Impact of Specialty and Nonspecialty Palliative Care on Quality of Dying With Alzheimer's Disease or Related Dementias

Abstract

Background:

Older adults with Alzheimer's disease and related dementias can benefit from palliative care (PC). Whether specialty and nonspecialty PC have the same effect on outcomes is unclear. We examined the effects of these 2 interventions on comfort, symptom management, satisfaction with care, and potentially burdensome transitions, including hospital admission, emergency department visit, intensive care unit admission in the end-of-life, and in-hospital death.

Methods:

This PRISMA-adherent systematic review involved a search of PubMed, Medline, EMBASE, Cochrane Library, ProQuest, and CINAHL for studies published from January 1, 2013, to November 4, 2024. Primary studies that reported at least one of the 7 patient-level outcomes were included: Comfort Assessment in Dying with Dementia (CAD-EOLD), Symptom Management at the End-of-Life (SM-EOLD), Satisfaction with Care at the End-of-Life in Dementia (SWC-EOLD), hospital admissions, emergency department visits, intensive care unit admissions, and in-hospital death.

Results:

Nineteen articles involving 142,772 participants were included. The evidence, comprising studies of adequate to strong quality, revealed that both specialty and nonspecialty PC did not differ in terms of comfort, symptom management, or satisfaction with care. However, both approaches significantly reduced the likelihood of intensive care unit admissions and in-hospital deaths. Specialty PC was associated with decreased emergency department visits (OR 0.53, 95% CI 0.28–1.00; I ² =86%).

Conclusions:

Future research is needed to understand factors influencing PC interventions that can improve comfort, symptom management, and care satisfaction for these individuals and their families.

Estimates indicate that the population aged ≥85 years will more than double between 2014 and 2040. This growing aging population is expected to contribute significantly to the increasing burden of dementia, with many individuals requiring palliative care (PC) as their disease progresses, making dementia a major contributing factor.¹ Research found that people living with Alzheimer's disease and related dementias (ADRD), including those with vascular dementia or dementia with Lewy bodies/Parkinson's, mixed dementia, or Alzheimer's disease, corresponded to over 9 years of life lost compared with the general population.² These groups of older adults often suffer from various medical comorbidities, such as pneumonia, urinary tract infection, and urinary and stool incontinence.^3,4 They also have an altered presentation of illness during the end-of-life (EOL) care due to loss of cognitive function, including inability to localize pain, minimal verbal communication with important others, inability to perform activities of daily living, poor oral intake, and memory deficits.³ In addition, it was common for persons with ADRD to have behavioral and psychological symptoms, including affective symptoms, anxiety, psychosis, agitation, aggression, disinhibition, and wandering, which is a leading cause of placement in care facilities. These factors not only negatively impact the quality of life but also increase the burden of care, particularly during the final weeks or months of EOL care. They can result in high rates of acute care utilization and a greater likelihood of in-hospital deaths due to burdensome transitions.⁵ Given the typical median survival time of 1 year for these older adults, long-term support care services are often essential.⁶

Indicators of burdensome and aggressive treatment in this population include unscheduled hospital visits, emergency department (ED) visits, and in-hospital deaths. Assessments of symptom control and comfort are also used to evaluate the quality of dementia-specific EOL care.^7,8 It has been found that older patients are less likely to die in a hospital if they have discussed their EOL care preferences with health care professionals.⁹ Family satisfaction reflects effective communication about prognosis and treatment goals.⁷

PC can reduce acute care use by improving symptom management and patient comfort.^4,10 PC is defined as “an approach that improves the quality of life of patients and their families who are facing problems associated with life-threatening illness,” according to the World Health Organization (WHO).¹¹ PC, which is often confused with EOL care, is often delivered by a multidisciplinary team of clinicians who specialize in addressing complex symptom management and emotional support needs of patients and families. Such care can be provided by both clinicians with specialist PC training, as well as non-PC specialists who provide primary care, such as geriatricians, family medicine physicians, and general internists.

Access to PC is challenging due to limited PC specialist availability. Further, given that the trajectory of ADRD is both prolonged and unpredictable, accessing specialty PC for extended periods of time may be unsustainable.¹ Research shows that only 5.9% of patients with dementia in the community consulted a specialty PC team, while 2.0% each had consultations with specialist nurses or hospice services in the last 3 months of their lives.¹² Further, patients and their families often face difficult decisions about specialty PC for many reasons. These may include uncertainty about life expectancy, the patient’s ongoing difficulty in expressing their care preferences, and a lack of clarity regarding how coexisting medical conditions might influence their prognosis.¹³

There have been conflicting results regarding the benefits of PC for persons with ADRD. For example, a study of 25,278 individuals who died of dementia found conflicting results regarding the impact of PC services. It showed that PC did not affect ICU admissions, but increased rates of admissions to hospital in the last 6 months of life and reduced the likelihood of dying at home or in a nursing home.¹⁴ A recent systematic review also found that changing organizations to access specialty PC for patients with ADRD had very low certainty of improving the quality of dying.¹⁰ Given the low utilization of specialty PC, it is necessary to evaluate whether specialty or nonspecialty PC has a differing impact on persons with ADRD approaching the endoflife. To address this need, this review aimed to explore the effectiveness of specialty and nonspecialty PC on comfort, symptom management, satisfaction with care, and potentially burdensome transitions among older adults dying with ADRD. This area of study may provide insight into which care model would best benefit from integrated EOL dementia care.¹⁵

METHODS

A systematic review and meta-analysis were conducted to determine the impact of specialty and nonspecialty PC, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Protocol (PRISMA-P) guidelines.¹⁶

Search Strategy and Selection Criteria

Electronic searches were performed using 6 databases PubMed, Medline, EMBASE, Cochrane Library, ProQuest, and CINAHL from January 1, 2013, to November 4, 2024, in which a series of coordinated and patient-centered dementia care programs in various settings had been implemented over time. Synonyms, combined with Medical Subject Heading (MeSH) and Boolean modifiers such as “OR,” “AND,” and “NOT” were employed. Subsequently, a manual search was performed using Google Scholar, and WHO website. The search terms are presented in Supplement 1, Supplemental Digital Content 1, http://links.lww.com/MLR/D47.

In this study, “specialty PC” refers to any PC services primarily provided by clinicians or nurses who are either PC certified or subspecialty trained. Conversely, PC or EOL care services, as well as comprehensive geriatric interventions involving ACP, that are led by clinicians or teams not trained in specialty PC are classified as “nonspecialty PC.” Studies with a mixed population were included if (1) over half of the participants were diagnosed with dementia of any type, or (2) the data were reported separately. Studies on training the health care team in knowledge and skills to improve comprehensive geriatric EOL care or PC were also included.

The researchers compared PC exposure and at least one of the outcomes of the effect using the validated scales Comfort Assessment in Dying with Dementia (CAD-EOLD), Symptom Management at the End-of-Life (SM-EOLD), and Satisfaction with Care at the End-of-Life in Dementia (SWC-EOLD),¹⁷ as well as hospital admissions, ED visits, intensive care unit (ICU) admissions, and in-hospital deaths. Evaluators of the quality of care could be people with ADRD, their family members, clinicians, or nurses. Publications in English and Mandarin were considered based on the language competencies of the reviewers. Studies were excluded if they did not focus on addressing the holistic needs of patients, for example, by implementing bereaved family caregivers’ feedback¹⁸ or only ACP communications.¹⁹ Cross-sectional descriptive studies and qualitative studies were excluded.

Study Selection

The 3 reviewers independently screened articles retrieved from the database search based on their titles and abstracts. Full-text articles were assessed according to the inclusion and exclusion criteria. Disagreements were resolved by a consensus between the 2 authors. The process is presented in a PRISMA 2020 flow diagram (Fig. 1).

FIGURE 1.

PRISMA 2020 flow diagram.

Quality Assessment

The risk of bias was independently rated by 2 of the 3 investigators using the Joanna Briggs Institute appraisal checklists for different study types. The checklist assessed the quality and bias of the studies in terms of design, procedures, and analyses. Scores were given as 2, 1, and 0 for yes, unclear, and no answers, respectively. Percentages were derived from the sum of the scores divided by the total score of the relevant items. Scores were expressed as a percentage, where >80% was considered strong, >70% good, >50% adequate, and ≤50% limited.²⁰ Studies were not excluded based on quality assessment.

Data Extraction

The main outcomes and measures included the effectiveness of comfort, symptom management, health care satisfaction, health care utilization during EOL care (eg, admission to hospital, ED visit, and ICU admission), and in-hospital death. Two of the 4 investigators extracted relevant data. When there was missing data, the study authors were contacted to request the missing information.

Data Synthesis and Analysis

The effects of specialty and nonspecialty PC models were reported using the standard mean difference (SMD) for continuous outcomes, odds ratio (OR) for dichotomous outcomes, and 95% CI. Statistical heterogeneity was measured by applying the χ² and I ² tests. If I ² >30%, indicating moderate to substantial heterogeneity, data were analyzed with the random effects models; otherwise, the fixed effects models were adopted due to heterogeneity.²¹ Sensitivity analyses were performed if statistical heterogeneity was present. A meta-analysis was performed in Review Manager, Version 5.4, The Cochrane Collaboration, 2020, and the statistical significance level (P-value) was set at 0.05.

RESULTS

The database search identified 8124 articles. After deduplication, 5350 studies remained for title and abstract screening, and 119 full-text studies were further assessed for eligibility. After assessing eligibility and citation searching, 19 articles that involved 142,772 participants were retained: 6 RCTs,^22–27 2 quasi-experimental,^28,29 1 cross-sectional,³⁰ 8 cohort,^14,31–37 and 2 mixed-methods studies.^38,39 Of the 2 mixed-methods studies, one involved a quasi-experimental design,³⁸ while the other involved a cohort study.³⁹ Table 1 contains the study characteristics. Eight studies were conducted in the United States,^{26,29–31,34,35,37,39} 3 in Belgium,^24,27,36 3 in Canada,^14,29,32 2 in England,^25,38 and one each in Australia,²³ Finland,²² and Sweden.³³ Out of 19 included studies, 3 studies focused on persons with ADRD staged as Functional Assessment Staging Tool (FAST) ≥6a,^23,25,29 2 focused on persons with Global Deterioration Scale ≥5,^26,29 and one focused on those with Cognitive Performance Scale score ≥3.³⁵ Ten studies evaluated specialty PC services,^{23,26,27,29,31–33,35–37} 8 studies evaluated nonspecialty PC services,^{22,24,25,28,30,34,38,39} and one focused on both.¹⁴

TABLE 1.

Characteristics of the Studies Included (n=19)

Study, country, year	Participants with dementia and sample size for meta-analysis	Setting	Intervention description	Personnel who provided PC program, enrollment duration	Relevant methods; sources of outcomes measured	Patient characteristics*: PC vs. usual care	PC outcomes: PC vs. usual care
Agar, Australia, 201724	Residents with advanced dementia (staged FAST 6a or above), and Australia-modified Karnofsky Performance Status ≤50; living in residential care. PC n=67; Usual care n=64	Twenty nursing homes in 2 major Australian cities.	Facilitated case conferencing. A registered nurse was trained as a PC planning coordinator, who worked for 2 d per week or equivalent to: (1) identify residents with advanced dementia likely to benefit from a case conference; (2) organize case conferences with optimal participation by family, multidisciplinary nursing home staff, and external health professionals; (3) develop and oversee implementation of PC plans; and (4) train nursing and direct care staff in person-centered PC.	Specialty PC coordinators; Median (IQR)=6 (10) months	Cluster RCT; nursing home and medical records, face-to-face or telephone interviews with nurses and family members.	Age, mean±SD: 84.7±7.9 vs. 85.8±8.2; Female, n (%): 41 (61.2) vs. 37 (57.8) Died in a nursing home, n (%): 58 (88) vs. 55 (89)	Family-rated, mean±SD CAD-EOLD during the last 7 days of life: 34.7±5.9 vs. 35.5±5.9, ns SM-EOLD in the last 90 d of life: 29.0±9.5 vs. 31.7±7.4, ns SWC-EOLD during the last 90 d of life: 31.0±5.3 vs. 30.3±4.2, ns Nurse-rated, mean±SD CAD-EOLD 32.1±6.1 vs. 33.3±5.7, ns SM-EOLD 22.4±9.6 vs. 23.2±8.3, ns Care in the last month of life ≥1 hospital admission visit n (%): 13 (19) vs. 11 (18), ns ≥1 ED presentation without hospital admission, n (%): 6 (9) vs. 6 (10), ns
Beernaert, Belgium, 201725	Persons admitted to the acute geriatric ward, died on the ward, and had been in the hospital for >48 h. PC n=164; usual care n=118	Acute geriatric wards in 10 hospitals.	Training the geriatric health care staff: (1) implementation guide; (2) care guide for the last days of life; (3) supportive documentation such as information leaflets for family carers about entering the dying phase.	Geriatric team; >4 wk: 33%; ≥2 wk to <4 wk: 38%; ≥1 wk to <2 wk: 17%; ≥72 h to <1 wk: 10%; <72 h: 3%	Cluster RCT; medical or nursing records, questionnaire collected from the nurse and physician most closely involved in the patient’s care.	Mean age at death: 85.8±6.8 vs. 84.0±7.5 Female, n (%): 62 (47) vs. 48 (44) Had dementia at the time of death, n (%): 74/93 (80) vs. 35/80 (44) Cancer, n (%): 14/108 (13) vs. 19/89 (21) Residence of a nursing home or other institution before hospital admission, n (%): 34/110 (31) vs. 28/91 (31)	Family-rated, mean (95% CI) CAD-EOLD: 29.8 (27.3–32.3) vs. 32.0 (28.9–35.0), ns SM-EOLD (modified): 5.2 (4.0–6.5) vs. 5.0 (3.3-6.7), ns SWC-EOLD: 29.6 (28.1–31.5) vs. 33.5 (31.2–35.8), P=0.04 Nurse-rated, mean (95% CI) CAD-EOLD: 34.6 (33.3–35.9) vs. 31.3 (29.9–32.7), P<0.0001 SM-EOLD (modified): 4.4 (3.5–5.4) vs. 5.9 (4.9–6.9) ns
Cassel, United States, 201632	Medicare Advantage insurance holders who agreed to participate in the Transitions program and had dementia. PC n=92; Usual care n=276	Community in southern California.	A home- and clinic-based PC program provided by a multidisciplinary team, which has 4 components: in-home medical consultation, ongoing evidence-based prognostication of further survival, caregiver support, and ACP.	Specialty PC team; average 215.3 d (1–18 mo)	Observational, retrospective study using propensity-based matching; medical records data and billing and claims data.	Age, mean±SD: 87.0±5.9 vs. 87.0±6.1; Female, n (%): 57 (62.0) vs. 185 (67.0)	Care in the last month of life Number of hospitalizations/month, mean±SD: 0.1±0.3 vs. 0.3±0.3, P<0.001 Number of hospital days/month, mean±SD: 0.8±2.1 vs. 1.7±2.6, P<0.001 Admitted within 30 d before death, %: 17.4 vs. 63.0, P<0.001 30-day readmission rate, mean: 0.1 vs. 0.4, P<0.01 ICU admission in the last 30 d of life, %: 8.7 vs. 34.4, P<0.001 Dying in hospital, %: 5.4 vs. 51.1, P<0.001
Catic, United States, 201330	Hospitalized individuals aged ≥65, admitted to any clinical service for 48 h or less, advanced dementia based on Global Deterioration Stage 7 criteria. English-speaking proxy is available. PC n=5; Usual care n=24	A 631-bed teaching hospital in Boston.	Geriatricians and a PC nurse practitioner conducted consultations, which consisted of structured consultation, counseling, and provision of an information booklet to the family, and postdischarge follow-up with the family and primary care providers.	Specialty PC team; not reported	Before-and-after feasibility pilot study; electronic medical record and interviews.	Age, mean±SD: 90.0±2.8 vs. 84.4±7.2; Female, n (%): 3 (60.0) vs. 15 (62.5) Residence in a nursing home, n (%): 4 (80.0) vs. 13 (54.2)	One month after discharge Referred to hospice n (%): 2 (40.0) vs. 6 (25.0) ≥1 ED visits n (%): 0 (0) vs. 8 (33.3) ≥1 rehospitalization, n (%): 0 (0) vs. 7 (29.2) SM-EOLD (patient), mean±SD: 28.4±9.5 vs. 28.5±9.7 SWC-EOLD (proxy), mean±SD: 30.8±7.0 vs. 29.9±4.5
Earp, Canada, 202133	Adults deceased from dementia, vascular dementia, Alzheimer's disease, senility between 2007 and 2016. PC n=1,098 (Early ≥90 d before death: n=321; Late ≥8 but <90 d: n=457; Very late <8 d: n=320); no PC n=3,912	In the Calgary Zone of Alberta Health Services.	An integrated program that includes PC consult teams (institutional, community-based, and cancer pain and symptom clinic), a tertiary PC unit, palliative home care, and hospices (institutional and community-based).	Specialty PC team; ≥90 d: 321(29) ≥8 but <90 d: 457(42) <8 d: 320 (29)	Retrospective cohort study; administrative data.	Female, n(%): 3256 (65.4)	In the last month of life, compared with no PC: very late PC, RR (95% CI) >1 ED visits: 1.04 (1.01–1.06), P=0.01 >1 hospital admission: 1.04 (1.01–1.07), P=0.004 ≥1 ICU admission: 0.99 (0.99–1), P=0.001 >14 d in hospital: 1.16 (1.12–1.21), P<0.001 Hospital death: 1.26 (1.21–1.3), P<0.001 Late PC, RR (95% CI) >1 ED visits: 0.99 (0.97–1.01), P=0.21 >1 hospital admission: 1.04 (1.01–1.06), P=0.002 ≥1 ICU admission: 1 (0.99–1.01), P=0.72 >14 d in hospital: 1.19 (1.15–1.24), P<0.001 Hospital death: 1 (0.97–1.04), P=0.998 Early PC, RR (95% CI) >1 ED visits: 1 (0.98–1.02), P=0.83 >1 hospital admission: 0.99 (0.98–1), P=0.16 ≥1 ICU admission: 1 (0.99–1), P<0.001 >14 d in hospital: 0.97 (0.94–1), P=0.04 Hospital death: 0.94 (0.91–0.98), P=0.001 Early PC reduced the risk of hospital death.
Froggatt, England, 202026	Residents with advanced dementia (FAST 6 or 7), their informal carers, and nursing home staff were recruited between August and November 2017. 4 wk Namaste Care n=17, Usual care n=14 24 wk Namaste Care n=10, Usual care n=7	Eight nursing homes in England.	Namaste Care is a complex dementia group intervention that provides structured personalized care in a dedicated space, focusing on enhancements to the physical environment, comfort management, and sensory engagement.	Primary care team; 1–6 mo	Feasibility cluster RCT; questionnaires, observation, interviews, completion of a session activity log, and device use.	Namaste care + PC vs. PC Age, mean±SD: 79.0±10.5 vs. 84.7±6.4 Female, n (%): 10 (55.6) vs. 5 (35.7) Residence in a nursing home (%): 17 (100) vs. 14 (100)	CAD-EOLD (staff) after 4 wk, mean±SD: 36.4±4.0 vs. 33.4±3.4 CAD-EOLD (staff) after 24 wk, mean±SD: 37.6±2.9 vs. 33.6±1.9 SWC-EOLD (informal carer) after 4 wk, mean±SD: 31.3±5.9 vs. 29.5±2.4
Fürst, Sweden, 202234	Persons over the age of 65 y who had died between 2015 and 2019 with a diagnosis of advanced cancer and dementia. PC n=253; Usual care n=348	Acute hospitals, nursing homes.	Specialty PC is offered mainly in the form of advanced home PC, or the form of hospital inpatient PC. Both types of care are staffed by physicians, registered nurses, physiotherapists, occupational therapists, dietitians, and others, of which physicians and nurses are available around the clock.	Specialty PC team; not reported	Retrospective observational registry study; health care consumption data from the Stockholm Regional Council.	On the basis of n=605 (lost n=4) Age, mean±SD: 82.9±7 Female, n (%): 319 (53) Malignant neoplasm, n (%): 605 (100) Residence in a nursing home, n (%): 325 (54)	ED visits last month, n (%) 106 (42) vs. 153 (44), P=0.61 Hospitalizations last month, n (%) 117 (46) vs. 142 (41), P=0.18 In-hospital death, n (%) 8 (3) vs. 73 (21), P<0.0001
Gotanda, United States, 202331	A random sample of fee-for-service Medicare beneficiaries with dementia who died in 2016–2018 was studied. Beneficiaries were excluded if they (1) died before July 1, 2016, (2) were long-term residents of nursing homes, or (3) had outcomes that could not be attributed to a physician. Physicians with geriatric training n=8340; Usual care n=91,291	Long-term residents in Medicare- or Medicaid certified nursing homes in the United States.	Had received EOL care from physicians trained in geriatrics with self-reported specialties in geriatric medicine or board certification in geriatric medicine.	Physicians with primary care specialties with geriatric training	Cross-sectional study; linked 3 databases: (1) 20% of Medicare claims data 2016–2018, (2) long-term care Minimum Dataset 3.0 2016–2018, (3) Medicare Data on Provider Practice and Specialty data, and (4) Doximity data.	Care provided by physicians with vs. without geriatric training Age, mean±SD: 87.1±7.7 vs. 85.3±8.0 Female, n (%): 5,005 (60.0) vs. 52,345 (57.3) Cancer, n (%): 1,269 (15.2) vs. 14,910 (16.3) Residence in a nursing home (%): 0 vs. 0	In the last month of life Adjusted proportion ED visits, %: 47.5 vs. 51.5, P<0.001 Hospital admissions, %: 41.9 vs. 45.2, P<0.001 ICU admissions, %: 23.8 vs. 26.8, P<0.001
Hanson, United States, 201927	Dyads of hospitalized patients with late-stage dementia (dementia stage 5–7 on the Global Deterioration Scale) and family decision-makers PC n=30; Usual care n=32	A single site with robust clinical resources in PC and geriatric care.	(1) Receive specialty PC consultation during hospitalization with postdischarge telephone support by a PC nurse practitioner (2) Usual hospital care with educational material on dementia caregiving	Specialty PC team; not reported	RCT; postdischarge telephone family interviews.	Age, mean±SD: 83.0±8.8 vs. 84.7±8.7, P =0.434 Female, n (%): 20 (66.7) vs. 15 (46.9), P =0.116 Preadmission residence, n (%): P=0.107 Home 13 (43.3) vs. 19 (59.4) Assisted living 9 (30.0) vs. 2 (6.3) Nursing home 7 (23.3) vs. 9 (28.1) Other 1 (3.3) vs. 2 (6.3)	CAD-EOLD (family), mean±SD: 34.8±4.2 vs. 34.0±4.1, P=0.52 60-day hospital/ED visits (no. of events/follow-up days): 0.68 vs. 0.53, P=0.42
Jennings, United States, 201935	Decedents enrolled in a comprehensive dementia care management program from 2012 to 2016. PC n=184; Usual care n=138	Urban, academic medical center.	A comprehensive dementia care management program involved dementia service nurse practitioners supervised by a physician dementia specialist, partnering with primary care providers and community organizations (1) structured needs assessments of patients and their caregivers; (2) creation and implementation of individualized dementia care plans; (3) monitoring and revising care plans, and specifically addressing ACP; (4) referral to community organizations for dementia-related services and support; (5) access to a clinician 24 h a day, 7 d a week for assistance and advice.	Geriatric team; 14.8 mo on average (SD=10.1 mo; range=1–44 mo)	Observational, retrospective cohort; electronic health record data.	Age at death (range, 56–102), mean±SD: 86.7±7.9 Female sex, n (%): 174 (54)	In the last 6 mo of life Any hospitalization, n (%): 79 (43) vs. 43 (31), P=0.04 Any ED visit, or observational stay, n (%): 53 (29) vs. 32 (23), P=0.27 Hospital length of stay (range, 2–51 d), median (IQR), d: 5.8 (3.7–8.3) vs. 4.1 (3.1–8.9), P=0.22 Any ICU stay, n (%): 11 (6) vs. 6 (4), P=0.62 ICU length of stay (range, 0.3–27.3 d), median (IQR), d: 2.0 (1.0–3.4) vs. 5.8 (0.4–11.7), P=0.41 Place of death, n (%) Hospital ward or ED: 27 (15) vs. 22 (16), P=0.75 ICU: 3 (2) vs. 4 (3), P=0.45 Hospital: 30 (16.3) vs. 26 (18.8), P=0.555
Lamppu, Finland, 202123	Long-term care facility residents via proxy were judged likely to have a prognosis of less than 12 mo. PC n =151; Usual care n =173	20 Long-term care wards in Helsinki.	Four 4-hour educational workshops for staff on PC principles: (1) ACP, (2) adverse effects of hospitalizations, (3) symptom management, (4) communication, (5) supporting proxies, (6) challenging situations.	Primary care team; ≥12 mo: 249(76.9) ≥6 but <12 mo: 42(13.0) <6 mo: 33(10.2)	Single-blind, cluster RCT using a questionnaire survey.	Age, mean±SD: 83±8 vs. 84±8 Female, n (%): 115 (76) vs. 130 (75) Severe dementia, n (%): 91 (60) vs. 112 (65) Cancer, n (%): 11 (7) vs. 10 (6)	SWC-EOLD (proxy), mean±SD: 28.7±5.2 vs. 27.8±5.1, P=0.14
Livingston, UK, 201339	Residents with dementia living in a care home for >1 mo when they died. PC n=42 No PC n=56	A 120-bed nursing home in London, UK	Ten-session manualized, interactive staff training program. The topics were: the challenges of dementia EOL care; the physical process and symptoms; emotional and psychological needs; planning for care; care planning and communication with residents and relatives; religion and spirituality; holistic care for people with dementia at the EOL and finally a summarizing and reflective session.	A consultant physician and care home senior managers, who were trained in the Gold Standard Framework (a program for enhancing EOL care).	Mixed-methods study; interviews, questionnaires.	Mean age at death: Age, mean±SD: 88.1±7.1 vs. 85.5±7.9 Female, n (%): 29 (69.0) vs. 30 (53.0)	In the last 3 mo of life Hospital length of stay: median (IQR), day: 1.3 (9.5) vs. 4.0 (15.8), P=0.22 Hospital death, n (%): 7 (24.1) vs. 16 (53.3), P =0.02
Miller, United States, 201736	Residents with consultations, moderate to very severe dementia, with the Cognitive Performance Scale score ≥3 (moderate to advanced cognitive impairment), and deaths in 2006–2010. PC n=203 (Earlier PC n=91; Later PC n=112); no PC n=2,756	31 nursing homes in 2 states in the United States.	Initial specialty PC consultations made by nurse practitioner PC specialists, under the supervision of certified PC physicians. Interdisciplinary PC team members visit residents when needs exist, reviewing advance directives, and discussing goals of care.	Specialty PC team; ≤1 mo, n (%): 112 (55.2) >1 but ≤6 mo: 91 (44.8)	Retrospective cohort study; nursing home consultation data from 2 PC organizations.	Age at baseline, %: 40–74 y: 9.4 vs. 5.9 75–79 y: 13.8 vs. 9.2 80–84 y: 26.1 vs. 18.9 85–89 y: 19.7 vs. 26.7 90–94 y: 23.2 vs. 25.9 ≥95 y: 7.9 vs. 13.5 Female, %: 61.6 vs. 68.5 Cancer, %: 15.3 vs. 15.8 Residence in a nursing home, %: 100 vs. 100	Hospitalization in last 7 d of life, %: PC vs. no PC: 16.0 vs. 38.8 Later PC vs. no PC: 11.6 vs. 20.5 Earlier PC vs. no PC: 4.4 vs. 18.3 Hospitalization in last 30 d of life, %: Earlier PC vs. no PC: 7.7 vs. 24.5 ED visits in last 30 d of life, %: Earlier PC vs. no PC: 5.5 vs. 17.6
Miranda, Belgium, 202137	All home-dwelling older people, who resided at home and did not have a recorded nursing home stay in the last 6 mo of life, died with dementia in 2010–2015 in Belgium and received palliative home care support for the first time between 360 and 15 d before death. PC n=2,839; Usual care n=2,839	Palliative home care support in Belgium.	Receiving home visits by a multidisciplinary palliative home care team; home visits by a PC nurse or physiotherapist, or receiving financial allowance for PC.	Specialty PC team; 15–360 d	Matched decedent cohort study; 8 routinely collected nationwide administrative databases.	Age, mean±SD: 83.4±6.5 vs. 83.5±6.2 Female, %: 48.7 vs. 47.5 Cancer, %: 28.3 vs. 27.2 Care dependence, %: 76.8 vs. 78.1 Residence in nursing home (%): 0 vs. 0	In the last 14 d of life, % ≥1 ED visits: 10.3 vs. 23.9; RR=0.37 ≥1 Hospital admissions: 17.5 vs. 50.5; RR=0.21 ≥1 Admissions to ICU: 0.5 vs. 4.4; RR=0.12 Died in hospitals: 24.3 vs. 67.5; RR=0.15
Nguyen, United States, 202240	Patients with ADRD diagnosis code at the Kaiser Permanente Southern California, Panorama City Medical Center from January 1, 2015, to July 31, 2020. Home-based primary care n=287 (decedents n=159) Usual care n=861 (decedents n=423)	One medical service area in the United States.	Primary care services are offered in private homes, assisted living facilities, or board and care homes by an interdisciplinary team. Team includes (1) nursing case management; (2) social worker available through the clinics or other programs; (3) pharmacist from the geriatric clinic; (4) after-hours triage with PC nurses and on-call physicians for urgent needs.	Geriatric and/or PC trained physicians and nurse practitioners; 357 d (IQR: 162–743)	Mixed-methods study used a retrospective observational cohort design; electronic health records and claims.	Age, mean±SD: 85.9 (8.1) vs. 85.0 (9.5) Female, n (%): 190 (66.2) vs. 570 (66.2)	Hospital death, n (%): 16 (10.1) vs. 78 (18.4)
Quinn, Canada, 202015	Adults who died from dementia and reside in a nursing home in Ontario between 2010 and 2015. PC n=9,255; Usual care n=16,023	Hospitals and the community in Ontario, Canada.	Specialty or primary PCs are provided in hospitals, outpatient clinics, and the home.	Generalists (eg, family doctor or other non-PC specialists) or specialty PC team; >7 d but ≤6 mo	Population-based matched cohort study; linked clinical and health administrative databases.	Age, mean±SD: 84.3±9.0 vs. 84.1±8.9 Female, n (%): 13,700 (58.9) vs. 23,590 (58.9) Cancer, n (%): 1,494 (6.4) vs. 2,117 (5.3)	In patients dying from dementia Admissions to hospital, crude rate±SD: 3.6±8.2 vs 2.8±7.8 per person-year; adjusted rate ratio=1.33; 95% CI: 1.27–1.39 ED visits, crude rate±SD: 1.2±4.9 vs 1.3±5.5 per person-year; adjusted rate ratio=1.06; 95% CI: 1.01–1.12 ICU admission, crude rate±SD: 0.2±2.1 vs. 0.2±2.1 per person-year; adjusted rate ratio=1.03; 95% CI: 0.96–1.11 Dying at home or in a nursing home, n (%): 6667 (72.1) vs. 13384 (83.5); adjusted OR 0.68; 95% CI: 0.64-0.73
Sharda, United States, 202038	Older adults aged 65 years or older living with dementia, were hospitalized in 2015. PC n=157; Usual care n=157	Tertiary academic medical center and affiliated community hospital.	Inpatient PC consultations	Specialty PC team; not reported	Retrospective study; electronic medical record review.	Age, mean±SD: 84.3±8.3 vs. 83.9±8.0, P=0.414 Female, n (%): 98 (62.4) vs. 99 (63.1), P=0.907 Residence of a nursing home or assisted living facility, n (%): 72 (45.9) vs. 48 (30.6)	Hospital death, n (%): 21 (13.4) vs. 6 (3.8)
Van den Block, Belgium, 202028	All residents who had died in a list of nursing homes within a specific geographical area in Belgium, England, Finland, Italy, the Netherlands, Poland, and Switzerland were recruited between 2015 and 2017. PC n=425; usual care n=558	73 nursing homes.	The PACE Steps to Success Program: (1) ACP with residents and families; (2) assessment, care planning, and review of resident needs and problems; (3) coordination of care via monthly multidisciplinary PC review meetings; (4) high-quality care with a focus on pain and depression; (5) care in last days of life; (6) care after death.	Specialty PC team; 6 months implementation of the PACE Steps to Success Program, and 4 mo consolidation with ongoing support where needed	Multifacility cluster-randomized clinical PACE trial; questionnaire survey.	Age at time of death, unadjusted mean±SD: 85.9±8.6 vs. 85.6±8.8 Female sex, unadjusted frequency, n (%): 265/414 (64.0) vs. 354/547 (64.7) Functional status 1 mo before death, unadjusted mean±SD: 18.8±5.1 vs. 19.0±4.9	In the last week of life CAD-EOLD (family) mean (95% CI): 29.9 (28.9 to 30.9) vs 30.2 (29.3 to 31.0) CAD-EOLD (staff), mean (95% CI): 30.9 (29.6 to 32.1) vs. 31.0 (29.8 to 32.2) SWC-EOLD (family), mean (95% CI): 32.3 (30.9 to 33.7) vs. 32.3 (31.0 to 33.6)
Verreault, Canada, 201829	Residents with terminal dementia (staged as FAST 7e, f) excluded the residents without any family or family did not complete the questionnaires. PC n=70; Usual care n=54	Four long-term care facilities in the 2 cities in Canada.	(1) Training of physicians and all nursing staff; (2) clinical monitoring of pain; (3) regular mouth care routine; (4) early and systematic communication with families about EOL care issues with the provision of an information booklet; (5) use of a nurse facilitator to implement and monitor the intervention.	Primary care team; ≤ 1 year	Quasi-experimental study; questionnaires mailed to family members after the death of their relative.	Age, mean±SD: 88.9±9.4 vs. 88.1±6.2; Female, %: 76.3 vs. 69.8	CAD-EOLD (family) in the last 48 h of life: 35.8±5.3 vs. 33.1±6.6, P=0.03 SM-EOLD (family): 34.7±11.0 vs. 29.8±10.2, P=0.03

^*Information on the relevant data source and sample included in the review.

ACP indicates advanced care planning; CAD-EOLD, Comfort Assessment in Dying with Dementia; ED, emergency department; EOL, end-of-life; FAST, Functional Assessment Staging Tool; ICD 10, International Classification of Diseases 10th revision; ICD 9, International Classification of Diseases 9th revision; ICU, intensive care unit; IQR, interquartile range; ns, not significant; OR, odds ratio; PC, palliative care; RCT: randomized controlled trial; RR, risk ratio; SM-EOLD, Symptom Management at the End-of-Life; SWC-EOLD, Satisfaction With Care at the End-of-Life in Dementia.

Risk of Bias

Ten of the 19 studies were judged to be of strong quality.^{14,26,27,29,30,34–37,39} The critical appraisal scores ranged from 61.5% to 100.0% (Supplement 2, Supplemental Digital Content 1, http://links.lww.com/MLR/D47).

Effect on Comfort Assessment in Dying With Dementia

Of the 19 studies, 6 that recruited a total of 974 participants were used to produce a meta-analytic model for CAD-EOLD.^23–28 The findings indicated that the impact of nonspecialty PC on patient comfort remains uncertain (SMD 1.46, 95% CI −0.05, 2.97; I ²=97%; P=0.06). This uncertainty encompasses a range of interventions aimed at improving the physical environment, managing comfort, engaging the senses, and training health care professionals in areas such as pain management, mouth care, and communication, specifically for older adults in the FAST stage ≥6 in nursing homes or long-term care facilities.^24,25,28 After excluding the study by Verreault²⁸ due to the clinical heterogeneity—since it only involved people with terminal dementia classified as FAST stages 7e and 7f—nonspecialty PC was found to be associated with higher CAD-EOLD scores (SMD 2.12, 95% CI 1.24, 3.00; I ²=60%; P<0.001). In addition, providing training and care guide aimed at improving the quality of dying and EOL care for health care professionals in acute geriatric hospital wards led to significant improvements in the comfort of dying patients, as assessed by nurses, though families did not report similar improvements.³² Specifically, there were significant reductions in discomfort, pain, restlessness, shortness of breath, choking, difficulty swallowing, and fear in the last 48 hours of life. In contrast, feelings of serenity, peace, and calm increased during this time. However, instances of gurgling, anxiety, crying, or moaning did not improve.²⁴ On the other hand, specialty PC had little to no effect on comfort during EOL care in both nursing homes and hospital settings (P=0.62)^31,34,35 (Fig. 2).

FIGURE 2.

Forest plot of effect: (A) comfort assessment in dying with dementia, (B) symptom management at the end-of-life, (C) satisfaction with care at the end-of-life in dementia.

Effect on Symptom Management at the End-of-Life

Two studies, involving 310 participants, investigated the effects of nonspecialty PC on SM-EOLD scale and found the results to be uncertain (P=0.59).^24,28 One study²⁸ reported that training care teams in long-term care facilities could significantly improve their scores assessed by family members on SM-EOLD (SMD 0.46, 95% CI 0.10, 0.82). However, another cluster-randomized controlled trial noted no significant differences in SM-EOLD as assessed by nurses (baseline-adjusted mean difference −0.41; 95% CI −1.86, 1.05; P=0.58) or by family members (baseline-adjusted mean difference −0.59; 95% CI −3.75, 2.57; P=0.71).²⁴

Two studies assessed the impact of specialty PC and found that scores on the SM-EOLD showed little or no difference in both nursing home and medical center settings (SMD −0.28, 95% CI −0.60, 0.04; I ²=0%).^23,29 There is not enough evidence to support the claim that PC nurse practitioners, who collaborate closely with geriatricians to care for older adults with ADRD at a Global Deterioration Scale score of 7,²⁹ can improve symptom control. Similarly, there is no evidence that a PC planning coordinator initiating a facilitated case conference in nursing homes for individuals who have reached stage of ≥6a on the FAST can enhance symptom management.²³

Effect on Satisfaction With Care at the End-of-Life in Dementia

Three studies examined the effects of nonspecialty PC on the increased level of SWC-EOLD but observed no differences between those who received nonspecialty PC and those who did not (SMD 0.02, 95% CI −0.38, 0.42; I ² =43%).^22,24,25 Similarly, findings from 3 studies^23,27,29 suggested that specialty PC had no impact on satisfaction with care for older adults living with ADRD (SMD 0.04, 95% CI −0.12, 0.19; I ²=0%).

Effect on Hospital Admission

Two studies examined the relationship between nonspecialty PC and hospital admissions at the end-of-life.^30,34 The first study found that completing the Physician Orders for Life-Sustaining Treatment (POLST) as part of a comprehensive dementia disease management program increased the likelihood of hospital admissions in the last 6 months of life by 66% (95% CI 1.05, 2.64, P=0.03).³⁴ In contrast, older adults not enrolled in hospice and under the care of physicians who had received geriatric training were significantly linked to a lower proportion of hospital admissions in the last month of life (P<0.001).³⁰ Another study included services delivered by both specialty and nonspecialty PC, making it difficult to determine the effect of each individually. In this study, PC was also found to increase rates of admissions to hospital in the last 6 months of life, with an adjusted rate ratio of 1.33 (95% CI 1.27, 1.39).¹⁴

Six studies involving 9654 participants indicated that specialty PC had no significant effect on hospital admissions during EOL care (OR 0.41, 95% CI 0.16, 1.02; I ²=96%).^{23,29,31,33,35,36} After excluding clinical heterogeneity³³ from the sensitivity analysis because all patients in this study had advanced cancer and concomitant dementia, the results showed that there was a significant decrease in hospital admissions, but the heterogeneity remained (OR 0.28, 95% CI 0.15, 0.53; I ²=78%) (Fig. 3). In studies where specialty PC was primarily through home visits, there was a clear reduction in hospital admissions.^31,36 However, when the intervention took place mainly in nursing homes, involved in-hospital consultations, or occurred across various care settings without specifying the services provided, the impact on hospital admissions remains uncertain.^23,29,32,35

FIGURE 3.

Forest plot of effect: (A) hospital admission, (B) ED visit, (C) ICU use, (D) in-hospital death.

Effect on Emergency Department Visit

In the last 6 months of life, the impact of nonspecialty PC that includes a completed POLST on the risk of ED visits remains uncertain (OR 1.34, 95% CI 0.81, 2.23).³⁴ However, in the final month of life, patients cared for by physicians with geriatric training had a significantly lower rate ED visits (P<0.001).³⁰

In a meta-analysis of 5 studies involving a total of 9286 participants found that specialty PC was linked to a lower likelihood of one or more ED visits during the last month of life (OR 0.53, 95% CI 0.28, 1.00; I ²=86%).^{23,29,33,35,36} When excluding clinical heterogeneity³³ in a sensitivity analysis—since all patients in that study had advanced cancer and concurrent dementia—the association appeared even stronger (OR 0.37, 95% CI 0.32, 0.43; I ²=0%). Of the 5 studies, 2 focused on residents in nursing homes and home-dwelling older adults. These studies found that home visits by a specialty PC team were associated with a reduced risk of ED visits (OR 0.36, 95% CI 0.31, 0.42; I ²=0%).^35,36 In contrast, hospital-based specialty PC initiatives, such as including POLST and referrals to postdischarge services,²⁶ as well as PC consultations and postdischarge follow-ups,²⁹ did not demonstrate a significant reduction in ED visits within 60 days after discharge²⁶ or during EOL care.²⁹ Similarly, interventions across various care settings, which did not specify the services provided, were found to have no impact or even an increased rate of ED visits.^14,32

Effect on Intensive Care Unit Admission

A study demonstrated that comprehensive dementia care does not significantly affect the likelihood of ICU admissions during the last 6 months of life and the duration of ICU stays.³⁴ However, patients cared for by physicians who had received geriatric training were significantly associated with a lower rate of ICU stays in the last month of life (P<0.001).³⁰

On the basis of 2 studies, older adults who received home- or clinic-based specialty PC experienced an 87% lower risk of ICU admission in the last month of life (OR 0.13, 95% CI 0.08, 0.21; I ²=15%).^31,36 In contrast, interventions implemented across various care settings, which did not specify the services provided, showed an uncertain impact on ED visits.^14,32

Effect on In-Hospital Death

Three studies found that nonspecialty PC significantly reduced the likelihood of in-hospital death (OR 0.55, 95% CI 0.32, 0.95; I ²=43%).^34,38,39 These studies focused on interventions designed to enhance ACP^34,38 and EOL care skills, as well as providing support for generalists and home-based primary care teams.^38,39

In contrast, 3 studies on specialty PC services, all of which provided specialized home-based PC support, demonstrated an 88% reduction in the odds of in-hospital death compared with usual care (OR 0.12, 95% CI 0.07, 0.20; I ²=60%).^31,33,36 However, studies that offered specialty PC only through inpatient consultations found that patients receiving this type of care were 3.8 times more likely to die in a medical setting than those who did not receive specialty PC (OR 3.89, 95% CI 1.52, 9.91).³⁸ Earp et al³² found that specialty PC was effective in reducing the risk of in-hospital death only if the intervention began at least 90 days before death (P<0.001). Quinn et al¹⁴ did not conduct subgroup analyses of older adults who received PC from a nonspecialty or a specialty PC team. Nevertheless, their study found that PC provided in hospitals, outpatient clinics, and homes—including home care services—was associated with decreased odds of dying at home or in a nursing home rather than in a hospital (adjusted OR 0.68, 95% CI 0.64, 0.73), particularly for those residing in nursing homes.¹⁴

DISCUSSION

This systematic review and meta-analysis found that PC services had little measurable effect on comfort, symptom management, and satisfaction with care regardless of whether provided by specialty or nonspecialty PC providers. In contrast, there is evidence that both specialty and nonspecialty PC are associated with reductions in ICU admissions and in-hospital deaths. In addition, specialty PC was associated with a lower risk of ED visits. Most of the nonspecialty PC examined in this study was provided by teams with geriatrics training, which appeared to have a stronger impact on reducing hospital admissions and ED visits than completing the POLST.

Various factors can influence the effectiveness of interventions on the same outcome measures. First, among the 19 studies included in the analysis, only 6 focused specifically on individuals with moderate to very severe ADRD, specifically those with a Global Deterioration Scale score equal to or greater than 5 or a FAST Scale staged as 6a or above.^{23,25,26,28,29,35} Uncertainty about a person’s prognosis will affect the consideration of the benefits of introducing specialty PC, especially at an earlier stage in the disease trajectory. This uncertainty also poses challenges for conducting research with individuals in the late stages of ADRD.¹⁰ The timing of referrals or to initiate specialty or nonspecialty PC is crucial for ensuring patient satisfaction with EOL care and addressing unmet needs.^40,41 Furthermore, none of the studies included in this review were conducted in Asian countries, highlighting a gap in research. The overall low priority given to PC for older adults with ADRD in many countries complicates the measurement and comparison of EOL care outcomes.¹⁰ In addition, the effects of comfort and satisfaction with care were evaluated only by family members or nurses in this review. Future research should explore the effectiveness, delivery, and acceptability of different PC models and consider the perspectives of persons with ADRD when evaluating care outcomes.

Care teams that focus on the specific needs of older adults can deliver the principles of PC assessment and management, but geriatric care emphasizes the preservation and restoration of function and autonomy, and providing holistic care in the social and physical context, ultimately preserving patient comfort and dignity.^34,42,43 However, similar to a recent study,²¹ the perception and satisfaction of care among families were not significantly improved in the 2 PC models for older adults dying with ADRD assessed in this review. These individuals often require a range of expertise, including acute, rehabilitation, mental health, and PC, which needs to adapt to their changing circumstances.⁴³ Health care professionals outside of geriatric units may lack familiarity with the specific needs of dementia patients, particularly in areas such as pain assessment, communication, and nursing care. This gap in knowledge can result in lower satisfaction with care and inadequate symptom management. In addition, those with ADRD often experience behavioral and psychological symptoms, for which there is no cure or FDA-approved treatment. Despite the availability of specialty PC, lack of adequate knowledge and availability of certain medications in certain settings were considered as main obstacles in providing palliative sedative in many countries, causing hesitation among physicians in using it.⁴⁴ Previous reviews have highlighted the need to improve the understanding of health care professionals’ and families’ experiences of EOL care to ensure adequate PC intervention for symptom control, especially when there is failure to conceptualize dementia as a terminal condition, even in advanced stages.^45,46 The interventions discussed in this review primarily focused on professional training, consultations, or case conferencing, benefiting persons with ADRD more than family caregivers. Future research should prioritize identifying services that best satisfy the needs of family caregivers and include the perspectives of persons with ADRD and their families.

Despite the comprehensive nature of the program, with almost all participants or their proxies having ACP discussions and referral to community organizations for dementia-related services in this review, only specialty PC reduced the risk of ED visits. In contrast, nonspecialty PC did not have this effect. The uncertainty of the prognosis of advanced dementia may contribute to health care professionals struggling to timely withhold or withdraw life-sustaining treatments even though individuals have early ACP discussions. Goals of care and treatment may be more discussed in specialty PC; therefore, family caregivers are more prepared for EOL care. Future exploration of reasons for burdensome transitions among older adults receiving nonspecialty PC may help improve services.

One major strength of this work is our focus on studies published between 2013 and 2024, which enhances the consistency of progress in PC and public health policies. PC outcomes arise from complex interactions, and these developments have not only transformed how specialty and nonspecialty PC are delivered but have also fostered a more global consensus on the definitions of quality PC for older adults with ADRD. In addition, none of the studies included in our collection took place during the COVID-19 pandemic, a period during which strained medical resources and limited access to care could have impacted EOL care outcomes and the preferred place of death for patients with dementia.⁴⁷ Seven out of the 19 included study samples comprised nursing home or long-term care facility residents.^{22,23,25,27,28,35,38} This can strengthen the generalizability of the results to real-world care settings. The limitations of this study are related to the relatively small number of studies that met the inclusion criteria for nonspecialty PC.

CONCLUSION AND IMPLICATIONS

Both specialty and nonspecialty PC showed no difference in comfort, symptom management, or satisfaction with care. However, both approaches significantly reduced the likelihood of ICU admissions and in-hospital deaths. Specialty PC also decreased hospital admissions and ED visits, although these findings were associated with low certainty. Nonspecialty care teams that receive geriatrics training and collaborate with specialty PC services can improve comfort and alleviate the disease burden for older adults with ADRD. Currently, there is insufficient evidence to confirm that PC improves EOL symptom management and overall satisfaction with care. Therefore, future research should prioritize achieving consensus among persons with ADRD, family members, and professionals on the wishes for EOL care. More high-quality studies evaluating different models of PC in various settings for older adults with ADRD are essential to enhance their quality of dying.

Supplementary Material

mlr-63-851-s001.docx ^{(57.4KB, docx)}