Differences in Physician Burnout by Specialty: A Cross-Sectional Survey

Abstract

Background

The prevalence of physician burnout increased notably during the COVID-19 pandemic, but whether measures of burnout differed based on physician specialty is unknown. The authors sought to determine the prevalence of burnout, worklife conflict, and intention to quit among physicians from different specialties.

Methods

This is a cross-sectional online survey of physicians working at 2 urban hospitals in Vancouver, Canada, from August to October 2021. Responses were categorized by specialty (including surgical and nonsurgical), and data about whether physicians provided frontline patient care during COVID-19 were also included. Physician burnout was measured using the Maslach Burnout Inventory.

Results

The survey response rate was 42% (209/498). The overall prevalence of burnout was 69%. Burnout was not significantly different by specialty or between frontline COVID-19 specialties compared with other specialties. Physicians in surgical specialties were more likely to report work–life conflict than those in nonsurgical specialties (p = 0.012). Differences in intention to quit among specialties were not statistically significant.

Conclusion

During the COVID-19 pandemic, physician burnout was high across physicians, without significant differences between specialties, highlighting the need to support all physicians.

Introduction

Burnout in physicians is increasing worldwide and can result in inferior quality of care provided to patients.¹ Maslach and Jackson defined burnout as a psychological syndrome that includes emotional exhaustion, depersonalization, and a sense of low personal accomplishment.² The COVID-19 pandemic has been associated with increased prevalence of burnout in physicians.³ However, several studies have suggested that not all physician groups experience burnout to a similar degree. In a recently published study, Myran et al compared visits for mental health and substance use care by physician specialty during the first year of the COVID-19 pandemic to the year before in the province of Ontario, Canada.⁴ Although a remarkable increase in visits was reported across all included specialties (critical care/emergency, internal medicine, family medicine, pediatrics, surgical specialties, and anesthesia), these rates differed widely between specialties.⁴ Reported increases in visits ranged between 20% and 30% for critical care and emergency, psychiatry, surgery, and family medicine and were > 30% for internal medicine and pediatrics. Physicians in the specialty of anesthesia experienced the highest increase (74%) in mental health care visits.⁴ In another recent study from the US, Melnikow et al conducted a national survey to assess physician burnout among frontline specialties (family physicians, internists, critical care physicians, hospitalists, emergency physicians, and infectious diseases physicians) at 2 time points during the first year of the COVID-19 pandemic.⁵ From May/June 2020 to December 2020/January 2021, increased burnout was reported in all frontline specialties, with hospitalists and primary care respondents experiencing the highest increase.⁵

Although increased prevalence of burnout during the COVID-19 pandemic has been reported across the world, there are limited data on differences in prevalence of burnout among a broad range of specialties. In the present study, the authors surveyed physicians from a wide range of specialties, including anesthesia, emergency medicine, internal medicine, obstetrics and gynecology, orthopedics, psychiatry, radiology, general surgery, laboratory medicine, and other surgical specialties at 2 urban hospitals in Vancouver, Canada, for burnout. The authors also compared the components of burnout, perceived work–life conflict, and intention to quit or if they had quit among these groups.

Methods

A cross-sectional online survey was administered to physicians in all hospital departments at 2 teaching hospitals in Vancouver, Canada, between August and October 2021. The study was approved by the Providence Health Research Ethics Board and reporting followed the Strengthening The Reporting of Observational Studies in Epidemiology checklist.⁶

Participants and setting

Division and department email lists were used to identify active members of each department at one tertiary care (400-bed) and one community care (100-bed) hospital in Vancouver, Canada. All participants worked at one or both hospitals, although they may have also been active in other community hospitals, outreach sites, and private practices. Respondents provided informed consent and were not provided any incentive for participating in the study. Physicians were compared by specialty. As there were only a few physicians in several of the surgical subspecialties (urology, plastic surgery, vascular surgery, and ophthalmology), these specialties were combined and categorized as “other” surgical specialty. Physician groups were also categorized as surgical and nonsurgical specialties, as well as frontline COVID-19 physicians [emergency department (ED), general internal medicine, or Intensive Care Unit] vs nonfrontline COVID-19 physicians.

Survey administration

The survey was administered online using the Qualtrics survey platform (Qualtrics, Provo, Utah). Participants were able to complete the survey through both web- and mobile-based modalities. A modified Dillman approach⁷ was utilized to reach the participants for this study. From August to October 2021, online survey links were sent out to email lists from each department, followed by 2 reminder emails. Responses to the questionnaires were kept anonymous. There were no incentives provided to complete the survey.

Survey

Details of the survey, including pilot testing, have been published elsewhere.⁸ Briefly, the survey included questions on demographics; practice characteristics, including specialty type; and measures of burnout, work–life conflict, and intention to quit.

Demographic and work characteristics

Demographic items were age, sex, self-reported ethnicity, sexual orientation, and number of children. Work characteristics included specialty, clinical hours worked per week, number of weekend days working in a month, frequency of electronic medical record use, and frequency of on-call duties per month.

Burnout, work–life conflict, and intention to quit

Burnout was measured using the 22-item Maslach Burnout Inventory (MBI)–Human Services Survey for Medical Personnel, the most commonly used measure of burnout for health care professionals.² The three component scales of the MBI are emotional exhaustion, depersonalization, and personal accomplishment, which are composed of 9, 5, and 8 unique items, respectively. Emotional exhaustion refers to one’s feelings of chronic fatigue and being emotionally overextended by their work. Depersonalization means an impersonal response and loss of empathy toward patients. Personal accomplishment measures feelings of competence and success in one’s work, reduction in which results in negative self-assessment. Each of the 22 items are scored on a 7-level frequency scale in the following manner: 0 = “never,” 1 = “a few times a year or less,” 2 = “once a month or less,” 3 = “a few times a month”; 4 = “once a week,” 5 = “a few times a week,” and 6 = “every day.” Utilizing the sum of reported scores in each domain, the MBI provides a quantitative rating in each category, as well as a holistic measure of burnout. Consistent with commonly used criteria in the literature, burnout was defined as an emotional exhaustion score of ≥ 27 or a depersonalization score of ≥ 10.⁸ The same cutoffs were used to define the presence or absence of emotional exhaustion and depersonalization. Also consistent with other published studies, personal accomplishment scores were assessed separately. A score of ≤ 33 indicated a feeling of low personal accomplishment. Panagioti et al illustrated that single-point alterations in burnout scores are associated with significant differences in self-perceived major medical errors, work hour reductions, and rates of suicidal ideation.¹

In addition to evaluating the components of the MBI, work–life conflict and intention to quit were also assessed, both of which have been shown to be associated with burnout.^9,10 For work–life conflict, participants were asked to select their agreement or disagreement with the following statement: “My work schedule leaves me enough time for my personal/family life.” Participants were also asked if they have ever left or considered quitting a work position because of burnout and their options were “No, I have never considered quitting or left a position,” “Yes, I considered quitting but did not leave,” “Yes, I left a position,” and “Yes, I am considering quitting a position now.”

Patient and public involvement

Participants in this study included physicians only and no patients were involved in any stages of the study process.

Statistical analysis

Baseline characteristics were compared between those who did or did not experience burnout, including specialties, using the Fisher’s exact test. The authors used multivariable logistic regression models to evaluate the independent association between physician specialty and the presence of burnout with and without adjusting for potential personal and work confounders. The authors also evaluated the associations between physician specialty and the individual components of burnout, work–life conflict, and intention to quit in separate models. All adjusted models included the following covariates that have been associated with increased chance of burnout¹¹: age, sex, ethnicity (dichotomized to White or visible minority physician due to sample size), number of clinical hours worked per week, weekend days worked per month, and frequency of on-call duties, as well as perceived effect of the COVID-19 pandemic on burnout.

Missing values ranged from 9% to 12% across all survey questions (12% of physicians had incomplete MBI scores). Respondents with missing values were not significantly different from those without missing data in terms of specialty, sex, age, and years of clinical practice, and these physicians were excluded from the analyses. As a sensitivity analysis, the authors compared burnout, work–life conflict, and intention to quit among specialties that had a < 50% response rate vs a > 50% response rate, and there were no significant differences. All tests were 2-sided, and the level of significance was 0.05 using STATA 12.0 (College Station, Texas).

Results

There were 209 respondents of the 498 physicians to whom the survey was sent (42% response rate), of which 12% did not provide MBI scores, leaving an overall 37% complete response rate with MBI data. Response rates varied by specialty: 76% for emergency medicine, 62.5% for radiology, 53% for surgery, 47% for anesthesia, 39.7% for psychiatry, 37% for laboratory medicine, and 23.4% for internal medicine.

Baseline characteristics

Most participants identified as women (104/184) (Table 1), were between the ages of 36 and 50 years and were White, with 30% of the respondents identified as a visible minority. More than two-thirds of respondents had children. Of the respondents, 27% worked in surgical specialties (50/184), 40.5% reported having on-call duties > 4 times per month, and 92% reported working at least 1 weekend day per month. Almost all (91%) reported frequent use of electronic medical records.

Table 1:

Baseline characteristics according to physician burnout, % (n)

Characteristics	Burnout N = 135	No Burnout N = 61	P Value
Age, y
25-35	13.95 (18)	5.26 (3)	0.006
36-50	58.14 (75)	47.37 (27)
51-65	27.13 (35)	38.60 (22)
≥66	0.78 (1)	8.77 (5)
Women	62.70 (79)	44.64 (25)	0.05
Ethnicity
White	64.34 (83)	68.97 (40)	0.40
Asian/Pacific Islander	13.18 (17)	8.62 (5)
South Asian	5.43 (7)	12.07 (7)
Other	11.63 (15)	6.90 (4)
LGBTQ	6.92 (9)	0 (0)	0.06
Parity
No children	29.13 (37)	22.41 (13)	0.60
1–2 children	48.03 (61)	50.00 (29)
≥3 children	22.93 (29)	27.59 (16)
Specialty
Nonsurgical	72.22 (91)	74.14 (43)	0.08
Surgical	27.78 (35)	25.86 (15)
Anesthesia	11.90 (15)	20.69 (12)
Emergency medicine	15.87 (20)	13.79 (8)
Internal medicine	20.63 (26)	13.79 (8)
Obstetrics and gynecology	7.94 (10)	1.72 (1)
Orthopedics	7.94 (10)	5.17 (3)
Psychiatry	12.70 (16)	6.90 (4)
Radiology	3.97 (5)	13.79 (8)
General surgery	6.35 (8)	6.90 (4)
Laboratory medicine	7.14 (9)	5.17 (3)
Other surgical specialties	12.07 (7)	12.07 (7)
Divisions primarily responsible for	34.1 (46)	26.2 (16)	0.32
COVID care (internal medicine,
critical care and ED)
Clinical duty hours
< 40 h/w	46.5 (72)	52.1 (38)	0.04
< 60 h/w	21.9 (34)	8.2 (6)
Weekend days worked/ mo
None	5.38 (7)	13.79 (8)	0.17
1–2	59.23 (77)	56.90 (33)
3–4	23.85 (31)	24.14 (14)
≥5	11.54 (15)	5.17 (3)
On-call days/mo
None	17.32 (22)	13.79 (8)	0.56
1–3	44.88 (57)	39.66 (23)
≥4	37.80 (48)	46.55 (27)
View COVID as affecting burnout	75.2 (127)	45.6 (36)	< 0.0001

ED, emergency department; LGBTQ, lesbian, gay, bisexual, transgender, queer.

Physician burnout based on specialty

Prevalence of burnout ranged from 36% in radiology to 91% in obstetrics and gynecology (Figure 1). However, these differences were not statistically significant, and physician specialty was not independently associated with burnout after adjustment for other covariates [adjusted odds ratio (aOR) = 0.93; 95% confidence interval (CI): 0.85–1.02] (Table 2). Overall burnout prevalence was similar between surgical specialties (70%) and nonsurgical specialties (68%) (p = 0.84) and between those who worked at the frontline for COVID-19 care (74%) vs those who did not (66%) (p = 0.28).

Figure 1:

Prevalence of burnout by specialty. Obs & Gyne = obstetrics and gynecology.

Table 2:

Adjusted odds ratios of physician specialty and burnout, quitting or intention to quit, and work–life conflict

Category	Burnout		Quit or intention to quit		Work–life conflict
	aOR (95% CI)	P value	aOR (95% CI)	P value	aOR (95% CI)	P value
Specialty	0.93 (0.85–1.02)	0.12	0.96 (0.89–1.05)	0.41	1.02 (0.94–1.10)	0.67
Age	0.50 (0.27–0.92)	0.03	1.40 (0.82–2.38)	0.22	1.18 (0.71–1.95)	0.52
Women vs men	0.76 (0.40–1.47)	0.42	0.78 (0.40–1.51)	0.45	0.52 (0.27–1.02)	0.06
Ethnicity	1.85 (1.08–3.17)	0.03	1.27 (0.81–1.98)	0.30	1.16 (0.76–1.78)	0.48
Clinical hours	2.72 (1.29–5.72)	0.008	0.85 (0.47–1.55)	0.6	3.19 (1.71–5.96)	0.001
High frequency of on-call duties	0.52 (0.27–1.00)	0.05	1.26 (0.70–2.29)	0.45	0.80 (0.45–1.43)	0.45
Weekend work	1.95 (1.10–3.43)	0.02	1.20 (0.74–1.94)	0.46	1.79 (1.09–2.93)	0.02
Perception COVID-19 associated with burnout	9.40 (3.74–23.7)	< 0.001	2.56 (0.98–6.66)	0.05	3.59 (1.50–8.58)	0.004

aOR, adjusted odds ratio; CI, confidence interval.

Emotional exhaustion, Depersonalization, and Personal Accomplishment

Components of burnout, including emotional exhaustion, depersonalization, and feeling low personal accomplishment, were observed in a large proportion of respondents across all specialties (Figure 2). In surgical specialties, 70% (35/50) reported emotional exhaustion, 49% (25/51) reported depersonalization, and 25% (12/48) reported low personal accomplishment. Among physicians in nonsurgical specialties, 57.5% (77/134) reported emotional exhaustion, 47% (63/134) reported depersonalization and 31.8% (42/132) reported feelings of low personal accomplishment. In multivariate analyses adjusting for covariates and work characteristics, there was no independent association of specialty on prevalence of each of these subscales (Figure 3).

Figure 2:

Proportion of physicians who experienced high emotional exhaustion. Obs & Gyne = obstetrics and gynecology.

Figure 3:

Multivariate association of specialty and high emotional exhaustion, high depersonalization, and low perception of personal accomplishment (dots represent odds ratios and whiskers, 95% confidence intervals).

Work–life conflict and intention to quit

There were no significant differences in reporting work–life conflict or quitting or intention to quit across all specialties (Figures 4 and 5) including after adjustment for covariates (Figure 6 and Table 2) and comparing frontline workers with nonfrontline workers. However, the prevalence of work–life conflict was significantly higher in surgical specialties than in nonsurgical specialties (66.7% vs 35.9%, respectively, p = 0.01) even after adjusting for differences in clinical working hours, weekend duties, and frequency of call shifts (aOR = 2.88; 95% CI: 1.22–6.82, p = 0.016). Surgical and nonsurgical specialties similarly reported a 25% prevalence of quitting a position or considering quitting a position, and there was no significant difference in the multivariate analysis.

Figure 4:

Prevalence of perceived work–life conflict (%). Obs & Gyne = obstetrics and gynecology.

Figure 5:

Prevalence of physicians who quit or intend to quit a position. Obs & Gyne = obstetrics and gynecology.

Figure 6:

Multivariate association with specialty and burnout, quit or intention to quit, and perceived work–life conflict (dots represent odds ratios and whiskers, 95% confidence intervals).

Discussion

The present study’s findings show that physician burnout among various medical specialties was high 18 months into the COVID-19 pandemic and that there were no significant differences between specialties, surgical and nonsurgical groups, and frontline and nonfrontline physicians. Similarly, there were no significant differences between specialties in the components of burnout, namely, emotional exhaustion, depersonalization, and feelings of low personal accomplishment. Work–life conflict was significantly higher among surgeons compared to those in nonsurgical specialties even after adjusting for personal and work characteristics. There were also no significant differences in intention to quit or having quit a position across specialties, surgical and nonsurgical specialties, or between COVID-19 frontline physicians and other physicians.

Few studies have compared burnout across a wide range of surgical and nonsurgical specialties in hospitals during the pandemic.¹² The findings of the present study extend previous findings¹² to highlight the pervasiveness of burnout in physicians during the COVID-19 pandemic across a broader range of specialties. A national survey in the US evaluated burnout of various frontline physicians during the first and second waves of the pandemic.¹³ That study administered the Professional Fulfillment Index Burnout Composite scale to 2000 family physicians, 2000 internists, 2000 critical care physicians, 1000 hospitalists, 2000 ED physicians, and 1000 infectious disease specialists.¹³ Comparing the first and second waves, burnout increased across all surveyed specialties except for emergency medicine, for whom the initially high rates of burnout at the start of the pandemic were followed by a slight improvement.¹³ Similarly, Gupta et al found that prevalence of anxiety was 78% in medical specialists, 87% in surgical specialists, and 92% in ED physicians who worked in Bathinda, India.¹⁴ Houdmont et al administered the MBI to surgeons in the UK during the pandemic.¹⁵ Respondents reported high prevalence of emotional exhaustion (57%) and depersonalization (50%) and a low prevalence of low personal accomplishment (15%).¹⁵ A comparable pattern was observed in the present study. A national survey of Canadian emergency medicine physicians reported a 60% prevalence of burnout during the second wave of the pandemic, as opposed to < 20% during the first wave.¹⁶ A similar survey of emergency medicine physicians in the US also reported exhaustion at work and in their personal lives.¹⁷ These findings are consistent with the present study. However, the present study’s findings extend to multiple specialties not previously examined, including radiology, psychiatry, and laboratory medicine, demonstrating the high prevalence of burnout in many medical and surgical specialties.

Work–life conflict has previously been identified as a major issue among surgeons, particularly surgeons who are women. Comparing feelings of work–life conflict between physicians in surgical and nonsurgical specialties in the present study, the authors found a significantly higher prevalence in surgeons during the pandemic. This finding is consistent with those of a study performed prior to the pandemic, indicating substantial work–life conflict in surgical specialties. In that survey of 7197 American surgeons, 47% reported work–life conflict.¹⁸ Some of the suggested reasons for this finding include low autonomy in scheduling, limited allowance for job sharing, insufficient systemic support with child care, and concerns about medical lawsuits.^18,19 On average, surgeons work 60 hours per week and are on-call 2 nights per week,²⁰ leaving limited time to spend on self-care or other activities. In another survey of 24,922 surgeons from the American College of Surgeons, Shanafelt et al reported that although rates of job satisfaction were high among surgeons, most (64%) felt that their work schedule did not leave enough time for their personal and family lives.²⁰ Physicians’ work–life conflict is multifactorial, potentially stemming from a combination of factors, including caring for children, heavy workload, frequent overtime work, and inflexible work hours.²¹ In the present study, physicians attributed the pandemic as a significant source of physician burnout.

Although physicians primarily responsible for caring for patients with COVID-19 experienced psychological and physical distress during the pandemic,²² physicians in all specialties reported high rates of burnout irrespective of their specialty. This finding demonstrates the widespread impact of the pandemic, thereby emphasizing the need to address causes of burnout for physicians in all specialties. Some of the potential stressors include chance of contracting COVID-19 and fear of spreading infection to family members; social isolation, telemedicine, and lack of personal interaction with patients; limited supply of personal protective equipment and medical equipment; reductions in salary, benefits, and vacation time; lack of access to reliable and up-to-date information; and increased demands for child care at home due to school closures.^23–25

Strengths and limitations

Strengths of this study included use of a validated, standardized assessment tool (the MBI) and the broad inclusion of multiple specialties from both a community and a tertiary hospital, which increased physician representation. However, there were also limitations in this study. First, considering a final response rate of 37%, nonresponder bias is one of the main limitations of this study. However, this response is consistent with other physician survey studies, and the present study’s high prevalence of burnout in several specialties was consistent with other studies.²⁵ Further, there were no significant differences in burnout, work–life conflict, or intention to quit among specialties that had a < 50% response rate vs a > 50% response rate. Second, due to the differences in specialties and sizes of the departments, data collected from smaller departments and divisions (eg, ophthalmology; ear, nose, throat, and neck; plastic surgery; vascular surgery; urology) had to be combined for meaningful comparisons to be made. Hence, the experience of physicians in these individual groups may not be consistent with the aggregate data in the combined group. Third, the authors did not have data from before the COVID-19 pandemic to compare with the present study’s findings. Hence, the authors could not draw any inferences about changes due to the pandemic from this study. A meta-analysis of the prevalence of burnout in medical and surgical residents prior to the COVID-19 pandemic reported an average burnout prevalence of 51% among all specialties.²⁶ Varied burnout rates were reported ranging from ~ 36% in family medicine residents to as high as 77% in radiology residents.²⁶ Similar to the present study, the differences in burnout between specialties in that study were not statistically significant, with no major difference between residents in surgical and medical specialties. To the authors’ knowledge, however, similar studies have not been conducted in attending physicians prior to the COVID-19 pandemic. Last, there may be unique pandemic-related factors that were not captured by the MBI. Over the past few years, there have been specific COVID-19 burnout views scales developed that could be a valuable supplement to the standardized MBI and should be considered in future studies.²⁷ Such factors include but are not limited to implemented preventive measures established by local governments, level of adherence to public health policies, each individual’s health status, and fear of COVID-19.²⁷

Conclusion

Overall, burnout measured during the COVID-19 pandemic was high among physicians across all specialties. Surgeons reported higher work–life conflict during the pandemic compared to nonsurgeons. Overall, these findings further highlight the universally high levels of burnout among physicians during the global pandemic, irrespective of the nature of their specialty. Interventions to reduce burnout and promote wellness should target all physician specialties.