Impact of moral distress on burn-out: a cross-sectional survey of front-line family physicians during the initial wave of the COVID-19 pandemic

Abstract

Objectives

Primary care clinicians are especially prone to burn-out. The primary objective of this study was to investigate factors contributing to burn-out and moral distress and their relationship among practising family physicians (FPs) in California early in the COVID-19 pandemic.

Design

Cross-sectional study, online survey evaluating burn-out, moral distress and associated factors.

Setting

California FPs between July and August 2020 practising in community health centres, hospital systems, private clinics and university systems were surveyed with a 22-item online questionnaire.

Participants

FPs practising in California were eligible. The final sample included 218 physicians.

Primary and secondary outcome measures

The primary independent variable was frequency of moral distress and the primary outcome variable was worsening burn-out. Moderator variables included gender and employer support.

Results

FPs experiencing higher burn-out and moral distress were more likely to report concerns regarding personal COVID-19 risk and lack of personal protective equipment. Practising self-care and personal wellness were associated with decreased moral distress. Female physicians were 3.86-fold more likely to report worsening burn-out compared with male physicians. Employer support was associated with a 59% reduced burn-out risk and 54% reduction in frequent moral distress. Frequent moral distress was associated with a 3.12- fold higher burn-out risk. Gender moderated the relationship between moral distress and burn-out. Moral distress was associated with a 3.55-fold increase in burn-out risk among females.

Conclusions

Female FPs experienced greater levels of moral distress and burn-out than male physicians. Moral distress was differentially associated with increased burn-out among female physicians. Employer support was a protective factor against moral distress and burn-out.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• A main strength of this study was that it assessed burn-out and moral distress among physicians on the front lines in real time which minimised recall bias and increased accuracy of reporting.

• This study describes the relationship between moral distress and burn-out during the peak of uncertainty of the COVID-19 pandemic among family physicians in California in 2020.

• Differences in burn-out and moral distress were evaluated in a stratified analysis according to gender and demonstrated the specific influence of moral distress on worsening burn-out among female family physicians.

• A main limitation of the study is the low response rate, however, healthcare surveys usually have a low response rate; this did not impact our ability to make statistical conclusions regarding our results.

• The cross-sectional design of the study cannot definitively determine causality.

Introduction

The novel COVID-19 pandemic emerged in December 2019¹ and has created significant stressors on all aspects of society. Health professionals and systems faced significant strains given increased workloads, uncertainties and rapidly changing conditions to care for patients with suboptimal resources and knowledge.^{2 3} One of the first papers describing the experiences of healthcare workers on the front lines during the pandemic in China showed increased anxiety, depression, insomnia and distress. Those directly caring for patients with COVID-19 and women reported the most severe symptoms.⁴

Burn-out, a mental state characterised by feelings of exhaustion, cynicism, depersonalisation and reduced job efficacy,^{5 6} is unfortunately commonplace among physicians. High occupational stress, time pressures, intense workloads and inadequate occupational support contribute to burn-out in medicine.⁷ Primary care doctors, serving as the first point of contact with the medical system, are especially prone to burn-out.^{8 9} Female physicians^{10 11} and physicians earlier in their training are disproportionately prone to burn-out.¹² Other factors associated with burn-out are younger age, long work hours (>40 hours), poor job satisfaction and conflict in the home.¹³ Engaging in work that interfered with home life, life-changing decisions and exposure to patients with COVID-19 were identified as factors contributing to physician burn-out in an international study on COVID-19.² An additional important risk factor was assuming new roles with inadequate preparation.^{2 3 14}

Multiple studies, the majority of which examined nurses, have identified associations between increasing moral distress and burn-out.^{15 16} Moral distress was first defined in 1984 by Andrew Jameton as an internal state that arises when an individual identifies the ethically correct course of action but faces constraints to perform such actions. Peter and Liaschenko¹⁷ further emphasised that moral distress arises when the individual has agency to make moral decisions, however, is constrained from acting on these decisions.¹⁷ Competing demands and constraints during the pandemic contributed to moral distress given the ethical obligations of healthcare professionals while facing barriers such as insufficient staff, personal protective equipment (PPE) and resources leading to challenges in providing quality care to patients.¹⁸ Limited visitation policies for patients who were critically ill or dying contributed further to moral distress.¹⁹ Experiencing moral distress can threaten an individual’s sense of self-worth because of incongruences between one’s values and one’s actions.²⁰ During the pandemic, the shift from a patient-centred to a community-centred focus of care also contributed to situations leading to moral distress.²¹

Factors hypothesised in our study to contribute to providers’ moral distress during the pandemic were based on the research team’s experience, literature review^{2 13 21 22} and our initial pilot interviews.²³ Factors identified included inadequate resources, working under conditions that posed risk to clinicians and their families’ lives, restricting patient visitors during end-of-life care, limited availability of PPE and the rapid transition to providing virtual patient visits that might compromise quality of care.²¹ As early as medical school, health professionals who experience higher moral distress tend to become more burned out.²⁴ One study conducted in Canada in 2017 with internal medicine residents’ identified a significant association between moral distress and depersonalisation, a key component of burn-out.²⁵

We aimed to evaluate specific risk and protective factors for burn-out and moral distress, such as lack of PPE, personal COVID-19 risk and degree of employer support among family physicians (FPs) working on the front lines during the initial wave of COVID-19 in California in 2020. The primary goal of the present study was to evaluate whether more frequent moral distress is associated with worsening burn-out during this time of increased demand for FPs. Female gender and employer support were investigated as moderators in the relationship between moral distress and burn-out.

Methods

Study design and sample

This cross-sectional study included FPs practising in California between July and August, 2020. An invitation to participate in an online Qualtrics survey was emailed to FPs through mailing lists obtained from the California Academy of Family Physicians, University of California Family Medicine Department Chairs and California Residency Directors. The survey was sent to 2177 FPs and the final sample included 218 FPs practising in a variety of clinical settings (online supplemental figure S1).

Patient and public involvement

There was no involvement of patients or the public in design, conduct, reporting or dissemination of the research.

Based on the power analysis to detect an OR of 2 with a power of 0.08, alpha of 0.05, relative proportion of worsened burn-out to not worsened burn-out of 2 and worsening burn-out proportion in the population of not reporting moral distress to those reporting moral distress of 0.3, there needed to be a total sample of 111.

Instrument

A survey including 22 multiple choice and Likert scale questions (questionnaire in online supplemental appendix) was developed by the researchers based on review of literature, our qualitative interviews and clinical experience of faculty, residents and medical students. The survey was piloted on two FP attendings, two FP residents and five medical students who had knowledge about the project and volunteered for interviews. Pilot testing resulted in rephrasing questions for greater clarity and precision as well as the deletion and addition of items.

Measures

In the pilot interviews, we asked FPs the following questions about burn-out and moral distress, respectively: ‘How has the pandemic affected your level of burnout?’; ‘Has your level of burnout improved, gotten worse, or stayed the same since the start of the pandemic?’; ‘What, if anything, would you attribute that to?’ and ‘Have you experienced moral distress or outrage in relation to COVID-19?’ These questions were used to formulate the questions in the survey about burn-out and moral distress. Additional questions in the survey included ‘What are your chief concerns during this pandemic?’ and ‘What coping strategies or resources have you used to preserve your wellbeing?’ These questions were used in conjunction with review of the literature to determine covariates to evaluate below (see online supplemental information for wording of all interview and survey questions).

While some studies published during a similar time frame incorporated validated measures, these studies were restricted to investigation of a single construct (eg, burn-out). Our interest was not in conducting an in-depth assessment of any one area but rather in identifying the relationships among these constructs. It was impractical to use existing validated scales due to the additional time and effort they would require of respondents. Knowing how FPs were overwhelmed during this early phase of the pandemic, our priority was to obtain a ‘snapshot’ of their experiences in the briefest, most efficient way possible. For this reason, we asked respondents to recall their burn-out levels prior to the pandemic and in a separate question to compare their burn-out levels currently with their prepandemic levels.

As with other studies performed during the same time period,26,28 we relied on a questionnaire format in which respondents could rapidly endorse items or not; and the use of brief Likert-type scales to capture continuous gradations. Examination of existing literature and our qualitative interviews with FPs guided our choice of items. From these interviews conducted by two medical students over zoom, certain themes became apparent such as motivations to continue practising, fear of the virus and increasing burn-out levels.

Demographic data

We collected information on gender (male, female); training level/setting (resident physician, academic-based attending, community-based attending); years in practice (0–10, 11–20, >21) and practice setting (any inpatient vs outpatient).

Motivation for treating patients with COVID-19

We inquired about whether participants cared for patients with COVID-19 (yes/no). Motivating factors based on initial interviews were assessed using multiple choice questions, similar to Alabri and Bte Siron²⁹ that asked about reasons for engaging in care of patients with COVID-19, including volunteered; job requirement; sense of duty to others/community; leadership; personal fulfilment and other.

Concerns and coping/support

Four survey questions inquired about concerns identified in the literature^{27 30 31} regarding personal risk (yes/no), transmission to others (yes/no), lack of PPE (yes/no) and lack of knowledge (yes/no).

Questions about support were divided into professional support and personal support. Professional support questions inquired about employer support (yes/no); colleague support (yes/no) and patient positive feedback (expressions of appreciation and gratitude) to physicians (yes/no). Personal support/coping questions inquired about topics identified in the literature^{26 30} including self-care (yes/no) and wellness practices (yes/no), support from family and friends (yes/no) and religious/spiritual practices (yes/no).

Positive and negative emotions experienced

Participants were asked to select from a list of 14 emotions including seven positively valenced emotions (excitement, gratitude, hope, joy, relief, connection and compassion) and 7 negatively valenced emotions (anger, anxiety, depression, fear, frustration, grief and helplessness). These emotions were generated from our qualitative pilot data as well as informal discussions with colleagues. Participants were asked to recall their top three emotions from the first 2 weeks of the pandemic and during the most recent week prior to completing the survey (online supplemental figure S2). Emotions in the week leading up to survey completion were used as a covariate in the models. Many of the items in this list were also included in an online study of physician emotions during the early phase of the pandemic.³²

Primary independent variable

Moral distress

Frequency of moral distress was assessed on a 5-point scale (‘never’, ‘rarely’, ‘sometimes’, ‘frequently’ and ‘all the time’). Previous research³³ demonstrated that it is possible to measure moral distress as a single construct using a 5-point scale. To maximise the effect size and create similar cell sizes, we further categorised frequency of moral distress into three levels: (1) ‘never to rarely’, (2) ‘sometimes’ and (3) ‘frequently to all the time’ when evaluating the relationship between moral distress and burn-out. Because our interest was in evaluating how moral distress influenced pandemic-related burn-out level, the survey assessed this construct specifically in connection to physicians’ patient care experiences during the COVID-19 pandemic.

Factors assessed as contributors to moral distress were assessed through a ranked order multiple-choice question that included factors suggested by other studies evaluating moral distress during the pandemic^{21 34} including limited resources, patients unable to say farewell to loved ones, futile care, witnessing patient suffering, low staffing, lack of PPE, lack of hardship compensation and lack of employer support.

Outcome

Burn-out

Burn-out prior to and during the pandemic was assessed from participants’ perspectives. Prepandemic burn-out was assessed on a four-level scale from ‘not burned out’ to ‘very burned out’. This forced choice format (eliminating the midpoint) resembles the approach taken by Dolan et al,³⁵ with primary care physicians. A follow-up question used three options to determine whether respondents’ level of burn-out had improved, worsened or remained the same compared with prior to the pandemic. A 3-point scale is considered valid for a determination of specific experience.³⁶ For the statistical analyses, burn-out was binarised to ‘no’/’yes’ based on the following groups: prior to the pandemic (‘not’—‘somewhat’) versus (‘fairly’—‘very’); at survey completion (‘same’—‘improved’) versus (‘worsened’). The ‘no’ burn-out group was the reference group in all analyses.

Statistical analysis

Descriptives

Cross-tabulations using χ² and Monte-Carlo statistics were used to determine statistically significant (p<0.05) differences in demographics and characteristics according to burn-out level (‘worsened burn-out’ vs ‘improved/similar burn-out’) and frequency of moral distress (‘rarely or never,’ ‘sometimes’ and ‘frequently/all the time’). Similarly, differences in frequency of negative and positive emotions, concerns and professional and personal coping strategies were determined based on burn-out level and frequency of moral distress.

Univariate logistic regression was used to determine the magnitude of effect of each variable on burn-out level (prepandemic and how burn-out level changed during the first wave of the pandemic) and frequency of moral distress. Univariate logistic regression was also used to estimate the risk of worsening burn-out according to frequency of moral distress experienced.

Multivariate models

Covariates that were found to be associated with either worsening burn-out level or more frequent moral distress (p<0.1) were included in the multivariate logistic regression model to reduce confounding. Two multivariate unconditional logistic regression models were constructed: (1) adjusting for descriptive characteristics of the sample: gender, training level and prepandemic burn-out and (2) further adjusting for predictors associated with worsening burn-out level or increasing moral distress frequency (p<0.1) during the pandemic including lack of PPE, lack of knowledge, COVID-19 risk, number of negative emotions and coping parameters including self-care, colleague support and employer support. The number of positive emotions was not included as a covariate given it was the inverse of number of negative emotions and would therefore be redundant. Two moderation analyses were performed for ‘gender’ and ‘employer support’ in the relationship between frequency of moral distress and burn-out level. Linear χ² trend analyses were performed by including moral distress frequency as a continuous variable in all analyses.

All data analyses were performed with IBM SPSS V.27.

Results

Risk and protective factors for moral distress and burn-out

Burn-out during the pandemic was more common among female than male physicians (79.4% vs 20.6%; p<0.001). Conversely, there were no observed gender differences in moral distress. Burn-out prior to the pandemic was associated with higher frequency of moral distress and worsened burn-out during the pandemic. More frequent moral distress and worsening burn-out were each associated with concerns about personal risk and lack of PPE. Higher frequency of moral distress was associated with concerns for a lack of knowledge.

Physicians who endorsed feeling supported by their employer demonstrated reduced frequency of moral distress and lower burn-out (p<0.05). There were no statistically significant differences in burn-out or moral distress reported across level of training, years in practice and practice setting (p>0.05) (table 1).

Table 1. Demographics and characteristics according to change in burn-out and moral distress frequency during the initial surge of the COVID-19 pandemic in California in 2020.

	Burn-out level			Frequency of moral distress
	Improved /same	Worsened		Never /rarely	Sometimes	Frequently /all the time
	N (%)	N (%)	P value	N (%)	N (%)	N (%)	P value
Gender
Male	65 (50.0)	14 (20.6)	<0.001	38 (46.9)	24 (37.5)	17 (35.4)	0.347
Female	65 (50.0)	54 (79.4)		43 (53.1)	40 (62.5)	31 (64.6)
Training level/setting
Resident	84 (63.6)	34 (50.7)	0.080	46 (56.8)	42 (64.6)	26 (55.3)	0.529
Attending	48 (36.4)	33 (49.3)		35 (43.2)	23 (35.4)	21 (44.7)
Years in practice
Early career (0–10)	85 (63.9)	38 (55.9)	0.269	46 (56.1)	45 (70.3)	26 (55.3)	0.151
Senior (11+)	48 (36.1)	30 (44.1)		36 (43.9)	19 (29.7)	21 (44.7)
Practice setting
Outpatient	55 (41.7)	30 (44.1)	0.740	39 (48.1)	22 (34.4)	25 (53.2)	0.104
Inpatient*	77 (58.3)	38 (55.9)		42 (51.9)	42 (65.6)	22 (46.8)
Burn-out prior to the pandemic
Not-somewhat	117 (88.0)	52 (76.5)	0.035	65 (85.5)	56 (90.3)	31 (70.5)	0.021
Fairly/very	16 (12.0)	16 (23.5)		11 (14.5)	6 (9.7)	13 (29.5)
Moral distress
Never/rarely	55 (47.0)	21 (32.3)	0.001
Sometimes	44 (37.6)	18 (27.7)
Frequently/all the time	18 (15.4)	26 (40.0)
Concerns
Personal COVID-19 risk	88 (73.9)	60 (90.9)	0.006	58 (70.7)	52 (80.0)	43 (89.6)	0.039
Transmission to others	107 (89.9)	64 (97.0)	0.144	71 (86.6)	60 (92.3)	44 (91.7)	0.476
Lack of PPE	43 (36.1)	38 (57.6)	0.005	26 (31.7)	25 (38.5)	32 (66.7)	<0.001
Lack of knowledge	54 (45.4)	36 (54.5)	0.232	35 (42.7)	29 (44.6)	31 (64.6)	0.039
Personal coping strategies
Self-care and wellness practices	113 (86.3)	53 (77.9)	0.135	68 (89.5)	48 (77.4)	35 (77.8)	0.113
Support from family and friends	114 (87.0)	53 (77.9)	0.098	62 (81.6)	54 (87.1)	35 (77.8)	0.439
Religion and spiritual practices	48 (36.6)	25 (36.8)	0.986	28 (36.8)	21 (33.9)	15 (33.3)	0.903
Professional coping strategies
Employer support	83 (63.8)	28 (41.8)	0.003	49 (65.3)	31 (50.0)	19 (41.3)	0.026
Colleague support	113 (86.9)	53 (79.1)	0.153	66 (88.0)	51 (82.3)	22 (71.7)	0.078
Patient feedback	54 (41.5)	32 (47.8)	0.404	32 (42.7)	29 (46.8)	23 (50.0)	0.724
Number of negative emotions†
0	13 (10.1)	1 (1.5)	0.006‡	8 (10.8)	3 (5.0)	0 (0.0)	0.016†
1	29 (22.5)	12 (17.6)		22 (29.7)	7 (11.7)	7 (15.9)
2	38 (29.5)	13 (19.1)		14 (8.9)	19 (31.7)	11 (25.0)
3	49 (38.0)	42 (61.8)		30 (40.5)	31 (51.7)	26 59.1)
Number of positive emotions†
0	51 (39.5)	42 (61.8)	0.011‡	31 (41.9)	31 (51.7)	27 (61.4)	0.031†
1	37 (28.7)	14 (20.6)		15 (20.3)	19 (31.7)	10 (22.7)
2	28 (21.7)	11 (16.2)		20 (27.0)	7 (11.7)	7 (15.9)
3	13 (10.1)	1 (1.5)		8 (10.8)	3 (5.0)	0 (0.0)

Majority of p values were based on χ2 statistics.

Bold values indicate p<0.05.

^*Any inpatient care.

^†Out of the top three emotions experienced in the past week.

^‡P value calculated by Monte-Carlo Statistics given small cell sizes.

PPE, personal protective equipment

The most common participant-identified reasons for increased moral distress were related to the following statements: ‘patients did not receive optimal care due to lack of resources’, ‘dying patients could not say goodbye to loved ones because of isolation precautions’, ‘inadequate compensation for altered work hours’, ‘duties and/or personal risk’ and ‘lack of proper PPE’ (data not shown).

Results of the uinivariate logistic regression showed that the likelihood of being burned out at time of survey was 3.8-fold higher for female physicians in comparison to male physicians (OR 3.86; 95% CI 1.95 to 7.62) (table 2). Consistently, female physicians were more likely to be burned out prior to the pandemic (online supplemental table S1). Physicians who were burned out prior to the pandemic were 2.25-fold (OR 2.25; 95% CI 1.05 to 4.84) more likely to be burned out during the pandemic (table 2). Physicians reporting more frequent moral distress in relation to COVID-19 were 3.78-fold (OR 3.78; 95% CI 1.73 to 8.28) more likely to report worsening burn-out during the pandemic (table 2).

Table 2. Effect size of demographics and characteristics associated with worsening burn-out level in the first wave of the COVID-19 pandemic in California.

	Worsening burn-out level
	OR	95% CI	P value
Gender
Male	Reference
Female	3.86	1.95 to 7.62	<0.001
Training level/setting
Resident	Reference
Attending	1.70	0.94 to 3.08	0.081
Years in practice
Early career (0–10)	Reference
Senior (11+)	1.40	0.77 to 2.54	0.270
Practice setting
Outpatient	Reference
Inpatient*	0.91	0.50 to 1.63	0.740
Burn-out prior to the pandemic
Not-somewhat	Reference
Fairly/very	2.25	1.05 to 4.84	0.038
Moral distress
Never/rarely	Reference
Sometimes	1.07	0.51 to 2.26	0.856
Frequently/all the time	3.78	1.73 to 8.28	0.001
Concerns
Personal COVID-19 risk	3.52	1.39 to 8.96	0.008
Transmission to others	3.59	0.78 to 16.55	0.101
Lack of PPE	2.4	1.30 to 4.44	0.005
Lack of knowledge	1.44	0.79 to 2.64	0.233
Personal coping strategies
Self-care and wellness practices	0.56	0.26 to 1.20	0.138
Support from family and friends	0.53	0.25 to 1.14	0.102
Religion and spiritual practices	1.01	0.55 to 1.85	0.986
Professional coping strategies
Employer support	0.41	0.22 to 0.74	0.003
Colleague support	0.57	0.26 to 1.24	0.157
Patient feedback	1.29	0.71 to 2.33	0.405
Number of negative emotions†	1.71	1.22 to 2.41	0.002
Number of positive emotions†	0.59	0.42 to 0.83	0.002

P values are calculated based on χ2 statistics; ‘Improved/same’ served as the comparison group for OR analyses; For ‘Yes’/’No’ variables, ‘No’ was the comparison group. Bold values indicate p<0.05.

^*Any inpatient care.

^†Out of top three emotions experienced in past week.

PPE, personal protective equipment

Physicians experiencing worsening burn-out were significantly more likely to report concerns regarding personal COVID-19 risk (OR 3.51, 95% CI 1.39 to 8.96) and lack of PPE (OR 2.39; 95% CI 1.29 to 4.44) (table 2). Employer support was associated with a 59% reduction in risk of worsening burn-out (OR 0.41; 95% CI 0.22 to 0.74) (table 2). Endorsement of more frequent negative emotions was associated with a 71% (OR 1.71; 95% CI 1.22 to 2.41) increase in worsening burn-out level. Conversely, endorsement of more frequent positive emotions was associated with a 41% decrease in burn-out level (OR 0.59; 95% CI 0.42 to 0.83) (table 2). Magnitude of effect pertaining to factors associated with more frequent moral distress can be found in online supplemental table S2.

Motivating factors

The majority of respondents (n=125) reported that their motivation to care for patients with COVID-19 stemmed from ‘professional responsibility to patients, colleagues and/or community’. Other common motivating factors were ‘job requirement’ (n=125) and ‘commitment to underserved communities’ (n=112) (data not shown).

Relationship between moral distress and burn-out

Participants reporting worsening burn-out during the pandemic were 3.78-fold more likely to report experiencing moral distress ‘frequently to all the time’ (OR 3.78; 95% CI 1.73 to 8.28). After adjusting for demographic variables including gender, level of training and prepandemic burn-out (p<0.1), the effect size was attenuated to 3.12 (OR₁ 3.12; 95% CI 1.36 to 7.13). After further adjusting for positive and negative predictors achieving significance level of p<0.1, including lack of knowledge, lack of PPE, COVID-19 risk, number of negative emotions, self-care, employer support and colleague support, the association between the highest level of moral distress and burn-out level was attenuated to 1.47 (OR₂ 1.47; 95% CI 0.56 to 3.89) and was no longer statistically significant (table 3).

Table 3. Relationship between moral distress frequency and worsening burn-out level during the initial COVID-19 surge in California in 2020.

	Worsening burn-out level
	Univariate model		Multivariate model adj.1		Multivariate model adj.2
Moral distress frequency	OR	95% CI	ORadj.1	95% CI	ORadj.2	95% CI
Never/rarely	Reference		Reference		Reference
Sometimes	1.07	0.51 to 2.26	0.99	0.45 to 2.16	0.69	0.28 to 1.72
Frequently/all the time	3.78	1.73 to8.28	3.12	1.36 to7.13	1.47	0.56 to 3.89
P trend	<0.001		0.004		0.279

adj.1: adjusted for sex, training level, prepandemic burn-out; adj. 2: adjusted for sex, training level, prepandemic burn-out, lack of PPE, lack of knowledge, COVID-19 risk, negative emotion count, self-care, employer support, colleague support; P_trend: calculated based on incorporating moral distress frequency into the model as a continuous variable. Bolded values indicate statistical significance (CI excluding 1).

PPEpersonal protective equipment

In the stratified analysis by gender, the association between moral distress frequency and worsening burn-out persisted only in females in the univariate analysis (OR 3.89; 95% CI 1.42 to 10.67) and after adjusting for level of training and prepandemic burn-out (p<0.1), higher frequency of moral distress was associated with a 3.55-fold increase in risk of worsening burn-out level (OR₁ 3.55; 95% CI 1.43 to 7.39). After further adjustment for positive and negative predictors achieving <0.1 significance level, the effect size was attenuated to 1.36 and was no longer statistically significant (OR₂ 1.36; 95% CI 0.37 to 5.03). In all analyses, there was no significant association between moral distress frequency and burn-out level among male physicians (table 4). Results of the exploratory analysis are presented inonline supplemental appendix and did not demonstrate employer supports as a moderator in the relationship between moral distress and worsening burn-out (online supplemental table S3).

Table 4. Relationship between moral distress frequency and worsening burn-out level during the initial COVID-19 surge in California in 2020 stratified by gender.

	Burn-out level now
	Female physicians			Male physicians
	Improved/same	Worsened		Improved/same	Worsened
Moral distress	N (%)	N (%)	P value	N (%)	N (%)	P value
Never/rarely	25 (42.4)	15 (29.4)	0.010	29 (51.8)	6 (42.9	0.277*
Sometimes	25 (42.4)	15 (29.4)		18 (32.1)	3 (21.4)
Frequently/all the time	9 (15.3)	21 (41.2)		9 (16.1)	5 (35.7)
	Model 0: female physicians			Model 0: male physicians
Moral distress	OR	95% CI	P value	OR	95% CI	P value
Never/rarely	Reference		0.013	Reference		0.270
Sometimes	1.00	0.40 to 2.47	1.000	0.81	0.18 to 3.63	0.778
Frequently/all the time	3.89	1.42 to 10.67	0.008	2.69	0.66 to 10.92	0.168
Ptrend	0.010			0.085
	Model 1: female physicians			Model 1: male physicians
Moral distress	ORadj.1	95% CI	P value	ORadj.1	95% CI	P value
Never/rarely	Reference		0.030	Reference		0.671
Sometimes	1.04	0.41 to 2.61	0.939	0.92	0.19 to 4.44	0.913
Frequently/all the time	3.55	1.43 to 7.39	0.016	1.87	0.41 to 8.57	0.419
Ptrend	0.017			0.192
	Model 2: female physicians			Model 2: male physicians
Moral distress	ORadj.2	95% CI	P value	ORadj.2	95% CI	P value
Never/rarely	Reference		0.399	Reference		0.687
Sometimes	0.60	0.19 to 1.85	0.373	1.06	0.10 to 11.23	0.959
Frequently/all the time	1.36	0.37 to 5.03	0.643	0.37	0.02 to 5.65	0.474
Ptrend	0.629			0.865

Model 0: unadjusted; Model 1: adjusted for training level, prepandemic burn-out; Model 2: adjusted for training level, prepandemic burn-out, lack of PPE, lack of knowledge, COVID-19 risk, negative emotion count, self-care, employer support, colleague support.

All other p values are calculated by χ2 test; P_trend: calculated based on incorporating moral distress frequency into the model as a continuous variable.

^*P value calculated by Monte-Carlo statistics.

PPE, personal protective equipment

Discussion

Many studies investigating moral distress have focused on nurses.^{16 20 37} However, recent research is revealing the experience of moral distress among physicians, specifically during the COVID-19 pandemic.^{38 39} This study explored the experience of moral distress and burn-out and their relationship among California FPs early in the COVID-19 pandemic.

Like other studies in the literature,^{25 40 41} this study demonstrated an association between moral distress and burn-out. Recent models point to the complexity of this relationship and indicate that burn-out both amplifies moral distress and is increased by moral distress.⁴² Work–home imbalance is a mediator in this relationship.⁴³

Our findings are consistent with those of other studies during the COVID-19 pandemic.^{2 21 43} Women physicians are disproportionately impacted by moral distress. A previous study suggested that women are more affected by moral distress because they have higher levels of moral sensitivity.⁴⁴ This heightened moral sensitivity may contribute to moral distress and worsening burn-out during the pandemic specifically among women physicians.

Importantly, in the multivariate model further adjusting for coping and concern variables, the effect size between moral distress and burn-out was attenuated and no longer reached statistical significance, suggesting that the relationship between moral distress and burn-out can be at least in part explained by variables including PPE, knowledge gap, COVID-19 risk, negative emotions, self-care, employer support and colleague support. Future research into how to mitigate the effects of negative predictors and enhance positive predictors is expected to improve our ability to establish interventions to limit the impact of moral distress on burn-out, specifically among women.

The importance of employer support cannot be overstated. In the present investigation, employer support was associated with a statistically significant 54% decrease in likelihood of being morally distressed and a 59% decrease in likelihood of worsening burn-out. Similarly, a study evaluating the relationship between moral distress and burn-out in the general US physician population during the COVID-19 pandemic showed that moral distress was a mediator in the relationship between employer support and burn-out.⁴⁰ According to Shanafelt et al,³ physicians at Stanford University and Mt. Sinai acknowledged the importance of leadership inquiring about their needs and well-being during the pandemic and addressing those needs. ‘Hear me’, ‘protect me’, ‘prepare me’, ‘support me’ and ‘care for me’ were common themes identified.³ These themes were echoed in the present study and in the interviews performed with a subset of 20 respondents.²³

Our findings also confirm increased levels of burn-out among female physicians both prepandemic and in the first wave of the pandemic. This is consistent with prior work demonstrating burn-out disproportionately effecting women in medicine¹³ and during the COVID-19 pandemic specifically.⁴ Since women constitute more than half of family medicine trainees, it is important to understand factors specifically contributing to their burn-out.⁴⁵ Women physicians were found to earn lower salaries at times than their male counterparts for equivalent positions.⁴⁶ Women physicians are more likely to experience harassment and mistreatment by colleagues, staff and patients, with 30% of women physicians reporting sexual harassment.⁴⁷ Additionally, women in dual-career partnerships spend on average more time in the mental organisation of home responsibilities and schedules.⁴⁸ Women physicians often manage more home responsibilities and childcare.¹¹ Increased demands on female physicians may lead to increased conflicts at home which can increase burn-out.² More research is needed to implement effective interventions specific to gender needs in the workplace of FPs.¹¹

Intervention implications

Consistent with other studies,^{2 18 38} employer support emerged as a key factor in mitigating levels of moral distress and burn-out. Addressing structural modifiable factors in the work environment such by reducing personal contagion risk, providing adequate PPE, increasing access to up-to-date information and a providing a place to voice concerns are critical.^{43 49} Encouraging self-care practices and providing peer debrief and counselling for physician employees may also be beneficial in reducing the negative emotions arising from moral distress.^{43 49} Ensuring equitable working conditions among genders would likely improve moral distress and burn-out and will help women physicians in particular feel supported through future public health crises.

Strengths and limitations

This is the only study to the authors’ knowledge to directly assess the relationship between moral distress and burn-out during the COVID-19 pandemic among FPs in California. Our assessment was done in real time, which minimises the possibility of recall bias. We demonstrated that after controlling for demographic data, the relationship between moral distress and burn-out persisted only in female physicians. Our results are consistent with another study investigating the association between moral distress and burn-out in physicians working in New York City with COVID-19 patients.⁵⁰ Our findings encourage further research to better understand how to protect front-line workers from moral distress and burn-out during public health crises, to identify what employer support is most needed, and how to specifically address the needs of female physicians who are disproportionally affected by moral distress.

One of the major limitations of this study is the low response rate. Despite our best efforts to improve our response rate, a robust return of surveys proved elusive. We attribute this to the fact that we were in the early days of a global pandemic and what many believed to be an existential crisis. Widespread anxiety and fear among these front-line FPs likely contributed to their giving a low priority to survey completion. Given the time-sensitivity of questions at time of data collection, the team proceeded with data analysis since there was an adequate sample to power statistical comparisons for our main outcome. Nevertheless, additional studies are needed to validate our findings. We hypothesise that those experiencing moral distress and burn-out may have had a lower likelihood of response given the time needed to complete the survey, therefore, our results could be an underestimation of moral distress and burn-out. If this were the case, the significance of the results would be unlikely to change. There is a possibility that FPs affiliated with University of California were more likely to respond to the survey, however, we attempted to survey all California FPs and believe that our results are generalisable to the State of California and especially to academic medical centres.

Another limitation was that validated measures of burn-out and moral distress such as the revised Moral Distress Scale⁵¹ and the Maslach Burnout Tool⁶ were not incorporated into the survey, a decision guided by our desire to make our survey as simple, short and easy to complete as possible. Further, we only surveyed California FPs possibly limiting generalisability to other states. However, we do believe that the experience of California FPs would not be different than elsewhere in the USA. Lastly, our survey did not ask respondents to identify their race/ethnicity.

Conclusion

This study identifies sources of distress and resources recommended to improve conditions for California FPs early in the COVID-19 pandemic. This study is unique because of the investigation of the association between acute moral distress and burn-out in real time during the COVID-19 pandemic among FPs in California. We cannot predict future public health disasters, yet we can protect health professionals who risk their own well-being to treat patients with increased support, adequate PPE, flexibility, childcare, debriefing, increased compensation and access to mental health and wellness resources.

Data availability statement

Data are available on reasonable request.