The World Health Organization has declared dementia a public health crisis and issued a call for global nations to make dementia an international public health priority [1]. Although dementia is a serious illness that progresses over many years, little is known about the everyday suffering endured over time by patients in the early stages of the disease, as most research focuses on patients who are already institutionalized and at the end of life. However, patients early in the dementia trajectory are beset with distressing symptoms. An estimated 85% [2] of persons with dementia live in the community (eg, private residences or non–nursing home residential care facilities) with unmet needs [3]. Proactive, patient-centered primary palliative care will likely benefit early-stage patients living in the community. Implementing early palliative care (EPC) will facilitate ongoing assessment and management of the distressing symptoms experienced by persons with dementia from very early in the illness trajectory, result in identification and management of primary palliative care needs, and facilitate timely advance care planning [4].

The aim of this study is to test the impact of a virtual, nurse-led early primary palliative care intervention (ELICIT) on older adults living in the community who are chronically ill and have a diagnosis of cognitive impairment or who are at risk of it. We hypothesize that, compared to the usual care (UC) arm, more participants in the intervention arm will (1) express supportive care needs to the blinded evaluators, (2) complete conversations on goals of care and advance care planning documentation, and (3) have lower end-of-life resource use (hospital admissions and days and emergency visits) and higher levels of goal-concordant care.

Participants randomized to the UC arm received care from their primary care provider (PCP) and their care team. In this case, UC refers to the standard or routine outpatient care that participants received from their primary care team. This included any appointments, consultations, and procedures that were standard for outpatient primary care.

Participants in the intervention arm received UC + a virtual, nurse-led ELICIT.

All study nurses were trained using a training program provided through a learning management system followed by immersive simulation training with standardized patients. The feedback of the standardized patients and the preceptor was used to refine their skills. Ongoing training for skill building and education was provided through nurse peer support, weekly meetings with the study physician, and the weekly trial team meetings.

The nurse-led intervention was designed to focus on the four key components of primary palliative care: (1) assessing and palliating distressing physical symptoms; (2) assessing and supporting psychological, social, cultural, and spiritual aspects of care; (3) establishing goals of care, identifying a surrogate decision-maker, and completing advance care planning documentation; and (4) providing care coordination support.

It is important to note that the nurses delivering the intervention were registered nurses, and the care they provided was within their scope of practice. The nurses assessed the participants using several tools (Multimedia Appendix 1 [6-16]).

The intervention comprised 12 virtual encounters (1 per month) with the participants over a 1-year period, including an initial encounter (session 1) followed by 11 monthly encounters (sessions 2-12).

In session 1, a complete assessment of the participants’ early primary palliative care needs using a set of measures (Table 1) and an assessment of participants’ preferences for care were completed. During this session, the nurses introduced advance care planning and the concept of goals of care. The nurses worked with the participants (and their caregivers if present) to create a list of all the issues they needed help with.

After session 1, the nurse presented the participant information and preferences to the study palliative care physician and worked to craft a personalized care plan based on the participants’ needs and expressed wishes.

In sessions 2 to 11, the nurse reviewed the participants’ situation over the previous month and identified any concerns. They completed the Distress Thermometer and pain assessment, used these to explore any underlying issues to be palliated, and discussed those with the study palliative care physician. The study nurses worked with the PCP of each participant to implement the early primary palliative care plan. For example, if the participant reported pain and needed any medications, the study nurse would first consult with the study palliative care physician and then work with the participant to craft a personalized care plan. If the participant and their care partner wanted to contact their PCP, they could do so themselves. If the participant or caregiver wanted the study nurse to contact their PCP on their behalf, the nurse did so. All medical therapies could be prescribed by the individual participants’ PCPs. The nurse continued the discussion about goals of care and advance care planning and helped complete the documentation. If the participant was interested, the nurse discussed the Physician Orders for Life-Sustaining Treatment (POLST) form choices with the participant and helped them identify their preferences. The participant was encouraged to discuss their goals of care with their PCP and finalize their choices. Once the participant and their PCP signed the POLST form, it was uploaded into the electronic medical record, and the original was sent to the participant to post on their refrigerator.

At the end of each encounter, the nurse worked with the participants to identify any issues that needed attention and make plans for follow-up in the next few days. The nurse also scheduled the date for the following month’s session.

On the last follow-up visit (session 12), the nurse reviewed the intervention encounters over the previous year and summarized the issues identified (if any). They provided the study team’s contact information to the participants in the event that they needed further assistance and thanked them for taking part in the study.

Potential eligible participants were identified by working with the neurology team. The study coordinators contacted the participants via phone to schedule a screening call to determine eligibility. If eligible, and if the participants and caregivers consented to take part in the study, we scheduled an appointment with the IT expert to help the participants access the video portal through their computer or smartphone. The blinded study evaluators consented all participants and completed all evaluation measures virtually. A total of 200 enrollees who met the inclusion criteria were randomly assigned to the intervention (UC + virtual, nurse-led EPC) or to UC. Consenting caregivers were also allowed to take part as observers along with the participants.

Participants were eligible to take part if they were older adults with chronic illnesses living in the community and they had undergone a full evaluation by a multidisciplinary neurology team of experts. The Stanford Alzheimer’s Disease Research Center team assessed the participants’ cognitive status and determined their Clinical Dementia Rating (CDR) scores through consensus. Participants with a composite CDR score of 0 (no symptoms), 0.5 (very mild), 1.0 (mild), or 2.0 (moderate) were eligible to take part in the study.

Participants were excluded from the study if they had severe dementia (CDR=3.0) or if they lived in an institutional setting.

After obtaining informed consent, participants were randomized to either the intervention (UC + virtual, nurse-led EPC) or to UC. The study nurses worked with participants randomized to the EPC intervention or their proxies to schedule session 1 of the intervention as a secure virtual video encounter. Sessions 2 to 12 were scheduled monthly video or phone encounters based on participant preference. To the extent possible, the same nurse was assigned to each participant across all sessions to ensure continuity. Evaluators, blinded to the treatment condition, collected the outcome measures from participants in both arms at baseline (at the time of study enrollment) and months 4 and 12. Long-term observational follow-up is currently ongoing for all participants through the electronic health records to track their health care resource use and determine whether the participants received preference-sensitive care at the end of life.

The first primary outcome measure is the number of participants who report supportive care needs in both arms to the blinded evaluators (time points: day 0 and months 4 and 12).

The second outcome measure is completion of conversations on goals of care and documentation of advance directives and the POLST [8] in the electronic health records (time points: daily during the intervention period).

The third outcome measure is end-of-life resource use (hospital admissions and days and emergency visits). We hypothesize that this will be lower for the participants in the intervention group and that, compared to those in the control group, the participants in the intervention group are more likely to receive goal-concordant care at the end of life. In both arms, blinded evaluators will track the participants’ end-of-life outcomes by monitoring their electronic health records and through interviews with their caregivers after the death of the participants (time points: at 1 year and then annually).

To compare the EPC intervention and UC groups, for ordinal outcome measures, we will use a hierarchical linear model that assumes that the outcome O for each participant p at a time t is a linear function with an intercept i and slope s: O(p,t) = i(p) + s(p)ln(t+1) + e(t,p), where t is the time of measurement (measured from baseline) and e is an error term. The intercept for each participant (i[p]) is the true baseline response, with randomization having the same distribution in both groups. The slopes of the groups reflect the impact of the intervention or of UC over the course of treatment. Both intercepts and slopes are assumed to have a bivariate distribution in the population of participants, and the error terms within participants are assumed to be intercorrelated according to an autoregressive covariance function (outcomes closer together are assumed to be more highly correlated than those more distal to each other). In addition, ln(t+1) is used rather than t because outcomes in randomized clinical trials are more likely to have a “fishhook” relationship to time t, with more rapid changes early on than later.

For the power calculation, the null hypothesis to be tested is that the slopes in the EPC group have the same distribution as those in the UC group (ie, that there is no effect of EPC relative to UC). The significance level is 5%, with no adjustment for multiple testing because the outcomes chosen tend to be independent of each other. For 80% power to detect a moderate effect size with a 2-tailed 5% test, the sample size per group is approximately 63. However, we recruited a total of 100 participants per arm (200 in total) as we anticipated that the number of participants with a CDR score of 0 and at risk of cognitive impairment and those with a CDR of 0.5, 1, or 2 would be similar, allowing for within-group analysis.

There is a growing shortage of palliative care workers. There is also an escalating need for early implementation of primary palliative care for older adults in the community with dementia and other serious illnesses. The registered nurse workforce can be trained using the ELICIT intervention model to provide early primary palliative care within the scope of nursing.

Our trial is notable because most existing research focuses on the palliative needs of patients who have severe dementia or are in an institutional setting [17-21]. To our knowledge, this is the first nurse-led, virtual intervention that provides early primary palliative care for older adults with cognitive impairment living in the community. Registered nurses provide the bulk of care for patients enrolled in hospice. However, currently, registered nurses do not provide early primary palliative care to patients. While a registered nurse cannot function as an independent provider, trained nurses are well able to provide early primary palliative care in collaboration with a patient’s PCP. Once we identify the scope of practice for registered nurses in providing early primary palliative care, nurses can be trained using our intervention. Training the nurse workforce to provide early primary palliative care will greatly expand access to these services and enable patients to receive primary palliative care early in their illness trajectory.

Older adults are often excluded [22] from randomized clinical trials. This randomized controlled trial has recruited 200 community-dwelling older adults with chronic illnesses and varying levels of cognitive impairment, and currently, we are following the participants on an ongoing basis.

There is limited research and understanding regarding the primary palliative care needs of patients in the early stages of cognitive impairment, making it difficult to provide timely and effective care interventions. Experts have noted that primary palliative care is a fundamental aspect of high-quality care for patients with dementia [23]. While primary palliative care is appropriate for persons living with dementia at any stage, challenges exist in identifying and assessing their unique physical, psychological, social, and spiritual needs, particularly due to the delayed diagnoses, uncertain prognosis, and difficulties in communication and decision-making. In this trial, we tracked the supportive care needs expressed by all the participants to the blinded evaluators. On completion of this study, we will have a better understanding of the primary palliative care needs of patients with mild and moderate cognitive impairment who are living in the community.

Older adults with moderate and severe dementia often lose their ability to participate in discussions on goals of care and shared decision-making. Eliciting the patients’ values and goals of care and obtaining anticipatory guidance for their future care is an integral part of high-quality care [24]. Studies have shown that patients tend to choose less aggressive interventions when they have an opportunity to participate in shared decision-making and document their preferences [25-27]. It is also known that patient-designated and next-of-kin surrogates incorrectly predict patients’ end-of-life treatment preferences [28]. There is a window of opportunity early in the trajectory of cognitive decline when the patient retains mental clarity and decisional capacity. Discussions on goals of care should be conducted during this window while patients are still able to voice their values and preferences for care [29]. While the patients’ goals and preferences may change over time, documenting goals of care and completing advance care planning will serve to establish a baseline of their preferences while they still have the ability to express them.

This study had the following limitations. First, the COVID-19 pandemic started at the exact time when we enrolled our very first participant, and we rapidly redesigned our trial into a virtual intervention. If COVID-19 had struck even a few months later, we would have been in the middle of our trial and forced to stop the study as our original intervention and evaluation were supposed to be in person. In preparation for the virtual study, we piloted all the study measures with a few participants before the start of enrollment. Our pilot participants reported challenges with completing the Edmonton Symptom Assessment Scale, PAM, and QOL-AD measures virtually as some questions had reverse wording and all had Likert-scale response options that made it hard for them to answer. As the measures were part of our original proposed study, we retained them as secondary measures. Second, in this study, early primary palliative care was provided by registered nurses. Trained registered nurses can assess participants and identify their medical and psychosocial needs, provide care coordination and education, and conduct conversations on goals of care; however, they cannot independently make a diagnosis or prescribe medications. While the nurses were able to fulfill the psychoeducation and care coordination needs of the participants, they had to work with the participants’ PCPs to fulfill all their medical needs. This can be seen as a study limitation, but we believe that this is also a strength as defining the scope of registered nurses as encompassing the provision of primary palliative care will inform the path forward as we work to expand the primary palliative care workforce. If effective, this intervention might be an innovative solution to offset the growing shortage of palliative care workers. Nurses trained in this intervention may be able to become the backbone of the primary palliative care workforce in the future. Third, this study was a virtual, ELICIT instituted early in the trajectory of cognitive impairment, and its long-term impact is currently unknown. We are conducting long-term observational follow-up of study participants to better understand whether the intervention has any impact on the end-of-life care of the study participants and whether it resulted in goal-concordant care.

Currently, very little is known about the early primary palliative care needs of patients with cognitive impairment living in the community. If successful, this trial will delineate the common early primary palliative care needs of patients at risk of cognitive impairment and those with mild cognitive impairment and mild and moderate dementia. Such knowledge will allow future interventions to better support persons living with cognitive impairment in the community.