (1).(27) Studies also targeted different persons in the healthcare experience. Most studies targeted patients or health care recipients only (7).(27,31,35,38–41) The remaining nine studies targeted caregiver-patient dyads (4),(28–30,42) caregivers only (3),(32,36,37) volunteers and patients (1),(34) and health care providers (1).(33)

Intervention Strategies

Of the 16 studies reviewed, 13 studies utilized an educational intervention,(27–30,32,33,35–41) two studies utilized community support interventions,(31,34) and one study evaluated the effectiveness of a clinical home-based palliative care program.(42)

Educational Interventions

The 13 studies that included an educational intervention focused primarily on advance care planning (ACP) and/or advance directives (AD). The educational interventions were all designed to improve utilization or access to palliative and end-of-life care for persons from racial and ethnic URGs by targeting barriers including language, knowledge, attitudes, beliefs, health literacy, and culturally appropriate information delivery. All of the studies but one targeted patients and/or caregivers. The only exception focused on healthcare providers.(33) Using the modified Downs and Black checklist for quality assessment, two of the 13 educational studies were evaluated as being “low” quality,(38,41) five were “moderate” quality,(32,36,37,39,40) four were “high” quality,(28,29,33,35) and two as “very high” quality.(27,30) Six of the 13 educational studies were single group in design and therefore at increased risk for bias.(36–41)

In regard to the overall effectiveness of the educational interventions, all but one of the studies reported that the intervention had significant associations with some of the outcome measures of interest, which included engagement, completion, attitudes, and knowledge of AD and ACP, and congruence in goals of care. For the one study directed at health care professionals,(37) outcomes examined competency and knowledge of health literacy, and cultural differences at end-of-life.

Of the two studies evaluated as being of “very high” quality, Lyon et al.’s (2019) RCT involved two, 1-hour ACP educational sessions targeting African American adults living with HIV and their surrogate decision-makers.(30) The ACP educational intervention took place at 5 HIV clinics and was compared to 2 educational sessions on developmental history and nutrition. This study measured documentation of AD and ACP in the medical record. A significant increase was found in AD completion (p < 0.0001) in the intervention versus control groups.

In Sudore et al.’s (2018) RCT, Spanish-speaking patients were targeted in the utilization of the “PREPARE for Your Care” intervention or a 5-step online interactive skill-building program designed especially for diverse patients and surrogates utilizing video stories, narratives, and testimonials to model how to engage in ACP.(27) The multi-lingual intervention was delivered in five, 10-minute videos and took place at four safety-net primary care clinics in San Francisco. The intervention was compared to easy-to-read AD and was found to significantly increase documentation of AD (p < 0.001) and ACP engagement scores (p < 0.001).

Community Support Interventions

Two of the 16 studies included in the review utilized volunteer peer(34) or patient navigators.(31) Navigators were trained to engage and empower patients to seek primary palliative care(31) or offer practical, emotional, and spiritual support to patients in the community.(34) These navigator interventions aimed to increase accessibility of palliative care services for persons from racial or ethnic URGs. More specifically they aimed to overcome barriers of accessibility associated with mistrust of medical providers, the acceptability of services, and geographic accessibility of palliative or end-of-life care services. Using the modified Downs and Black checklist, these studies were found to be of “low” to “moderate” quality.

In regard to the reported effectiveness of the navigator interventions, Fischer et al.’s (2018) intervention resulted in higher rates of ACP documentation (p < 0.001) and higher rates of ACP engagement scores (p < 0.001).(31) In Hanson et al. (2015), the intervention resulted in increased patient communication about cancer needs (p = 0.04), and increased awareness of Hospice (p=.04), although quality of life scores remained unchanged.(34) Hanson et al. (2015) included a non-randomized comparison group, patients who did not want a support team, and therefore the effectiveness of the intervention cannot be fully evaluated.

Clinical Care Interventions

One of the studies included in this review utilized an intervention of home-based palliative care, targeting an underserved, largely immigrant community in Honolulu, Hawaii.(42) The home-based palliative care program aimed to improve access to and deliver culturally sensitive palliative care. This study measured symptom relief, quality of life, community resource use, hospitalizations, ED visits, ACP, and care satisfaction. The study reported an improved sense of well-being (p = 0.03) in quality of life scale while no change was found in other quality of life domains. There was also a reported increase in discussion and documentation of end-of-life care preferences (p < 0.01), a decrease in hospitalization (p = 0.002), and no change in ED visits. The study lacked a comparison group and therefore the effectiveness of the intervention must be considered with caution.

Intervention Outcomes

Of the eight domains outlined by the NCP guidelines, the reviewed studies addressed seven. All of the studies addressed structure and process of care by measuring outcomes related to patient care plans (ACP or AD) and/or communication between patient and clinician or patient and caregiver. In addition, all of the studies addressed the domain of cultural aspects of care in some capacity as the interventions were culturally tailored or aimed at specific groups. Outcomes related to physical aspects of care were examined by Fernandes et al. (2010) and Fischer et al. (2018) who used the Edmonton Symptom assessment scale(42) and Brief Pain Inventory.(31) Five studies examined outcomes related to psychological and psychiatric aspects of care by looking at bereavement support,(36) quality of life,(34,42) decisional conflict and confidence scales, and self-perception tools.(28,29) One study,(34) examined spiritual aspects of care by including spiritual support as part of a patient support checklist. Six studies examined outcomes related to care of the patient nearing the end of life,(29,31,35–37,41) including openness to hospice, home death, and funeral burial plans,(36) willingness to accept end-of-life services,(37) hospice length of stay and ICU at the end of life,(31) attitudes and comfort with end of life,(35) quality of communication about the end of life,(28) and preferences for end-of-life care.(41) Finally outcomes related to ethical and legal issues were addressed by the eight studies using outcomes related to AD or ACP documentation.(27,30,32,35,38–40,42) No studies examined outcomes related to social aspects of care or the influence of social factors such as determinants of health.

The Socioecological Model

Of the 13 educational interventions, 12 targeted the individual level of the socioecological model,(27–30,32,35–41) and the remaining study, on health care professionals,(33) targeted the interpersonal level of the framework. One of the educational studies(40) targeted barriers to equity at the individual and community levels, including a community partnership with a local church and church leader. The two community support interventions(31,34) targeted barriers to equity in care at the community level, and the one clinical intervention(42) targeted barriers at the organizational level. There were no interventions targeting the policy level of the socioecological model.

Discussion

This review found a small number of serious illness care interventions aimed at increasing accessibility or improving outcomes for persons from racial or ethnic URGs. Of the sixteen (16) studies that met the inclusion criteria, most studies were quasi-experiments. Despite the breadth of research confirming disparities in care and health outcomes for URGs with serious illness, studies that utilize RCTs to assess the effectiveness of palliative care or ACP interventions at improving outcomes for URGs are lacking. Reviewed studies were designed for various racial or ethnic URGs. However, the absence of studies on Native American populations must be noted. While there were no existing intervention studies targeting Native American populations at the time of this review, there has been some formative work performed.(43)

The quality rating of the studies reviewed varied with only two randomized controlled trials rated “very high” quality demonstrating a clear need for the development of high-quality intervention research in the area. There was significant variability in the outcomes measured by included studies, making it difficult to compare the effectiveness of the interventions. Although most included studies reported significant findings in AD or ACP related outcomes (e.g., engagement, completion, attitudes, and knowledge of AD/ACP, etc.), there were limited findings related to the other measured outcomes such as quality of life, psychosocial-spiritual wellbeing, or pain management.

The reviewed interventions utilized three main strategies: education, community support, and clinical care. When considering these interventions within the context of the socioecological framework, the educational interventions target barriers to equity primarily at the individual level, while community support and clinical care interventions target barriers at the community and organizational levels, respectively. Overall, interventions targeting the interpersonal and policy levels of the socioecological model were lacking.

Education interventions were by far the most common and used various strategies in ways that intended to be responsive to the diversity of the population studied. These interventions worked at the individual level of the socio-ecological model, targeting individuals’ health behaviors to influence ACP or AD knowledge, attitudes, and completion. To address the issue of URGs completing ACP and AD at lower rates than non-Hispanic whites, the reviewed educational interventions utilized strategies of providing URGs with bilingual and ‘culturally sensitive’ ACP or AD materials. Strategies were creative and diverse in their approach, ranging from a video soap opera (telenovela) (33,37) to a bilingual, easy-to-read, interactive, online AD and ACP program.(27) Although our findings suggest possible effectiveness of educational interventions, there are limitations to individual-level models of health behavior change that may be particularly problematic when applied to improving outcomes for diverse populations. A person’s efforts to change their health practices are mediated by various educational, economic, cultural, or social factors.(44) Despite individual efforts, beliefs, or attitudes, structural inequities such as a poorly resourced hospital, may limit what palliative care services are available to patients.

Community-based interventions can improve accessibility and acceptability of health care services by building partnerships and trust within communities of URGs. The reviewed studies used laypersons, either patient navigators or volunteers, to build rapport with and reach URGs. The two reviewed studies utilizing peer navigation interventions were either a pilot trial or lacked a comparison group and did not show high effectiveness. Additional high-quality research is needed to explore how community-based interventions such as peer navigation may improve palliative or end-of-life care outcomes for persons from racial or ethnic URGs.

Clinical care interventions have the advantage of examining how existing clinical practice might be leveraged to improve outcomes for large groups of people. Often, they are scalable, making them good designs for pragmatic trials. While we found one, it lacked a comparison group, indicating an opportunity for future research in this area.

Implications for future research

It is recommended that future intervention research aiming to improve health disparities for URGs in serious illness care consider the complex range of variables that influence health disparities and health behaviors.(18) High quality randomized controlled trials are needed that engage both stakeholders and community members with the utilization of theoretical frameworks that target various levels of the socioecological model. Currently, the majority of intervention studies in this area of research utilize strategies that target individual-level factors influencing health disparities and consider outcomes related to ACP and AD knowledge, attitudes, and completion.

To better meet the needs of URGs and address issues of mistrust and the impact of systemic racism on health and health care, it is of particular importance that intervention models consider oppressive practices and structures that create barriers to equitable care at all levels of the socioecological framework. Future educational interventions may consider ways to include other levels of the framework, besides the individual, in their design. An approach that considers multiple levels of the socioecological framework will be better suited to incorporate outcomes that cover the eight domains determined important for intervention research by The National Consensus Project (NCP) for quality palliative care.(20) Of the eight domains, in particular, research is needed that examines the influence of social factors, such as determinants of health, on outcomes. A recent report from the National Academies of Science, Engineering, and Medicine (NASEM) confirmed the importance of addressing social factors in health equity research.(45) The report highlights the potential for integrating health services that address social needs and the social determinants of health to achieve health equity. It emphasizes the need to move upstream, addressing the impact of the social determinants of health on health outcomes.

Future research should consider the unique needs and barriers of persons from different URGs and specific disease presentations. Research should strive to target interventions, tailoring them for specific groups such as the Native American population, who are largely absent in research sampling, or undocumented immigrants who face unique barriers in accessing hospice and palliative care services.(46) Future research may consider what intervention components and strategies (other than language) are relevant to each unique URG and explore and incorporate clearly defined recruitment strategies for study samples. Future research may also consider how interventions to improve outcomes for persons from racial or ethnic URGs may need to be tailored for specific disease presentation. Interventions to improve serious illness care for persons living with dementia, a growing population(47) who often experience extended end-of-life trajectories,(48,49) experience difficult-to-manage symptoms,(50) and are disproportionately non-Hispanic Black and Hispanic,(47) are of particular importance. However, targeted interventions can focus on the unique needs and barriers of populations experiencing a broader range of the leading causes of serious illness and death, such as heart failure and COPD.

Lastly, some existing studies focused on broader populations have been successful in recruiting large numbers of URGs. However, the interventions are not targeted and have not performed sensitivity analyses by race.(51) To better understand the effectiveness of large scale interventions for populations from URGs, sensitivity analysis by race or ethnicity must be included in research analysis.

Strengths and limitations of the current review

A comprehensive systematic review was conducted to include relevant publications on serious illness care intervention research for URGs in the United States. Anticipating the limited research in this area, the inclusion criteria were broad to include a range of experimental study designs, participants, interventions, and outcomes that shared the goal of improving care for racial and ethnic underrepresented populations in palliative care. In doing so this review was able to compare, evaluate, and highlight gaps in the intervention strategies being employed to reduce health disparities for URGs in serious illness care. Due to the heterogeneous nature of the studies, formal evaluation of intervention effectiveness was not possible. Although attempts were made to be as inclusive as possible in the data search strategy, it is conceivable that intervention studies whose terms, title, or abstract did not match this review’s search terms, or whose publications were reports from programs with results that were not searchable, may have been missed. In addition, the review was limited to studies published in the English language, thus it may be missing other potentially relevant studies written in non-English languages. Finally, all of the included studies were US-based in order to focus on the unique health care system and related challenges occurring in the US. As a result, the findings of this review may not be generalizable to other countries and do not cover potentially effective strategies to reduce disparities being employed internationally.

Conclusion

The effectiveness of ACP and serious illness care interventions in improving outcomes for URGs is uncertain. Two RCTs rated “very high quality” indicate that interventions targeting URGs can have significant and positive effects on AD and/or ACP-related outcomes. Additional high-quality educational interventions report significant and positive effects on ACP-related outcomes and caregiver stress. The majority of these studies were, however, evaluated as “moderate” or “low” quality mostly due to a lack of a comparison group. High-quality intervention studies that address the complex multi-level factors of health disparities for URGs with serious illness are desperately needed to increase equitable access, use, and quality of life.

Key Message.

This review article summarizes and evaluates existing palliative and end-of-life care interventions to improve outcomes for racial and ethnic underrepresented populations in the US. Results indicate the interventions’ effectiveness in improving outcomes for these populations remains uncertain. High-quality intervention studies addressing racial and ethnic inequities in palliative care are needed.

Search Strategy - Medline Ovid.

1. palliative care.mp. or Palliative Care/ or terminal care.mp. or Terminal Care/ or advance care planning.mp. or Advance Care Planning/ or end-of-life.mp. or Hospices/ or hospice.mp. or goals of care.mp. or Advance Directives.mp. or Advance Directives/

   AND

2. Minority Groups/ or Ethnic Groups/ or ethnic minorities.mp. or african american.mp. or African Americans/ or African Continental Ancestry Group/ or african americans.mp. or hispanic.mp. or Hispanic Americans/ or latino.mp. or asian american.mp. or Asian Americans/ or asian americans.mp. or Asian Continental Ancestry Group/ or Oceanic Ancestry Group/ or American Native Continental Ancestry Group/ or minority groups.mp.

   AND

3. ((randomized controlled trial or controlled clinical trial).pt. or randomized.ab. or placebo.ab. or drug therapy.fs. or randomly.ab. or trial.ab. or groups.ab.) not (exp animals/ not humans.sh.)