Changes in Burnout and Moral Distress Among Veterans Health Administration (VA) Physicians Before and During the COVID-19 Pandemic

Between 2019 and 2020, burnout and moral distress among a national sample of VA physicians decreased among those who experienced improvements in workload, organizational satisfaction, and psychological safety. However, medical center-level COVID-19 case rates and state-level restrictions were not related to either burnout or moral distress.

Background

We investigated the impacts of workload, resources, organizational satisfaction, and psychological safety on changes in physician burnout and moral distress among physicians during the early pandemic.

Methods

We obtained national administrative and survey data on burnout, moral distress, organizational satisfaction, psychological safety, COVID-19 burden, and state-level restrictions for 11,877–14,246 Veterans Health Administration (VA) physicians from 2019 and 2020. We regressed the changes in burnout and moral distress on the changes in reasonable workload, appropriate job resources, organizational satisfaction, and psychological safety, controlling for COVID-19 burden and restrictions, and individual and medical center characteristics.

Results

Burnout and moral distress were not related to COVID-19 cases or restrictions but were reduced by improvements in workload, organizational satisfaction, and psychological safety.

Conclusions

Health systems should be conscious of factors that can harm or improve physician well-being, especially in the context of external stressors.

LEARNING OUTCOMES

1. Identify rates burnout and moral distress among VA physicians in 2019 and 2020.

2. Understand how burnout and moral distress among VA physicians were related to workload, resources, organizational satisfaction, and psychological safety during the COVID-19 pandemic.

Before the COVID-19 pandemic, physician burnout rates in the Veterans Health Administration (VA) were persistently high, with rates of 43%–49% from 2013 to 2017. ¹ In contrast, a recent systematic review estimated that only approximately 35% physicians in all healthcare systems in the United States were burned out on average. ² Other national studies of VA healthcare found similarly high burnout rates: 31%–38% of behavioral health providers were burned out from 2015 to 2018 ³ and 48%–66% of primary care providers were burned out from 2013 to 2019. ^4–6 New workplace conditions imposed by COVID-19 (ie, limited resources and treatment, extended shifts, pandemic-induced stress) may have taken a psychological and emotional toll on physicians. The pandemic has increased work-related stressors, with some studies showing increased burnout and mental health issues, ^7–9 particularly in higher-risk settings, such as the emergency department, intensive care units, and COVID-19 wards. ^10–12 However, limited and inconsistent research exists on the longitudinal effects of COVID-19 on physician well-being, making it challenging to address physician well-being issues in the workplace during this period.

These additional workplace demands and stressors of the pandemic likely harmed well-being among physicians, increasing not only their burnout but also their moral distress, measured burnout and moral distress. Burnout is a work-related syndrome characterized by emotional exhaustion, depersonalization, and a reduced sense of personal accomplishment. ¹³ Moral distress is a complex concept that can include guilt and anxiety from experiencing stressful events counter to one’s values, ¹⁴ such as when healthcare workers are unable to provide proper care. ¹⁵ Moral distress has been represented as three factors involving family-, infection-, and work-related factors in one study that reported more than half of frontline COVID-19 healthcare workers experienced it. ¹⁶ Furthermore, burnout was influenced by moral distress related to helping loved ones and workers not feeling they were able to do enough for patients. Other work shows a cumulative effect of moral distress building over time leading to burnout. ¹⁷

Some studies report that surgeons and primary care providers experienced decreased or minimally changed burnout compared with prepandemic rates, ^18,19 while others show varying effects on burnout for physicians working in COVID-19 patient wards, ²⁰ an area where stress and job demands could be among the highest. Studies on intensive care unit nurses report greater burnout during the pandemic, ²¹ especially compared with physicians within the same care setting. ²² Furthermore, clinicians who were not treating patients with COVID-19 reported similar levels of moral distress as those who were, attributed to guilt from not helping manage the crisis.

Most work examining burnout and moral distress among clinicians has used a single medical center or a limited number of participants and settings ^23–26 making it difficult to understand how clinicians in different geographical areas were affected by the local COVID-19 prevalence and other geographic changes in response to the pandemic, such as mask mandates or other restrictions. These mixed findings underscore a major knowledge gap of how the COVID-19 pandemic has impacted burnout and moral distress.

Factors from the job-demands resource (JDR) model have commonly been included to examine characteristics affecting well-being during the pandemic. ^27–29 The JDR model theorizes that two types of factors directly affect employee well-being—job demands and job resources. ³⁰ These demands and resources include literal job demands and resources, like workload or staffing, as well as psychological or environmental demands and resources, like interpersonal conflict or a positive organizational culture. ³¹ Previous work in VA among behavioral healthcare providers and primary care physicians has shown that reasonable workload, ^3,5 adequate resources, ³ and organizational satisfaction ⁵ are all linked to lower burnout. During the COVID-19 pandemic, healthcare staff were challenged with limited resources, such as a lack of personal protective equipment, shifting workloads, and concern for personal and loved one’s safety. It is likely that increased workloads and decreased resources increased burnout and moral distress among physicians during the pandemic, but good organizational resources may have blunted some of the pandemic’s impact. Organizational satisfaction or a sense of psychological safety (the concept of allowing or encouraging measured risk taking in the workplace ³²) may have decreased burnout and moral distress despite increased workloads and decreased resources. The objective of the current study is to examine the impact of the COVID-19 pandemic, through its effects on job demands and resources, on physician burnout and moral distress using a national geographically distributed sample of VA physicians.

METHODS

Study Design

We examined secondary data sets using responses to an annual survey of all VA employees during 2019–2020. The VA has a physician workforce of more than 25,000 with more than 140 medical centers across the United States. We examined how the changes in job resources and job demands were associated with changes in burnout and moral distress. We aggregated individual survey responses to the service level within medical centers and merged with publicly reported data on state policies and COVID-19 prevalence among the veteran population for the medical center. The study was approved by the VA Boston Healthcare System.

Sample

The All Employee Survey (AES) is an annual voluntary and anonymous online survey sent to all active employees within the VA. ³³ The total number of eligible and invited employees from medical centers was 328,535 (63.4% response rate) for the survey administered in June 2019 and 341,261 (69.5% response rate) for the survey administered in September–October 2020. A total of 11,877 physicians responded in 2019 and 14,246 responded in 2020 in service areas that met our criteria for inclusion. Sample characteristics are presented in Supplemental Table 1, http://links.lww.com/JOM/B330.

We aggregated data to understand changes at the site level. The service area was the unit of analysis.

Measures

Outcomes included measured burnout and moral distress while predictors included items on job demands, resources, psychological safety, and organizational satisfaction from the AES, workgroup composition, service area, and publicly available datasets to represent prevalence of COVID-19 and state restrictions.

Outcomes

Burnout was computed as the average response to two items based on the Maslach Burnout Inventory ³⁴ and represented the concepts of exhaustion and depersonalization: two commonly used themes to understand burnout. The exhaustion item asked, “I feel burned out from my work” and the depersonalization item asked, “I worry that this job is hardening me emotionally.” Respondents rated items using a seven-point Likert scale by choosing one of the following response options: (1) “never”, (2) “a few times a year or less,” (3) “once a month or less,” (4) “a few times a month,” (5) “once a week,” (6) “a few times a week,” or (7) “every day.” Those who indicated having feelings of either exhaustion or depersonalization once a week or more (>4) were coded as having burnout, consistent with previous studies of VA burnout. ^1,3 This scoring method of using select burnout items has been validated against use of the full instrument. ³⁵ Moral distress was assessed using a single item developed a priori for the survey, “In the past year, how often did you experience moral distress at work (ie, you were unsure about the right thing to do or could not carry out what you believed to be the right thing).” Respondents also rated the moral distress item using a seven-point Likert scale ranging from (1) “never” to (7) “every day.” Respondents who indicated having these feelings once a week or more (>4) were coded as having moral distress. This moral distress item was similar to a validated nursing moral distress item. ³⁶ For both burnout and moral distress outcomes, lower scores reflected lower levels of burnout and moral distress at the service level.

Predictor Measures

From the AES, reasonable workload values were based on responses to the item, “My workload is reasonable.” Appropriate job resources values were based on responses to the item “I have the appropriate supplies, materials, and equipment to perform my job well.” Psychological safety was assessed based on responses to the question: “It is worthwhile in my workgroup to speak up because something will be done to address our concerns.” The items were rated on a scale from (1) “strongly disagree” to (5) “strongly agree” with higher scores indicating more favorable working conditions. The question “Considering everything, how satisfied are you with your organization” was assessed using a scale ranging from (1) strongly dissatisfied to (5) “strongly satisfied.” Missing data among AES measures were between 0.8% and 3.2%.

We classified respondents into service areas based on self-report to the question: “What is the main type of service you provide?” and restricted our analysis to focus on the most represented services with at least five respondents per medical center in both years. The services represented included the following: administrative, emergency medicine, medical specialties, mental health, primary care, surgical specialties, and acute care/intensive care units. For respondents who did not fit into one of those service categories, we combined responses to create an “all other area” category. Primary care was used as the referent group in regression models.

We obtained publicly available data on prevalence of COVID-19 reported by VA ³⁷ as of September 15, 2020—the start of the AES administration. The data included VA confirmed total COVID-19 cases, number of active cases for that day, and number of known deaths among the VA population served. We created a cumulative COVID-19 case rate per 1000 unique patients at the medical center level. We also accounted for geographical region (West, South, Midwest, and Northeastern as a referent) and urban or rural area in the model. We accounted for gender differences in burnout, which may have an influence, ^8,38 by looking at the proportion of females in a service area over the study period.

To understand the influence that state practices, we used a publicly available index scoring states on COVID-19 restrictions for August 2020. ³⁹ The data for the index were compiled from various sources, including the federal program offices, such as US Census Bureau, Centers for Disease Control and Prevention, and other agencies, such as Kaiser Family Foundation and National Restaurant Association. The index assigned weights based on the extent that each state mandated public restrictions. Practices (with weight assigned) given the most weight of 100 points included the following: requirements to wear a face mask in public (9.30), public gathering size limits (13.95), school closures (9.3), restaurant/bar (18.6) and nonessential business operations (9.3), and “shelter in place” orders (13.95). We reverse scored the index so that higher scores indicated greater state restrictions.

Analysis

Based on respondent’s selection of service area, we computed average scores for the measures from the AES. To ensure aggregation was appropriate, we required at least five respondents per service area for each study year. We computed the difference between each service area survey measure in 2020 compared with 2019. Thus, increased prevalence in burnout or moral distress indicated that the measure increased over time. We used a difference-based approach and generalized linear model that regressed the change in either burnout or moral distress on the changes in independent variables of reasonable workload, job resources, psychological safety, organizational satisfaction, and other control measures. Analyses were conducted using SAS Software 9.4 (SAS Institute, Inc.; Cary, NC). We report unstandardized parameter coefficients and standard errors along with using a criterion of P < 0.05 to indicate significance.

RESULTS

We found minimal change in burnout or moral distress before versus after the start of the pandemic overall. We observed a small increase in other AES measures indicating that respondents had more favorable views of working conditions after the pandemic (Table 1).

TABLE 1.

Descriptive Characteristics of the Study Sample

Variable	Range	Level	2020	2019	2020–2019
			Mean (SD)	Mean (SD)	Mean (SD)
Burnout	0–1	Service	0.31 (0.18)	0.32 (0.18)	−0.01 (0.18)
Moral distress	0–1	Service	0.12 (0.10)	0.12 (0.10)	0.00 (0.14)
Reasonable workload	1–5	Service	3.78 (0.51)	3.66 (0.59)	0.12 (0.45)
Appropriate job resources	1–5	Service	3.92 (0.38)	3.79 (0.47)	0.14 (0.41)
Psychological safety	1–5	Service	4.08 (0.39)	4.05 (0.42)	0.03 (0.41)
Organizational satisfaction	1–5	Service	3.79 (0.44)	3.67 (0.51)	0.12 (0.45)
Percent female*	0–100	Service	0.40 (0.16)	—	—
Cumulative COVID-19 case rate	Cases per 1000 unique veterans	Medical center	7.73 (6.04)	—	—
Active COVID-19 case rate	Cases per 1000 unique veterans	Medical center	0.39 (0.38)	—	—
State restrictions	0–100	State	57.37 (12.85)	—	—

*Percent female based on combined demographics for 2019 and 2020.

In multivariable burnout model (Table 2), physicians who perceived that their workload became more reasonable (b = −0.09) or who were more satisfied with their organization (b = −0.12) were less likely to be burned out. Workgroups with more female physicians saw an increase in burnout (b = 0.10). Mental health groups also reported an increase in burnout (b = 0.05). In the model looking at moral distress, physicians who reported increases in organizational satisfaction (b = −0.05) and psychological safety (b = −0.05) were less likely to be morally distressed. The case rate of COVID-19 or amount of state restrictions seemed to have no influence on changes in either burnout or moral distress.

TABLE 2.

Changes in Measured Burnout and Moral Distress From 2020 to 2019

	Burnout	Moral Distress
	b (SE)	b (SE)
Intercept	−0.06 (0.05)	0.02 (0.04)
Δ Reasonable workload	−0.09** (0.02)	−0.02 (0.01)
Δ Appropriate job resources	0.01 (0.02)	−0.02 (0.01)
Δ Organizational satisfaction	−0.12** (0.02)	−0.05** (0.02)
Δ Psychological safety	−0.03 (0.02)	−0.05** (0.01)
Cumulative case rate	0.00 (0.00)	0.00 (0.00)
Rural	−0.01 (0.02)	−0.02 (0.02)
Region: Midwest	0.02 (0.02)	0.00 (0.02)
Region: South	0.00 (0.02)	−0.02 (0.01)
Region: West	−0.01 (0.02)	−0.03 (0.02)
State restrictions	0.00 (0.00)	0.00 (0.00)
Workgroup female percent	0.10* (0.04)	0.05 (0.04)
Service: administrative	0.03 (0.03)	−0.01 (0.02)
Mental health	0.05* (0.02)	0.01 (0.02)
Surgical specialty	0.02 (0.02)	−0.01 (0.02)
Emergency medicine	0.00 (0.03)	0.02 (0.02)
Acute care/intensive care	0.01 (0.02)	0.01 (0.02)
Medical specialty	0.01 (0.02)	0.01 (0.02)
All other areas	0.02 (0.02)	0.03 (0.02)

* Indicates P < 0.05, ** indicates P < 0.01; unstandardized β coefficients and standard errors reported in table.

DISCUSSION

We found minimal changes in measured burnout and moral distress before and after the start of the COVID-19 pandemic. Across service areas, we saw a small increase in ratings of reasonable workload, job resources, organizational safety, and psychological safety in 2020 from 2019 (ie, respondents indicated that these working conditions improved during the pandemic). The fact that respondents did not report worsening working conditions during the pandemic was contrary to our a priori expectations, but it is encouraging that so many physicians were able to experience better working conditions during such a challenging time in medicine. Consistent with the JDR model, increased ratings of reasonable workload, organizational satisfaction, and psychological safety were related to lower odds of burnout or moral distress.

Burnout and moral distress did not worsen during the pandemic in our sample. Among VA physicians, burnout has been generally declining since 2014, ^1,40 which was the height of the “access crisis,”⁴¹ during which the agency received intense public scrutiny about the lack of appointment availability for veterans. There is some emerging evidence, among VA primary care physicians and non-VA physicians in multiple specialties, that burnout hit a low point in 2020, ^40,42 and then subsequently increased. It is possible that at the start of the pandemic, physicians both in the VA and in the community, were engaged and rose to the challenge to provide care during the crisis, but for some physicians those feelings of engagement waned, and burnout increased as the pandemic wore on. High engagement has consistently been linked to lower burnout across professions before the pandemic. ⁴³ Moral distress may have remained low in our sample for similar reasons and then may have waned throughout the pandemic as resources became more scarce and public support for the pandemic dropped. We also found that perceptions of a reasonable workload did not change during the early pandemic. There is evidence that in-person VA outpatient ⁴⁴ and emergency department ⁴⁵ visits dropped significantly in 2020, presumably as patients skipped nonurgent care to avoid exposure to the virus, so it is likely that workload did remain reasonable for the average physician.

Increased workload ^46,47 and hours worked ^47,48 have long been associated with higher burnout among physicians. Previous studies of behavioral health providers and primary care physicians in VA also reported a similar relationship between workload and burnout. ^3,5 Our findings are consistent with this work and highlighted the impact that the pandemic had on perceived workload for different physicians. For some service areas, like administrative areas, the COVID-19 pandemic may have led to more complexity in managing and acquiring resources, such as personal protective equipment, creating and implementing new policy guidance on safety, space for treating patients who need special precautions, staffing, and interfacing with the community. ^49,50 For clinical areas, perceptions of more reasonable workload may have been due to more patients using telehealth or virtual health services or decreasing the number of requested appointments or demand for services. ^51–53

Adequate job resources ⁵⁴ have been less well studied in terms of change during COVID-19, but in one study of primary care providers and staff, full staffing was related to lower burnout, while panel overcapacity was related to higher burnout. ⁵⁵ Physicians across service areas may have greater exposure and contact with hospitalized patients for COVID-19, making the need for obtaining resources, like N95 masks, sterile gowns, and gloves for in-person encounters. Physicians who were initial points of contact for COVID-19 patients may have been particularly challenged with needing to manage patients without having full knowledge or resources available, or having to withhold treatment from certain COVID-19 patients as resources, like ventilators, became limited.

Organizational satisfaction and psychological safety have been previously linked to lower burnout in VA-⁵ and non-VA ⁵⁶ primary care. A small early pandemic study of two VA primary care clinics also showed the association between positive perceptions of the working environment across multiple dimensions (eg, community and fairness) and lower burnout. ¹⁹ There is some emerging evidence that engagement within one’s workplace can be connected to better well-being, even in the midst of a crisis like the COVID-19 pandemic. ⁴⁰ In our study, physicians who were more satisfied by their workplace during the pandemic were less likely to experience burnout or moral distress. It is possible that approval of the organization’s structure or mission drove this increase in organizational satisfaction, thereby giving these physicians hope and confidence as they worked throughout the pandemic. Psychological safety seems to have a similar relationship to moral distress. Physicians who were given the flexibility in their workplace to take risks and try out new ideas during the pandemic may have been less likely to be morally distressed in their decision making. Thus, empowering physicians to make more of their own decisions during crisis situations may improve their well-being.

Physician measured burnout and moral distress did not seem to be strongly influenced by active case rate measures or by state restrictions to help curb the pandemic. COVID-19 cases likely had a large effect on the working lives of physicians, but its effect on burnout and moral distress seems to have been mediated by workload, resources, organizational satisfaction, psychological safety, and other factors. It is encouraging that large increases in burnout and moral distress from external stressors like COVID-19 can be potentially mediated by improving working conditions.

State-level restrictions also had no relationship with burnout or moral distress. However, we were unable to assess differences in local policies, such as school closures, that may have been obscured by state-level data and may not have impacted workplace perceptions. It is possible that the local approach to the COVID-19 pandemic was similar to other public health crises before the pandemic, and physicians may not have been surprised by the state regulations. Some physicians may have supported state policies that diverged from scientific and medical recommendations. Approval and roll-out of vaccinations for COVID-19 occurred several months later, and the local response to vaccinations may have had a more substantial effect on physician burnout and moral distress if there was a perception that local populations were disregarding widely recommended preventative measures.

Reasonable workload, organizational satisfaction, and psychological safety were related to lower burnout and moral distress, even after accounting for COVID-19 cases and state restrictions. These relationships are likely to persist in a postpandemic environment, highlighting the need to create and support supportive and psychologically safe workplaces with sustainable workloads. While reducing workloads or redesigning workplaces may seem expensive, there is a substantial business case for combating burnout. ⁵⁷ Burned-out physicians are more likely to leave their jobs, ⁵⁸ and replacing one physician can cost a healthcare system up to $1,000,000 ⁵⁹ after accounting for recruitment, training, and lost revenue. Healthcare workplaces should seek to ensure reasonable workloads and positive working environments prevent physician burnout and costly turnover in the long run.

The study had several strengths, including the use of a large, geographically distributed sample of providers, examining changes over time, and connecting medical center–level COVID-19 data to geographic areas to understand the potential impact. We had a robust response rate of greater than 60% to the survey.

Limitations of the study include focusing solely on VA, which may not be representative of other healthcare systems and medical centers across the United States. It is unclear whether we found improved favorable ratings on the survey measures because VA medical centers were able to secure resources and manage demands relatively better. We only studied physicians, so our estimates may not reflect the levels of burnout among other professions who were also affected by the pandemic. In addition, we only examined burnout, moral distress, and working conditions during the first 6 months of the pandemic; these factors may have deteriorated significantly as the pandemic continued. We were unable to track individual responses over time, which may have helped improve precision of our estimates and interpretations regarding changes in measured burnout and moral distress because of the changing workplace and pandemic features. Finally, the measure we used to understand moral distress has not been validated, although the survey measure directly asks about perceived moral distress, so it has high face validity. Our survey measures may not have been sensitive enough to detect changes or asked in a way that may reflect the experiences of respondents.

Extending our analysis to compare differences among other job roles, such as nursing, social work, and administrative staff would be important given the team-based nature of care. Given that burnout and moral distress are complex concepts, interviews with providers about their experiences during the time would help frame findings.

Results suggest that physician well-being measures did not change greatly based on local COVID-19 case rates or state-based restrictions. Changes in measured burnout and moral distress seemed to be affected more so by changes in reasonable workload, organizational satisfaction, and psychological safety. Health systems need to be conscious of factors that improve or impair physician well-being, especially in the context of systemic changes that could improve organizational culture or increase work-related stressors.