Navigating a “Good Death” During COVID-19: Understanding Real-Time End-of-Life Care Structures, Processes, and Outcomes Through Clinical Notes

Abstract

Background and Objectives

The coronavirus disease 2019 (COVID-19) pandemic severely disrupted hospice care, yet there is little research regarding how widespread disruptions affected clinician and family decision-making. We aimed to understand how the pandemic affected structures, processes, and outcomes of end-of-life care.

Research Design and Methods

Retrospective narrative chart review of electronic health records of 61 patients referred and admitted to hospice from 3 New York City geriatrics practices who died between March 1, 2020, and March 31, 2021. We linked longitudinal, unstructured medical, and hospice electronic health record notes to create a real-time, multiperspective trajectory of patients’ interactions with providers using directed content analysis.

Results

Most patients had dementia and were enrolled in hospice for 11 days. Care processes were shaped by structural factors (staffing, supplies, and governmental/institutional policies), and outcomes were prioritized by care teams and families (protecting safety, maintaining high-touch care, honoring patient values, and supporting patients emotionally and spiritually). Processes used to achieve these outcomes were decision-making, care delivery, supporting a “good death,” and emotional and spiritual support.

Discussion and Implications

Care processes were negotiated throughout the end of life, with clinicians and families making in-the-moment decisions. Some adaptations were effective but also placed extraordinary pressure on paid and family caregivers. Healthcare teams’ and families’ goals to meet patients’ end-of-life priorities can be supported by ongoing assessment of patient goals and process changes needed to support them, stronger structural supports for paid and family caregivers, incentivizing relationships across primary care and hospice teams, and extending social work and spiritual care.

Background and Objectives

Home hospice offers comprehensive, interdisciplinary services that enable individuals to be cared for and die at home (Carlson et al., 2007). Hospice is associated with improved symptom control, quality of life, and satisfaction with end-of-life care (Harrison et al., 2022; Teno et al., 2007). However, hospice remains underutilized and often misunderstood (Cagle et al., 2020; MedPAC 2018), with its benefits undermined by systematic barriers to timely referral and enrollment and care coordination challenges between medical and hospice settings (Teno et al., 2007; Wang et al., 2019). Only half of eligible Medicare beneficiaries receive hospice benefits, and a third of hospice recipients are enrolled for less than a week before their death (MedPAC, 2018). In addition, honoring an individual’s preference to die at home is a complex and labor-intensive process, requiring close coordination between a number of interdisciplinary team members across nursing, home health, spiritual care, social work, and more (de Graaf et al., 2024), as well as substantial amounts of family caregiving as end-of-life care needs intensify (Bhagianadh & Arora, 2023).

Although these challenges existed prior to the coronavirus disease 2019 (COVID-19) pandemic, the public health emergency profoundly altered the structures that shape the delivery of home hospice care, highlighting existing gaps that can harm end-of-life care quality and disrupt the complex interdisciplinary teamwork required to provide it. Public health mitigation policies such as social distancing, infection control, and self-quarantine helped to contain the virus, but also disrupted caregiving networks and increased caregiver strain (Aldridge et al., 2023; Federman et al., 2021; Franzosa et al., 2023). Infection control policies, such as limiting in-person visits, disrupted usual medical, and end-of-life care (Aldridge et al., 2023). Shortages of staff and supplies, coupled with an unprecedented increase in the volume of end-of-life patients delayed hospice admission, disrupted care delivery and coordination, and threatened to overwhelm an already strained system (Gergerich et al., 2021; Shugarman et al., 2023). Hospices had to quickly adopt new protocols reflecting the new reality and state and federal regulations, including implementing COVID-19 screening and training for staff and patients; transitioning to telehealth services; managing admissions paperwork virtually; and determining how to manage COVID-positive patients (Shugarman et al., 2023).

Yet despite this rapidly changing environment, there is little research about how clinician and family decision-making was shaped by the significant barriers to care delivery during the COVID emergency. A key reason is the lack of data on real-time decision-making. We conducted a retrospective narrative chart review leveraging data from detailed medical and hospice clinical practice notes, a robust source of information as healthcare providers worked to share patient information across teams fragmented by COVID disruptions (Franzosa et al., 2023), to create longitudinal trajectories of patient care. This approach enabled us to conduct a chronological, multiperspective analysis of patients’ and families’ interactions with their primary care and hospice clinicians during the first year of the pandemic. Our goal was to identify how the structures in which care was delivered affected care processes and patient and providers’ priority outcomes and decision-making at the end of life. Although the COVID-19 pandemic was extraordinary, the ongoing fault lines it highlighted in end-of-life care provision can inform future practice and policy.

Research Design and Methods

Conceptual Framework

We grounded our study in the Donabedian model, which provides a framework for evaluating the quality of care (Donabedian, 1966, 1988, 2005). In this model, structures are the conditions under which care is provided, such as the organizational policy and resource environment. Processes are the activities that contribute to care, including specific types of care provided and communication between team members. Structures and processes then shape outcomes, or the results of care, and how these results are experienced by patients and families. Our team had previously examined the ways in which structural aspects of hospice organizations (e.g., for-profit ownership, chain affiliation) and regional healthcare supply affect hospice care processes, including National Quality Forum (Zheng et al., 2018) preferred practices for hospice care and their impact on patient and family outcomes (Aldridge, 2021; Aldridge et al., 2022; Harrison et al., 2022). We also identified changes in care processes during the early pandemic, including more fractured communication between healthcare teams, and difficulty providing care due to the workforce and medical supply shortages (Aldridge et al., 2023; Franzosa et al., 2023). For this study, we specifically considered public health containment and mitigation policies—including social distancing, infection control, and self-quarantine—as significant shifts in the structure of end-of-life care delivery with implications for hospice processes of care (staffing; visits; care coordination; supplies) and in turn, outcomes (timely hospice enrollment; missed hospice visits; unmet need; caregiver strain).

Study Population and Setting

Our setting included three geriatric practices at a large urban, academic health system in New York City: an outpatient geriatrics and palliative care practice; a home-based primary care practice; and a community-based palliative care practice. These practices offer primary and consultative palliative care to approximately 6,000 older adults with complex medical needs and multiple chronic conditions. Almost two thirds of hospice referrals from the system are made to our study partner, a large, nonprofit home care and hospice provider caring for more than 6,000 patients annually throughout the New York metro area.

Sample and Data Collection

The sample included practice patients who were referred, admitted to hospice, and died between March 1, 2020, and March 31, 2021. We identified patients who were either referred to hospice or died during the study period (as practices do not systematically record referrals) and matched them to hospice electronic health records (EHRs) using date of birth, last name, first name, and hospice referral or admission dates, if available. Adapting procedures from a previous study (Franzosa et al., 2023), we extracted sociodemographic information from structured fields in the health system EHR (age, race/ethnicity, gender, Medicaid eligibility, public housing), health status (dementia, receipt of home health aide care, date of death), and hospice utilization (enrollment status, admission date, length of admission). For our content analysis, we extracted unstructured, qualitative medical and hospice record notes describing in-person visits, phone calls, televisits, and emails to capture care structures, processes, and outcomes. Medical and hospice notes typically included providers’ activities and perceptions during in-person or virtual/phone visits (e.g., exams, wound care, medication administration, educating and counseling patients and caregivers); care coordination activities within and between medical and hospice teams (e.g., scheduling visits, ordering supplies and tests, sharing changes in patient health status, managing health insurance, home health aide scheduling), and text of emails and phone calls from, and responses to, patients and caregivers through the patient portal. Hospice notes also included admission notes; templated assessments for intake, COVID-19 screening and infection prevention protocols; and descriptions of spiritual care visits. Using a unique patient ID, we linked medical and corresponding hospice practice notes and organized each patient’s notes chronologically to capture their care trajectory from referral through death. Study coordinators (P.K., L.L.) reviewed linked notes for accuracy augmented by a quality check by a second team (E.F., L.M.) before being entered into Atlas.ti (v 23.2.1) for analysis.

Data Analysis

We analyzed each patient’s care trajectory across all clinical sites between March 1, 2020, and the last clinical note following their death. We used directed content analysis (Elo & Kyngäs, 2008; Hsieh & Shannon, 2005) as our analytic approach using the key variables and interrelationships in our prior work (Aldridge et al., 2023; Franzosa et al., 2023). We used the structures, processes, and outcomes described in our conceptual framework as our a priori codes, and then used an inductive approach to further refine and expand our framework to specifically examine COVID-19 impacts on care delivery.

To start, four coders (E.F., K.F., P.K., L.M.) independently reviewed four patients’ abstracted notes chronologically to gain insight into their care trajectory and applied our a priori codes to medical notes that matched our operational definitions. In analytic memos, coders noted which a priori codes were and were not found in the data, and highlighted data excerpts that challenged or were not captured in our existing codes but expanded our understanding of structures, processes, and outcomes of end-of-life care during COVID and the relationships between them. For example, although our a priori framework considered staffing and supplies as part of the delivery “process” of care, as the pandemic drove staffing and supply shortages these were more reflective of the “structure” in which care was delivered. We then jointly reviewed each patient case to refine existing code definitions and propose new codes. Coders flagged clinical or policy questions (e.g., whether a process was “usual care” or related to the pandemic) for review by our team’s clinician (L.M.) and/or hospice organization partners. The coders then independently coded six additional cases and jointly reviewed the codes to ensure consistency and refine operational definitions. Once the codebook stabilized, coders divided the remaining cases for coding and analysis. To ensure rigor, we maintained coding memos and flagged emergent questions about code definitions or applications, which were discussed in weekly meetings (Rolfe, 2006). As an additional quality check, a senior researcher (E.F.) reviewed 10% of the final coded sample to ensure that codes were applied consistently. Discrepancies in coding were resolved by group consensus.

Once initial coding was complete, we situated codes within our three main domains (structures, processes, and outcomes) using a structured template. The template guided us to reflect closely on whether the code captured causes of changes in care (structures); changes themselves (processes); or impacts (outcomes), using anchor exemplar quotes. We used these templates to guide the discussion of our domains and refine and extend our framework. We then reviewed the groups of codes captured under each domain to construct higher-level themes, constantly returning to our data to support our interpretations and conclusions. For instance, under the “process” codes, we identified four distinct process types, which we mapped to the most commonly observed outcomes. Throughout the analysis, we maintained analytic memos and an ongoing shared document detailing our discussions and decision-making to build a thorough, reflexive audit trail of our process. We also engaged in analyst triangulation and peer debriefing with the full study team and outside experts in hospice and clinical care to verify trustworthiness. We returned to this audit trail throughout our analysis to confirm our rationale and interpretations and to ensure that no relevant data was inadvertently excluded (Morrow, 2005; Rolfe, 2006).

Research activities were approved by the Institutional Review Board of Icahn School of Medicine at Mount Sinai (Protocol #21-00716-CR001). This paper adheres to the Standards for Reporting Qualitative Research (O’Brien et al., 2014; Supplementary Table 1).

Results

Patient Demographics

We identified 342 patients who were referred to hospice or died during the study period. Sixty-one patients met our inclusion criteria (matching hospice record; admitted/died during the study period; Figure 1). Most patients in the sample were older, had dementia, and were briefly enrolled in hospice before dying (Table 1). Median hospice enrollment was 11 days. Fifty-two percent were referred to hospice and 41% died between March 1 and June 1, 2020, corresponding to the peak of NYC COVID-19 hospitalizations and deaths (Thompson et al., 2020). Notably, 28 patients were excluded from our sample because they were referred but not admitted to hospice; most died within a day or 2 of hospice referral.

Figure 1.

Sample selection.

Table 1.

Characteristics of Patients Admitted to Hospice Who Died Between March 1, 2020, and March 31, 2021 (N = 61)

	Value
Sociodemographics
Age, mean (SD)	90.4 (9.71)
Female, %	75.4
Race/ethnicity, %
Black/African-American	7.3
Hispanic/Latinx	5.5
Asian	0.0
White	50.9
Other	36.4
Primary language English, %	83.6
Medicaid, %	63.9
Lives in public housing, %	16.4
Householda
Lives alone	16.4
Lives with 24/7 aide services	39.3
Lives with family	50.1
Clinical characteristics
Any home care aide services in study period %	100.0
Has dementia, %	93.4
Period of death, %
March 1 to June 30, 2020	41.0
July 1 to October 31, 2020	21.3
November 1, 2020 to March 31, 2021	37.7
Hospice characteristics
Period of admission, %
March 1 to June 30, 2020	52.5
July 1 to October 31, 2020	26.2
November 1, 2020 to March 31, 2021	21.3
Time from referral to admission days, median (range)	2 (0, 44)
Enrolled within 48 hr, %	52.5
Duration of hospice enrollment in days, median (range)	11 (1, 280)

^aFour patients lived with family and also received 24/7 home care aide services.

Qualitative Findings: COVID-19 Impact on the Relationship Between Structures, Processes, and Outcomes

Below, we present (1) the structural challenges that shaped possibilities for care delivery; (2) outcomes that were prioritized within this context; and (3) how care processes were adapted to achieve these outcomes. (See Table 2 for further examples.) We present the domains in this order as our analysis showed that these domains were not linear, and processes were frequently adapted to achieve emerging priority outcomes.

Table 2.

Processes Used to Meet Patients’ Priority Outcomes at End of Life and Illustrative Quotes

Priority outcomes and observed processes used to achieve them	Exemplar quote
Priority outcome: protecting clinician and patient safety Observed processes: Decision-making
Weighing virtual versus in-person care	Daughter asked about having the patient seen, but when I shared my thoughts that [urgent visit] was likely more risk than help—better to limit visitors in the apartment, daughter agrees. (Primary Care MD note, ID 658) I explained that trips to the hospital bring burdens and discomfort that remaining at home would not, and that for someone at the end of life, recurrent trips to the ED is burdensome and risky, especially now (Primary Care MD note, ID 477)
Weighing the value of hospice care (self-triage; science vs intuition)	Patient confused and very [hard of hearing]. She did not seem to recognize [spiritual care provider]. Daughter agreed that in-person visits not worth the risk of COVID at this point. (Hospice spiritual care note, ID 680)
Priority outcome: maintaining high-touch care Observed processes: care delivery
Coordinating across care settings	Overall situation seems to me that patient’s children, though clear they want to take care of their mom at home, have limited understanding of her condition or impending death and that video visits aren’t sufficient for this. Called and talked to [hospice care coordinator] at length. Discussed concerns that family need more support, that if as per [paid caregiver’s] assessment she is close to death family really need someone to assess her and communicate that in person, as well as teach them how to use meds. (Primary Care MD note, ID 598) Contacted [primary care practice] and collaborated with NP who knows patient well. Received verbal orders. Communicated to MD that patient appears imminent. NP will be visiting tomorrow (Hospice RN note, ID 276)
Advocating as a caregiver	Discussed colonization vs. infection but daughter very concerned mother will get worse. Agree to [antibiotic] to have in house if symptoms do not improve or begin to worsen. Also asking for labs—discussed likely not needed given we have [recent] labs—but she insists—so will order (Primary Care MD, ID 496) [Daughter] states [patient] was denied from hospice per intake nurse. Was told he needs to show more signs of decline … [daughter] doesn’t feel these are accurate measures of declines. She feels that he has more falls, recurrent UTI, weight loss and overall is weaker … she is inquiring how this decision can be overturned … Will reach out to social worker to discuss options (Primary Care Coordinator note, ID 126)
Doing what needs to be done	Daughter mentioned that aide was supposed to be relieved at 8am but aide never came; supposedly agency was looking for a replacement. Agency states that replacement will be at 5:30 p.m. Aide was not too happy but has no choice in the matter. Daughter was very understanding and knows that aide did a full shift the night before and is tired. (Hospice note, ID 66)
Expanding caregiving roles	Family, namely husband, is not comfortable giving her meds, but could be persuaded. We agreed that the better part of valor seems to be to hold off on administering meds unless patient seizes at which time (Primary Care MD, ID 367) Noted stage 2 wound to left buttock. Instructed [aide and daughter] to turn and reposition q2hr to prevent further skin breakdown. Instructed new wound care order. Daughter is able and willing to perform daily wound care when [skilled nurse] is not available. (Hospice RN, ID 137)
Priority outcome: honoring patient beliefs, values and goals of care Observed processes: “Good death” practices
Honoring final wishes	Issues with having [grandchild] visit as she [out of state] and was febrile this past weekend, tested and COVID negative. We discussed the need for a 14-day quarantine to protect both [patient] and her HHAs and I would suggest against travel for the grandchild right after a febrile illness in case the test is false negative. (Primary Care MD, ID 680)
Honoring religious and cultural practices	Because of COVID-19, the autopsy won’t be performed, so the family won’t know what the ultimate cause of death was, they are frustrated by this … They are frustrated because people ask them “what can we do for you,” though no one can come visit (Primary Care MD note, ID 756)
Supporting patients emotionally and spiritually Observed processes: supportive care
Supporting the full caregiving team (family, aides)	I encouraged your brother to call our office or home hospice if he has any questions. I don’t want him or any of you to worry alone (Primary Care MD note, ID 214) Social Worker offering EOL support for [family-paid aide] who is anxious and has worked with patient for 9 years. She has experienced death from other patients but remains frightened. SW offered her ongoing support, EOL ed and reinforced what to do at time of patient’s death (Hospice social work note, ID 29)
Continuing trusted relationships	I can only imagine how hard it is to not be able to visit your dad when that was the plan, but hopefully its reassuring to you that he seems to be ok now. Please stay safe and be in touch with any questions (Primary Care MD note, ID 1135)

Structures: Staffing, Supplies, and COVID-19 Policies

As with other documented reports, our data captured significant staffing and supply shortage challenges, with home care worker shortages the most common and urgent issue. In one case, a frustrated caregiver reported that an agency supervisor, “told her that she has no aides who want to work and that [the caregiver] should take care of her mother herself” (ID 659). Medical supplies were also in short supply as patients experienced delays for morphine, oxygen, hospital beds, and other necessities, in some cases requiring primary care or hospital staff to call multiple vendors, switch medications or dosages, or lean on personal relationships with vendors to procure supplies. One primary care doctor emailed a family caregiver about the need to proactively order medications, noting, “it might be difficult to obtain this morphine at the moment it is needed and, if goals of care are comfort-focused … [medication] will be crucial” (ID 184). Although these issues were most prevalent in the early months of the pandemic, they remained evident throughout our study period.

Changes in medical and hospice protocols, particularly around COVID-19 mitigation policies, also shaped care delivery. New regulatory and payment flexibilities allowed practices in our sample to employ electronic consent and virtual assessments and deliver social work and spiritual care virtually. These changes facilitated admission and allowed social workers and chaplains to quickly schedule urgent visits. The medical and hospice practices also implemented organization-level protocols, including decision tools in the EMR to assess the need for an in-person visit and requirements for COVID-19 screening and testing for staff and patients. However, protocols between organizations did not always align; for example, medical notes recorded delays initiating or re-starting home care worker services following hospital discharges because patients’ home health agencies required a negative COVID-19 test, a particularly challenging rule due to limited test availability early in the pandemic. In other cases, funeral homes declined to pick up COVID-positive patients, leaving families with few alternatives.

Outcomes: Safety, High-Touch Care, Honoring Goals of Care, Emotional, and Spiritual Support

Given these structural constraints, clinicians and families chose to prioritize certain outcomes. These included (1) protecting clinician and patient safety; (2) maintaining high-touch care; (3) honoring patient desires, values, and goals of care; and (4) supporting patients and families emotionally and spiritually. Some outcomes, like keeping patients and providers safe from infection and maintaining high-touch, hands-on services despite a shift to virtual care, represented new priorities specific to the pandemic. Others, such as honoring patients’ end-of-life goals and emotional and spiritual support, were already core hospice values that gained new importance as providers attempted to maintain care despite structural disruptions.

Processes: Decision-Making, Care Delivery, Supporting a “Good Death,” Emotional, and Spiritual Support

We identified four types of processes that clinicians and families used in trying to achieve each outcome, including (1) decision-making; (2) care delivery; (3) supporting a “good death”; and (4) enhanced emotional and spiritual support. Although we tie processes to their most commonly observed outcomes later, we also note processes were not mutually exclusive and were frequently used to achieve more than one of the identified outcomes.

Decision-making

Decision-making processes included weighing virtual versus in-person care and weighing the value of receiving care at all. As COVID-19 knowledge evolved, records showed that clinicians and family often made risk–benefit decisions based on their own expertise and instinct. Although the health system and hospice had policies in place to elect in-person or virtual care, incomplete evidence left leeway for clinicians to exercise their own judgment. EMR notes showed clinicians advising families on whether a diagnostic test was worth potentially exposing a patient or technician to COVID-19, or whether a bruise or wound needed to be evaluated in person, or if photos or video would suffice. In these cases, providers often talked through risks and benefits with families, most often erring on the side of lower-risk care, as such care generally aligned with hospice philosophy and patients’ goals. In one case, a provider reported a family requesting an X-ray for a patient after a fall. The primary care physician (PCP) advised instead monitoring symptoms through telehealth, noting that, “we need to be particularly thoughtful about the pros and cons of treatments/diagnostics during this pandemic” (ID 214). Three months into the pandemic, notes show that while study practices’ internal policies for in-person visits relaxed, in-person care was still often reserved for necessary hands-on services such as skilled nursing visits or home health aide services, whereas social work, spiritual care, and supervisory doctor visits remained largely virtual through the study period.

Family members also expressed worry for the safety of their loved ones. In some cases, families “self-triaged,” choosing not to engage the healthcare provider, virtual or otherwise, until there was an urgent need or caregiving responsibilities became too great. Notes show some families declined alternatives such as virtual last rites or hourly phone check-ins offered in lieu of continuous in-person care during the active dying phase, fearing it may confuse or upset the patient or simply not seeing the value in a virtual visit. Patients and families also turned to a trusted provider to help make decisions, often the PCP with whom they had a longer-term relationship than the hospice provider. In one instance, a family member emailed their PCP to express concerns that their loved one was experiencing withdrawal from psychotropic medication and was unhappy with how hospice was managing symptoms. After the PCP agreed with the hospice recommendations, the family member felt more confident proceeding with the plan of care. Finally, anxious families sometimes trusted their own intuition over “expert” guidance; one social worker noted that a family member refused to accept delivery of a hospital bed despite the delivery agent wearing the required protective equipment.

Care delivery

Despite restrictions on in-person visits, clinicians, families, and patients adapted care delivery processes to maintain as much of the high-touch care central to hospice as much as possible. This required increased coordination across care settings, family advocacy, doing what needed to be done, and expanded caregiving roles.

Medical and hospice providers worked closely together, particularly during the delicate period between hospice referral and admission. Primary care and hospice often work to bridge gaps during the hospice transition, and these efforts were particularly intensive during the first few months of the pandemic when staffing and medical resources were scarce. Primary care teams, in consultation with hospice, organized the delivery of oxygen, morphine, and other supplies to bridge the gap and provided anticipatory guidance to families about what to expect as patients entered the final stages of life. After admission, many PCPs continued as the patient’s provider of record, and had ongoing doctor-to-doctor and nurse-to-nurse communication with the hospice team to reconcile medications, manage symptoms, and address urgent issues like falls or COVID-19 infection. Coordination was not always smooth; in several cases, primary care team notes conveyed frustration about the lack of information or communication from the hospice team. Misaligned institutional policies also presented challenges; for instance, a PCP wrote that they had requested an in-person hospice visit for family caregivers who were anxious about administering morphine and wanted hands-on training but were told this did not rise to the level of urgency required for an in-person visit.

Families regularly advocate for patients, and in the pandemic context, they more actively became “squeaky wheels” to advocate for services they felt their family members needed. Our data showed persistent phone calls requesting additional home care worker services, service referrals, and appeals of insurance denials. In some cases, family members pre-emptively emailed PCPs for a hospice referral to manage gaps in care that could otherwise necessitate inpatient admission or asked for a second opinion (e.g., requesting a diagnostic test or medication change) if they were unsatisfied with the response from the hospice team. Despite these efforts, families routinely encountered administrative barriers such as insurance requirements or difficulty reaching providers.

As in-person care was disrupted, both family and paid caregivers took on new roles and responsibilities, providing medical support such as administering IV medications or wound care that would normally be performed by a clinician. Nurses frequently reported talking to family members and home care workers through these tasks by phone or video. Home care workforce shortages created extra demands for both aides and families. One primary care coordinator described an aide working a second shift after an overnight when a replacement worker did not show up. When the agency informed them a replacement wouldn’t be available until the evening, “the aide was not too happy but has no choice in the matter. Daughter was very understanding and knows that aide did a full shift the night before and is tired” (ID 137). In another case, a patient tested positive for COVID-19; her PCP wrote that the “aide will have to leave and [daughter] will need to be caregiver and this will be really hard for her but she can do it” (ID 615).

Supporting a “good death”

Primary care and hospice teams worked closely together to provide a “good death” as defined by the patient and to honor final wishes, although infection control policies and resource shortages frequently challenged their efforts. We observed that family members often accepted in-person visit restrictions earlier in the care trajectory, but reported distress when they were unable to have an in-person goodbye closer to the end of life. One patient’s daughter emailed that she was distraught when travel restrictions prevented her traveling from Europe to be with her dying mother. Another reportedly became tearful after testing positive for COVID-19 and realizing she would not have the closure of an in-person goodbye. Some families, particularly close to death, chose to take risks. In one case, a PCP noted that a patient was determined to make a last visit to family in another state despite social distancing and travel restrictions. Hospice and primary care notes showed team members working with her family to find a temporary out-of-state hospice provider and arranging COVID-19 testing and transport.

Both provider notes and patient and caregiver emails demonstrated patients’ and families’ strong desire to avoid hospital and nursing home admission, citing media “horror stories,” and expressing distress at the thought of patients dying alone. In these cases, medical providers worked closely with families to meet patients’ goals of care to die at home. Honoring religious and cultural practices also became a priority during the disruptions of the first pandemic surge as funeral homes were inundated, at times making timely collection of bodies impossible. One daughter (ID 128) expressed sadness in an email over the family’s inability to honor the patient’s wishes to donate their body to science, because the patient was COVID positive and the donation could not be accepted. Another family was unable to have an autopsy to determine whether their parent had a long-suspected health condition.

Emotional and spiritual support

Spiritual care and emotional support are essential aspects of hospice and were particularly valued by clinicians and families during this time of heightened anxiety and stress. Clinicians, social workers, and spiritual care providers actively tried to support the whole household by caring for the patient. In one case, a PCP offered reassurance to siblings struggling with how to best care for their mother and avoid inpatient admission after multiple falls and emergency visits from the fire department. Via email, their PCP validated their feelings of being overwhelmed, praised their care, and acknowledged that “it’s a lot to be a caregiver, especially at this time” (ID 214). Families frequently reported appreciating these efforts. As one long-distance caregiver emailed to her father’s PCP, “I am so grateful for your words, updates, and constant supervision of my father. They are so precious to me, especially now in these challenging times” (ID 1135). Hospice social workers and chaplains also recorded in their notes that home care workers expressed feelings of exhaustion, anxiety, and fear. As with families, staff reported providing these workers with validation, active listening, and reassurance.

Medical and hospice teams also worked to ensure continuity of care and trusted provider relationships during transitions to hospice, reassuring families and patients that the medical team would remain available and, in many cases, continue caring for them until the end of life. In one case, a PCP reassured a dying patient via email that he would, “follow [him] closely in the coming days and would call twice a day” until the end (ID 155). Another PCP emailed her patient’s children that, “I will remain her primary MD. I work closely with hospice and am a hospice and palliative care specialist. As I have told her again and again, I love her dearly and have no plans of giving her up as my patient” (ID 214).

Dynamic Relationships Between Structure, Process, and Outcome Domains

The Donabedian model and our a priori framework indicated that structures and processes would impact outcomes of hospice care (Donabedian, 1966) in a linear form. However, our data revealed a more dynamic, iterative relationship across these domains. In a reversal of the expected directionality, outcomes prioritized by families and patients drove changes in care processes that enabled them to achieve their goals. For instance, the greater dependence on family caregivers, which was driven by structural staffing and supply shortages, made emotional support for families a higher-priority outcome. Providers responded to this need by adapting processes to increase social work and spiritual care visits and to have nurses provide virtual hands-on skills training. As we anticipated, public health containment and mitigation policies substantially changed the structure of hospice care, as did staffing and supply shortages, which we had initially considered process factors. Our analysis also showed our a priori outcomes (e.g., missed hospice visits, unmet needs) inadequately captured patient and family experiences, so we adjusted our model to broaden these concepts (Figure 2).

Figure 2.

Relationships among care delivery structures, processes, and priority outcomes.

Discussion and Implications

In the early COVID-19 pandemic, clinicians, families, and patients in our study faced structural pressure to provide a “good death” under extraordinary circumstances. Hospice is a system of care with defined structures, processes, and outcomes that require substantial adaptation to meet new challenges. We found that care processes were negotiated and iterated throughout the end-of-life, with clinicians and families making in-the-moment decisions about the care they were comfortable receiving and providing and how it could best be delivered to achieve patients’ and families’ changing goals. Yet many of these challenges, while intensified by the pandemic, were not new. Healthcare organizations have long struggled to support timely hospice admission, ease transitions between medical and hospice care, and adequately support and engage family caregivers. The dynamism required by the pandemic forced hospices to look closely at their model and processes, allowing for innovation in a space with a long-set model of care provision.

We are not the first to argue that structures, processes, and quality outcomes of end-of-life care are not fixed or independent, but dynamic and fluid. Krikorian and colleagues found in a systematic review that “what constitutes a good death is … highly individual, changeable over time, and based on personal perspectives” (Krikorian et al., 2020) with patient and family priorities fluctuating within and between groups and at different stages of the dying process. Bhagianadh and colleagues similarly advocate for a multidimensional conception of end-of-life care quality, noting that rather than distinct “silos” it is critical to consider simultaneous interrelationships across domains (Bhagianadh & Arora, 2023). For example, although most individuals want to die at home, supporting this goal requires intensive contributions from caregivers (Nowak, 2016). Focusing on narrow quality outcomes, such as the number of visits, may obscure the complexity of what it takes to help an individual die at home and does not fully capture the nuanced patient and family experiences apparent in our data. Future studies should consider the interplay between these domains.

Considerations for Patients and Families

Our analysis supports existing evidence that caregivers experience substantial emotional, financial, and physical challenges in managing complex end-of-life care at home (Campling et al., 2017). Families’ need for support is not unique to the COVID-19 context, although our study demonstrates that it became more urgent in the face of resource shortages and public health restrictions. Our notes showed caregivers performing tasks with which they were uncomfortable, taking on extra caregiving shifts, and advocating for patients. Caregivers who are less comfortable managing patients’ symptoms report higher caregiver burden and lower satisfaction (Quigley et al., 2020), whereas family support and emotional and spiritual support are associated with higher hospice care ratings (Anhang Price et al., 2020; Hotchkiss et al., 2024). Providers’ deliberate efforts to provide emotional and spiritual support, in addition to hands-on skills training, was a key support process for many families. In addition to social workers and spiritual care providers, we also found that nursing staff proved instrumental in bridging communication between care teams and families, including supervising complex tasks such as wound care remotely. Efforts to provide caregivers with skills training and access to emotional and mental health services may help alleviate some strain (Burgdorf, Reckrey et al., 2023). Additionally, although COVID-19 was an extreme case, family and paid caregivers regularly encounter challenges navigating the health system and advocating for services (National Academies of Sciences & Medicine, 2016). That persistent caregiver advocacy was sometimes vital to accessing services demonstrates that patients without a caregiver, or whose caregivers lack knowledge about the health system, will face even greater barriers (Burgdorf et al., 2023). This may further drive inequities in access to and quality of hospice services and deserves attention from healthcare providers and policy-makers in considering strategies to enhance case management and patient navigation for targeted patient populations.

Considerations for Healthcare Organizations

COVID-19 mitigation policies also created new considerations in balancing safety with hands-on care needs and preferences. We found that risk–benefit calculations varied across cases and end-of-life stages, and that clinician and family priorities were not always aligned. Clinicians frequently advised against in-person visits, particularly in the first pandemic months. Although circumstances sometimes made an in-person goodbye impossible, in some cases, a physical visit or last rites may have alleviated families’ guilt and distress despite the potential risk. These findings suggest that patient and family priorities may change rapidly during the last stages of life, highlighting a need for healthcare organizations to create defined processes that weigh safety concerns against regulatory constraints, and the value of in-person versus remote care at specific touchpoints. These changes are also important signposts for care delivery processes in the future. Although new flexibilities like enhanced telephonic support to patients and families have continued, multiple patients in our study declined the spiritual and social work visits they were entitled to because they were offered virtually. Given these findings and documented access barriers to telehealth (Franzosa et al., 2021; Kalicki et al., 2021), research examining the most effective modalities and preferences for these services across the care trajectory is important to ensuring patients and families receive the full benefit of hospice care and that existing disparities in care delivery and quality are not exacerbated.

Supporting providers and other care team members is also essential as the end-of-life care field grapples with workforce shortages. Many hospice social workers reported significant moral distress at not being able to provide the level of care they felt necessary during the pandemic (Gergerich et al., 2021; Head et al., 2019) and more social workers left hospice care than any other health sector, exacerbating a trend (Vossel, 2022). Given the importance that patients and families place on these services as part of quality care, prioritizing investments in social work and spiritual care to help address patient priorities may help improve perceptions of hospice value and quality outcomes. The home care workforce is similarly facing crisis shortage levels exacerbated by COVID-19 (Tyler et al., 2022) as our data reflected. In addition to the strain on clinicians, our data highlighted emotional pressures on home care workers, who rarely have access to grief and bereavement support offered to clinical staff (Tsui et al., 2019). Hospice providers might consider ways to support these important front-line workers going forward. For example, during the pandemic, VNS Health, a large home care provider serving the New York metro area, began a phone-based, professionally facilitated group support call for aides which continued throughout the public health emergency (Tsui et al., 2022). Developing enhanced training for paid caregivers specifically around navigating end-of-life care and managing its emotional impact may better prepare them in providing high-quality care to dying patients and their families.

Considerations for Policy and Payment

End-of-life care is easily fragmented, as it is delivered across multiple teams engaged in different care processes between various organizations and care settings (de Graaf et al., 2024). Gaps in medical and hospice care coordination, as well as ensuring optimal timing of hospice enrollment, are an ongoing challenge, and late enrollment can prevent families and individuals from receiving the full benefits of hospice care and result in poor outcomes (Aldridge et al., 2015). Our analysis showed that some families hesitated to enroll in hospice for fear they would lose access to their doctors, and that efforts to maintain “warm handoffs” and ongoing communication between primary care and hospice were particularly important to patients and families during this uncertain time. Our sample included medically complex patients, nearly all with dementia and with long-standing relationships with primary care teams. These teams were accustomed to working closely with hospice, and many continued as providers of record through death, reducing care fragmentation at the end of life. Even when transitions were not seamless, having a known provider bridging the transition to hospice was reassuring to families and patients. It is also notable that patients in our study had a median hospice stay of only 11 days, and many of those referred but not enrolled in hospice died within 1–2 days of referral (Figure 1). The national median length of stay remains stubbornly at 18 days (MedPAC, 2022) despite findings that repeatedly demonstrate the value of early receipt of hospice services for patients and families (Adams et al., 2009; Teno et al., 2007; Yamagishi et al., 2015). Clearly, there is ample opportunity to improve more timely and beneficial admission. Payment models that incentivize coordination across medical and hospice care could help bridge this gap. For instance, the CMS Medicare Care Choices Model demonstration allowed beneficiaries to enroll in hospice without forgoing other care and was shown to improve hospice enrollment and reduce racial and ethnic disparities (Kranker et al., 2022; Niedzwiecki et al., 2024). Similar models to incentivize care across the medical and hospice settings could help facilitate and maintain trusting provider relationships. In addition, allowing billing for transitional care could support earlier hospice enrollment and improve care coordination (Kumar et al., 2020).

Study Strengths and Limitations

Our innovative study design captured objective, real-time choices and decision-making processes of care teams, patients, and families during a uniquely chaotic time, providing the understanding of how end-of-life care decisions are made in the moment. Additionally, the practices in our sample were particularly attuned to end-of-life care. A majority of geriatric providers in our three practices are certified in Hospice and Palliative Medicine and have expertise with patient transitions to hospice, which is not the case for many others. One study limitation is that our sample included only patients admitted to hospice. Given the low rates of hospice enrollment nationally and particularly in New York State (Dhingra et al., 2022), specific examination of patients who never receive hospice care is important. We could also only analyze data that were reported and documented in patient records. Hospice notes were largely templated and included fewer unstructured comments, which may not capture some nuances of the patient experience. Although our study design did capture family perspectives through the text of their emails and phone calls, we did not solicit their opinions and perceptions directly, and many interactions with families were interpreted through the provider lens.

Challenges in end-of-life care coordination and quality are not new. The stress of COVID-19 identified opportunities to improve hospice access and quality moving forward. Aligning practice and policy with what matters most to patients and families, and considering multidimensional aspects of care, can improve the quality of and satisfaction with hospice in the long term.

Funding

This work was supported by the National Institute of Nursing Research of the National Institutes of Health under Award Number R01NR019792 to M. D. Aldridge.

Conflict of Interest

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health, the Veterans Health Administration, or the U.S. Government. The authors have no conflicts of interest to report.

Data Availability

Data sharing is not possible due to IRB restrictions and ethical considerations. This study was not preregistered. Examples of the authors’ analytic process are included in the manuscript and are available upon request from the corresponding author.

Author Contributions

Emily Franzosa (Conceptualization [Equal], Data curation [Equal], Formal analysis [Lead], Methodology [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]); Patricia Kim (Data curation [Equal], Formal analysis [Equal], Project administration [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]); Laura Moreines (Data curation [Equal], Formal analysis [Equal], Methodology [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]); Margaret McDonald (Conceptualization [Equal], Data curation [Equal], Formal analysis [Equal], Resources [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]); Daniel David (Formal analysis [Equal], Methodology [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]); Jonelle Boafo (Formal analysis [Equal], Project administration [Equal], Writing—review & editing [Equal]); Dena Schulman-Green (Conceptualization [Equal], Formal analysis [Equal], Investigation [Equal], Methodology [Equal], Writing—review & editing [Equal]); Abraham Brody (Conceptualization [Equal], Data curation [Equal], Funding acquisition [Equal], Methodology [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]); Melissa D. Aldridge (Conceptualization [Lead], Data curation [Lead], Formal analysis [Lead], Investigation [Lead], Methodology [Lead], Supervision [Equal], Writing—original draft [Equal], Writing—review & editing [Equal])