Prevalence and Predictors of Burnout Among Hospice and Palliative Care Clinicians in the U.S

Abstract

Context

Many clinical disciplines report high rates of burnout, which lead to low quality of care. Palliative care clinicians routinely manage patients with significant suffering, aiming to improve quality of life. As a major role of palliative care clinicians involves educating patients and caregivers regarding identifying priorities and balancing stress, we wondered how clinician self-management of burnout matches against the emotionally exhaustive nature of the work.

Objectives

We sought to understand the prevalence and predictors of burnout using a discipline-wide survey.

Methods

We asked American Academy of Hospice and Palliative Medicine clinician members to complete an electronic survey querying demographic factors, job responsibilities, and the Maslach Burnout Inventory. We performed univariate and multivariate regression analyses to identify predictors of high rates of burnout.

Results

We received 1357 responses (response rate 30%). Overall, we observed a burnout rate of 62%, with higher rates reported by nonphysician clinicians. Most burnout stemmed from emotional exhaustion, with depersonalization comprising a minor portion. Factors associated with higher rates of burnout include working in smaller organizations, working longer hours, being younger than 50 years, and working weekends. We did not observe different rates between palliative care clinicians and hospice clinicians. Higher rated self-management activities to mitigate burnout include participating in interpersonal relationships and taking vacations.

Conclusions

Burnout is a major issue facing the palliative care clinician workforce. Strategies at the discipline-wide and individual levels are needed to sustain the delivery of responsive, available, high-quality palliative care for all patients with serious illness.

Introduction

Many clinicians practice palliative care to create a meaningful difference in the lives of persons with serious illness and their caregivers. Along with the sense of fulfillment garnered by improving the quality of life of those with need is a simultaneous awareness that clinical hospice and palliative care practice may be more inherently stressful and challenging than previously appreciated. Challenges, such as managing a clinical program that is in high demand but under-staffed,^1,2 addressing the multiple and complex needs of patients and caregivers often in crisis, and facilitating resolution of conflicts between any combination of patients, caregivers, and the health system, may lead to significant stress.

Contrastingly, a high degree of meaning and personal satisfaction felt by palliative care clinicians may obviate this job-related stress, not allowing it to evolve into a more distressing state such as feelings of burnout. Potentially, the naturally rewarding work of palliative care clinicians, alongside the natural selection of those who comprise the field (i.e., those who stay in the field must inherently know how to resolve the emotional challenges), may prevent escalation of normal stress to something more. We do know that in other medical fields, job stress does frequently evolve into overt burnout, which is described as loss of enthusiasm for work (emotional exhaustion [EE]), feelings of cynicism (depersonalization [DP]), and a low sense of personal accomplishment.³ This phenomenon is increasingly studied in health care, both to describe the penetration of this issue across all fields of medicine, while also informing interventions to allay the downstream, negative consequences of unmanaged burnout.⁴

Shanafelt et al. recently described the comparative results of burnout in various physician populations.⁵ In addition to reporting an overall rate of 46% of all physicians, they noted the highest rates in physicians who must address multiple areas of medical need simultaneously (e.g., family medicine, internal medicine) or those that are asked to focus on a specific condition but often in high-acuity situations (e.g., emergency medicine, obstetrics). Furthermore, others have described higher rates of burnout among critical care physicians,⁶ where high-acuity situations are the norm, and outcomes not always successful.

Palliative care shares many characteristics with these other medical disciplines that experience high rates of burnout, including balancing distress across multiple diseases or domains, while frequently being called to high-acuity or rapidly evolving situations. Previous evaluations of burnout in palliative care have either described burnout in qualitative ways or using small samples sizes^7,8 or were conducted in a different era of clinical palliative care.^9,10 To complement these efforts, we planned a large, nationwide survey to evaluate the prevalence, severity, and predictors of burnout in clinical palliative care professionals.

In partnership with the American Academy of Hospice and Palliative Medicine (AAHPM), one of the largest palliative care membership organizations in the U.S., we conducted a membership-wide survey to understand the prevalence and correlates of burnout among clinicians. The purpose of this research was to ultimately inform the development of future interventions to reduce the burnout experienced by the dedicated palliative care professionals who care for at least six million persons and their caregivers every year.¹¹

Methods

We conducted an electronic survey of demographics and burnout among specialty hospice and palliative care clinicians in the U.S. Participation was voluntary, and no remuneration was offered. The study was reviewed and approved as exempt by the Duke University institutional review board (Pro00045381). The survey concept was approved by the AAHPM Board of Directors, but members did not have influence over the content of the survey itself.

Participants

We invited all members of the AAHPM with available electronic mail addresses to participate in the survey (N = 4456). Nonclinician respondents were excluded from the analysis. AAHPM provided a roster of member e-mail addresses who were active members as of June 1, 2013.

Survey Procedures

The electronic survey was conducted over six months using an initial electronic invitation letter and two follow-up letters. Additionally, invitations via Facebook posts, Twitter messages, blog posts, and electronic newsletters were sent throughout the survey period. Other participating social media outlets used to disseminate invitations to the study included the Palliative Care Network, National Hospice and Palliative Care Organization, PalliMed, GeriPal, and the Society of General Internal Medicine End-of-Life Interest Group.

Survey Development and Validation

Our palliative care–focused burnout survey was modeled after similar surveys conducted by the American Society of Clinical Oncology and the American College of Surgeons.^12,13 The survey included 52 questions. Eight questions involved clinician demographics or practice setting, four involved career choice and work/life balance, three questions from the SF-12¹⁴ queried quality of life, nine addressed clinical experience and credentials, five involved job characteristics, and one queried self-care. To measure burnout, we used the Maslach Burnout Inventory Human Services Survey³ (license/permission was obtained from Mind Garden, Inc., Menlo Park, CA). This 22-question survey (MBI-22) has been used frequently in other studies examining burnout in health care workers, including physicians and nurses.

The MBI-22 uses a seven-point Likert scale to score the frequency of feeling each of 22 components used to measure burnout. Thus, for each of the 22 questions, a score of 0 through 7 is recorded. For example, one question within the MBI-22 Human Services Survey is, “I feel emotionally drained from my work.” Respondents can answer a frequency of feeling this way, from “never” (scored as 0) to “every day” (scored as 7). Thus, the minimum score possible is 0 and the maximum is 154. Our survey did not explain the scoring system within the text of the questions.

To assess for each subdomain of burnout (e.g., depersonalization and emotional exhaustion), we used the scoring and cutoff criteria of the MBI-22. To calculate a score, each domain had a selected set of questions on which to focus; additive scores from these questions represented a total subdomain score. Depersonalization was scored as “high” for a total score of 13 or higher across five questions; seven or higher constituted “moderate.” Emotional exhaustion was scores as “high” for a total score of 27 or higher across nine questions; 17 or higher constituted “moderate.”

The electronic survey was twice reviewed for face and content validity by six palliative care physicians, one expert in health care–associated burnout, and one data analyst. After the first review, iterative updates were made to reflect suggested changes and improvements.

Statistical Analysis

Descriptive statistics were used to investigate the demographics, job characteristics, and burnout symptoms of palliative care clinicians. We investigated the pattern of data missingness and found that it was random and not informative. We imputed missing responses from the MBI-22 questionnaire for those who left more than 33% of the MBI-22 questionnaire blank, or participants who had less than 10 answered responses (note that the EE and DP subscales used 14 of the 22 MBI-22 questions) were removed from the analyses.

We modeled burnout by dichotomizing our outcome variable to high EE or DP. We used Spearman correlation to check for correlation between high EE or DP, EE, DP, and the MBI-22 scales.

We implemented a multivariable logistic regression model to predict risk of burnout (high DP or EE). We used variables from the literature and those hypothesized to ameliorate burnout including marital status, children living at home, numbers of colleagues in practice (three or less vs. four or more), physicians versus nonphysicians, and hours per week worked (>50 hours vs. <50 hours), age (<50 vs. >50 years). For the purposes of this analysis, we use the term “non-physician clinicians” to designate non-MD/DO clinicians such as registered nurses, advanced practice professionals, social workers, chaplains, and the like.

To identify a concise set of variables for modeling variables associated with burnout, we used a bootstrap multivariable stepwise logistic regression technique to select and validate the significant variables. We accounted for random sampling and multiple testing by creating 1000 new random samples (based on our original data set), then selected the variables that most frequently were selected to be significant across the 1000 bootstrapped samples. All analyses were performed using either R version 2.15.1 (graphics) or SAS version 9.4 (SAS Institute, Inc., Cary, NC).

Results

We received surveys from 1357 specialty hospice and palliative care clinicians, yielding a response rate of 30%. After imputation, we had 1056 complete MBI questionnaires.

Demographics

Demographic characteristics of responding clinicians are summarized in Table 1. A majority of the respondents were over age 50 (57%), female (65%), married or partnered (82%), did not have children at home (55%), and were physicians (68%). We found that most clinicians worked in palliative care for less than 10 years (67%), planned to stay for at least 10 or more years (53%), dedicated more than 75% of their weekly effort toward clinical responsibilities (42%), and worked exclusively in clinical palliative care (31%). We investigated job characteristics for clinicians and found that a majority worked over 40 hours per week (59%) and had at least four colleagues (57%).

Table 1.

Demographic Characteristics

Variable [Missing]	Total (N = 1357), n (%)
Age, yrs [341]
20–30	16 (1.6)
31–40	195 (19.2)
41–50	225 (22.1)
51–60	387 (38.1)
61–70	176 (17.3)
>71	17 (1.7)
Gender [352]
Male	355 (35.3)
Female	650 (64.7)
Marital status [343]
Married	780 (76.9)
Single, never married	67 (6.6)
Partnered	52 (5.1)
Divorced	93 (9.2)
Widowed	22 (2.2)
Children living in the home [346]
Yes	459 (45.4)
No	552 (54.6)
Highest clinical degree [337]
MD/DO	691 (67.7)
NP/PA	113 (11.1)
MSN/CNS	34 (3.3)
RN	66 (6.5)
MSW	36 (3.5)
Chaplain (MDiv, or equivalent)	21 (2.1)
Other	59 (5.8)
How long have you worked in hospice or palliative care, yrs [319]
0–5	354 (34.1)
6–10	340 (32.8)
11–20	230 (22.2)
21–30	83 (8.0)
31–40	31 (3.0)
Clinical responsibility as component of total effort [329]
≥75%	430 (41.8)
<75%	598 (58.2)
Clinical settings of care [319]
Hospice Only	228 (22.0)
Palliative care only	319 (30.7)
Mixed	491 (47.3)
Average hours worked per week [319]
0–20	143 (13.8)
21–40	278 (26.8)
41–60	507 (48.8)
61 or greater	110 (10.6)
Number of colleague clinicians in practice [318]
0	86 (8.3)
1–3	361 (34.7)
4–10	343 (33.0)
≥10	249 (24.0)
Chance of leaving palliative care at 10 yrs [336]
0%–33%	507 (49.7)
34%–66%	161 (15.8)
67%–100%	353 (34.6)

Burnout Prevalence and Severity

The characteristics of responding clinicians’ burnout prevalence, career satisfaction, and feelings about work environment are summarized in Table 2. We found that 60% of palliative care clinicians had high EE, and 24% had high DP. We found that 62% of clinicians experienced at least one symptom of burnout based on reporting high EE or high DP. When comparing among clinician types, we found that the burnout rate (either having high EE or high DP) was significantly higher in nonphysician clinicians (66% vs. 60%; P < 0.001).

Table 2.

Prevalence of Burnout

Variable	Total (N = 1056), n (%)
Emotional exhaustion
Low	113 (10.7)
Moderate	315 (29.8)
Higha	628 (59.5)
Depersonalization
Low	260 (24.6)
Moderate	544 (51.5)
Higha	252 (23.9)
Burnout
High EE or high DPb	657 (62.2)
My work schedule leaves me enough time for my personal/family life
Missing	1
Strongly disagree	132 (12.5)
Disagree	306 (29.0)
Neutral	178 (16.9)
Agree	329 (31.2)
Strongly agree	110 (10.4)
How much of the time during the past four weeks …
Felt calm and peaceful
Missing	2
All of the time	21 (2.0)
Most of the time	331 (31.4)
Good bit of the time	274 (26.0)
Some of the time	258 (24.5)
A little of the time	155 (14.7)
None of the time	15 (1.4)
Did you have a lot of energy
Missing	3
All of the time	32 (3.0)
Most of the time	289 (27.4)
Good bit of the time	239 (22.7)
Some of the time	278 (26.4)
A little of the time	170 (16.1)
None of the time	45 (4.3)
Have you felt downhearted and blue
Missing	2
All of the time	9 (0.9)
Most of the time	28 (2.7)
Good bit of the time	79 (7.5)
Some of the time	239 (22.7)
A little of the time	510 (48.4)
None of the time	189 (17.9)

EE = emotional exhaustion; DP = depersonalization.

^aAs assessed using the full Maslach Burnout Inventory. Per standard scoring of the Maslach Burnout Inventory for health care workers, clinicians with scores on the emotional exhaustion subscale of at least 27, scores on the depersonalization subscale of at least 13 are considered to have a high degree of burnout in that dimension.

^bHigh EE or High DP is related to having a high risk of burnout (see Methods).

Regarding work-life balance, less than half of the clinicians reported that their work schedule left them enough time for personal and family life (42%). At work, 33% of the clinicians reported being calm and peaceful most of the time, 30% reported they had a lot of energy most of the time, and 4% experienced being downhearted and blue most of the time.

Finally, we performed a multivariable logistic regression analysis to see the relationship between burnout and the associated variables found during the bootstrap stepwise variable selection method (Table 3). We found that younger nonphysicians who worked more than 50 hours per week, worked with fewer colleagues, and worked weekends often were at the greatest risk for burnout.

Table 3.

Multivariate Logistic Regression of Factors Associated with Burnout

Odds Ratio Estimates
Effect	Point Estimate	95% Wald Confidence Limits
≥50 hours per week versus <50 hours per week	1.672	1.245	2.246
Work with three or less colleagues versus four or more	1.657	1.253	2.192
<50 years old versus ≥50 years old	1.914	1.454	2.521
Other versus MD/DO	1.959	1.439	2.665
Work weekends often or all the time versus other	1.648	1.240	2.188

Personal Management of Burnout

We queried the types of management strategies reported by clinicians and the importance of each. Strategies rated the most important were as follows: “talking with family, friends, or significant others for support,” “participating in recreation/hobbies/exercise,” and “taking vacations.” All three were considered “essential” by at least 50% of all respondents. Reading and “nurturing spiritual/religious aspects of myself” also were rated highly. Journaling and reflective writing alongside regular meetings with counselors or therapists were reported as the least used and least important among the options provided.

Comparative Analysis

We compared the respondents of our survey to the AAHPM member roster from October 2013. We compared two overlapping demographic variables between the two years in palliative care practice and age. AAHPM data on years in practice were incomplete, with data only available on 1821 of 4608 members (40%). Regarding years in practice, data were 28% complete (1306/4608). Compared to the available AAHPM data, the survey population was significantly different, with more survey respondents having practiced 0–5 or 6–10 years versus the AAHPM roster (P < 0.01). Furthermore, we observed significant differences in ages, with more survey respondents being over the age of 51 years than would be expected by the AAHPM roster (P < 0.01).

Discussion

Palliative care clinicians report a burnout rate of 62%. This is one of the highest published rates reported when compared to national samples of physicians from other specialties.⁵ Burnout is more likely reported by clinicians who are younger, not physicians, those who work with fewer colleagues, work longer hours, and work on weekends.

These results highlight some important conclusions for the field. First, we observe a clear relationship between clinician age (and not experience) and risk for experiencing burnout. This may be caused by several reasons, including selection bias of those who remain in clinical medicine past their junior years (those who experience significant burnout have already left) or a commonly reported phenomenon of higher stress and lower job satisfaction among clinicians¹⁵ and professionals in the early stages of a career.¹⁶ Second, our results demonstrate different rates of burnout among clinician types. The combined group of advanced practice professionals, nurses, social workers, and chaplains reported higher rates of burnout than their physician colleagues, even when accounting for differences in age and gender among the groups. Third, of the factors associated with higher rates of burnout, some are clinician-related and thus not modifiable by the organizations in which they work, but some potentially are under the control of palliative care organizations. For example, local palliative care organization management can carefully and strategically manage the number of hours worked per week and after-hours responsibilities of each of its clinicians. This is balanced with the progressively increasing workload of clinical groups, who have to balance asking for clinical business from referring colleagues with providing available services when those requests are made.

Our results also emphasize the need for local and discipline-wide strategies to manage clinician burnout. Interventions developed, piloted, and implemented within our discipline are few but evolving. Recent data from semistructured interviews of palliative care clinicians reported system, patient, and personal factors and challenges related to work stressors. Furthermore, this study demonstrated that palliative care clinicians are interested in further training to manage burnout, including mind-body skills, education, and cognitive strategies.⁸ Colleagues also have reported self-management strategies important to palliative care clinicians to address burnout, including focusing on physical well-being, personal boundaries, taking time away, focusing on professional relationships, and taking a transcendental perspective.⁷ Additionally, we must explore resilience strategies that have worked in other disciplines, paying particular attention to factors unique to our field such as size and training background, including career fit/meaning, personal strategies, physician engagement groups, structural changes in practice, and others.^17–19

Ultimately, prevention and management of clinician burnout will help our discipline provide sustainable, responsive, and available palliative care for all patients across the serious illness continuum, and in all geographic locations of care. Fulfilling this promise is critical, as the number of patients eligible for services is growing at a historic pace. Current projections estimate the number of persons diagnosed with cancer in the U.S. growing by 45% during the next 20 years.²⁰ For those with chronic illness, estimates put growth at an additional 20 million persons between 2015 and 2030, totaling the prevalence of those living with chronic illness to around 170 million Americans.²¹ When matching these findings with the sobering data from the current palliative care clinician workforce shortage,¹ we are faced with a looming crisis. Solutions to address this crisis must simultaneously focus on supporting the current clinicians in the field through management of burnout and job stress, building the skills of our primary palliative care partners,²² and exploring alternate strategies to introduce new clinicians to the specialty.

Regarding future directions, there are several follow-up investigations related to the relationship between job responsibilities, job stress, and mismatched regulatory and reimbursement policies that are needed. No doubt, the landscape of health care delivery has changed significantly, with increasing demands placed on clinicians related to medical documentation, reporting data on quality, and omnipresent threats to clinician reimbursement. Whether these challenges faced by our colleagues in other fields is similar to our own, and thus shared drivers of burnout, is unknown. In many ways, palliative care is similar to other fields. We draw from the same clinician workforce pool, are subject to the changes inherent to reimbursement reform, and are increasingly in demand as the population with chronic and serious illness grows. But from a policy perspective, we are also unique. For example, we are champions of multidisciplinary care, yet often have several team members who provide care “in kind,” mismatching care provided with the necessary reimbursement to support validation and sustainability of those services. We also focus on the needs of caregivers, yet suffer from a lack of policies and reimbursement that explicitly support this mission. Burnout is a global challenge across all of health care, which highlights likely systems issues that require policy-level changes. Which of these issues should be the focus of our advocacy efforts needs to be defined.

This study has several limitations. First is the low response rate. We used a crowdsourcing technique to maximize the number of palliative care clinicians who had access to the survey, as has been described by others.²³ Implications of the low response rate on the findings relate to the nonresponse bias. We could imagine that the 70% of the field who were nonresponders may not experience burnout and thus had little motivation to participate. If true, the true burnout prevalence rate for the entire field may be lower than our findings. Additionally, we are limited by our inability to determine causality or direction of effect for the associations observed. Furthermore, we did find that our respondents differed significantly from the AAHPM membership roster, limiting the generalizability of the findings to the entire membership. Furthermore, our nonphysician sample size was small, limiting our ability to definitely characterize this population. Despite these challenges, our study stands as one of the largest U.S.-based research surveys of palliative care clinicians. The survey respondents and the conclusions drawn from the data largely focus on the experience and characteristics of physician clinicians in palliative care. Despite the discrepancy in the sample sizes of respondents, we were able to identify important differences in burnout rates between the clinical groups.

Conclusions

Burnout is a common and pervasive phenomenon. The high prevalence highlights an urgent need for strategies to mitigate its effects and sustain the work-force. Such solutions would address modifiable organization-level characteristics associated with higher burnout, such as team size and hours worked. Furthermore, they would build and practice resilience skills, instead of focusing on “being tougher.” We must reach a delicate balance between calls for expanding the role of palliative care within serious illness care and ensuring a workforce that does not overpromise and underdeliver responsive palliative care.