Exploration of Burnout, Emotional Thriving, and Emotional Recovery in an Academic Medical Center: a Mixed Methods Quality Improvement Project

Thriving (ET), and Emotional Recovery (ER). Qualitative responses were reviewed for themes using mixed induc-tive-deductive analysis. The initial coding was done by small teams with consensus obtained through large group discussions. This study was IRB-approved as non-human subjects research. Results: Of the 671 respondents, 54% met criteria for burnout (Burnout+), 65% for ER (ER+), and 61% for ET (ET+). ER+ and ET+ were present in nearly half of Burnout+ respondents (53% and 43% respectively). Several themes emerged in the qualitative analysis: workload and expectations; tangible resources; work culture; and salary/benefits, with leadership influencing each of the domains. Conclusion: Burnout, ET, and ER can co-exist within the same individual. Employee wellbeing is not adequately reflected by the binary of whether or not an individual is experiencing burnout. All employees at academic medical centers, including staff, researchers, and clinicians, are vulnerable to the same workplace factors driving burnout. Our findings have been used to target areas of intervention during the COVID-19 pandemic at our institution. We propose that other academic medical centers may have similar workplace stressors that they could assess and target for improvement. ABSTRACT


Study Design
In the summer of 2019, we distributed an anonymous, cross-sectional survey to members of the Department of Medicine at the University of Colorado (n = 1,994) with two email reminders sent over two months.This survey included both original questions and questions adapted from Adair's Thriving Culture Survey and Krol's Net Promoter Score [7,8].The total project period (including survey design, implementation, and analysis) was from August 2019 to November 2020.This quality improvement (QI) project was reviewed and approved as non-human subject research by the Colorado Multiple Institutional Review Board (# .

Setting and Participants
This survey was distributed to all members of the Department of Medicine (staff, administrators, researchers, and clinicians) across five sites, including four health systems within the University of Colorado system: the University of Colorado School of Medicine, the University of Colorado Hospital, Denver Health, Rocky Mountain Regional Veterans Administration, and National Jewish Health.

Data Collection
Our survey included 34 close-ended questions and four open-ended questions (Appendix 1).Survey data were collected and managed using a secure, web-based application electronic data capture tool (Research Electronic Data Capture [REDCap], Nashville, TN), hosted at the University of Colorado [9].

Variable Definition
The survey included questions within the domains of emotional thriving (ET), emotional recovery (ER) and Burnout, adapted from previously validated metrics by Adair et al. [7] Respondents with an average of ≥4.0 on a 5-point scale (1-strongly disagree to 5-strongly agree) to four respective domain questions would be designated as scoring positively for the ET and ER domains.For the five burnout questions, an average of ≥3.0 would indicate various gradations of burnout and are designated as Burnout+.For this analysis, we defined "positive responses" as the combination of ET+, ER+, and Burnout-and "negative responses" as the combination of ET-, ER-, and Burnout+.

Quantitative Analysis
Taking into account the survey design, we used the adjusted Wald chi-square test to explore whether ET, ER, and Burnout were associated with each other; whether the workplace environment questions were associated with ET, ER, or Burnout; and whether reports of ET, ER, or Burnout were associated with female gender; being in a clinical role; being in a research role; having a mentor or being a mentor.We used a Wilcoxon-Mann-Whitney test to identify associations between ET, ER, or burnout and years in the Department of Medicine, an ordinal variable.Odds ratios with 95% confidence intervals are reported.
Data were analyzed using SAS Enterprise Guide 8.1 (SAS Institute, Inc., Cary, North Carolina).If a variable was included in a specific statistical test, subjects with missing data were excluded from that analysis (i.e., complete case analysis).Given the number of statistical tests conducted, Bonferroni's correction for multiple comparisons was applied, with p < 0.001 considered statistically significant.

Qualitative Analysis
Each of the open-ended survey responses (Appendix 1 -Survey Instrument) were coded by a team of two to three individuals (KMJ and KTM, RN, AG, and KND; ZAC and LM; and LK-R and ZAC), including clinicians, one professional research assistant, and two statisticians.After the initial coding process, the coding teams presented their impressions to the larger qualitative study team for further discussion and development of consensus around themes.Multiple members (AK, KMJ, LM, RN, and ZAC) had prior qualitative research experience.Qualitative content analysis was conducted to categorize and interpret the text responses [10].We utilized a mixed inductive-deductive approach using both conventional content analysis and directed content analysis.Study team members assigned to a given open-ended question read all of the responses to immerse themselves in the content.The responses were then reviewed and coded independently by each coding team member.The codes applied to each response were then organized into conceptual categories.Discrepancies in coding were discussed within the coding team until a consensus was reached.Evaluation and refinement of the conceptual categories and codes grouped therein continued until consensus across all open-ended responses was reached between coding team members.The full qualitative team (AG, AK, KND, KMJ, KTM, LK-R, LM, RN, and ZAC) met to discuss the findings from each coding team.Consensus across all team members was reached through independent review of the responses and discussion of themes identified.A synthesis of results emerging from each analysis was summarized and question-specific concepts were compared across the survey questions to identify overarching themes.

Study Participants
Of the 1,994 invited, 671 persons completed this survey (response rate 34%) (Table 1).Thirty-three percent of respondents identified as male, 61% as female, with 5% who preferred not to answer, identified as non-binary, or did not respond (Table 1).Respondents included clinical faculty, advanced practice providers, research faculty, research staff, administrative staff, graduate students, and clinical fellows.Sixty-four percent of respondents served in a clinical role; 49% served in a research role.Two-thirds of respondents were in a mentorship relationship.Among respondents, 61% endorsed measures consistent with emotional thriving (ET+), 65% endorsed measures consistent with emotional recovery (ER+), and 54% endorsed measures consistent with Burnout+.

Quantitative Results Emotional Thriving, Emotional Recovery, and Burnout
When comparing the subgroups of respondents who endorsed ET+, ER+, or Burnout+ with those who did not, we observed no significant differences in demographics.Specifically, we did not observe a significant association between female gender; years in the department; serving in a clinical role; serving in a research role; or being in a mentorship relationship with report of ET, ER, or Burnout (Table 1).

Perceptions of Workplace Environment
Responses to queries surrounding workplace environment are depicted in Figure 1 (select workplace environment questions) and Appendix 2 (all questions).The statement with the  1).The remainder of the workplace environment questions ranged between 50-80% agree/strongly agree among the ET+ and ER+ subgroups.
An example of a mixed response is seen with "We have a 'We are in it together' attitude," in which ET+ and ER+ subgroups had high rates of agree/strongly agree (79% and 72% respectively) while Burnout+ had low agreement/strongly agree (49%).
The Burnout+ subgroup rated all statements with <50% Agree/Strongly Agree except for "The people I work with care about me as a person" (73% Agree/Strongly Agree as above) and "Being a part of [this department] is a source of professional pride" (51%) (Appendix 2).We observed that so-called "positive responses" (ET+, ER+, Burnout-) tended to show very similar rates of agree/strongly agree with each other.Similarly, "negative responses" (ET-, ER-, Burnout+) showed similar rates of agree/strongly agree with each other.

Explicit Recognition of Workplace Contributions
When asked how recently they had received explicit recognition of the value of their workplace contributions, respondents endorsing "negative responses" had slightly less recent experiences than those with "positive responses" (Figure 2).ET+ was significantly associated with more recent explicit workplace recognition (p < 0.0001).We observed a signal towards significance among persons Burnout+ and ER-reporting less recent recognition from their colleagues (p = 0.0085 and p = 0.0142, respectively).

Qualitative Results
Qualitative analysis of the four open-ended questions in this survey expanded our understanding of these quantitative findings.Table 2 highlights representative quotes for each of these themes, both positive and negative.

Workload and expectations
Excessive workload and unrealistic expectations emerged as common stressors amongst all respondents regardless of clinical, research, staff, or administrative roles.Significant tensions were identified between job expectations and what the respondents were actually capable of completing.The sense of feeling overwhelmed from unrealistic expectations and workload was pervasive.Conversely, clear expectations and reasonable workloads (when present) led to greater job satisfaction and fulfillment.

Tangible Resources
The perceived lack of necessary resources was another commonly identified stressor (e.g., inadequate staffing, paucity of administrative support, insufficient physical workspaces).Respondents expressed the importance of adequate funding and protected time to support other initiatives essential to the academic mission of these institutions.While most comments reflected a request for more resources, some respondents reported improved workplace satisfaction when appropriate resources were in place.

Work Culture
Respondents highlighted the importance of a work culture in which individuals are recognized for their contributions and feel valued by their coworkers and leaders.Respondents noted that opportunities for growth and career development led to professional fulfillment.Conversely, some individuals reported a lack of support from leaders and colleagues; increased stress; and dissatisfaction.

Salary/benefits
Noncompetitive salaries and the lack of generous parental and family leave policies were presented as notable stressors.Staff respondents discussed the challenges presented by salaries not reflecting the high cost of living in this metropolitan area.Faculty respondents described inadequate financial compensation for non-clinical work that is expected by the academic mission of their institutions.Conversely, some comments cited job flexibility and good benefits at the institution as contributing to their workplace satisfaction.

Leadership
Respondents shared the importance of leadership in influencing workload, communication, resource allocation, and expectation setting.Additionally, respondents highlighted the need for transparency in hiring and promotions processes to ensure fairness and equity.Leadership was identified as a crucial component in whether departmental culture was supportive, safe, and collegial.As one respondent aptly stated: "Poor leadership at [their workplace] has a corrosive effect on morale."(Respondent 346, Man/he/him, MD/PHD faculty)

DISCUSSION
This survey reflects the views of nearly 700 members of a large clinical department spanning 5 collaborating institutions, including clinical providers, researchers, administrators, and staff.Nearly two-thirds endorsed measures consistent with ET and ER, while over half endorsed symptoms of Burnout.We observed that ET, ER, and Burnout can simultaneously co-exist in the same individual, reinforcing Adair et al's findings that thriving and burnout are not necessarily opposite sides of the same coin [7].Additionally, our respondents cover a wide variety of roles, yet they report similar rates of ET, ER, and Burnout as each other, suggesting that what appears to be an individual's own experience of ET, ER, and Burnout highlight the impact of a shared environment [5,11,12].
The degree of uniformity to responses of workplace environment questions can provide leaders with unique insights regarding domains upon which to focus interventions.In this survey, the statements of "we have enough staff to handle the workload" and "our culture makes it easy to learn from the mistakes of others" both had similarly low levels of agreement regardless of the presence of ET, ER, or Burnout.Targeting these issues that were perceived as problematic by a great majority is a natural next step to improve the wellbeing of all persons in this department.Conversely, the workplace environment domains that reveal mixed experiences suggest potential vulnerabilities within subpopulations necessitating further evaluation to develop tailored interventions to improve their wellbeing.
Our qualitative and quantitative data highlight the overlap between individuals feeling valued and recognized by their colleagues and their wellbeing metrics.Recognition has been cited as an important factor for employee retention and to protect against burnout [13].Positive recognition experiences and a culture of respect can mitigate the impact of burnout among employees and foster wellbeing [14].Our findings show that leaders can use the combination ET, ER, and Burnout as indicators of workplace culture to better quantify the need for increased recognition and support in their departments.
Our qualitative analysis identified several important domains which impact wellness and burnout.Those domains include workload and expectations; tangible resources; work culture; salary and benefits, and lastly leadership which impacts all the other domains.Our findings are consistent with Shanafelt and Noseworthy's 2017 study which found that "workload and job demands", "efficiency and resources", and "culture and values" are among a larger list of drivers of burnout [2].Our respondents described increased workplace satisfaction when workload and expectations were reasonable.More commonly, however, respondents described experiences of over-work and unattainable expectations in concert with inadequate resources and compensation as contributors to their burnout.This reinforces the importance of balancing clear and reasonable expectations with fair compensation, appropriate resources, and supportive leadership [2,[15][16][17].
Organizational interventions (e.g.promoting effective leadership) are far more effective in decreasing provider burnout than individually-focused interventions [17,18].Strong leaders need to advocate for appropriate resources, competitive salaries, and comprehensive benefits for their departments.Leadership crucially sets the tone for workplace culture, ensuring it remains supportive, safe, and collegial [2,16,17,19].
Studies published prior to the COVID-19 pandemic identified burnout of clinical providers as deeply prevalent and impactful, posing a threat to the quality of patient care, organizational health, and provider wellbeing [3,20,21].This is even more relevant with increasing rates of burnout and distress experienced throughout all employee populations in AMCs as a result of the pandemic.
Mounting pressures in U.S. healthcare systems have been identified as contributing to an imbalance between the demands of clinical jobs and the resources available to those clinicians [22].However, this survey highlights the importance of balancing demands and resources across AMCs, extending beyond clinicians [21] to include researchers, administrators, and staff.
Our study is unique in that it reflects the experience of all of these subpopulations.Prior to this, much of the literature surrounding wellness in healthcare settings has typically focused upon clinical providers [1,5,15].Our findings suggest that no single metric is sufficient to capture the wellbeing of the respondent; these metrics should be considered as a composite.While it is important to address wellbeing at an individual level, it is also incumbent upon leadership to move the discussion on wellness and thriving away from individual self-care and towards progressive organizational change, addressing resources, supporting a safe culture, and communicating clear and fair expectations to their employees [17,23].

Limitations and Strengths
This QI survey reflects the experiences of employees in a Department of Medicine across 5 sites, including 4 health systems, and thus may not be generalizable to other settings.This survey did not include racial or ethnic demographics, which prevents identifying differences amongst underrepresented minority members.Second, this survey was conducted before the COVID-19 pandemic, which has increased workloads and financial strain on our healthcare systems and is disproportionately impacting racially diverse communities [24][25][26].Thus, our data represents a pre-pandemic baseline, but does not reflect the impact of this pandemic on the members of our department.Due to the voluntary nature of this survey, we may be missing the sentiments of many individuals.Additionally, this survey was not designed to allow for subgroup analyses by role (e.g., staff or faculty, clinical, research, education, or administration) or by site as many faculty members work in multiples sites and hold multiple roles simultaneously in this department.Those limitations notwithstanding, this survey captured the experiences of 671 employees spanning 5 sites (including 4 health systems) across ranks and roles including administrative staff, research staff, clinicians, and researchers.
To our knowledge, this is among the first published surveys to include all roles in a large academic medical department.Thus, our findings add the important perspectives of non-clinicians and non-researchers to the medical literature.Additionally, the combination of qualitative and quantitative data provides a more nuanced understanding of the experiences of our workforce.We believe that the results of this study are even more relevant with the ongoing COVID-19 pandemic.This pandemic has worsened and revealed pre-existing weaknesses in healthcare systems while simultaneously increasing workload, personal stressors, and systemic stressors [22][23][24].Investment in the wellbeing of all department members is now even more important to allow these institutions to continue to provide much-needed healthcare and scientific advancement at this time.

CONCLUSION
Inclusion of all roles is necessary to fully understand the wellbeing of employees and the workplace experience.The quantitative responses to this survey showed high levels of the positive attributes of ET and ER simultaneously coexisting with burnout within individuals and across this department.
Wellbeing metrics are more complex than a simple binary of whether someone is experiencing burnout.We observed that directly assessing the workplace environment can provide valuable insights regarding domains to target (more uniform responses) and domains requiring further evaluation (more varied responses).
Qualitative analysis of free answer questions allow a deeper understanding of the experience of employees.The themes arising from these questions highlight the importance of leadership to address the balance of workload and expectations, being transparent and equitable regarding salary and benefits, and exploring the availability and distribution of tangible resources.We saw that these themes are important universally in our AMC amongst clinicians, researchers and staff.These themes existed prior to the onset of the COVID-19 pandemic and have become even more relevant in the presence of increasing burnout and distress expressed by healthcare workers since the onset of the pandemic.Our hope is that the lessons learned from this QI survey will resonate with other AMCs, informing interventions to improve their workplace culture and the wellbeing of all their employees.
Funding Source: The project was supported by the University of Colorado Anschutz Medical Campus Department of Medicine.
Conflicts of Interest: JAC has participated as an educator and conference planning committee member for the Institute of Functional Medicine, payments for which were made directly to his institution.He is also the owner of NYOC, LLC, which is an inactive corporation initially established to provide educational tools and consulting for patients, but has had no work done nor financial activity for the past 3 years.
Otherwise, to the best of our knowledge, no conflicts of interest exist, financial or otherwise.

Table 1 :
Characteristics of Faculty and Staff Respondents C Cl li in ni ic ca al l R Ro ol le e R Re es se ea ar rc ch h R Ro ol le e M Me en nt to or rs sh hi ip p R Re el la at ti io on ns sh hi ip p ( (M Me en nt to or r o or r M Me en nt te ee e) ) 1 Clinical Counseling or Social Work, Clinical Coordination or other Clinician, Education and QI, Management, Researchers and Principal Investigators 2 Fellow, Clinician MD/DO/APP, MD/PHD Faculty, Physician Scientist 3 Research assistant, Lab personnel, PhD candidate/Post-Doc/PhD, MD/PHD Faculty, Physician Scientist MD/DO: medical doctor/doctor of osteopath; APP: advanced practice provider; PhD: doctor of philosophy

Table 2 :
Representative Quotes by Theme from Open-Ended Survey Question