Experiences of a Health System’s Faculty, Staff, and Trainees’ Career Development, Work Culture, and Childcare Needs During the COVID-19 Pandemic

Key Points Question What are the associations of the COVID-19 pandemic with career development and what are the work culture and childcare needs of employees and trainees? Findings In this survey study, most participants with children did not have childcare fully available and many considered leaving the workforce and were worried about their career. Being female with children or having a clinical job role was associated with consideration for leaving the workforce and reducing hours. Meaning These findings suggest that a substantial number of employees and trainees experienced major stress and work disruptions because of the COVID-19 pandemic.


Introduction
As a result of the COVID-19 pandemic, 42% of US workers, including many in academic medical centers, were quickly transitioned to working from home in March 2020, and many were simultaneously required to provide childcare and substantial assistance with schoolwork for children during their workdays. 1 Employed women, in particular, were likely to face greater burdens because they spend 22% more time on unpaid household and care work compared with their male counterparts, with Black or African American and Latina mothers spending nearly twice as much time as men on unpaid housework. 2 The pandemic has led mothers in heterosexual relationships to reduce their work hours 4 to 5 times more vs fathers because of the pressures of having children at home. 3 Notably, women comprise 74.9% of hospital employees, 4 many of whom are essential clinical workers; the extent of the needs and difficulties for these workers during the pandemic remain largely unknown. Employees and trainees who must present to work or training in person may face new childcare expenses for school-age children, resulting in a higher financial burden.
In addition, for those who work in a clinical setting, several studies have shown that there is substantially higher stress for health care workers during the pandemic compared with before the pandemic. [5][6][7][8][9][10] These life changes are associated with serious concerns about the impact on the careers and well-being of faculty, staff, and trainees. Several recent studies 5 have examined the effects of burnout, stress, depression, and anxiety on frontline medical staff during the global pandemic. Numerous studies 6,7 across the globe have demonstrated the substantial burnout of frontline workers, and additional studies [8][9][10] conducted in the US have shown similar themes.
However, many of these studies have been limited by only examining either frontline workers or physician trainees. Most studies also do not address important family-work balance issues, such as childcare needs during the pandemic, which contribute greatly to the stress and burnout of staff. 8 To our knowledge, no previous studies have examined the work-life needs of both the clinical and nonclinical staff of a medical center.
The primary aim of this survey study was to evaluate the association of the global pandemic caused by SARS-CoV-2 with productivity, career development, and likelihood of leaving the workforce or reducing hours by employees in a tertiary care academic medical center. This study also identified factors (ie, gender, race/ethnicity, job role, and working in a clinical setting) and moderators associated with the aforementioned outcomes. The secondary aim was to describe and identify dependent care and work culture needs to inform institutional policy changes.

Methods Participants
A Qualtrics survey was distributed via email to all 27 700 faculty, staff, and trainees (eg, professional and graduate students, residents, medical fellows, and postdoctoral fellows) at University of Utah Health between August 5 and August 20, 2020, as a quality improvement initiative to inform institutional leadership on how best to support employees. It was sent via a university-wide listserv in which recipients clicked a link to opt into the study anonymously. Because anonymity was imperative to allowing employees to speak freely, there was no system to track differences between responders and nonresponders. The initial email was sent from the CEO and SVP of the University of Utah Health, with 1 reminder email. The University of Utah institutional review board classified the study as exempt and informed consent was waived because the responses were anonymous and because the study presented minimal risk to the participants. Our results follow the American Association for Public Opinion Research (AAPOR) reporting guideline for survey studies.

Survey Instrument
The survey was developed by a multidisciplinary team of psychologists, physicians, trainees (students, residents, and fellows), and staff. Before launching the survey, a convenience sample of approximately a dozen faculty, staff, and trainees (3-4 people per role) took the survey and provided feedback to identify confusing wording and to suggest additional questions. The survey was mostly quantitative but also included several open-ended questions. The survey took 10 to 20 minutes to complete (contents included in this article are in the eAppendix in the Supplement).
Primary outcome measures were 4-fold, each scored on a Likert scale from 1 (low) to 5 (high).
Participants were asked if, as a result of the pandemic, they had considered leaving the workforce, considered reducing their hours, or experienced reduced productivity and whether their career had been impacted. Secondary outcomes included what type of work culture adaptations they would find helpful while working during the pandemic (participants rated their perceived effectiveness of 8 solutions on a 4-point scale from 1 [not helpful at all] to 4 [extremely helpful], with an option for other or not applicable). For those who indicated they had dependent children, additional questions inquired about childcare needs, again with participants rating the effectiveness of potential services (assuming they were provided at an affordable cost) on a 4-point scale from 1 (extremely unlikely) to 4 (extremely likely), with an option for not applicable. Finally, participants completed standard demographic measures. Participants self-reported their race and ethnicity using US Census questions (multiple responses were allowed). Race and ethnicity were measured because we had a priori hypothesized that individuals from groups that are underrepresented in medicine and science would experience the effects of the pandemic differently. For race and ethnicity, we categorized selfreported answers into the following groups for analyses: (1) Asian or Asian American, (2) White or European American, and 3) all racial/ethnic groups that are underrepresented among faculty and trainees, including people who are Hispanic or Latino or Latina of any race, Black or African American, American Indian or Alaskan Native, or Native Hawaiian or Pacific Islander. We separated the aforementioned underrepresented groups from Asian Americans, given that those underrepresented groups are also more likely to be disproportionately affected by adverse health effects of COVID-19. 11

Statistical Analysis
Descriptive statistics (proportions or means) are reported for respondent characteristics and primary and secondary outcomes. Four multivariable linear regressions were conducted on the continuous outcome measures. Interaction terms were selected a priori to be entered in the regression analyses.
Unstandardized coefficient estimates are presented for regression analyses to represent how much the mean of the dependent variable shifts given 1-unit change in the independent variable (holding other variables in the model constant). To interpret significant moderator effects, the mean estimated value of each outcome was reported for each group. 12 In the Results section, we discuss the findings of the combined main and interaction effects when interactions are statistically significant and main effects only when interaction terms are not significant. Two-tailed tests were conducted with a significance level of P < .05. SPSS statistical software version 26 (IBM) was used to analyze the data. Data analysis was performed from August to November 2020.

Results
Email invitations to take the survey were sent to 28 000 individuals; 300 emails were returned to sender. Of the remaining 27 700 potential respondents, 5951 (21%) accessed the survey and 5030 (18%) completed the survey and were included in the analysis (mean [SD] age, 40 [12] years).
Although more women completed the survey than men (3738 respondents [75%] were women), the proportion is similar to that of women within the health system (across all roles, women typically compose 64% of the workforce, and 74.9% of hospital employees are female). 4 Respondent characteristics overall and stratified by job role are displayed in Table 1

Associations of Demographic, Clinical, and Job Role Factors With Career and Productivity Outcomes
Four multivariable linear regression models assessed associations and interactions related to each of the 4 primary outcomes ( Table 3). The full regression models are shown in eTable 1, eTable 2, eTable 3, and eTable 4 in the Supplement. Participants who were younger, married, a member of an underrepresented racial/ethnic group, and worked in a clinical setting reported greater consideration for leaving the workforce. Younger, married, and Asian American participants reported a greater consideration for reducing their work hours. Participants who were younger, married, Asian American, male, and had at least 1 child reported decreased productivity, whereas participants who identified as being a member of racial/ethnic group that is underrepresented in medicine reported increased productivity. Younger, Asian American, and underrepresented participants and participants with at least 1 child reported greater worry about the impact of COVID-19 on their career development.

Significant Moderators of Association Between Demographic, Clinical, and Job Role Factors With Career and Productivity Outcomes
Statistically significant interactions were found for each of the 4 outcomes (Table 3 and Table 4  positions. Being a faculty member or trainee with nonclinical responsibilities was also associated with a perceived decrease in productivity and increased worry about the impact of COVID-19 on their career development, compared with nonclinical staff.

Dependent Care and Work Culture Needs
Of 2412 participants with children aged 18 years or younger, 1589 (66%) reported that they did not have childcare fully available, and 783 of 2747 (33%) reported being comfortable taking their child to school or a childcare facility ( Table 2 and   a Given the low sample size for the third gender category (nonbinary, transgender, preferred to self-describe, or preferred not to say), these participants were excluded from this analysis and gender was coded such that 0 = cisgender male and 1 = cisgender female. Other categories were coded as follows: married (0 = single, divorced, separated, widowed, 1 = married or living with partner); Asian (0 = not Asian, 1 = Asian or Asian American); underrepresented ethnic group (0 = not reported underrepresented minority, 1 = Hispanic or Latino or Latina, Black or African American, American Indian or Alaskan Native, Native Hawaiian or Pacific Islander); clinical job role (0 = nonclinical, 1 = clinical); staff (0 = faculty or trainee, 1 = staff); trainee (0 = faculty or staff, 1 = trainee); and has child(ren) (0 = no child, 1 = has child). All of the outcomes are scored 1 (low) to 5 (high).

Discussion
In this survey of 5030 faculty, staff, and trainees of a US health system, we found widespread reported stress associated with caregiving, decreased productivity, concerns about career development, and consideration of either reducing hours or leaving the workforce 6 months after the beginning of the COVID-19 pandemic. These experiences were exacerbated for workers who provide clinical care, those with children at home, women, and people of color-especially those who identify as a belonging to a racial/ethnic group that is underrepresented among medical professionals, academics, and trainees.
Faculty and trainees (most notably women) and workers with nonclinical job roles in particular reported considering leaving the workforce and reducing hours, experiencing reduced productivity, and facing greater concern about the impact of COVID-19 on their careers.
Given the disproportionate impact COVID-19 has on employees of health systems, institutions must find ways to support their employees, both in terms of workplace cultural adaptations and assistance with familial responsibilities. Most participants indicated they wanted continued flexibility in terms of when and where they work and to receive their work schedules at least 1 month in advance (presumably to make arranging childcare easier). Among workers with children younger than 18 years, we found that although some would like to place their children in temporary childcare center settings, the vast majority preferred assistance in home childcare, tutoring (either in person or virtually), or finding groups of like-minded parents (to form pods).
Although academic centers cannot single-handedly relieve many of the stressors facing employees, they have substantial opportunities to influence the employee experience. Our findings suggest that institutional policies could be developed to support all employees, including families, by addressing telecommuting policies and schedule flexibility, as well as providing expanded support options to address psychological stress of employees and trainees and the educational and direct care needs of their children. The development of new, high-impact stopgap measures, such as tutoring and resource matching, provides an opportunity to meet acute needs related to COVID-19.

Limitations
Our ability to generalize these data is limited because the survey was sent to employees of only 1 health system; furthermore, the response rate was too low to be generalizable for the entire health system. However, this is one of the first studies to examine the needs of all employees, allowing us to understand the varying needs of different types of employees. Other limitations include selection bias with regard to individuals who chose to complete the survey. Although the email requesting survey completion emphasized the goal for all employees to participate regardless of dependent status, it is possible that more parents of children aged 18 years or younger completed the survey than those without children. The respondents also included a low portion of racial and ethnic groups that are not representative of the US population, although this is mostly accounted for by the overall low population of such groups in the state of Utah. Furthermore, unlike staff and trainees, faculty were not explicitly asked whether they provided clinical care. In our analysis, we included all participants who reported being on the clinical track as providing clinical care, but were unable to categorize physicians who are on tenure track as to whether they provide clinical care. Thus, we may have underreported the percentage of faculty providing clinical care.

Conclusions
In this survey of 5030 faculty, staff, and trainees of a US health system, our results suggest that respondents were struggling during the COVID-19 pandemic. As a result, even after investing substantial amounts of time in years of training, many were considering leaving the workforce because of stress and caregiving responsibilities related to the pandemic. Health systems must develop effective strategies to ensure that the workplace acknowledges and supports employees during this unprecedented time, not only within the work environment, but also in managing unanticipated childcare responsibilities due to lack of childcare or in-person school. In doing so, health systems will improve the likelihood of retaining generations of well-trained clinicians, scientists, and staff.