Barriers and Facilitators to End-of-Life Care Delivery in ICUs: A Qualitative Study

Abstract

OBJECTIVES:

To understand frontline ICU clinician’s perceptions of end-of-life care delivery in the ICU.

DESIGN:

Qualitative observational cross-sectional study.

SETTING:

Seven ICUs across three hospitals in an integrated academic health system.

SUBJECTS:

ICU clinicians (physicians [critical care, palliative care], advanced practice providers, nurses, social workers, chaplains).

INTERVENTIONS:

None.

MEASUREMENTS AND MAIN RESULTS:

In total, 27 semi-structured interviews were conducted, recorded, and transcribed. The research team reviewed all transcripts inductively to develop a codebook. Thematic analysis was conducted through coding, category formulation, and sorting for data reduction to identify central themes. Deductive reasoning facilitated data category formulation and thematic structuring anchored on the Systems Engineering Initiative for Patient Safety model identified that work systems (people, environment, tools, tasks) lead to processes and outcomes. Four themes were barriers or facilitators to end-of-life care. First, work system barriers delayed end-of-life care communication among clinicians as well as between clinicians and families. For example, over-reliance on palliative care people in handling end-of-life discussions prevented timely end-of-life care discussions with families. Second, clinician-level variability existed in end-of-life communication tasks. For example, end-of-life care discussions varied greatly in process and outcomes depending on the clinician leading the conversation. Third, clinician-family-patient priorities or treatment goals were misaligned. Conversely, regular discussion and joint decisions facilitated higher familial confidence in end-of-life care delivery process. These detailed discussions between care teams aligned priorities and led to fewer situations where patients/families received conflicting information. Fourth, clinician moral distress occurred from providing nonbeneficial care. Interviewees reported standardized end-of-life care discussion process incorporated by the people in the work system including patient, family, and clinicians were foundational to delivering end-of-life care that reduced both patient and family suffering, as well as clinician moral distress.

CONCLUSIONS:

Standardized work system communication tasks may improve end-of life discussion processes between clinicians and families.

Approximately 20–30% of people die in the ICU in the United States (1–4). Furthermore, nearly 60% of ICU admissions result in death (1). End-of-life care delivery presents particular challenges because of clinical heterogeneity among ICU patients ranging from sudden-onset catastrophic illness or injury in young people to acute exacerbation of chronic disease among people who survived previous critical illness. This clinical heterogeneity leads clinicians to communicate uncertainty of treatment benefit and recovery potential to patients and families through level-setting, leading to patient and family denial or anger as a normal response to grief (5). The uncertainty coupled with miscommunication about treatment and recovery leads to worse patient experiences and family satisfaction with end-of-life care delivery (6). Less patient-centered decisions and patient-clinician communication has also been associated with family perceptions of poor end-of-life care delivery (7–9). Wide discrepancies have been documented between patient stated preferences and the end-of-life care delivered (10–12). These conflicts are exacerbated by disagreement between clinicians on patient clinical goals (13–16).

ICU end-of-life quality metrics consensus statements have been released to standardize end-of-life care delivery (17–20). These quality metrics include patient-centered decisions and patient-clinician communication, both defined by expert opinion. Quantitative studies have demonstrated ICU-level variation in achieving these quality metrics in end-of-life care delivery (21–23). Yet, frontline ICU clinicians perspective of how ICU work system components (people, environment, tools, tasks) produce end-of-life care delivery process metric variability has not been studied.

The overall objective of this study was to identify which ICU work system components act as barriers or facilitators of end-of-life care delivery process metrics (10) from the perspective of frontline ICU clinicians across seven ICUs in three hospitals. The study hypothesis was that the work system results in end-of-life care delivery process variability leading to perceived worse outcomes for patients, families, and clinicians.

MATERIALS AND METHODS

Study Design

A qualitative observational cross-sectional study was conducted to identify which ICU work system components act as barriers or facilitators of end-of-life care delivery from the perspective of frontline ICU clinicians across seven ICUs in three hospitals in a large academic health system. The Consolidated Criteria for Reporting Qualitative Studies checklist was used to describe the process of qualitative data collection, analysis, and reporting of the findings (24). This study was approved by the University Institutional Review Board Committee (Study Number: STU00215198, Approval Date: May 26, 2022, Study Title: Understanding the Differences in Drivers of End-of-Life Care Delivery Across Intensive Care Units). Informed consent was obtained from all participants. All procedures were followed in accordance with the Helsinki Declaration of 1975.

Theoretical Framework

Systems Engineering Initiative for Patient Safety (SEIPS) model was the theoretical framework used for this analysis. SEIPS builds on the classic Donabedian quality metrics of structures, process, and outcomes in healthcare (25). SEIPS provides a robust definition of structure into work system components. These work system components include the people, environment, tools, and tasks.

End-of-life care delivery process was defined across the six quality domains of end-of-life care (10). These six domains included: 1) patient and family-centered decision-making; 2) communication within the team and with patients and family; 3) end-of-life care continuity; 4) emotional and practical support for patients and families; 5) symptom management and comfort of care; and 6) spiritual support for patients and families and emotional and organizational support for ICU clinicians (10).

Setting

This study was conducted in seven ICUs across three hospitals within an academic urban healthcare system located in the Midwestern United States between May 2022 and November 2022. These hospitals were selected to ensure transferability of the findings and included a large urban teaching hospital, a suburban tertiary care community hospital, and a suburban community hospital.

Participant Selection

First, the relative number of participants with different training were purposively sampled using an initial sampling strategy to select respondents most likely to yield a wholistic understanding of the study question in identifying which ICU work system components act as barriers or facilitators of end-of-life care delivery process quality of care metrics. This included frontline ICU clinicians involved in end-of-life care such as physicians (intensivists, anesthesiologists, surgeons, palliative care physicians, fellows, and residents), critical care nurses, advanced practice providers (physician assistants, nurse practitioners), social workers, and chaplains (26). Overall, 330 frontline ICU clinicians were invited to participate in virtual, semi-structured interviews via email. Up to three follow-up emails were sent if no response was received. Following initial purposive sampling, snowball sampling was conducted which included referrals from initially sampled respondents to ensure a heterogeneous study population, all with broad expertise in end-of-life care delivery as well as representative participation from each clinical discipline, training level, type of ICU, and hospital setting (26). The final numbers of participants with different training background for the study was decided once new participant enrollment no longer increased the diversity information power across clinician roles and hospitals. Qualitative research studies are not designed to seek generalizable findings, as understood through statistical-probabilistic generalizability. We do, however, expect “naturalistic generalizability” which is the recognition of similarities and differences to the results with which the reader is familiar and “transferability” which is whenever a person or group in one setting considers adopting something from another that the research has identified (27, 28). Other aspects of our study design contributed to increasing the rigor of our qualitative research, in particular, the many interdisciplinary roles, the variety of types of ICU, and the variety of hospital settings of participants who agreed to an interview. Neither patients nor family members/caregivers were involved in this study, as the goal was to assess clinician perceptions of work system determinants of end-of-life care delivery.

All participants provided written consent. All interviews were conducted in dyads consisting of full-time postgraduate physician researchers (B.K., E.A., A.C., L.M.J., K.T., S.Y.). Interest in the research topic was motivated by personal experience with the challenges of end-of-life care during their medical training. Two interviewers had several years of experience in qualitative methods. Interviewers were assigned to conduct participant interviews to ensure no previous working relationship. Participants were informed of the stated aims of the study and reasons for conducting the research. All interviews were conducted virtually with only the participant.

Data Collection

A semi-structured interview guide was developed by a post-doctoral research fellow (B.K.) and intensivist (J.M.K.), at the university where the study was conducted (Supplemental Digital Content, http://links.lww.com/CCM/H505). The interview guide was iteratively reviewed by the study team made up of clinicians and researchers with backgrounds in health services research (J.K.K., A.M.S), qualitative research (J.K.J.), and educational research (E.A.). The interview guide was pilot tested and iteratively revised with input from nonparticipant experts to understand how quality of end-of-life care delivery resulted from the SEIPS work system components (people, environment, tools, tasks) (25, 29). One interviewer led the interview while the other interviewer took field notes. Interviews lasted between 30 and 60 minutes depending on participant responses. The final number of interviews and sample size of the study was determined by thematic saturation grounded in the concept of information power (30). Information power depends on: 1) the aim of the study (how quality of end-of-life care delivery resulted from the work system), 2) sample specificity (high), 3) use of established theory (SEIPS), 4) quality of dialogue (rich), and 5) analysis strategy (inductive and deductive coding). New participant enrollment ceased when subsequent interviews no longer generated new information or increased diversity in participation across hospitals and clinician roles, including clinician specialty and level of training, to maximize transferability of findings (30). No repeat interviews were conducted. Interviews were audio-recorded, transcribed verbatim, and de-identified.

Data Analysis

Transcripts were uploaded into MAXQDA 2022 (VERBI Software, 202, Berlin, Germany). Six team members coded transcripts (L.M.J., B.K., E.A., A.C., K.T., M.L.T.N.). All transcripts were coded in dyads. Conflicts and discrepancies in coding were resolved with iterative revisions to reach consensus within the coding dyads. A codebook (i.e., coding tree) was inductively generated and revised by the entire study team after reviewing three transcripts, then used to analyze all remaining data. Thematic analysis was conducted through coding, category formulation, and sorting for data reduction to identify central themes (31).

Deductive reasoning was applied for data category formulation and thematic structuring anchored on the SEIPS model, which was developed to understand the structures, process, and outcomes in healthcare as well as their relationships to each other (25). Disagreements were discussed until consensus was reached under the supervision of an intensivist with qualitative experience (A.M.S.) and a qualitative methodologist (J.K.J.). The central themes were data derived with specific quotes selected which most adequately represented each theme. Member checking was conducted with the last handful of participants to receive feedback on findings, ensure the study results adequately summarized their responses, and then incorporated changes based on participant feedback (32).

RESULTS

A total of 27 semi-structured interviews with ICU clinicians were conducted including a wide representation based on participant age, gender, and clinician roles (Table 1). Critical care nurses were the most represented clinician participants (25.9%) followed by equal representation from intensivist attendings (14.8%), palliative care attendings (14.8%), and social workers (14.8%). This sample represented cardiothoracic, cardiac, surgical, medical, and neuro ICUs across three hospitals including a large academic urban hospital (59.3%), a tertiary care community hospital (18.5%), and a small suburban community hospital (22.2%). Four main themes were identified from the interviews (Fig. 1). First, work system barriers delayed end-of-life care communication among clinicians as well as between clinicians and families; clinician-level variability existed in end-of-life communication tasks; poor communication tasks exacerbated the misalignment of clinician-family-patient priorities and treatment goals; and this misalignment leads to clinician moral distress from providing nonbeneficial care.

TABLE 1.

Demographic and Professional Characteristics of ICU Clinicians Who Participated in End-of-Life Care Delivery Interviews, May 2022 to November 2022

ICU Clinician Roles	n (%)
Age
< 35 yr old	6 (22.2)
36–50 yr old	16 (59.3)
> 50 yr old	5 (18.5)
Female gender	15 (55.6)
Clinician role
Critical care nurse	7 (25.9)
Critical care fellow	2 (7.4)
Resident	1 (3.7)
Critical care attending	4 (14.8)
Surgery	1 (3.7)
Cardiology	1 (3.7)
Pulmonology	1 (3.7)
Anesthesiology	1 (3.7)
Advanced practice providera	3 (11.1)
Palliative care attending	4 (14.8)
Social work	4 (14.8)
Chaplain	2 (7.4)
Hospital
Large urban teaching hospital	16 (59.3)
Tertiary care community hospital	5 (18.5)
Suburban community hospital	6 (22.2)

^aOf the three advanced practice providers, two worked directly on ICU teams and one worked with a palliative care team.

Figure 1.

Systems engineering initiative for patient safety conceptual model of central themes for end-of-life care delivery in the ICU, illustrating the themes and model structure derived from deductive reasoning of ICU clinician interviews.

Theme No. 1: Work System Barriers Delayed End-of-Life Care Communication Among Clinicians As Well As Between Clinicians and Families

ICU clinicians reported difficulties in discussing end-of-life care with families when all care teams waited until a patient’s prognosis progressed to very poor. Delays in holding family meetings were common to end-of-life care delivery (Table 2, Quote No.

TABLE 2.

Representative Quotes of the Central Theme: Work System Barriers Delay End-of-Life Care Communication Among Clinicians As Well As Between Clinicians and Families, From ICU Clinicians Who Participated in End-of-Life Care Delivery Interviews, May 2022 to November 2022

Quote No.	Representative Quote
No. 1	“The bigger barrier is deciding when to have a family meeting. We don’t get to the point of having a family meeting until it’s obvious that prognosis is very poor… we’re delayed in admitting that to the family members and admitting that to ourselves.” ICU Nurse
No. 2	“After we do interdisciplinary rounds in the morning, then we decide who needs the goals-of-care interventions. And then we try to get those going as soon as possible. You can have goals-of-care discussion on and off through the entire stay, you don’t have to wait until the patient’s dying to have a goals-of-care discussion.” ICU Nurse
No. 3	“More intentional and early training for trainees. You might be there late that day and your fellow is not available and your attending’s gone home. One of these situations can come up as early as the first day.” ICU Attending
No. 4	“Some [specialists] don’t think it’s in their scope or purview to have to have those conversations. There are a lot of people who do take it on, but there’s this binary, ‘Okay, we’ll take care of the patient and then if they’re dying, let’s defer to palliative care.’” ICU Fellow
No. 5	“There’s a lot of cooks in the kitchen… who are invested in this patient’s care, and so they might have all differing focuses. So it’s just hard because they might not always communicate, or what they communicate to the patient might be different if another provider comes.” ICU Nurse
No. 6	“The interdisciplinary rounding is extremely helpful. Oftentimes the nurse knows more than anybody, so that gives him or her a really good forum to do that. It’s really beneficial for the patient” Palliative Care Attending
No. 7	“There’s some barriers to this sometimes with people with communication disabilities or different primary languages, and there’s barriers in cultural differences.” ICU Attending
No. 8	“Communication works well when there’s regular updates and there’s compassion that’s provided. But usually compassion where you’re trying to understand what the patient’s family is feeling rather than assuming. Even if you think they’re culturally the same, all people have a different history. So assume too much, you’re going to get yourself in trouble.” ICU Attending
No. 9	“The physical space issue is a big issue… There’s not a lot of places outside the room for a family meeting that are accommodating and private.” Palliative Care Attending
No. 10	“You’re having a life altering event and there’s this alarm that won’t stop from the room three doors down… plus families do not take good care of themselves in the midst of an ICU setting. The ICU is not the ideal setting for end of life care.” Chaplain

1. Conversely, tasks engaging in end-of-life discussions throughout a patient’s ICU stay were facilitators to end-of-life care delivery (Table 2, Quote No.

2. Early training of residents in end-of-life discussions was also a facilitator because learners participated in these discussions even on their first day (Table 2, Quote No. 3). Perceived over-reliance on other people such as palliative care in end-of-life discussions prevented timely end-of-life care delivery (Table 2, Quote No. 4).

Participants reported that the multidisciplinary team structure created work system barriers to end-of-life care communication. ICU clinicians described challenges in clinician-to-family care coordination when patients had multiple people who were subspecialty clinicians with different or conflicting treatment goals (Table 2, Quote No. 5). Conversely, environment spaces for care communication such as interdisciplinary rounding, facilitated end-of-life care communication (Table 2, Quote No. 6).

Additionally, people or patient/family experiences, language, and culture could lead to barriers in care communication between clinicians and families (Table 2, Quote No. 7). ICU clinicians reported that acknowledging cultural differences was a task that facilitated care communication among some clinicians and families. Tasks such as spending time to understand a patient’s culture was highly valued thus facilitated end-of-life care delivery (Table 2, Quote No. 8).

Environments within the ICU were also barriers to care communication between clinicians and families, and thus, end-of-life care delivery. For example, when there was not enough physical room to comfortably accommodate a family meeting in a private setting (Table 2, Quote No. 9). Furthermore, frequent alarms in the environment prevented the ICU clinicians from delivering uninterrupted end-of-life care or end-of-life discussions (Table 2, Quote No. 10).

Theme No. 2: Clinician-Level Variability in End-of-Life Communication Tasks

ICU clinicians described different styles in end-of-life care communication tasks with patients and their families that was a barrier to end-of-life care delivery. They noted that “the quality of the conversation depends very highly on who is having it” (Table 3, Quote No. 1). Participants reported that consistent clinicians, such as advanced practice providers, played a pivotal role in end-of-life discussions and palliative care teams. Advanced practice providers had rich institutional memory and were reported as “lynch pins” to end-of-life discussions because of their consistent role in care teams, close relationships and trust they built with patients and families over time as a function of their schedule as opposed to ICU nurses who switched off every few days, attendings that switched off every week, and learners who switched off every month. One identified aspect of this variability was clinician willingness and effectiveness engaging in end-of-life communication tasks, as many clinicians were uncomfortable (Table 3, Quote No. 2). Conversely, some ICU clinicians reported that variability in end-of-life care communication tasks required skills specific to certain providers (Table 3, Quote No. 3, Quote No. 4). However, many ICU clinicians reported that discomfort stemmed from inadequate training. Although one ICU attending stated that trainees receive some education in how to have end-of-life care communications tasks, it continued to vary by patient due to a lack of standardized protocols (Table 3, Quote No. 5).

TABLE 3.

Representative Quotes of Clinician-Level Variability in End-of-Life Communication Tasks, From ICU Clinicians Who Participated in End-of-Life Care Delivery Interviews, May 2022 to November 2022

Quote No.	Representative Quote
No. 1	“People have such different styles. The quality of the conversation depends very highly on who is having it. There’s no standardization.” ICU Fellow
No. 2	“Knowing how to communicate willingness, meaning people may not be comfortable especially [with] end of life discussions. Providers are terrible at [communicating prognosis] to begin with. That can be a barrier.” Palliative Care Attending
No. 3	“Everybody is not equally talented as communicators. Some people are just better at it than others.” ICU Attending
No. 4	“That goes again with the individual. It depends on how well or how great they are with communication. It really depends on the person. We try to communicate everything as best we can, but some people are better than others.” Palliative Care Attending
No.5	“It varies patient by patient. Our trainees do rotate on our palliative care team… so they do get training in how to have discussions and establishing the decision maker. But there’s no set protocol that we follow.” ICU Attending

Theme No. 3: Misalignment of Clinician-Family-Patient Priorities and Treatment Goals

Participants frequently described misaligned priorities between clinicians, and patients and their families as a barrier to end-of-life care delivery. Several ICU clinicians discussed how this misalignment could be overcome by a task such as pre-meeting debriefs, which provided an opportunity to ensure all care teams had the same understanding of the patient’s current clinical status and treatment goals before meeting with family (Table 4, Quote No. 1, Quote No. 2, Quote No. 3).

TABLE 4.

Representative Quotes of Misalignment of Clinician-Family-Patient Priorities and Treatment Goals From ICU Clinicians Who Participated in End-of-Life Care Delivery Interviews, May 2022 to November 2022

Quote No.	Representative Quote
No. 1	“Being clear about prognosis, how someone’s not doing well and having a goals-of-care conversation rather than sliding into a recommendation for hospice. Not having a debrief with Palliative care before family entered the room [leads to] this disconnect.” ICU Fellow
No. 2	“It’s really important before the family meeting to make sure that everybody who’s going to be at the meeting is on the same page. Things do not go well when one person who’s part of the meeting has a different view than another person who is part of the meeting.” ICU Attending
No. 3	“A lot of these conversations are extremely open ended and if we all came together beforehand to talk about what we want out of this… that way we can communicate and convey to the patients that these are your options.” ICU Social Worker
No. 4	“It can be team distress where they’ve tried to have the conversations multiple, multiple times. They feel like maybe they’re not getting anywhere and sometimes it could just be their goals may not be in line with what the medical team’s recommendation would be.” Palliative Care Attending
No. 5	“There is a tendency in medicine for us to think of what a good death is. I often feel frustration from other team members when, ‘Oh, you didn’t get a DNR order. Oh, they’re not withdrawing care yet. How could this family do that? This is cruel.’ We could do a better job of supporting families when their decisions or goals do not align with our own.” Advanced Practice Provider
No.6	“It’s so hard, because family members and patients are all at different levels of healthcare literacy. Some patients don’t want to know if they’re dying. It’s very hard when we come from a place of knowledge, we crave science, we crave medical knowledge.” ICU Nurse

Participants identified people’s knowledge and psychosocial characteristics as another barrier to end-of-life care because there was misalignment when family and patient priorities did not coincide with clinician treatment goals (Table 4, Quote No. 4, Quote No. 5). Specifically, ICU clinicians noted the tendency in medicine to decide what a “good death is” for patients, although this definition may not be in line with the patients. Additionally, some patients may not know their diagnosis or clinical status, for cultural or literacy reasons, which caused further misalignment between clinicians, patients, and families (Table 4, Quote No. 6).

Theme No. 4: Clinician Moral Distress

Misalignment and conflict between the patient, family, and care teams led to poor outcomes perceived by participants for patients, families, and clinicians. ICU clinicians frequently described feelings of distress when providing medical care they felt was causing patient suffering (Table 5, Quote No. 1, Quote No. 2). Although the existence of hospital-based resources for clinician support, such as employee assistance programs and debriefings, was acknowledged, they were reportedly rarely used (Table 5, Quote No. 3, Quote No. 4). Most people relied on their social support system inside or outside their professional network to cope (Table 5, Quote No. 5). Participants reported a lack of staff reflection in difficult end-of-life situations (Table 5, Quote No. 6).

TABLE 5.

Representative Quotes of Clinician Moral Distress From ICU Clinicians Who Participated in End-of-Life Care Delivery Interviews, May 2022 to November 2022

Quote No.	Representative Quote
No. 1	“Her family was really pushing us to do everything. Everything is such a difficult term because what does that mean for each patient? We ended up keeping her trapped in ICU for a couple of months as she withered away and it was really hard to watch.” ICU Fellow
No. 2	“There are circumstances where there’s real moral angst, moral anxiety amongst the ICU providers, because they feel like they’re doing things that are just unkind to patients.” ICU Attending
No. 3	“Debriefs… we’re not good at it, full disclosure… Capturing it in real time with our current emotions and just getting it all out in the moment, is what we prefer, but we don’t have a good track record of doing so.” ICU Nurse
No. 4	“Everybody has employee assistance programs and we have debriefings… I’m not sure that there’s a good process in place to unpack the situations that we go through.” ICU Nurse
No. 5	“You know what, I would tell you that it’s really your friends that support each other. You know who you can go to, if that situation has stressed you out. There are some resources through the hospital, but I’m not one of those kind of people who utilizes that.” ICU Nurse
No. 6	“We try to create a culture that is mutually supportive. There’s nothing that’s structured or built into this. We try to recognize when there are things that are damaging. It’s very ad hoc.” ICU Attending

DISCUSSION

This study sought to identify which ICU work system components act as barriers or facilitators of end-of-life care delivery process metrics from the perspective of frontline ICU clinicians. The study found that work system barriers delay end-of-life care communication among clinicians and between clinicians and families, where clinician-level variability in end-of-life communication tasks exacerbated the peoples misalignment of clinician-family-patient priorities and treatment goals leading to clinician moral distress from providing nonbeneficial care.

There remain inconsistencies in end-of-life care delivery processes despite published consensus statements of process quality metrics (9, 10, 17–19, 33). Studies have demonstrated considerable variability in end-of-life symptom management, use of cardiopulmonary resuscitation before death, and decisions to forgo life-sustaining treatments between and within ICUs (9, 13, 15, 34). The findings of the current study fill a critical knowledge gap regarding which ICU work system components act as barriers or facilitators of end-of-life care delivery process metrics from the perspective of frontline ICU clinicians.

The implications of this study underscore the critical need for developing a sustainable work system that tackles barriers to end-of-life care delivery. People such as 24/7 available translators, environments such as dedicated meeting spaces, tasks such as scheduled early end-of-life communication and pre-family meetings debriefs could lead to better goal alignment among people in care teams and patient/families.

This study has several limitations that are mitigated through a robust study design. First, selection bias was present because not all specialties or levels of training involved in the care of ICU patients were captured. Snowball sampling was used to minimize bias from purposive sampling. Second, sampling bias may have been introduced because the study was limited to a large healthcare system in a single state, reducing transferability (35). However, the study deliberately selected participants from multiple different types of ICUs across three diverse types of hospitals in three different settings in a geographically diverse state (6, 14, 36). Third, participation in these qualitative interviews was voluntary, thus, introducing selection bias. However, three attempts were made to invite participants and virtual interviews were offered to minimize demands on participants. Fourth, response bias may also have been introduced as 330 frontline ICU clinicians were initially approached through purposive sampling while our final sample size consisted of 27 frontline ICU clinicians. However, we expect high “transferability’ which is when a person or group in one setting considers adopting something from another that the research has identified (27, 35). Other aspects of our study design contributed to increasing the rigor of our qualitative research, in particular, the many interdisciplinary roles, the variety of types of seven different ICUs and the variety of three hospital settings of participants who agreed to an interview. Last, the focus of this study was limited to the perceptions of frontline ICU clinicians within the ICU. This was done because frontline clinicians are most familiar with the ICU work system and have lived experience in how that translates to end-of-life care delivery processes and outcomes. Future work will incorporate the patient and family perspective on end-of-life care delivery in the ICU.

CONCLUSIONS

Work system barriers delay end-of-life care communication among clinicians and between clinicians and families, where clinician-level variability in end-of-life communication tasks exacerbated the people’s misalignment of clinician-family-patient priorities and treatment goals leading to clinician moral distress from providing nonbeneficial care. There is a need to support ICU work system components including people, environment, and tasks to facilitate end-of-life care delivery processes and improve outcomes.

KEY POINTS.

Question:

Do ICU work system components act as barriers or facilitators to end-of-life care delivery process metrics?

Findings:

In this qualitative observational cross-sectional study including 27 frontline ICU clinicians from seven different ICUs, four themes emerged: work system barriers delayed end-of-life care communication; clinician-level variability existed in end-of-life communication tasks; clinician-family-patient priorities or treatment goals were misaligned; and clinician moral distress occurred from providing nonbeneficial care.

Meaning:

Standardized work system communication tasks may improve end-of life discussion processes between clinicians and families.