Longitudinal Prospective Study of Emergency Medicine Provider Wellness Across Ten Academic and Community Hospitals During the Initial Surge of the COVID-19 Pandemic

Background: While the coronavirus (COVID-19) has had far-reaching consequences on society and health care providers, there is a paucity of research exploring emergency medicine (EM) provider wellness over the course of a pandemic. The objective of this study was to assess the well-being, resilience, burnout, and wellness factors and needs of EM physicians and advanced practice providers (APPs) during the initial phase of the COVID-19 pandemic. Methods: A longitudinal, descriptive, prospective cohort survey study of 213 EM physicians and APPs was performed across ten emergency departments in a single state, including academic and community settings. Participants were recruited via email to complete four weekly, voluntary, anonymous questionnaires comprised of customized and validated tools for assessing wellness (Well Being Index), burnout (Physician Work Life Study item), and resilience (Brief Resilience Scale) during the initial acceleration phase of COVID-19. Univariate and multivariate analysis with Chi-squared, Fisher’s Exact, and logistic regression was performed. Results: Of 213 eligible participants, response rates ranged from 31 to 53% over four weeks. Women comprised 54 to 60% of responses. Nonrespondent characteristics were similar to respondents. Concern for personal safety decreased from 85% to 61% (p<0.001). Impact on basic self-care declined from 66% to 32% (p<0.001). Symptoms of stress, anxiety or fear was initially 83% and reduced to 66% (p=0.009). Reported strain on relationships and feelings of isolation affected >50% of respondents initially without significant change (p=0.05 and p=0.30 respectively). Women were nearly twice as likely to report feelings of isolation as men (OR 1.95; 95%CI 1.82–5.88). Working part-time carried twice the risk of burnout (OR, 2.45; 95% CI, 1.10–5.47). Baseline resilience was normal to high. Provider well-being improved over the four-weeks (30% to 14%; p=0.01), but burnout did not significantly change (30% to 22%; p=0.39). Conclusion: This survey of frontline EM providers during the initial surge of COVID-19 found that despite being a resilient group, the majority experienced stress, anxiety, fear, and concerns about personal safety due to COVID-19, with many at risk for burnout. The sustained impact of the pandemic on EM provider wellness deserves further investigation to guide targeted interventions.


Background
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) aka COVID-19 pandemic has affected nearly every aspect of daily life. Beyond widespread stressors including social isolation, nancial hardship, and institutional disruptions, 1 health care providers have faced additional workplace demands like the need to synthesize the deluge of SARS-CoV-2 information and react to rapid clinical practice changes.
The drastic changes during a pandemic can impact the psychological and physical well-being of frontline emergency care providers. 2,3 As seen in prior pandemics, this can lead to negative psychological outcomes including acute and traumatic stress. 2,4−6 Reports from China and Canada during COVID-19 revealed high rates of distress, insomnia, anxiety, and depression among frontline health care providers. [3][4] These vulnerabilities were further articulated when the United Nations issued a policy brief "COVID-19 and the Need for Action on Mental Health." 7 For frontline emergency medicine (EM) physicians, wellness is complex, personal, and multifactorial, including constructs like well-being and burnout. 8,9 Prior to COVID-19, burnout among U.S. EM physicians was already high, in excess of 50%. [10][11][12] Physician burnout is associated with mental health disorders, interpersonal relationship di culties, substance abuse, and attrition from the profession. [13][14][15][16][17][18][19][20] Physician burnout negatively impacts quality patient care, patient satisfaction, and the healthcare system as a whole. 8,17−22 However, physician well-being is more than just the absence of burnout. 9,23 Well-being involves a myriad of in uences that produce an overall quality of work and life that contribute to the realization of one's full potential. 23 While studies are still emerging, there is a paucity of research exploring emergency medicine burnout and wellbeing during a global pandemic 3,4,24,25,43 Such information is necessary to inform the decisions of healthcare institutions as they build capacity, support their frontline workforce, and react to the impact of this pandemic. The objectives of this study were to assess the state of EM physician and advanced practice practitioner (APP) wellbeing, resilience, psychological distress, and burnout during the initial surge of the COVID-19 pandemic, to identify factors and needs associated with provider wellness in order to guide interventions.

Setting and Population
This longitudinal prospective cohort survey study assessed the demographic characteristics, wellness, psychological distress, burnout, resilience, and ongoing needs of EM physicians and APPs (i.e., nurse practitioners and physician assistants) on the frontlines of the COVID-19 pandemic. The eligible study population included 157 employed EM physicians and 56 APPs working in 10 emergency departments within a statewide healthcare system including 7 community hospitals and three academic teaching hospitals. APPs work collaboratively as a part of the emergency department team and do not have independent practice in Indiana. Nine of the participating hospital dashboards were available and accessed for comparative emergency department patient encounters for 2019 and 2020 during the study dates. The study protocol was reviewed and deemed exempt by the Indiana University IRB (Protocol l#2003971025) at Indiana University School of Medicine (IUSM).

Survey Development and Design
The survey instrument was a combination of customized questions, designed speci cally for this study, and three validated tools for assessing wellness (Well-Being Index), burnout (Physician Work Life Study item), and baseline resilience (Brief Resilience Scale) during the initial acceleration phase of COVID-19 [ Table 1]. After piloting a 22item draft survey of customized questions to 10 providers from the target population, a modi ed Delphi technique was used to ensure expert consensus for inclusion of both custom and validated survey tools to assess provider characteristics, COVID-19 related experiences (e.g., quarantine, safety concerns), well-being, burnout, and resilience. 26,27 Race/ethnicity demographic questions were omitted to ensure respondent anonymity given the lack of racial and ethnic heterogeneity of the population of interest. The nal 47-item survey was comprised of multiple choice, scaled rating, and yes-no questions with branching logic to minimize respondent burden [Supplement File 1]. The 6-items of this scale assess resilience, de ned as "the ability to bounce back or recover from stress." It uses a 5point Likert scale.

X
Well-Being Index (WBI) is a validated, 9-item instrument assessing six dimensions of distress and well-being. Higher WBI total scores [range: -2 (lowest risk) to 9 (highest risk)] re ect greater distress, lower meaning in work, and lower satisfaction with work-life balance. Scores at or above abnormal risk thresholds (≥ 3 for physicians and ≥ 4 for APPs) are associated with increased burnout, depression, decreased quality of life, and fatigue. WBI scores correlate other signi cant events such as medical error and intent to leave the job or profession. [28][29][30][31][32] Physician Work Life Study (PWLS) burnout item is a validated tool asking participants to rate their level of selfde ned burnout (1 = "I enjoy my work"; 5 = "I am completely burned out"). Scores were dichotomized (1 or 2 = no burnout; 3 to 5 = burnout symptoms present). Burnout measured via the PWLS predicts high emotional exhaustion, lower work satisfaction, higher self-reported medical error, and greater intent to leave the profession. 33 Both univariate and multivariate analysis was completed. Frequencies and percentages were summarized by group for categorical variables and continuous variables were summarized by group using the median and range. The proportions of subjects were compared using Chi-square, Fisher's Exact and Wilcoxon tests. A 5% signi cance level was used for all tests, p < 0.05. Correlation analysis using logistic regression was performed across time points and at the rst and last time points to evaluate the associations between PWLS (burnout), the WBI risk, and the odds of responding "Yes" to speci c COVID-19 survey questions, reported as odds ratio (OR) with 95% con dence intervals (CI).

Demographics
Of the 213 EM providers invited to participate in the study, 157 (74%) were physicians and 56 (26%) were APPs.
There were 348 total responses across 4 weeks of data collection. Weekly response rates ranged from 113 (week 1, 53%) to 66 (week 4, 31%). Females (range 54-60%) responded more frequently than males [ Table 2]. The participant group has similar occupational characteristics as the non-respondent group [Supplement Table 2]. There was a decrease in the number of overall emergency department patient encounters (i.e., volumes) during the study period in 2020 across nine study sites as compared to 2019 [ Figure 1]. In order to provide context for the wellness survey response trends, the daily COVID-19 positive case counts in the state as well as selected events and wellness initiatives during the study period are displayed in Fig. 2.

Concerns for Safety and Self-Care
Initially, 85% (96/113) of respondents reported acute concerns about their personal or family members' safety, which decreased to 61% (40/65; p = 0.0004) by week four. When polled for 'interventions that would make you feel safer,' additional Personal Protective Equipment (PPE) was the most frequent response, followed by hospital provided scrubs and showers [Supplement Table 3]. Basic self-care was impacted in 66% (75/113) of EM providers at week one, with signi cant weekly decreased over the study period (p < 0.001). Reduced ability to care for dependents ranged from 29-20%, without signi cant change over time (p = 0.21) [ Table 3].  Table 3].
The self-reported need for mental health resources was low (range: 4-12%). In contrast, 30% (28/113) of respondents reported a desire for stress reduction resources initially, athough this level dropped signi cantly by week four to 9% (4/65; p = 0.009). Most respondents reported having access to a mentor, colleague, friend, or family member to help them decompress (range: 81-93%). [Supplement Table 3].

Work Responsibilities and Academic Productivity
Initially, 59% (67/113) of respondents experienced additional work responsibilities or hours due to COVID-19. This decreased by week four to 34% (22/65; p = 0.001). Around one third of respondents experienced loss of academic productivity with no signi cant change over time (p = 0.14). The majority of EM providers felt supported by leadership, which was sustained during the study period, starting at 89% (100/113) and rising to 99% (64/65) by week three (p = 0.01) [ Table 3].

Resilience, Well-Being and Burnout Scales
The majority of respondents reported normal to high baseline resiliency on the BRS. At week 1, 30% (34/113) of respondents screened positive for burnout on the PWLS and "at risk" on the WBI. While burnout did not signi cantly change across 4 weeks (p = 0.39), the percentage of respondents "at risk" on the WBI decreased signi cantly to 14% (9/65; p = 0.01). [ Table 4].

Associations among Variables
In the study, female respondents were almost twice as likely to report feelings of isolation than their male colleagues [odds ratio (OR), 1.95; 95% con dence interval (CI), 1.82-5.88]. Academic hospital-based respondents were twice as likely to report increased work responsibilities compared to providers at community sites (OR, 2.04; 95% CI, 1.27-3.27). Mid-career EM providers were at three times greater odds of having their self-care impacted by the pandemic (OR, 3.03; 95% CI, 1.40-6.58) and over four times more likely to report stress, anxiety, or fear than their early-career counterparts (OR, 4.38; 95% CI 1.27-15.05). Respondents in practice 11-15 years had over twice the odds of screening "at risk" on the WBI than those in practice 1-5 years (OR, 2.50; 95% CI, 1.04-6.04). Working part-time carried twice the risk of burnout on the PWLS (OR, 2.45; 95% CI, 1.10-5.47) [ Table 5; Table 6].   Table 6].

Discussion
This prospective longitudinal survey of EM physicians and APPs at academic and community emergency departments was conducted during the acceleration phase of the COVID-19 pandemic in Indiana when cumulative positive COVID-19 cases increased statewide by over 4,400%. 39 The study had several interesting ndings. First, the majority of frontline EM providers experienced high baseline levels of stress, anxiety, fear, concerns for safety, and relationship strain due to COVID-19. Of note, despite coinciding with the acceleration phase of the pandemic, EM providers reported an improvement in each of these domains, although concerns persisted. Second, despite being a resilient group, many providers were at risk for burnout. Third, feelings of isolation endured during the study and were higher for women. Fourth, several subgroups, including women, part-time, mid-career, and academic providers had greater odds of COVID-19 impacting their wellness domains. And lastly, our study was able to identify speci c needs of our EM providers (e.g., PPE, scrubs, showers, childcare options, mental health resources) that guided the advocacy work and targeted interventions by our department and institution. [ Figure 2 and Supplement Table 1].
The high level of concern about personal safety and the safety of family and dependents found in our study is consistent with the ndings from prior pandemics and at other geographical locations affected by COVID-19. [41][42][43][44][45] Our study adds the early longitudinal perspective that safety concerns among our EM providers steadily improved. This reassuring nding and may be due, in part, to the fact that speci c wellness needs were addressed early and resources made available quickly via departmental, institutional, and community initiatives. Examples of interventions include the increased availability of PPE, hospital-supplied scrubs and onsite showers, access to sleep space either on site or at local hotels, and community-based laundry services. As Chen et al. reported, of these interventions, adequate PPE and rest were more important to frontline healthcare providers early in the pandemic to reduce stress than access to a psychologist. 46 Consistent with other studies, a majority of EM providers in our study reported feelings of psychological distress including anxiety, stress, and fear due to COVID-19. 3-4,43−46 In addition, about one third to one half reported increased strain on their relationships (with partner, children, and co-workers). Factors at the individual and systems levels may contribute to these ndings. Each EM provider is beholden to the pressures of their collective communities during a pandemic such as uctuations in childcare and school situations, nancial stress, and social isolation. Confounding this, EM providers may experience internal role con ict with regard to their work duties as frontline health care providers and their personal responsibilities to care for family or depedents. 2,46 Along with these pressures, EM providers often worry about the lack of treatment options or ventilator capacity for patients, bear the fear of infecting family or friends, and face thoughts of their own mortality or that of their colleagues and loved ones. 25,41,46 The persistent feelings of isolation experienced in our study is consistent with reports from the 2003 SARS outbreak. During SARS, frontline healthcare workers were at greater risk of feeling isolated than the general public, as well as the associated negative mental health consequences. 4 During COVID-19, the degree of social distancing and isolation is unprecedented on this generation, and the toll it will take on mental health is not fully evident.
Our study found that the well-being of certain subgroups of providers (i.e., women, part-time, mid-career, and academic providers) may be at greater risk during a pandemic. For example, in our study, women were twice as likely to a rm feelings of isolation. Other studies have shown that female gender is associated with more severe symptoms of depression, anxiety, or distress due to the COVID-19 pandemic. 3,43,49−52 It is imperative to consider the unique professional and personal situations of these subgroups in order to target support and resources. Additional investigation into how gender, employment status, or stage of career affects or is affected by the complex circumstances facing frontline providers is warranted.
Burnout is a syndrome resulting from chronic workplace stress and is characterized by emotional exhaustion, cynicism or depersonalization from one's job, and reduced e cacy, which suggests little malleability for positive change under conditions of heightened stress, such as a pandemic. 53 Our study con rmed this, as burnout, using the PWLS item, remained steady at about one quarter of EM providers. Interestingly, this burnout rate did not worsen and is less than the national average for the specialty of EM. This could be in part due to the normal to high baseline resilience scores or from other individual or system factors noted below.
It is curious that during our study the burnout rates remained steady using the PWLS, while provider well-being improved using the WBI, which also contains a burnout item. A likely explanation lies in the difference between the PWLS and WBI scales. The PWLS evaluates the self-determined presence of burnout via a single question. The WBI calculates burnout risk more broadly as one of many domains of wellness. Burnout amongst providers assessed by the WBI may not have changed over time, while other domains of wellness showed improvement, resulting in a lower "at risk" frequency.
Our study suggests that using brief validated scales, such as the PWLS and/or WBI, can provide valuable guidance to institutions before and during pandemics. Of particular note, the EM providers who endorsed burnout on the PWLS single item and screened "at-risk" on the WBI carried greater risks of endorsing concerns about personal safety, impact on dependent care, relationship strain, additional work responsibilities, and feelings of isolation due to COVID. Additionally, EM providers who screened at-risk on the WBI had a twenty times higher odds of reporting stress, anxiety, and fear due to COVID-19. These signi cant correlations support the utility of these tools in identifying early distress among frontline EM providers and guiding system-based interventions and resource allocation.
The signi cant improvement in provider well-being on the WBI may be related to both individual and system factors. The timely response and culture of the department and institution may account for mitigating factors, such as the presence of social support, leadership support, safety needs being met (e.g. PPE, scrubs, showers), nancial security, childcare options, and access to mental health support. Of particular note, 81-93% of providers reported having a friend, mentor, colleague, or family member to help them decompress, which offers the consistent presence of social support that is found to shield against negative life stress. 47,48 Additionally, the vast majority of respondents reported feeling supported by leadership (89-99%), which was sustained during the study period. The formation of a departmental wellness taskforce prior to the acceleration phase of the pandemic, whose objective was to evaluate wellness and elicit actionable items during the early stages of the pandemic, may have contributed, in part, to the decreased distress, and may be a key strategy to improve provider well-being.
The literature indicates that system, leadership, and community responses in conjunction with effective communication are crucial prior to and during a pandemic, as this can mitigate negative psychological responses. 2 In our provider group, no frontline health care providers experienced salary cuts or furloughs, despite decreased ED volumes, unlike numerous other hospital systems across the country. The tremendous outpouring of appreciation and support for frontline healthcare providers from the community may also have contributed to the improved wellness factors. Gratitude from patients, families and the community may have offset feelings of burnout and increased job satisfaction. 42 Emergency department volumes signi cantly decreased during the study period, which may have improved wellness factors. It is also possible, though not measured in this study, that providers' sense of control and perceived knowledge of the virus improved with time, which has been shown to mitigate negative effects of a pandemic on emotional wellness. 54 Limitations This study had several limitations. The study design used an online survey instrument, which is susceptible to response biases including self-selection (voluntary response bias), the sample size (nonresponse bias), as well as the survey length and competing surveys (fatigue bias). Although our response rate is similar to other online survey response rates, it remains a potential limitation. 55 Due to anonymous data collection we could not assess individual-level change over time. Another limitation is the lack of race/ethnicity demographic data, which is needed to further analyze the association of race/ethnicity on provider wellness. Due to the time-sensitive need for the survey, the COVID-speci c questions were not validated against other measures, and therefore only have face validity. The WBI asks questions regarding symptoms "over the last month," however, the instrument was used on a weekly basis and therefore may not be sensitive to that degree of change. 56

Conclusions
This study of frontline EM physicians and APPs during the initial surge of the COVID-19 pandemic in Indiana found signi cant levels of stress, anxiety, fear, concerns about safety, and relationship strain, all of which improved but endured. Additionally, while providers were a resilient group, feelings of isolation and burnout persisted, but did not signi cantly worsen. Well-coordinated departmental and institutional efforts that prioritize wellness can address the central need of frontline workers in ways that may mitigate or buffer acute stressors. The long-term effects of the pandemic on frontline providers warrants additional study, along with interventions that may mitigate psychological distress and burnout.