Interprofessional Clinician Perspectives on LVAD Deactivation Compared to Dialysis Discontinuation

Abstract

Background

Requests for left ventricular assist device (LVAD) deactivation may pose unique emotional and moral burdens on clinicians.

Objectives

The authors aimed to explore the perspectives of interprofessional clinicians regarding LVAD deactivation across clinical settings compared to cessation of hemodialysis.

Methods

Vignette-based interviews were conducted with a sample of interprofessional clinicians at a heart and vascular center from April 14, 2023, to June 5, 2023. Likert-scale responses to paired vignettes were analyzed descriptively and with inferential statistics.

Results

Eighty clinicians agreed to be interviewed for this study (27.5% physicians, 23.8% nurses, 17.5% physician trainees, 16.3% advanced practice providers, and 15% social workers, bioethicists, and chaplains). Clinicians uniformly felt that vignettes depicting LVAD deactivation were more ethically complex and less likely to be honored than dialysis discontinuation across clinical scenarios (P < 0.001). Clinicians were more likely to rate scenarios that described the patient without other medical comorbidities (“tired of life”) for either the patient with a LVAD or a patient on dialysis as ethically complex and were less likely to suggest honoring the patient's request (P < 0.001). These trends remained consistent across all demographic and clinician disciplines.

Conclusions

When compared to requests to discontinue hemodialysis, clinicians perceive LVAD deactivation as more ethically complex and were less likely to honor the patient's request. These differences persisted across clinical scenarios, from acutely ill to chronically ill to independent.

Central Illustration

The left ventricular assist device (LVAD) is a surgically implanted device that augments cardiac output in patients with advanced heart failure. The LVAD is unique in that it is currently the only durable mechanical device that can assist the function of a native organ, meaning the patient can live independently with the device in situ. With the ability to extend the duration and improve the quality of life for advanced heart failure patients, over 80% of LVADs implanted are now considered “destination therapy.”^1,2

As people living with an LVAD age and experience noncardiac-related illnesses (eg, malignancy), clinicians will encounter situations where their patients request LVAD deactivation. While most LVAD deactivation occurs in the setting of multiorgan failure in the intensive care unit (ICU), there are other clinical contexts where patients may request deactivation.3, 4, 5 LVAD deactivation is generally thought to be no different than withdrawal of other forms of life-sustaining therapies, ethically permitted as refusing medical therapy and premised on the principles of bodily autonomy and informed consent (or refusal).⁶ Regardless, LVAD deactivation continues to cause great distress in clinicians and caregivers, with some episodes of deactivation characterized as suicide.^7,8 Perhaps this is because LVAD deactivation is unique in some ways: the LVAD is mostly internal, acting as a replacement therapy, and LVAD deactivation reliably leads to death within minutes to hours.^3,5,9,10 These features lead to a varied clinician experience and perception of LVAD deactivation.¹¹

We sought to understand how perspectives of an interprofessional clinical team that cares for patients with LVADs might differ across life-sustaining therapies and clinical contexts.

Methods

This study was approved by the MedStar Health Research Institute Institutional Review Board (Study 2,569; 7/10/2023). Semistructured vignette-based interviews were conducted one-on-one with a convenience sample of an interprofessional cohort of clinicians between April 14, 2023, and June 5, 2023.

Vignette development

Compared to querying clinicians directly, a vignette-based methodology is considered less personal and therefore allows respondents to safely share their intuitions and perspectives on controversial or sensitive topics.¹² To create our 6 vignettes (Supplemental Table 1), we varied the prognosis and the type of life-sustaining therapy while keeping gender, patient request, and other details consistent. While distinct from the LVAD, we selected dialysis as a comparison to LVAD deactivation because it is the only other long-term organ replacement therapy that can be provided outside of the hospital. Dialysis and cessation of dialysis are also broadly familiar to clinicians. Vignettes describing a patient with a prognosis of hours to days (in the ICU) were presented first because these were expected to be least controversial. The next pair of vignettes described a patient with a prognosis of weeks to months (cancer), followed by a pair of vignettes depicting a patient with a possible prognosis of years who reports being “ready to die.” Before being used, vignettes were pilot tested on clinicians and students outside of the heart and vascular center. Vignettes underwent an iterative editing process for clarity.

Conducting interviews

Approximately 150 interprofessional clinicians in a quaternary heart and vascular center in Washington, DC, were invited to participate by email and approached during their clinical shifts. Clinicians were able to register for virtual or in-person interviews depending on availability. Interviews were conducted by a trained member of our study team who was not known to the clinical staff.

After reading each vignette, the interviewer asked the clinician to rate the following 2 questions on a 5-point Likert scale, “How likely are you to suggest we honor this person's request?” and “How ethically complex did you find this case?” The interview tool can be found in the Supplemental Figure 1. Demographic data were collected at the end of each interview.

Statistical analysis

Here, we report quantitative Likert-scale data of clinician responses. The qualitative results are analyzed and reported separately. Demographic data for clinicians and Likert-scale data of clinician response are reported using counts and percentages. Likert-scale data are reported using a five-point scale including clinician responses of “Not at All,” “Hardly,” “Somewhat,” “Very,” and “Extremely.” The categorical data are transformed into numerical responses of 1 to 5 such that a Wilcoxon signed-rank test could be used to evaluate a difference in response to distributions between paired vignettes. The Spearman correlation coefficient (ρ) was calculated to quantify the association between ordinal respondent demographics and Likert-scale responses to vignettes. Summary and inferential statistics were performed with RStudio (R Core Team [2022]).

Results

Eighty clinicians agreed to be interviewed, representing a wide range of backgrounds (65% female; 56.3% White, 21.3% Asian, and 13.8% Black) and disciplines (27.5% attending physicians, 23.8% nurses, 17.5% physicians-in-training, 16.3% advanced practice providers [APPs], and 15% of the cohort were social workers, bioethicists, and chaplains) (Table 1). While our cohort of clinicians was evenly distributed with regard to considering themselves religious, 63.8% answered that their religious tradition influenced “not at all” or “hardly” how they evaluated the clinical vignettes. Of clinicians, 66.3% were <40 years of age and reported being in practice for <10 years.

Table 1.

Clinician Demographic Data (N = 80)

Sex
Male	28 (35%)
Female	52 (65%)
Age
<30 y	13 (16.3%)
30-39 y	40 (50%)
40-49 y	14 (17.5%)
50-59 y	9 (11.3%)
>60 y	4 (5%)
Race/ethnicity
Black	11 (13.8%)
White	45 (56.3%)
Asian	17 (21.3%)
Identifies with more than 1 racial/ethnic backgrounda	5 (6.3%)
Prefer not to answer	2 (2.5%)
How religious are you?
Not at all	16 (20%)
Hardly	15 (18.8%)
Somewhat	29 (36.3%)
Very	17 (21.3%)
Extremely	2 (2.5%)
Prefer not to answer	1 (1.3%)
How much does religious tradition influence your answer?
Not at all	36 (45%)
Hardly	15 (18.8%)
Somewhat	14 (17.5%)
Very	7 (8.8%)
Extremely	5 (6.3%)
Prefer not to answer	3 (3.8%)
Discipline
Attending physician	22 (27.5%)
Nurse	19 (23.8%)
Physician trainee	14 (17.5%)
Advanced practice provider	13 (16.3%)
Otherb	12 (15%)
Years in practice
<5	29 (36.3%)
5-10	24 (30%)
11-20	18 (22.5%)
21-30	6 (7.5%)
31-40	3 (3.8%)

Values are n (%).

^aClinicians who identified with more than 1 racial/ethnic background included 3 participants who identified as White and Asian, 1 who identified as White and American Indian/Alaska Native, and 1 who identified as Caribbean.

^bIncluded social workers, bioethicists, and chaplains.

Figure 1 demonstrates how often clinicians reported being involved in the care of patients with LVADs and how often they were involved in LVAD deactivation. Of clinicians, 82.5% reported being either “sometimes,” “often,” or “always” involved in the care for patients with an LVAD. Sixty percent reported having been involved in either “some” or “many” LVAD deactivations.

Figure 1.

Care and Deactivation of an LVAD

How often clinicians report being involved in the care of patients with an LVAD and how often they have been involved in LVAD deactivation. LVAD = left ventricular assist device.

Figure 2 and 3 depict the Likert-scale responses for each vignette and P value calculations of vignette pairs using the Wilcoxon signed-rank test. In Figure 2 and Central Illustration, clinicians were more likely to suggest honoring the request for dialysis discontinuation than for LVAD deactivation in all paired scenario comparisons (P < 0.001 for ICU, cancer, and “tired of life” scenarios). Clinicians were just as likely to suggest honoring the patient's request for deactivation in the ICU and cancer scenarios for both dialysis (P = 0.70) and LVAD (P = 0.062). However, clinicians were more likely to suggest honoring the patient's request for deactivation in the cancer scenario when compared to the “tired of life” scenario regardless of the type of life-prolonging therapy (P < 0.001 for dialysis and LVAD).

Figure 2.

Honoring the Patient's Request

“How likely are you to suggest we honor this person’s request?” Likert scale responses and P value calculations of vignette pairs using the Wilcoxon signed-rank test. ICU = intensive care unit; LVAD = left ventricular assist device.

Figure 3.

Ethical Complexity of the Case

“How ethically complex did you find this case?” Likert scale responses and P value calculations of vignette pairs using the Wilcoxon signed-rank test. ICU = intensive care unit; LVAD = left ventricular assist device.

Central Illustration.

How Likely Are Clinicians to Honor a Patient's Request for Deactivation of Life-Sustaining Therapy? Comparing the LVAD to Dialysis Across Clinical Contexts

ICU = intensive care unit; LVAD = left ventricular assist device.

When comparing vignettes of dialysis discontinuation to LVAD deactivation across the various prognostic scenarios, clinicians uniformly found the LVAD deactivation scenarios to be more ethically complex (Figure 3) than dialysis discontinuation (P < 0.001 for ICU, cancer, and “tired of life” scenarios). No significant difference was present when comparing the ethical complexity of the cancer scenario vs the ICU scenario for either dialysis (P = 0.71) or LVAD discontinuation (P = 0.06). However, when comparing discontinuation across the prognostic scenarios, clinicians reported that the “tired of life” scenario was more ethically complex than the cancer scenario, regardless of life-prolonging therapy (P < 0.001 for dialysis and LVAD).

To determine how respondent demographic variables might influence clinicians' responses to vignettes, Spearman coefficients are reported in Supplemental Table 2. Increased religious adherence positively correlated with a perception of increased ethical complexity in the “tired of life” scenarios (P < 0.05). In the ICU and cancer scenarios, prior LVAD deactivation exposure was positively correlated with being more likely to suggest honoring the patient's request for dialysis discontinuation (P < 0.05) but not with LVAD deactivation. Prior LVAD deactivation exposure was also positively correlated with a perception of increased ethical complexity of requests for dialysis discontinuation in the ICU (P < 0.05). Otherwise, as seen in Supplemental Table 2, the age, number of years in practice, religious identity, or prior LVAD exposure was not correlated with the suggestion to honor the patient's request or with a perception of increased ethical complexity.

Discussion

Main findings

In this vignette-based study, we asked 80 clinicians representing a range of training disciplines, years in practice, and comfort with LVAD technology to rate ethical complexity and willingness to honor a patient’s request to deactivate life-prolonging technology. This cohort was more likely to rate LVAD deactivation as ethically complex and less likely to recommend honoring the patient’s request when compared to discontinuation of hemodialysis. This difference persisted across clinical situations, from acutely dying to chronically ill and independent.

Additionally, clinicians rated LVAD deactivation differently depending on the clinical acuity so that the more independent or “well” seeming the patient, the more likely the clinician was to report the patient’s request for deactivation as ethically complex (and the less likely to suggest honoring that request). This concept was echoed in a vignette-based interview of internal medicine physicians published in 2006. Respondents were more likely to withhold or withdraw life-sustaining treatments (ventilator, dialysis, antibiotics, or artificial nutrition/hydration) in patients who were terminally ill or incapacitated compared to those who were not terminally ill or were alert.¹³ It is noteworthy that our findings were consistent across a cohort of clinicians with a diversity of race, age, years in practice, discipline, religious adherence, and how often they provide care to patients with an LVAD.

What this study adds

Our novel study adds to prior work exploring clinician perspectives on LVAD deactivation. In a survey of cardiologists, most of whom care for patients with an LVAD, only 26% of cardiologists felt comfortable ordering an LVAD deactivation for a dying patient.^14,15 One in 3 cardiologists felt that there was a substantial difference between LVAD deactivation and withdrawal of life-sustaining therapy, and 13% regarded LVAD deactivation as euthanasia even for patients nearing death (27% in a subset of European cardiologists). Furthermore, in 1 qualitative study, caregivers and clinicians described the process of LVAD deactivation as more emotionally fraught than other procedures of withdrawing life-sustaining therapies.¹⁶

This growing body of evidence suggests that while western medical ethics and law make no distinction between clinical contexts and between different forms of life-sustaining therapies, clinicians report different experiences based on those 2 crucial variables. We believe this has important implications for practice: how clinicians discuss medical technologies with their patients will be colored by their experiences with those technologies, especially when a patient asks for a device to be deactivated at the end of life. While this phenomenon has been explored previously, we believe the LVAD carries added moral complexity.¹³

Strengths, weaknesses, and limitations

While our study did capture a cross-section of the interprofessional clinical team, our convenience sampling approach may have introduced a sampling bias into our results. Additionally, sampling clinicians from a single hospital center may limit transferability and reproducibility beyond this facility, region, or country. Sixty percent of respondents reported at least “some” previous direct involvement with LVAD deactivation; sampling clinicians from a high-volume LVAD implantation center may have yielded results that are not generalizable to smaller LVAD centers, where clinicians likely have even less personal experience to draw upon. Also notable was, how we included trainees and clinicians with less clinical experience. While these clinicians do not have as much experience with advanced therapies and technologies, they may be more familiar with the latest in cultural trends, morality, and ethics than their attendings who have been in practice longer. Since these clinicians are often at the front lines of patient care, we included them in our cohort.

It is also possible that clinicians' responses to vignette-based prompts do not translate directly to clinical practice given the complexities that arise in real-world patient cases. Clinicians tend to exchange views and seek out support in challenging situations, which may influence moral distress and clinical decision-making but is not reflected in our study of individual clinicians. Furthermore, vignette-based interviews may introduce biases due to differences in interpretation compared to asking clinicians directly whether they would honor a request to deactivate an LVAD without a clinical scenario to anchor on, as done by McIlvennan et al.¹⁵ However, using vignettes allowed us to vary the scenario on 2 axes: prognosis and type of life-prolonging therapy to test how respondents weighed the influence of these variables on their responses, which would not have been feasible with a direct query.

Also of note, we made no adjustment for a false discovery rate in our statistical analysis, so results should be interpreted with caution.

Finally, we believe there is important wisdom to be learned in the intuitions of front-line clinicians. We hope that the qualitative portion of our study will further this research into the moral experience of LVAD deactivation. As the LVAD is only the latest of our technologies to mimic the function of an internal organ, an updated phenomenological account of life-sustaining therapies is crucial.

Perspectives.

COMPETENCY IN INTERPERSONAL AND COMMUNICATION SKILLS: We used a novel vignette-based methodology to explore clinician intuitions regarding LVAD deactivation compared to discontinuation of hemodialysis. Clinicians were more likely to rate LVAD deactivation as ethically complex and less likely to recommend honoring the patient's request, a difference that persisted across clinical situations.

TRANSLATIONAL OUTLOOK: When compared to another form of life-sustaining medical treatment, our study demonstrates that requests to deactivate an LVAD may pose a greater moral challenge to clinicians.

Funding support and author disclosures

The authors have reported that they have no relationships relevant to the contents of this paper to disclose.