Study data were collected and managed using Research Electronic Data Capture (REDCap), hosted at the University of Chicago. REDCap is a secure, web-based software platform that supports data capture for research studies [18,19]. Qualifying participants for our study were at least 18 years of age and were frontline health care workers who were clinically active during the COVID-19 pandemic. The survey automatically terminated for respondents who did not meet these inclusion criteria. We asked the participants to frame all their responses for 1 month in the pandemic in which they experienced the greatest clinical intensity and risk of COVID-19 transmission. This time frame was selected as the most representative of the sleep of respondents while they were clinically active on the front lines caring for patients with COVID-19.

The survey contained the following categories: Sleep, Well-being, Career, and Baseline Demographics. The Sleep section included the Pittsburgh Sleep Quality Index (PSQI) [20] and the Insomnia Severity Index (ISI) [21]. The PSQI is a 19-item validated questionnaire that measures sleep quality over 1 month, generating 1 global score [20]. The ISI is a 7-item validated questionnaire that assesses the nature, severity, and impact of insomnia in adults [21]. The Sleep section also included questions about sleep disturbances due to device use. The Well-being section included the Mini-Z Burnout Survey, an 8-item validated questionnaire that investigates health care burnout and job satisfaction [22]. The Mini-Z also includes a 1-item burnout measure that is validated against the longer Maslach Burnout Inventory [23]. Other questions in this section related to pandemic-related worries, changing responsibilities, living situation, and health status. The Career section collected data on each participant’s job role, level of training, specialty, and career setting prior to the pandemic and whether they cared for patients with COVID-19 directly. Baseline demographics were collected on marital status, age, gender, race, ethnicity, type of practice, location of practice, and type of community served (eg, urban, suburban, rural). Two open-ended questions were also included in the survey: “What was the cause of your sleep disruption(s) during your reported month?” and “Is there anything else you would like to share?”

Upon completion of the survey, participants were invited to submit their email address to enter a daily raffle for a US $25 Amazon gift card (14 gift cards in total). Before deployment, the survey was piloted with a few health care workers who were not affiliated with the study. Changes made included deleting duplicate questions and removing one sleep scale to abbreviate the time to complete the survey to 10 minutes or less.

To disseminate the survey link, we created a Facebook page and a Twitter and Instagram account on behalf of the R.E.S.T. study. This enabled the use of engagement tracking built into these social media platforms to calculate the response rate. To market our survey through Instagram, we used paid advertisements. On Facebook, we posted on health care–specific social media groups (eg, Women Physician Wellness, StyleMD, and the Physician Collective). On Twitter, we tagged colleagues and organizations and employed health care–specific hashtags (eg, #MedTwitter, #NurseTwitter). Although the authors re-shared the R.E.S.T. survey posts on their personal Twitter accounts to increase visibility, the link remained connected to the original posts of the R.E.S.T. accounts to preserve the ability to track engagement. To promote the survey, an infographic was posted each day on the study’s social media accounts. Survey links were also placed in the biography of each social media account. To draw attention to the survey post and the account, we linked each post to standard information from national organizations (eg, the American Academy of Sleep Medicine and the Sleep Research Society) and to articles about sleep and the pandemic through the study’s social media accounts.

One of our study authors (SRD) used a customized bit.ly link [24] to track user engagement daily on Facebook, Twitter, Instagram, and email. The metrics analytics from the social media platforms enabled us to view specific metrics on reach (the total number of individuals who viewed the content) and link clicks (the number of times individuals chose to open the survey). At midnight on each day, we calculated the reach and link clicks of the study on each social media website. This enabled us to compare the performance of each platform and track our progress. We also calculated the number of survey responses per day while the survey was active.

Descriptive statistics were used to quantify the participants’ demographics and profession. Primary outcomes were poor sleep quality and moderate to severe insomnia. Poor sleep quality was defined as a PSQI score ≥5 [20]. Moderate-to-severe insomnia was defined as an ISI score of >14.21. The Mini-Z Burnout Survey was used to measure burnout [22]. Multivariate logistic regression was used to test for independent associations between age, gender, race (Black vs non-Black), ethnicity (Hispanic vs non-Hispanic), and profession (physician vs. nonphysician), with odds of each of the outcomes, after controlling for those with a preexisting sleep disorder. All statistical analyses were conducted using Stata Statistical Software, Release 16 (StataCorp LLC), with P<.05 used to indicate statistical significance.

There were two open-ended questions in the survey. For participants who responded to one or both questions, open coding of comments was used [25]. Sleep disturbance was mapped to four main themes: (1) children and family; (2) work demands affecting sleep; (3) personal health conditions; and (4) pandemic-related sleep disturbances. Subthemes were created for each of these themes. 

To our knowledge, this is the first study to evaluate self-reported sleep disturbances among a sample of predominantly US frontline health care workers during the COVID-19 pandemic with validated instruments using a survey distributed on social media. Our study demonstrates that nearly all (96%) of the surveyed health care workers reported poor sleep, over 30% reported moderate to severe insomnia, and over 50% reported burnout. A majority of respondents also reported sleep disruptions due to personal technology device use (eg, cell phone, iPad) or nightmares at least once per week. Additionally, survey participants reported a mean sleep duration of 6.1 hours per day, which is less than the recommended 7 hours of sleep for US adults [26]. Our study demonstrated that health care workers at highest risk of insomnia included those who were not physicians, were not male, were Hispanic, and directly cared for patients with COVID-19. We also demonstrate that sleep disturbances in health care workers during the pandemic are related not only to work demands and personal health but also to children and family as well as to the pandemic itself, including worry about the response to the COVID-19 pandemic. Given the strong association between sleep and burnout among health professionals, it is not surprising that a high number of the surveyed health care workers reported burnout [5].

The results of this study are consistent with prior studies demonstrating a high prevalence of insomnia in nurses and physicians caring for patients with COVID-19 on the front lines [8]. In a systematic review and meta-analysis, Salari et al [8] reported on the increase of insomnia in frontline nurses near 35% and in frontline physicians near 42%, however they did not report on a subgroup analysis of insomnia in women. In the meta-analysis by Pappa et al [7], the prevalence of insomnia in health care workers during the COVID-19 pandemic was near 40%, and the authors did not specifically report the prevalence of insomnia in women frontline health care workers. Our data are also in accord with those in a study by Trockel et al [5], who demonstrated that sleep-related impairment is an occupational hazard associated with increased medical-related errors and burnout. In comparing our sample to the general US physician population, we found higher rates of sleep disturbances and insomnia risk than in the general population of practicing physicians [12,27,28]. Our study findings of sleep disruption due to device use are consistent with findings demonstrating increased nightly screen time in health care workers and reports of doomscrolling [3,4].

Our findings have many clinical implications. Frontline health care workers involved in the direct care of patients with COVID-19 report decreased sleep time, increased insomnia, nightmares, fears for their safety, increased clinical workload, and concerns for their family, which reportedly impact their sleep. Health care workers are at high risk for the development of psychological distress and medical errors [5,29]. The professional environment for health care workers has drastically changed during the pandemic, which has brought challenges related to increased workload, reduced protective equipment and resources, rapidly evolving protocols, relocation of intensive care settings, fear of viral transmission, and social isolation from supportive networks [30,31]. Given this unfolding environment, the findings of sleep disturbances of frontline health care workers are not surprising. 

The study findings are particularly concerning given the natural course of insomnia and potential for long-term psychological impact, as seen in health care workers during prior severe acute respiratory syndrome (SARS) outbreaks [32-34]. In a recently reported longitudinal study on insomnia, nearly half (42%) of patients who had insomnia initially (as measured by the ISI) also had insomnia 5 years later, demonstrating a persistent course [35]. Likewise, studies on the mental health of health care workers who were active during the 2003 SARS epidemic suggest the psychological consequences may persist months to years after the epidemic ends [34]. While noting long-term impacts of mental health, it is also important to consider the implication of potential medical errors due to sleep loss by these health care workers [6]. Studies of sleep disorders in health care workers during prior infectious outbreaks are in concert with our study findings, and they forecast the potential longitudinal psychological impact on the sleep health of our frontline health care colleagues.

Although social media platforms are widely used, a limitation of this study may be that a sample recruited through social media platforms may not be representative of health care workers in the US and around the world. To combat time period bias, we asked participants to complete the survey while thinking of the month in which they were most clinically active in caring for patients with COVID-19. Although this approach may lead to biased results, we felt it would provide results that were most representative of the respondents’ experiences while they were serving on the front lines during the COVID-19 pandemic. Due to the nature of a worldwide pandemic, the incidence of COVID-19 is distributed differently across geographic areas; therefore, the time frame of reflection may differ between respondents. Because the focus of our survey was on sleep, it is possible that health care workers who were more likely to experience sleep disturbances completed the survey. Given the prevalence of sleep disorders in the general US population of 8%-19% [36,37], our sample does not appear to overrepresent people with sleep disorders. Our sample did not contain a sufficient number of nonphysician health care workers to characterize the sleep quality of nonphysician workers (eg, physician assistants, nurse practitioners, nurses, pharmacists, allied health workers) during the COVID-19 pandemic. We also asked the participants to complete the survey questions by reflecting on a time when they experienced the greatest clinical intensity and risk of COVID-19 transmission; this may have led to inconsistencies in our data. All data were self-reported, and we do not have objective measures of sleep duration and quality in this sample. We also do not have longitudinal data on our sample, limiting our ability to draw longitudinal conclusions. Furthermore, when using data obtained from social media, it is difficult to capture the accurate outreach and link clicks; for example, if an individual shares the link but not our original post, we cannot calculate the outreach and metrics. Therefore, the actual metrics may be higher than we were capable of measuring.

Prior to the pandemic, sleep and burnout issues existed among health care workers; however, strategies to improve and implement change are lagging behind. The reporting of these issues by frontline health care workers caring for patients with COVID-19 during the pandemic is not surprising. Studies demonstrating successful interventions to impact sleep and burnout are increasing, and they recommend interventions focused both at the individual and the organizational level. Examples of these recommendations include reducing work hours while maintaining the same salary, allowing schedule flexibility, behavioral wellness, and curricular interventions [38-41].

During the COVID-19 pandemic, almost all the health care workers we surveyed on social media reported poor sleep, with nearly half of respondents reporting moderate to severe insomnia and over half reporting burnout. Immediate interventions to improve the sleep and well-being of health care workers, discourage nocturnal social media use, and discourage doomscrolling are needed to strengthen the ability of these workers to continuously meet the daily demands of the COVID-19 pandemic. Concerns regarding clinician well-being are not new. Moreover, health systems and national organizations are placing greater emphasis on systemic change [38]. Practical frameworks for creating wellness exist; however, senior-level champions are critical for implementation [42]. Considering that sleep is a modifiable factor of our physical and mental health, prioritizing it may help mitigate the health risks associated with the pandemic and its potential longitudinal impacts on sleep and health.