Palliative care for persons with late-stage Alzheimer’s and related dementias and their caregivers: protocol for a randomized clinical trial

Background Limited access to specialized palliative care exposes persons with late-stage Alzheimer’s disease and related dementias (ADRD) to burdensome treatment and unnecessary hospitalization and their caregivers to avoidable strain and financial burden. Addressing this unmet need, the purpose of this study was to conduct a randomized clinical trial (RCT) of the ADRD-Palliative Care (ADRD-PC) program. Methods The study will use a multisite, RCT design and will be set in five geographically diverse US hospitals. Lead investigators and outcome assessors will be masked. The study will use 1:1 randomization of patient-caregiver dyads, and sites will enroll N = 424 dyads of hospitalized patients with late-stage ADRD with their family caregivers. Intervention dyads will receive the ADRD-PC program of (1) dementia-specific palliative care, (2) standardized caregiver education, and (3) transitional care. Control dyads will receive publicly available educational material on dementia caregiving. Outcomes will be measured at 30 days (interim) and 60 days post-discharge. The primary outcome will be 60-day hospital transfers, defined as visits to an emergency department or hospitalization ascertained from health record reviews and caregiver interviews (aim 1). Secondary patient-centered outcomes, ascertained from 30- and 60-day health record reviews and caregiver telephone interviews, will be symptom treatment, symptom control, use of community palliative care or hospice, and new nursing home transitions (aim 2). Secondary caregiver-centered outcomes will be communication about prognosis and goals of care, shared decision-making about hospitalization and other treatments, and caregiver distress (aim 3). Analyses will use intention-to-treat, and pre-specified exploratory analyses will examine the effects of sex as a biologic variable and the GDS stage. Discussion The study results will determine the efficacy of an intervention that addresses the extraordinary public health impact of late-stage ADRD and suffering due to symptom distress, burdensome treatments, and caregiver strain. While many caregivers prioritize comfort in late-stage ADRD, shared decision-making is rare. Hospitalization creates an opportunity for dementia-specific palliative care, and the study findings will inform care redesign to advance comprehensive dementia-specific palliative care plus transitional care. Trial registration ClinicalTrials.gov NCT04948866. Registered on July 2, 2021. Supplementary Information The online version contains supplementary material available at 10.1186/s13063-023-07614-4.


Background and rationale {6a}
Alzheimer's disease and related dementias (ADRD) affect 6.5 million Americans and their family caregivers [1]: three million live with late-stage ADRD [2,3].ADRD is the only major cause of death with no clinically relevant treatment to prevent, cure, or slow disease progression.ADRD prevalence is expected to double by 2030 [1,4].
Persons with late-stage ADRD suffer progressive dependency and distressing symptoms.Late-stage ADRD, defined as Global Deterioration Scale (GDS) 5-7, is characterized by moderately severe to very severe cognitive impairment and the need for daily caregiver support for activities of daily living (ADLs) [5].Though life expectancy may be months to years, distressing symptoms of ADRD and co-morbid conditions are prevalent [6][7][8].Neuropsychiatric symptoms-such as physical or verbal aggression, socially inappropriate behaviors, and sleep-wake cycle disruptions-occur in late-stage dementia and cause suffering for the person with ADRD and strain for caregivers [9,10].
Caregivers of persons with late-stage ADRD experience physical, emotional, and financial strain [11][12][13].Care in late-stage ADRD is demanding and requires substantial formal and informal resources.While most persons with ADRD at Global Deterioration Scale (GDS) stage 7 ADRD live in nursing homes, a majority of persons at GDS 5-6 live at home [3].Moreover, persons with latestage ADRD experience frequent transitions between home, long-term care facilities, and hospitals; thus, caregivers confront challenges coordinating care between settings and providers of care [14][15][16].
In prior research, caregivers of persons with late-stage ADRD prioritized the goal of comfort over life prolongation [24,25].For example, a majority of caregivers in the Goals of Care clinical trial prioritized the goal of comfort over function or life prolongation: 57% for GDS 5, 62% for GDS 6, and 84% GDS 7 [19].While ADRD caregivers face complex and morally distressing choices as surrogate decision-makers [14,26], many report poor quality communication and gaps in shared decision-making for late-stage ADRD [23,[27][28][29].Only 17% of nursing home residents have a living will, and only 38% of family decision-makers for persons with GDS 7 ADRD recall involvement in major medical decisions [25,30,31].
Acute illness is a sentinel event in ADRD, signaling an important opportunity to access palliative care teams, which are present in 72% of US hospitals [32][33][34].Comprehensive palliative care includes (1) counseling for prognostic awareness, (2) symptom management, (3) shared decision-making to align treatment with goals, and (4) enhanced support for emotional, spiritual, and practical needs [35].Interdisciplinary teams of physicians, advance practice providers, nurses, social workers, and chaplains provide palliative care.For patients with cancer and other serious illnesses and their caregivers, palliative care improves the quality of life, symptoms, and care [36,37]; it reduces treatment intensity without affecting survival [38][39][40][41], and unlike hospice, palliative care is appropriate even when life prolongation remains the goal.However, persons with ADRD and their caregivers have unique palliative care needs that do not match standard models of palliative care.

Objectives {7}
Our research objective is to conduct a multi-site efficacy randomized clinical trial (RCT) of the ADRD-PC program.We will enroll 424 dyads of hospitalized patients with late-stage ADRD (GDS 6-7 or GDS 5 with significant co-morbidity) with their family caregivers at 5 geographically diverse sites in the USA.Our primary hypothesis is that ADRD-PC will reduce hospital transfers (aim 1).The specific aims are as follows: Aim 1: To conduct a multi-site RCT of the ADRD-PC program (intervention arm) vs publicly available educational material for dementia caregivers (control arm) to compare 60-day hospital transfers (hospitalization and emergency room visits) for persons with late-stage ADRD (primary outcome).H1: 60-day hospital transfers will be lower in the intervention vs control arm.Aim 2: To compare patient-centered secondary outcomes between the intervention and control arms: (a) symptom treatment (Palliative Care Domain Index [23]), (b) symptom control (Symptom Management at the End of Life in Dementia Scale [50] and Neuropsychiatric Inventory Questionnaire [51]), (c) post-acute use of community palliative care or hospice, and (d) new nursing home transitions.Patients in the intervention vs control arms will have better symptom treatment (H2a) and symptom control (H2b), increased use of community palliative care or hospice (H2c), and fewer new nursing home transitions at 60 days post-discharge (H2d).Aim 3: To compare caregiver-centered secondary outcomes between the intervention and control arms: (a) communication about prognosis and goals of care, (b) shared decision-making, and (c) caregiver distress (Family Distress in Advanced Dementia Scale [52]).Caregivers in the intervention vs control arms will more often have documented communication about prognosis and goals of care (H3a) and report more shared decision-making about hospitalization and treatments (H3b) and less caregiver distress at 60 days post-discharge (H3c).

Trial design {8}
The design is an RCT conforming to SPIRIT and CON-SORT statements for trial methods and protocol [53,54].The approach uses NIA standards for a multi-site efficacy trial (stage II) of a protocolized behavioral intervention, while incorporating pragmatic features to support sustainability and future dissemination [55].

Study setting {9}
The study setting will be five US medical centers with interdisciplinary palliative care teams: University of North Carolina at Chapel Hill Medical Center (UNC), Massachusetts General Hospital/Harvard Medical School (MGH), Indiana University Health University Hospital (IU), UCHealth University of Colorado Hospital (UC), and Emory University Healthcare System (EU).Study sites are members of the Palliative Care Research Fig. 1 Alzheimer's disease and related dementias-palliative care (ADRD-PC) intervention model [46] Cooperative (PCRC) and utilize the Epic electronic health record system.Sites were recruited by a standardized PCRC process that seeks clinical trial sites with study-specific clinical practice resources matched to trial recruitment needs.

Site preparation
Before enrollment of study participants, the principal investigator (PI) and project manager will provide consultation for site-based teams to incorporate ADRD-PC tools in hospital electronic health records (EHR) systems.New tools will include the screening/case finding algorithm, the template for documenting clinical encounters, and the template for the Palliative Care Plan at discharge.The PI, project manager, and clinical research coordinator (CRC) will provide 4 h of research protocol training for site clinical research coordinators (site-CRCs) and site principal investigators (site-PIs) [56].The project manager and CRC will also provide 4 h of training for the site-CRCs.Training content will include implementation of EHR screening, procedures for recruitment and enrollment, interview and EHR review data collection, detection and adverse event data capture, and regulatory procedures [48].During this same time period, the PI will provide 4 h of training for the members of the interdisciplinary palliative care teams in each study hospital who will provide clinical consultation as part of the ADRD-PC program, focused on four domains: (1) prognostic awareness, (2) physical and neuropsychiatric symptoms, (3) shared decision-making for goals of care and key treatment choices in ADRD, and (4) transitional care to enhance post-acute support.

Eligibility criteria {10}
The study will enroll 424 dyads of hospitalized persons with late-stage ADRD and their family caregivers; dyads will participate for 60 days following the index hospitalization.The eligibility criteria for persons with ADRD are (1) aged 55 or older, (2) hospitalized, (3) have a physician-confirmed diagnosis of ADRD, and (4) staged GDS 6 or 7 or GDS 5 with additional co-morbidity defined by Charlson Comorbidity Index scored > 5.As in the pilot, the ADRD stage will be confirmed by the attending physician and caregiver.The eligibility criteria for caregivers are (1) the adult (aged 18 or older) legally authorized representative (LAR) for healthcare and has the capacity to serve in this role, (2) support the person with ADRD, and (3) can complete interviews in English.As in our prior ADRD research, other family caregivers may be present per primary caregiver request, but only the primary caregiver is a participant.Dyads will be excluded if (1) the LAR is not a family caregiver, (2) the patient currently receives palliative care or hospice, or (3) patient or caregiver would be unduly stressed.

Participant screening procedures
To recruit study participants, the site-CRCs will use rapid case-finding methods from the ADRD-PC pilot RCT, reviewing the lists generated by EHR screening algorithms to identify persons aged 55 and older with ADRD within 24 h of hospitalization [48].After a brief EHR review to confirm other eligibility criteria including evidence for ADRD stage and Charlson Comorbidity Index scoring, site-PIs will contact attending physicians and seek permission for the site-CRCs to approach caregivers of the person with ADRD.Attending physicians, who are in direct contact with the person with ADRD and their caregiver, confirm ADRD staging.

Participant recruitment and enrollment procedures
Family caregivers may be present in the hospital, barred from visitation due to COVID-19 restrictions and safety concerns, or may live at a great distance from the hospital; thus, recruitment and enrollment procedures will include both in-person and telephone/virtual communication methods.
Site-CRCs will introduce the study to eligible family caregivers using scripted telephone calls or in-person visits.When the site-CRC conducts recruitment and enrollment via telephone, he or she will ask if the individual is in a private location and feels comfortable talking about the study at that time.The site-CRC will schedule an alternative time if necessary.

Informed consent {26a}
The family caregiver will provide informed consent for themselves and for all persons with ADRD as they lack decisional capacity.If the family caregiver chooses to participate, the site-CRC will conduct the informed consent interview in person or over the phone.If the family caregiver agrees to participate, study informational materials and the informed consent form will be sent to them by mail or secure e-mail.Due to the low risk of study participation and the restrictions on immediate written consent described above, verbal consent will be accepted for participation in the enrollment caregiver interview and for randomization with initiation of the ADRD-PC or control conditions.

Additional consent provisions for collection and use of participant data and biological specimens {26b}
Not applicable as no additional consent provisions for ancillary studies are included.

Sequence generation {16a}
Randomization will occur following the enrollment interview, with the dyad as the unit of randomization and analysis.To ensure concealed allocation of randomization, the study statistician will create sitespecific, computer-generated random treatment assignments to assign dyads in a 1:1 ratio to intervention or control arms.

Concealment mechanism {16b}
The randomization will be generated by the study statistician within the REDCap software and will remain concealed until the evaluation of the study aims [57].

Implementation {16c}
After the allocation sequence is generated, each patient/caregiver dyad will be assigned a study ID number in each site.The site-CRC will reveal allocation to the site-PIs, who will communicate study arm assignment to the attending physician and to the palliative care team to initiate ADRD-PC consultation (intervention), or site-CRCs will deliver a copy of publicly available dementia caregiver educational material (control).

Intervention description {11a}
Patient-caregiver dyads that are randomized to the intervention will receive the ADRD-PC program that includes speciality palliative care consultation during hospitalization and telephone-based transitional care over 2 weeks after discharge.ADRD-PC dementia-specific palliative care, standardized caregiver education, and transitional care are described below.

Dementia-specific palliative care
Members of the site-based, interdisciplinary palliative care team, including a physician or advanced practice provider and at least one additional discipline (nurse, social worker, chaplain), will visit the dyad in the hospital until hospital discharge.Palliative care clinicians will address four domains: (1) prognosis-exploration and communication of stage, trajectory, and prognostic awareness; (2) symptoms-assessment and treatment for physical and neuropsychiatric symptoms; (3) shared decision-making-exploration of values and goals of care and discussion of treatment decisions; and (4) transitional care-assessment of needs, care planning, and recommendation for support services.Discussions may include exploration of the overall goals of care and decision-making about potentially burdensome treatments of resuscitation, ventilator use, feeding tubes, antibiotics, or future hospitalizations.When clinically appropriate, consultation will also include the completion of new advance directives, such as a Physician Orders for Life-Sustaining Treatment (POLST) [49], or comparable portable order set to document decisions for current care plans.Palliative care clinicians will document care in the EHR with the ADRD-PC Note Template.

Standardized caregiver education
Members of the site-based, interdisciplinary palliative care team will share and discuss the booklet, "Advanced Dementia: A Guide for Families, " which addresses common concerns and treatment decisions [45].In discussion with the dyad, clinicians will individualize the content of the booklet with counseling about topics including (1) dementia and its stages; (2) determining the primary goal of care; (3) approaches to decision-making, eating problems, decisions about hospitalization, and decisions for infections; (4) discussing how dementia affects the family; and (5) explaining differences between hospice and palliative care.

Transitional care
Members of the site-based, interdisciplinary palliative care team will also provide transitional care at three time points [58].Pre-discharge, a member of the palliative care team will create an individualized, templated palliative care plan and provide copies to the primary post-acute clinician and the family caregiver.This document will summarize recommendations in the four domains and provide contact information for the palliative care team for follow-up questions.Near discharge, the palliative care team will explore the adequacy of patient and caregiver support, such as community palliative care, hospice, and Alzheimer's disease caregiver support groups.Post-discharge, a designated palliative care team member (usually a social worker or case manager) will call the family caregiver within 72 h and again 2 weeks after discharge.The purpose of these calls will be to support the implementation of the palliative care plan and promote access to post-acute services.

Control condition
Patient-family caregiver dyads randomized to the control arm will receive educational materials from the Alzheimer's Association, specifically designed for late-stage ADRD caregivers, delivered by the site-CRCs [59].The patient will receive usual hospital and post-acute care.

Strategies to improve adherence to interventions {11c}
As described in Table 1, the NIH Behavior Change Consortium standards will be used to evaluate the degree to which clinicians use the ADRD-PC protocol as intended [60].
The PI will review monthly tracking reports of fidelity metrics and will use teleconferences to provide monthly coaching calls to site-CRCs, site-PIs, and clinicians.Coaching will be tailored and may include a review of study procedures, feedback on enrollment, retention and fidelity data, and strategies to overcome barriers.Sites failing to meet fidelity will be given site-specific feedback and re-training.

Assignment of interventions: blinding Who will be blinded {17a}
The lead investigator, a co-investigator (MT), a study consultant (SM), and outcome assessors will be masked.They will remain masked to the study arm assignments until analysis; however, site-PIs, including one study coinvestigator (CK), and the study statistician (FCL) will not be masked due to their roles in communicating study arm assignments and adverse event reporting.Site-CRCs will be masked to the study arm to collect outcome measures in family caregiver interviews.Since the EHR in study sites will have data regarding ADRD-PC palliative care encounters, site-CRCs will complete follow-up EHR reviews after 60-day follow-up interviews are completed and study participation ends for each dyad.

Procedure for unblinding if needed {17b}
If an unanticipated serious adverse event occurs, the IRB and the DSMB will consider unblinding prior to the final analyses if judged necessary to address participant safety.

Outcomes {12}
The description of study outcomes, measures and method of aggregation, data sources, and timing of data collection are reported in Table 2 and the narrative that follows.The primary outcome, 60-day hospital transfers, will be measured by the number of emergency room visits plus hospital admissions within 60 days after discharge from the index hospitalization.CRCs will collect this outcome data using 60-day EHR reviews and 30-and 60-day follow-up telephone interviews with family caregivers (Table 2).The 60-day hospital transfer outcome was selected owing to its importance to patients and caregivers and as a marker of healthcare cost [15,61].
Secondary patient outcomes will include (1) symptom treatment as measured by Palliative Care Domain Index items [23], which has 10 items scored present vs absent, ranging from 0 to 10 with higher scores indicating increased symptom treatment; (2) symptom control for physical symptoms as measured by the Symptom Management at the End of Life in Dementia (SM-EOLD) [50], scored from 0 to 45 with higher scores indicating better symptom control; (3) symptom control for neuropsychiatric symptoms as measured by the Neuropsychiatric Inventory Questionnaire (NPI-Q) [51], which has a range in two subscales of 0-36 and 0-60, with higher scores indicating worse symptom control; (4) access to hospice as measured by the percentage of people with ADRD who access hospice services; (5) access to community-based palliative care as measured by the percentage of people with ADRD who access community-based palliative care services; and (6) transition to nursing home level of care  Symptom management at the end of life in dementia-0-5 Likert-scaled measure of 9 symptoms during 30-day look-back.The range is 0-45 with higher scores indicating better symptom control [50].Method of aggregation-the mean score in intervention and control groups Neuropsychiatric Inventory Questionnaire-presence, frequency, and severity of 12 neuropsychiatric symptoms.The range is 0-36 and 0-60 with higher scores indicating worse symptom control [51] (2) documented discussion of goals of care as measured by the percentage of caregivers with documented discussion of goals of care during the index hospitalization; (3) shared decision-making regarding hospitalization as measured by the percentage of caregivers reporting shared decision-making about future hospitalization; (4) shared decision-making regarding burdensome treatment as measured by the percentage of caregivers reporting shared decision-making about resuscitation, ventilator use, tube feeding, and antibiotics for infection treatment; (5) caregiver distress as measured by the Family Distress in Advanced Dementia Scale [52], which ranges from 1 to 5 with higher scores indicating more distress; and (6) caregiver burden as measured by the Zarit Burden Scale short form [62], which ranges from 0 to 24 with higher scores indicating more burden.As described in Table 2, CRCs will collect secondary caregiver outcomes data in 60-day EHR reviews and/or follow-up telephone interviews with family caregivers 30 and 60 days after discharge from the index hospitalization.
In addition to these outcomes, data on covariates will be collected through EHR review and interviews with family caregivers.Patient demographics will include age, sex, race and ethnicity, and marital status.The preadmission residence will be measured as a private home, assisted living facility, nursing home, or others.Patient co-morbidity will be measured using the Charlson Comorbidity Index, a validated, widely used index with weighted scores for age and diagnoses [63].Functional status will be measured using the Bedford Alzheimer Nursing Severity (BANS) Scale for function in late-stage dementia [64].This scale ranges from 0 to 28, with higher scores indicating worse function (alpha = 0.80, Pearson's r 0.62-0.79)[64].ADRD stage will be measured using the GDS, and etiology will be defined by an attending physician during recruitment and enrollment [5].Advance directives will be measured as the presence or absence of a living will, Health Care Power of Attorney, portable DNR order, or POLST order set form [49]. Survival will be measured for each patient as days from study enrollment to death up to 60 days follow-up.Caregiver demographics will include health condition, age, sex, race and ethnicity, relationship to the patient, self-report of health status (excellent, very good, good, fair, poor), depression (PHQ-2), marital status, education level, current work situation, and social determinants of health [65].Family caregiver perception of prognosis will be measured using a single item asking what the caregiver expects will happen to the patient during the next 6 months, with response options of "get better, " "stay about the same, " "get worse, " or "likely to die."

Participant timeline {13}
Participant flow is shown in Table 3.

Sample size {14}
Sites will enroll N = 424 dyads of hospitalized patients with late-stage ADRD (GDS 6-7 or GDS 5 with significant co-morbidity) with their family caregivers.Statistical power is based on a 2-sided alpha of 0.05 significant tests and using standard deviation estimates from our pilot study (Table 4).Power calculations assume a "best guess" 15% dropout (death, withdrawal) rate based on prior ADRD research (n = 180 per group at 60-day follow-up) and a "worst case" 25% dropout rate (n = 159 per group).These rates consider 9% patient mortality and 92% caregiver retention rate in the pilot.In our pilot data, 0.52 is the lower bound of the 95% confidence interval for the incidence rate ratio (IRR) comparing intervention vs control for 60-day hospital transfers.This corresponds to an average absolute reduction of 0.26 transfers per 60-day period due to intervention, assuming the pilot study baseline control rate of 0.53.These power calculations consider reductions of 0.25 and 0.20 transfers, which are plausible values within the confidence bounds in the ADRD-PC pilot study [48].In Table 4, an "optimistic" reference power calculation is given assuming a Poisson model (i.e., variance equals the mean), and a second "realistic" calculation assumes overdispersion.In the latter case, the standard deviation of the number of transfers is assumed to be 30% greater than that in the Poisson model based on an estimated variance inflation factor in our pilot data of 1.69 (i.e.,√1.69 = 1.3).Allowing for overdispersion, we expect 81% power to detect a reduction in 60-day hospital transfers of 0.25, assuming a 15% dropout.For the number of palliative care domains for aim 2 (to cite one example of a secondary outcome), we expect more than 99% power to detect at least a 5-point improvement in the intervention group vs the control group at 60 days.

Plans for assessment and collection of outcomes {18a}
After informed consent is obtained, data collection will be identical for both arms.Data will be obtained from caregiver enrollment interviews and 30-and 60-day interviews post-discharge and from electronic health record (EHR) reviews encompassing the 60-day period post-hospitalization. Based on the pilot RCT, we estimate 10% of persons with late-stage ADRD will die during follow-up; in these cases, caregiver interview data collection will proceed using a Bereavement Interview adaptation.Outcome measures will focus on 60-day interviews; 30-day interim interviews are necessary to support retention, ensure data capture for persons with ADRD who die, and for valid recall of hospital transfers (primary outcome) and secondary outcomes.

Plans to promote participant retention and complete follow-up {18b}
Retention is supported by transitional care calls (on days 3 and 14 after discharge), appointments for follow-up

Data management {19}
Data collection will be the responsibility of the sitebased CRCs under the supervision of the site-based PI, with overall supervision from the PI and the project manager; the site-PIs will be responsible for ensuring the accuracy, completeness, legibility, and timeliness of the data reported.
All research data will be entered into REDCap, a 21 CFR Part 11-compliant data capture system provided by UNC [57].REDCap includes password protection and internal quality checks, such as automatic range checks, to identify data that appear inconsistent, incomplete, or inaccurate.Clinical data will be entered directly from the source documents [57].

Confidentiality {27}
Participant confidentiality and privacy will strictly be held in trust by the participating investigators, their staff, the safety and oversight monitor(s), and the sponsor(s) and funding agency.This confidentiality is extended to the data being collected as part of this study.No personally identifiable information from the study will be released to any unauthorized third party without prior written approval of the sponsor/funding agency.The study participant's contact information will be securely stored at each clinical site for internal use during the study.Data will be entered into a password-protected, secure database, and all paper documentation will be maintained in locked files.The Sheps Information Technology group, at UNC, enables standard operating procedures required to secure the network and databases, including operational and technical controls.All servers are located within a hardened data center.

Statistical methods for primary and secondary outcomes {20a}
Patient and caregiver dyads will be the primary unity of analysis.Descriptive analyses of variables will examine distributions, influential data points, and missing data.All a priori analyses, primary and subgroup, will use intention-to-treat analysis.The effectiveness of randomization will be evaluated by comparing intervention and control participants on baseline measures because variables that are not equally distributed between arms could potentially bias results.We will include these variables in each initial model and use a change-in-effect method for determining whether they are confounders.Given our randomized design, we anticipate little confounding.A priori exploratory analyses will examine the effects by sex, race/ethnicity, and GDS stage.

Aim 1 analysis
The control and intervention arms will be compared on hospital transfers (primary outcome) at 60 days postdischarge.This count outcome is defined as the total number of emergency room visits plus hospital admissions (including observation stays) per at-risk patientdays.Poisson regression with empirical "robust" standard errors allowing for overdispersion will be used to compare the transfer rates between groups.Length of followup will be used as the offset variable, which means that patients who have data only from the 30-day interview will be included.The primary analysis will be based on the model with the main effects for the treatment group, study site, GDS stage, and patient and caregiver sex as biological variables.Evaluation of the treatment effect will be based on a Wald test.The covariate-adjusted treatment effect will be quantified as an IRR with a 95% confidence interval.Considering that over 60% of patients in the pilot study did not have any hospital transfers, a sensitivity analysis will be conducted for the treatment effect on 60-day hospital transfers with a marginalized zero-inflated Poisson (MZIP) model [66], which may modestly increase power.Secondary Poisson and MZIP analyses will compare study arms for 30-day hospital transfer rates.
An exploratory analysis will be conducted to examine the interactions of the treatment group with study site, sex, race/ethnicity, and GDS Stage on 60-day hospital transfers [5].An interaction model will be used if an omnibus test for interactions is statistically significant at the 0.05 level.A second exploratory analysis will evaluate a dose-response effect of the intervention on the hospital transfer rate.A dose for patients receiving the intervention is defined as the number of key intervention components used in fidelity monitoring (range 0 to 4 with 0 for controls).The dose-effect is the incremental benefit of adding a single component; its rescaling gives the perprotocol effect of the full intervention.

Aim 2 analysis
As described below, the control and intervention arms will be compared on patient secondary outcomes at 60 days post-discharge on the following outcomes: • Symptom treatment: This ordinal outcome is the summation of presence (= 1) or absence (= 0) of 10 domains of palliative care [23].It will be treated as a continuous variable due to a high variation observed in the pilot data.Student's t-test will be used to compare the mean difference between the study arms unless violation of assumptions warrants the Mann-Whitney test.A multiple linear regression model will be used with main effects for treatment group, study site, GDS stage, and patient and caregiver sex as a biological variable.• Symptom control: The linear mixed models, with random intercepts for caregiver report of patients' symptom distress (SM-EOLD) [50] and caregiver report of patients' neuropsychiatric symptom distress (NPI-Q) [51], will be used with a time indicator (30 vs 60 days follow-up) with the above main effects.• Use of post-acute palliative care or hospice: This outcome will be defined as whether patients ever use hospice or outpatient palliative care during the time from discharge to 60 days follow-up.The proportions of those who do and do not use post-acute palliative care or hospice will be compared using Pearson's chisquare test.Logistic regression will be used for multivariable modeling with the same main effects as in aim 2 (symptom control, above).Final estimates will be reported as covariate-adjusted odds ratios.• Transition to nursing home care: This outcome will be defined by whether patients transition to nursing home care during the time from discharge to 60 days follow-up.The proportions of those who do and do not transition to nursing home care will be compared using Pearson's chi-square test.Logistic regression will be used for multivariable modeling with the same main effects as in aim 1. Final estimates will be reported as covariate-adjusted odds ratios.

Aim 3 analysis
The control and intervention arms will be compared to caregiver secondary outcomes at 60 days post-discharge.
• For the dichotomous outcome variables pertaining to communication about prognosis and goals of care and shared decision-making about hospitalization and shared decision-making about burdensome treatments, Pearson's chi-square test will be used to test for differences in proportions between the study arms.We will use logistic regression for the multivariable modeling.Treatment effect estimates will be reported as adjusted odds ratios.• To compare caregiver outcomes between intervention and control arms for caregiver distress (family distress in advanced dementia) [52], Student's t-test for the mean difference will be used since the outcome is continuous.Linear mixed models as described above will be used to test treatment effects for 30-day and 60-day outcomes jointly.

Interim analyses {21b}
No interim analyses are planned and this section is not applicable.

Methods for additional analyses (e.g., subgroup analyses) {20b}
An exploratory analysis will be conducted to examine the interactions of the treatment group with study site, sex, race/ethnicity, and GDS stage on 60-day hospital transfers and on secondary outcomes.

Methods in analysis to handle protocol non-adherence and any statistical methods to handle missing data {20c}
Missing data on independent variables are expected to be minimal.Simple mean and mode imputation for those variables will be included in regression models if less than 5% of patients have missing data; otherwise, conditional (i.e., regression) imputation will be used.Available case analysis will be conducted for 30-day outcomes.Missing 60-day dichotomous outcomes will be multiply-imputed if missingness exceeds 10%.

Composition of the coordinating center and trial steering committee {5d}
The research team will be led by the PI and managed by the project manager reporting to the PI, who will provide administrative leadership and study coordination.This multi-site RCT will utilize a single IRB (UNC) and a unified protocol for all study sites.The UNC Cecil G. Sheps Health Services Research Center will be the administrative home for the UNC research investigators and staff.The project manager and data manager (UNC research staff ) will facilitate data management, and data analysis will be led by the study biostatistician.

Composition of the data monitoring committee, its role, and reporting structure {21a}
The principal investigator (PI) will be responsible for ensuring the safety of participants daily.The Data and Safety Monitoring Board (DSMB), acting in an advisory capacity to the National Institute on Aging Director, will evaluate the progress of the study, review procedures for maintaining the confidentiality of data, the quality of data collection, management, and analyses.

Adverse event reporting and harms {22}
The occurrence of an adverse event (AE) or serious adverse event (SAE) may come to the attention of study personnel during study visits and interviews.Studytracked AEs will include major emotional distress for family caregivers and confidentiality risk events for the person with ADRD.Deaths, life-threatening health events, or acute illness episodes will be tracked for all participants; however, these are expected health outcomes for people with late-stage ADRD and will not be included in AE reporting.All AEs, not otherwise precluded per the protocol, will be captured on the appropriate case report form (CRF) and classified for severity and potential relationship to the study procedures.All harms and AEs occurring while on study will be documented appropriately regardless of relationship.Potential AEs and SAEs will be collected systematically for all study participants at each study visit, when research team members will inquire about the occurrence of AE/SAEs since the last visit.Potential AEs and SAEs will be collected non-systematically (i.e., spontaneous reporting from open-ended questions), when the occurrence of an adverse event (AE) or serious adverse event (SAE) may come to the attention of study personnel when a study participant presents for medical care or upon review by a study monitor.Any medical or psychiatric condition that is present at the time that the participant is enrolled will be considered as a baseline and not reported as an AE.Site-PIs will report any Unanticipated Problem Involving Risk to Subjects or Others (UPIRSO) or AEs to the PI and project manager in a timely manner.When a UPIRSO or AE is present, the PI and project manager will submit a report to the UNC IRB within 3 working days of receipt of this information.The reporting of any AE/SAE will be based on NIA and UNC IRB standards, on the severity of the AE (based on the Common Terminology Criteria for Adverse Events), its level of attribution to the intervention, and whether it is anticipated.All adverse events that are both serious (SAE) and unexpected would be reported to IRB, DSMB/SO, and NIA PO within 48 h of the study's knowledge of SAE.

Frequency and plans for auditing trial conduct {23}
Summary of SAEs will be reported to NIA, PO, and DSMB/PO quarterly, unless otherwise requested by the DSMB.Summary of all AEs regardless of classification will be presented for each DSMB meeting.

Plans for communicating important protocol amendments to relevant parties (e.g., trial participants, ethical committees) {25}
All study protocols and amendments, informed consent forms, and other study materials will undergo review by the Institutional Review Board at UNC prior to initiating research and will be subject to annual and other required reviews.Any protocol changes will be reported to site-PIs.Changes will also be reported to study clinicians when these changes affect their roles.

Dissemination plans {31a}
This study will be conducted in accordance with the National Institutes of Health (NIH) Public Access Policy.It requires scientists to submit final peer-reviewed journal manuscripts that arise from NIH funds to the digital archive PubMed Central upon acceptance for publication.Second, this study will comply with the NIH Data Sharing Policy and Policy on the Dissemination of NIH-Funded Clinical Trial Information and the Clinical Trials Registration and Results Information Submission rule.As such, this trial will be registered at ClinicalTrials.gov, and results information from this trial will be submitted to ClinicalTrials.gov.In addition, every attempt will be made to publish results in peer-reviewed journals.Data from this study may be requested from other researchers 2 years after the completion of the primary endpoint by contacting the Palliative Care Research Collaborative Group [67].Considerations for ensuring the confidentiality of these shared data are described above.

Biological specimens {33}
This section is not applicable.

Discussion
The ADRD-PC study will be the first test of comprehensive dementia-specific palliative and transitional care.This study addresses the significant public health impact of late-stage ADRD and suffering due to symptom distress, burdensome treatments, and caregiver strain.While many caregivers prioritize comfort in late-stage ADRD, shared decision-making and access to specialty palliative care are rare.The ADRD-PC program leverages hospital palliative care teams to deliver comprehensive, dementia-specific palliative care plus transitional care.By including persons with GDS 5-7, ADRD-PC attends to the suffering of persons with ADRD and their caregivers along the illness trajectory rather than only at end-of-life.
The study protocol is designed with consideration for potential problems and challenges.Palliative care cannot be ethically withheld from patients in the control group, and investigators anticipate the potential for contamination.Routine palliative care consultation will be permitted for controls, but contamination risk is decreased since in current care this is very uncommon.Even when control patients are seen by palliative care specialists, these clinicians will lack ADRD-PC dementia-specific training and tools.Any palliative care consultation for control dyads will be tracked.Furthermore, as indicated in the ADRD-PC pilot, most attending physicians determined the need for palliative care consultation in the first 1-2 days of admission.Thus, these patients are more likely to be excluded than to create bias to the null.Second, while not found in the single-site pilot, barriers to physician referral may exist at new sites.Site-PIs will function as opinion leaders and (supported by the PI) will provide in-person education to overcome any resistance.Third, sites may have difficulty implementing the EHR screening algorithm.The study team will work with them on alternative screening and case-finding strategies compatible with local practice norms.Fourth, sites may find that lengths of stay are too short for some persons with ADRD; if this occurs, we will consider alternative strategies to deliver some elements of the ADRD intervention via virtual visits.
For persons with late-stage ADRD and their caregivers, there is an urgent need to improve outcomes of burdensome hospital transfers, symptom distress, shared decision-making, and caregiver distress.This study is responsive to the National Alzheimer's Project Act, NIA priorities for geriatric palliative care, and quality care for ADRD.If efficacious, ADRD-PC has the potential for broad dissemination, as most US hospitals provide specialty palliative care.

Trial status
The study protocol (NCT03810534), version number 1, was registered on July 2, 2021.Key trial implementation milestones were achieved.First, in all sites, ADRD-PC tools were added to electronic health record systems (EPIC) between July 23 and November 15, 2021.Second, in all sites, ADRD-PC intervention training was completed between June 14 and July 9, 2021.Third, in four study sites, dyad enrollment was started between July 27 and November 16, 2021.
Over the period from October to December 2021, the study team determined that COVID-related surges in hospital census were causing delays in participant enrollment.It was recognized that the original plan to achieve the sample within the study period was not feasible.Thus, in consultation with the NIA and the study DSMB, the study team developed a plan to add a fifth hospital to the multi-site RCT.The ADRD-PC protocol was revised, and the revised protocol (version 2.0) was approved by the UNC IRB on April 13, 2023, and updated on Clini-calTrials.gov(April 13, 2023).Subsequently, the ADRD-PC protocol was implemented in the fifth study site, Emory University (EU) Hospital in Atlanta, Georgia.In the EU Hospital, EPIC modifications and staff training were completed between March 7 and May 31, 2023, and enrollment of study participants was started on July 24, 2023.With the implementation of the ADRD-PC protocol in five hospitals, all participant enrollment is expected to be complete in May 2024.
Finally, the ADRD-PC protocol was updated to incorporate a secondary study of a Spanish version of the ADRD-PC intervention.The Spanish language ADRD-PC protocol will permit enrollment of up to 50 additional dyads, generating data to be used in separate analysis to describe the feasibility and acceptability of the Spanish version of ADRD-PC.The revised protocol was approved by the UNC IRB on April 13, 2023, and updated on Clini-calTrials.gov(April 13, 2023).
• fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year

•
At BMC, research is always in progress.

Learn more biomedcentral.com/submissions
Ready to submit your research Ready to submit your research ?Choose BMC and benefit from: ? Choose BMC and benefit from:

Table 1
Strategies to improve adherence to interventions [60] SOP Standard operating procedures, Site-PI Site principal investigator, PC Palliative care, ADRD-PC Alzheimer's Disease and Related Dementias-Palliative Care Intervention, CRC Clinical research coordinator

Table 2
Primary and secondary outcomes

Table 2 ,
. Method of aggregation-the mean scores in intervention and control groups CRCs will collect secondary patient outcomes data in 60-day EHR reviews and/or follow-up telephone interviews with family caregivers 30 and 60 days after discharge from the index hospitalization.Secondary caregiver outcomes will include (1) documented discussion of dementia prognosis as measured by the percentage of caregivers with documented discussion of dementia prognosis during the index hospitalization;

Table 3
Schedule of enrollment, interventions, and assessments ADRD-PC Alzheimer's Disease and Related Dementias-Palliative Care, CRC Clinical research coordinator, EHR Electronic health record

Table 4
Power for comparing intervention and control arms based on the study design SD Standard deviation, PC Palliative care, SM-EOLD Symptom management at the end of life in dementia