Investigating the Psychological Impact of COVID-19 among Healthcare Workers: A Meta-Analysis

Previous meta-analyses were conducted during the initial phases of the COVID-19 pandemic, which utilized a smaller pool of data. The current meta-analysis aims to provide additional (and updated) evidence related to the psychological impact among healthcare workers. The search strategy was developed by a medical librarian and bibliographical databases, including Medline, Embase, CINAHL, PsycINFO, and Scopus were searched for studies examining the impact of the COVID-19 pandemic on the psychological health of healthcare workers. Articles were screened by three reviewers. Heterogeneity among studies was assessed by I2 statistic. The random-effects model was utilized to obtain the pooled prevalence. A subgroup analysis by region, gender, quality of study, assessment methods, healthcare profession, and exposure was performed. Publication bias was assessed by Funnel plot and Egger linear regression test. Sixty-five studies met the inclusion criteria and the total sample constituted 79,437 participants. The pooled prevalence of anxiety, depression, stress, post-traumatic stress syndrome, insomnia, psychological distress, and burnout was 34.4%, 31.8%, 40.3%, 11.4%, 27.8%, 46.1%, and 37.4% respectively. The subgroup analysis indicated higher anxiety and depression prevalence among females, nurses, and frontline responders than males, doctors, and second-line healthcare workers. This study highlights the need for designing a targeted intervention to improve resilience and foster post-traumatic growth among frontline responders.


Introduction
In December 2019, a novel coronavirus originated in Wuhan (China), which was later identified by the International Committee on Taxonomy of Viruses (ICTV) as SARS-CoV-2 causing the disease COVID-19 [1]. The spread of the virus was rampant, with the cases spiraling up to nearly 148,838 in March, which led to COVID-19 being designated as a pandemic by the World Health Organization [WHO] on 11 March 2020 [2]. Undoubtedly, pandemics have a long-standing history of impacting physical and mental health across all population groups, of which healthcare workers (HCWs) bear a disproportionate burden [3]. During Severe Acute Respiratory Syndrome (SARS) and Middle East Respiratory Syndrome (MERS) outbreaks, a sizable proportion of the HCWs experienced anxiety, emotional distress, and post-traumatic stress disorder (PTSD, aka fatigue battle syndrome) [4][5][6]. These adverse psychological outcomes persisted until 1-3 years in the post-pandemic periods [6]. Following the aggressive course of transmission, the COVID-19 pandemic has taken a firm grip worldwide and has surpassed the historical outbreaks in generating extraordinary amounts of pressure and psychiatric morbidities among healthcare workers (HCWs) [3,[6][7][8][9]. Physical and emotional exhaustion associated with managing large volumes of COVID-19 cases, shortage of personal protective equipment (PPE), risk of nosocomial infections, and fear of secondary transmission to family members, feelings of being rejected by others, and social isolation make frontline HCWs more vulnerable to the psychological corollaries of the COVID-19 pandemic [7,8,10]. Similar to previous outbreaks, the stress induced by this bio-disaster  has the potential to develop into PTSD [11][12][13]. In one Japan-based study, stress associated with deployment activities among the healthcare workforce was investigated [11]. The symptoms of PTSD were reported among Disaster Medical Assistance Team (DMAT) members, who were deployed to analyze, manage, and contain the transmission of COVID-19 on a Diamond Princess Cruise ship during the initial phases of the pandemic [11,12]. PTSD has already been cited as the main psychiatry disorder associated with disaster-related experiences or activities, especially among those being on the front lines of the battle against the pandemic [13]. In the wake of the COVID-19 pandemic, frontline responders are continuously working to meet the heavy healthcare demands and are exposed to higher levels of psychiatric morbidities. It is likely that these psychiatric issues will take a chronic course and will translate into PTSD in the repairing phases of the pandemic. Therefore, it is critical to assess the psychological impact among HCWs throughout the evolution of the COVID-19 pandemic to design early interventions to improve psychological outcomes. Several qualitative and quantitative studies have been conducted to explore all dimensions of the psychological spectrum among HCWs and provided valuable insights [11,[13][14][15].
Previous meta-analyses investigated the psychological impact on healthcare populations during the early episodes of this bio-disaster and utilized a smaller pool of data [14,15]. One (singlearm) meta-analysis of observational studies included 13 studies  with a sample of 33,062 participants [14]. This study was based on a literature search on April 17, 2020. Another meta-analysis attempted to expand the evidence (by adding 12 more studies) [29][30][31][32][33][34][35][36][37][38][39][40] by comparing the psychological impact among the general population, healthcare workers, and patients with preexisting conditions [15]. This study included a total of 19 studies [16][17][18][19][20][21][29][30][31][32][33][34][35][36][37][38][39][40][41] (based on a search on May 25, 2020), and the overall prevalence of psychological indicators among healthcare workers with no further distinction based on exposure (i.e., frontline and second-line workers) were reported [15]. With the rapid evolution of pandemic and increased number of hospitalizations, frontline HCWs are experiencing unprecedented emotional and psychological challenges [15,[33][34][35][36][37]. Along the course of the pandemic, new studies (n = 38) have rapidly been conducted in different parts of the world, [6,, which will provide additional insight into the existing literature. Additionally, it is imperative to make continuous efforts in collecting and synthesizing more data until the full picture of the psychological toll on healthcare workers emerges. Therefore, the purpose of this meta-analysis is to provide updated evidence (based on search prior to July 27, 2020, with 40 additional studies) across 24 countries to investigate the psychological impact of the COVID-19 pandemic on HCWs, with further stratification to the level of COVID-19 exposure.

Protocol Registration
To conduct a meta-analysis on studies investigating the psychological impact of the COVID-19 pandemic among health care workers, we followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses [PRISMA] guidelines [82]. This study protocol was registered in the

Assessment of Bias Risk
The quality assessment was performed by utilizing the National Institutes of Health (NIH) quality assessment tool (https://www.nhlbi.nih.gov/health-topics/study-quality-assessment-tools) to assess the quality of the studies. Two reviewers (K.B. and T.P.S.) independently evaluated the risk of bias and quality of the studies and rated them according to the tool's dictionary and guidelines. After assessing all the study components, the overall rating was determined using the criteria listed in the tool. Based on the number of "yes" answers, a rating of good (7)(8)(9), medium (4-6), or poor (≤3) was assigned to each study (Appendix B). Disagreements related to quality scores were resolved through a discussion among the reviewers, and a consensus was achieved upon the final judgement offered by the senior investigator (M.S.). The rating of two reviewers was compared, and the inter-rater agreement was calculated using Microsoft Excel.

Statistical Analysis
The overall prevalence and 95% confidence intervals of psychological outcomes were pooled using the Comprehensive Meta-Analysis Package (CMA version 3.0, Englewood, NJ, USA). The primary effect measure was the proportions of events, such as anxiety, depression, stress, psychological distress, insomnia, and impaired sleep quality. Due to methodologic variations and sample diversity across studies, the random-effects model was used to extract the pooled estimate [83]. Heterogeneity was assessed by the I 2 statistic, which measures the percentage of variance resulting from true differences in the effect sizes rather than the sampling error. Substantial heterogeneity [84] was defined as I 2 > 50%. Subgroup analysis (by potential sources of heterogeneity) was conducted according to the categorical moderating variables: country (China vs. other countries), gender, continent (Asia vs. other continents), quality of the study (good/medium), assessment methods, health care profession (doctors vs. nurses), type of exposure (high risk or frontline vs. low risk or second-line), and severity of psychological symptoms (mild/moderate/severe). Sensitivity analysis was conducted to identify studies which may severely affect the pooled prevalence. Funnel plot and Egger linear regression test was used to assess publication bias [85]. Significant level was set as two-sided and p < 0.05.

Study Screening
Our systematic search yielded 7255 potentially relevant papers (Figure 1), out of which 2768 duplicate studies were removed. The titles of the remaining 4487 records were screened, and 4052 studies were excluded because the studies were conducted on non-healthcare population (n = 1347), had a different outcome of interest (n = 1256), different study designs, including case series, reviews, perspectives or opinions, and interventional studies (n = 1013), and were irrelevant to the study's objective (n = 436). This resulted in 435 papers, which were advanced to abstract screening. A total of 321 papers were excluded after abstract screening because they were qualitative or mixed studies (n = 68), studies on a general population (n = 73), studies without prevalence data (n = 96), serological studies (10), or they were published as editorials, posters, or reviews (n = 74). Full-text screening of the remaining 114 papers generated 65 articles, which were included in the final review. Forty-nine articles were excluded in the final step because of following reasons: they were abstracts-only or in other languages (n = 6), used non-validated questionnaires (n = 16), had an unclear methodology or of low quality (n = 10), or had outcomes reported as means instead of the number of cases/or proportions (n = 17) (Figure 1).

Anxiety Prevalence
The pooled prevalence of anxiety in 46 studies with a sample size 51,596 was 34.4% (Table 1, Figure 2). The pooled prevalence of anxiety among good quality studies (n = 22) was 31.2% compared to 38.1% among medium quality studies (n = 24) ( Table 1). The pooled prevalence of anxiety in the continent of Asia across 34 studies was 32.7% compared to 39.3% found in studies among other continents. Twenty-two studies were conducted in China and had a pooled prevalence of 28.5% as opposed to the 40.4% (Table 1) pooled prevalence across studies (n = 24) conducted in other countries. The Generalized Anxiety Disorder survey questionnaire was used across 19 studies and a pooled prevalence of 36.8% was found (Table 1). Gender data were available in seven studies with a pooled prevalence of 46.9% for females and 44.2% for males. In groups by healthcare professions, the pooled prevalence was higher in nurses compared to doctors (39.3% vs. 32.5%). Anxiety by exposure with a pooled prevalence of 39.8% among frontline HCWs compared to 27.1% prevalence among second-line HCWs. Levels of anxiety with the highest pooled prevalence of 60.3% related to mild symptoms, followed by a 26.0% prevalence of moderate symptoms and a prevalence of only 14.3% for severe symptoms (Table 1).

Depression Prevalence
The pooled prevalence of depression in 46 studies with a sample size 53,164 was 31.8% (Table 2, Figure 3). The pooled prevalence of depression among good quality studies was 35.1%, compared to 28.6% among medium quality studies ( Table 2). The pooled prevalence of depression in the continent of Asia was 30.8% compared to 35.0% found in other continents. Twenty-three studies were conducted in China and had a pooled prevalence of 33.2% as opposed to the 30.4% pooled prevalence across studies conducted in other countries (n = 23). The Patient Health Questionnaire (PHQ) was used across 25 studies and a pooled prevalence of 29.7% was found. Gender data were available in seven studies, with a pooled prevalence of 43.4% for females and 40.9% for males. In the healthcare profession groups, examined in nine studies, the pooled prevalence of depression was higher in nurses compared to doctors (42.4% vs. 39.1%). Depression by exposure was reported in six studies with a pooled prevalence of 23.6% among frontline healthcare workers compared to 19.6% prevalence among second-line healthcare workers. Levels of depression were reported in 17 studies, with the highest pooled prevalence of 57.6% related to mild symptoms, followed by a 27.9% prevalence of moderate and only 10.4% of severe symptoms ( Table 2).

Stress Prevalence
The pooled prevalence of stress in 17 studies with a sample size of 16,235 was 40.3% (Table 3, Figure 4). The pooled prevalence of stress among good quality studies (n = 9) was 37.3% compared to 45.7% among medium quality studies (n = 8). The pooled prevalence of stress in the continent of Asia across 14 studies was 41.3% compared to 38.8% found in other continents. Seven studies were conducted in China and had a pooled prevalence of 44.2% as opposed to the 37.1% pooled prevalence in studies conducted in other countries (n = 10). The Perceived Stress Scale (PSS) was used across eight studies and a pooled prevalence of 61.4% was found. Levels of stress were reported in six studies, with the highest pooled prevalence related to moderate symptoms of 52.3%, followed by a 25.8% prevalence of mild and 18.9% of severe symptoms (Table 3). For stress, subgroup analyses by gender, healthcare occupations, and risk exposure were not conducted due to inadequate data. CI = Confidence Interval; GAD = Generalized Anxiety Disorder; SAS = Self-rating Anxiety Scale; Good quality score = 7-9; Medium Quality score = 4-6; I 2 statistic indicates the heterogeneity.    Forest plot for the studies on the prevalence of anxiety among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CIs of the overall population. The overall pooled anxiety using a random effects DerSimonian-Laird method was 34.4% (95% CI: 29.5-39.7).

Figure 2.
Forest plot for the studies on the prevalence of anxiety among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CIs of the overall population. The overall pooled anxiety using a random effects DerSimonian-Laird method was 34.4% (95% CI: 29.5-39.7).

Figure 3.
Forest plot for the studies on the prevalence of depression among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled depression using a random effects DerSimonian-Laird method was 31.8% (95% CI: 26.8-37.2).

Figure 3.
Forest plot for the studies on the prevalence of depression among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled depression using a random effects DerSimonian-Laird method was 31.8% (95% CI: 26.8-37.2).

Figure 4.
Forest plot for the studies on the prevalence of stress among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CIs of the overall population. The overall pooled stress using a random effects DerSimonian-Laird method was 40.3% (95% CI 31.4-50.0).

Publication Bias
The publication bias was assessed with Egger's test indices. As  Figure A13 in Appendix C), the publication bias was insignificant for all the psychological outcomes, except insomnia.

Discussion
This pooled analysis included a large data sample of studies (n = 65) with 79,437 participants, compared to previous meta-analyses that included 13-19 studies [14,15]. Notably, to our knowledge, this study is the largest to evaluate the psychological impact of COVID-19 among HCWs. Moreover, we extended the existing evidence by including other psychological outcomes of stress, psychological distress, burnout, and impaired sleep quality. The findings suggest that the overall prevalence of anxiety, depression, stress, insomnia was 34.4%, 31.8%, 40.3%, and 27.8%, respectively. Compared to previous meta-analyses [14,15], we report a higher prevalence for anxiety (34.4% vs. 23.2%-26.0%), depression (31.8% vs. 22.8%-25.0%), and PTSD (11.4% vs. 3%). The higher prevalence of anxiety, depression, and PTSD may be explained by the pervading climate of uncertainty generated with the advancing pandemic, limited signs of a workable vaccine, increased workload, lack of social support, and an intense fear of family transmission [86][87][88]. According to a recent Chinese report of 14,825 healthcare workers, depressive symptoms and PTSD were more common among HCWs with lower levels of social support and longer daily working hours (>12 h/day) [88]. We found a lower prevalence of insomnia compared to a previous analysis [14] (27.8% vs. 34.4%) despite having a greater number of studies in the current analysis (11 vs. 5). This may be due to the variance in the cut-off values (>14 vs. >8) of the ISI used by the recent studies, which were included in this updated meta-analysis [32,44,63,64]. Moreover, the direction of the etiological relationships between psychological morbidities and insomnia remains complex; for instance, anxiety disorders precede insomnia in nearly 70% of the cases [89]. Therefore, it is likely that significant changes in sleep architecture among HCWs will occur later and will be observed by prospective studies. In concordance with previous studies, our subgroup analysis by gender revealed that females had a higher prevalence of anxiety and depression compared to males [14,15,27,28,[89][90][91][92]. The current meta-analysis found higher levels of anxiety and depression among nurses compared to doctors, which may be because nurses have closer and prolonged contact with patients compared to doctors [3,93]. We investigated the prevalence of anxiety and depression by risk groups and found higher levels of anxiety and depression (as expected) among frontline responders as compared to the second-line workers. All of these results may be partly confounded by the fact that majority of the frontline workers' group constitutes nurses, who are responsible for providing direct care to the COVID-19 patients and for collecting sputum specimen for virus detection, and tend to be female [3,27,28,[94][95][96][97]. This repeats a finding from the SARS outbreak that nurses reported higher anxiety, depression, behavioral problems, and moral injuries (related to death and ethical dilemmas) [3,5,7,14,27,28,[92][93][94][95]. These intersecting factors make clear that it is imperative to develop interventions for the most vulnerable population: frontline nurses, who tend to be female and work long hours. Our study reported a higher prevalence of anxiety in other countries compared to China (40.4% vs. 28.5%). The reasons for this may be complex and could be influenced by how other countries are directing their medical resources towards the containment efforts or in treating infected patients rather than providing psychological services.
The higher prevalence of the psychosocial impact of COVID-19 on health care workers found in this study draws more careful attention to educational and policy interventions for this subgroup. Educational and behavioral interventions emphasizing hardiness, social support, positive thinking, a sense of coherence, and others have been advocated in the literature [96][97][98]. The role of self-care is also highlighted by some researchers [98][99][100]. Positive traumatic growth (a positive approach to the management of complex traumas) among health care workers has gained special attention, and interventions to build this have been suggested [96,99,100]. Some strategies, such as mindfulness interventions, can be instituted quickly to promote healthcare workers' mental health [97][98][99]. Organizational support is vital at the policy level and may need sufficient lead time to be enacted [97][98][99]. Work-based interventions, such as curtailing hours of work, having buddy support systems, having listening sessions between administration and health care functionaries, increasing coverage of tele counseling through employee assistance programs, having mental health consultants available to staff through telehealth, and other such measures, can go a long way in reducing the adverse psychosocial impact caused by COVID-19 [98,99].

Quality of Evidence
Compared to the previous reviews and meta-analyses, the current meta-analysis provides the most extensive evidence with a much bigger sample size of 79,437 participants. Although all studies included in our meta-analysis were cross-sectional, they were of high and medium quality. We performed subgroup analyses to account for the potential sources of heterogeneity and to identify additional vulnerabilities. Additionally, we investigated the potential for publication bias across all studies.

Study Limitations
There are a few limitations that merit discussion. First is the presence of heterogeneity across studies in terms of the survey tool and cut-off scores. Additionally, threshold criteria for defining levels of outcomes varied across studies; for example, some studies reported results as mild, moderate, moderate-severe, and severe, while others reported outcomes as mild, moderate, and severe. This may affect our subgroup analysis by severity. Second, data provided by the studies included in this meta-analysis depend on the self-reported psychological outcomes as recorded through assessment tools. Thus, there may be uncertainty related to the actual psychological illness or diagnosis. Third, sampling bias may exist (although lower than the previous meta-analysis), because nearly 48% (31/65) of the studies were conducted in China. This may also limit the generalizability of the results. Fourth, all studies included in this meta-analysis were cross-sectional, which only provided a snapshot of the existing situation with no exploration of longitudinal aspects. Last, we expect to have a language bias in the study because only studies published in the English language were included.

Research and Clinical Implications
The findings of this meta-analysis highlight the need to develop psychological interventions to promote the post-traumatic growth among HCWs. Higher prevalence estimates of stress provided by this study have important implications for developing early interventions to prevent PTSD, which may be higher in the repairing phases of the pandemic. Mental health and well-being interventions, such as education on coping techniques, online wellness activities, fostering post-traumatic growth, and opening channels for assistance in early signs of PTSS (Posttraumatic Stress Syndrome) before they manifest to PTSD, are essential.

Conclusions
This article represents, to our knowledge, the most extensive meta-analysis to assess the psychological impact of COVID-19 among HCWs. This meta-analysis provides additional evidence to the higher psychological impact among HCWs, particularly those who are female, nurses, and frontline responders. Furthermore, this adds a valuable insight to the existing meta-analysis findings by highlighting a significant difference in the psychological impact across frontline and second-line workers.  (stress* or "adaptation syndrome" or (caregiver adj4 (burden or fatigue)) or "compassion fatigue" or "reality shock" or "social defeat"    Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between exposure(s) and outcome(s)? Rating-Good, Medium or Poor), Good = (7-9 yes); Medium = (4-6 yes).)

Figure A2.
Forest plot for the studies on the prevalence of impaired sleep quality among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled impaired sleep quality using a random effects DerSimonian-Laird method was 64.3% (95% CI: 55.0-72.7).

Figure A3.
Forest plot for the studies on the prevalence of Post-traumatic stress syndrome (PTSD) among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled PTSD using a random effects DerSimonian-Laird method was 11.4% (95% CI: 3.6-30.9).

Figure A2.
Forest plot for the studies on the prevalence of impaired sleep quality among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled impaired sleep quality using a random effects DerSimonian-Laird method was 64.3% (95% CI: 55.0-72.7).

Figure A3.
Forest plot for the studies on the prevalence of Post-traumatic stress syndrome (PTSD) among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled PTSD using a random effects DerSimonian-Laird method was 11.4% (95% CI: 3.6-30.9).

Figure A2.
Forest plot for the studies on the prevalence of impaired sleep quality among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled impaired sleep quality using a random effects DerSimonian-Laird method was 64.3% (95% CI: 55.0-72.7).

Figure A1.
Forest plot for the studies on the prevalence of insomnia among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled insomnia using a random effects DerSimonian-Laird method was 27.8% (95% CI: 21.4-35.3).

Figure A2.
Forest plot for the studies on the prevalence of impaired sleep quality among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled impaired sleep quality using a random effects DerSimonian-Laird method was 64.3% (95% CI: 55.0-72.7).

Figure A3.
Forest plot for the studies on the prevalence of Post-traumatic stress syndrome (PTSD) among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled PTSD using a random effects DerSimonian-Laird method was 11.4% (95% CI: 3.6-30.9).

Figure A3.
Forest plot for the studies on the prevalence of Post-traumatic stress syndrome (PTSD) among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled PTSD using a random effects DerSimonian-Laird method was 11.4% (95% CI: 3.6-30.9).

Figure A4.
Forest plot for the studies on the prevalence of psychological distress among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled psychological distress using a random effects DerSimonian-Laird method was 46.1% (95% CI: 36.0-56.6).

Figure A5.
Forest plot for the studies on the prevalence of burnout among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIz). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled burnout using a random effects DerSimonian-Laird method was 37.4% (95% CI: 14.8-67.2).

Figure A4.
Forest plot for the studies on the prevalence of psychological distress among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled psychological distress using a random effects DerSimonian-Laird method was 46.1% (95% CI: 36.0-56.6).

Figure A4.
Forest plot for the studies on the prevalence of psychological distress among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIs). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled psychological distress using a random effects DerSimonian-Laird method was 46.1% (95% CI: 36.0-56.6).

Figure A5.
Forest plot for the studies on the prevalence of burnout among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIz). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled burnout using a random effects DerSimonian-Laird method was 37.4% (95% CI: 14.8-67.2).

Figure A5.
Forest plot for the studies on the prevalence of burnout among healthcare workers. The squares and horizontal lines correspond to the study-specific event (anxiety) rates and 95% confidence intervals (CIz). The diamond represents the pooled prevalence and 95% CI of the overall population. The overall pooled burnout using a random effects DerSimonian-Laird method was 37.4% (95% CI: 14.8-67.2).