Mental health of medical personnel during the COVID‐19 pandemic

Abstract Introduction The coronavirus disease 2019 (COVID‐19) pandemic caused significant changes in the everyday functioning of the general population, as well as medical workers. Medical personnel, especially those in direct contact with COVID‐19 patients, could have increased levels of stress, anxiety, and depression. The objective of this study was to explore the mental health status of medical personnel in Serbia during the pandemic by assessing stress levels, symptoms of anxiety, and depression. Methods This cross‐sectional study was conducted as an online‐based survey, in the period from 8 April to 14 April 2020, during the COVID‐19 pandemic. The study included 1678 participants, and the snowball sampling technique was used to reach healthcare professionals. The level of stress and symptoms of depression and anxiety were assessed among medical personnel in Serbia by the 10‐item Perceived Stress Scale (PSS), the Beck Depression Inventory IA (BDI‐IA), and the 7‐item Generalized Anxiety Disorder Scale (GAD‐7), respectively. Results A total of 1678 participants completed the survey, with a mean age of 40.38 ± 10.32 years, of which 1,315 (78.4%) were women, and 363 (21.6%) were men. Out of these, 684 (40.8%) participants were medical personnel, and 994 (59.2%) were people of other professions. Frontline medical personnel reported higher scores on all measurement tools than second‐line medical personnel (e.g., mean PSS scores: 19.12 ± 5.66 versus 17.53 ± 5.71; p = .006; mean GAD‐7 scores: 8.57 ± 6.26 versus 6.73 ± 5.76; p = .001; mean BDI‐IA scores: 9.25 ± 8.26 versus 7.36 ± 7.28; p = .006). Binary logistic regression showed that the probability of developing more severe anxiety symptoms doubles in frontline medical personnel. Conclusion Our findings suggest that frontline medical personnel is under an increased psychological burden during the COVID‐19 pandemic, having higher levels of stress, anxiety, and depression than second‐line medical personnel. Adequate measures should be taken to relieve this burden and preserve the mental health of frontline medical personnel.


| INTRODUC TI ON
The recent pandemic of coronavirus disease , caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), spread rapidly throughout the world after the first reported case in Wuhan, China (Lu et al., 2020). In the Republic of Serbia, the first case of COVID-19 was confirmed on 6 March 2020. On March 15, the Serbian government declared a nationwide state of emergency and a wide range of counterepidemic measures were progressively adopted (The Government of the Republic of Serbia, 2020).
Increased level of stress is associated with working in health care even in the usual circumstances. Depression, anxiety, sleep disturbances, and burnout were described in that context (Cleary et al., 2018). This pandemic put healthcare professionals in a challenging situation, especially those working in hospitals caring for patients with suspected or confirmed COVID-19. They were dealing not only with greater infection risk and fear of contagion and spreading the virus to their loved ones (Greenberg et al., 2020;Xiang et al., 2020) but also with emotional disturbances, sleep problems, isolation, lack of contact with their families, extended shifts, and physical exhaustion (Kang et al., 2020).
Previous studies have found that the COVID-19 pandemic has a severe impact on the mental health of healthcare workers as well as of the general population (Kang et al., 2020;Qiu et al., 2020).
According to research on similar outbreaks of severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS), anxiety and fear were the first symptoms to appear among medical personnel, but depression and post-traumatic stress symptoms appeared afterward causing severe consequences and had a long-term effect on the mental health of medical personnel (Chong et al., 2004;Lee et al., 2018).
The World Health Organization defines mental health as, "a state of well-being in which the individual realizes his or her own abilities, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to his or her community" (World Health Organization, 2005). Considering that the most frequent consequences of stressful life events on mental health are the occurrence of anxiety and depression (Fink, 2016), this paper aimed to explore mental health status by examining if medical workers who work with COVID-19 patients (frontline medical personnel) had higher levels of anxiety, depression, and perceived stress than medical workers who do not work with COVID-19 patients (second-line medical personnel).
To our knowledge, our study is the first of its kind conducted in Serbia, and it might be useful in planning and implementing appropriate mental health interventions, support, and follow-up for the frontline medical personnel.

| Study design, participants, and setting
This research was designed as a cross-sectional study, conducted during the fifth and sixth weeks of the COVID-19 outbreak in Serbia, and is a part of a larger study. Due to a nationwide lockdown, which was instituted as a counterepidemic measure, the study was conducted via a questionnaire on the online platform 1KA (Centre for Social Informatics, Faculty of Social Sciences, University of Ljubljana, Slovenia).
We used the snowball sampling technique to reach healthcare professionals and the general population. The link to the questionnaire was primarily sent by e-mail, and each participant was encouraged to disseminate the link further to their contacts, both professional and personal.

| Informed consent and anonymity
The survey was anonymous and in accordance with the ethical principles set by the Declaration of Helsinki. Participants were not asked for any information which could be used for identification. The informed consent was located on the introductory page, and participants agreed to it by clicking the "Begin" button. Participation was completely voluntary.

| Design of the questionnaire
Sociodemographic questionnaire and questionnaires assessing stress level, anxiety, depression, and resilience were used.
Sociodemographic data were collected on gender, age, marital status, occupation, children, and residence. The occupation was divided into the following categories: frontline doctors, second-line doctors, frontline nurses, second-line nurses, and other professions.
Since there are no paramedics in Serbia, that answer was not included as an option.
The level of stress was measured by the 10-item Perceived Stress Scale (PSS), which demonstrated good internal reliability and good construct validity (Cohen, 1988). Respondents were asked to score each of the ten items from 0 (never) to 4 (very often), according to their thoughts and feelings in the previous month. The scores were divided using tertiles, into low, medium, and high stress groups.  (Beck et al., 1988). The inventory consists of 21 groups of four statements, scored on a scale from 0 (normal or absent) to 3 (severe) (Beck & Beamesderfer, 1974). Results were divided into score groups as follows: minimal (<10); mild (10-18); moderate (19-29); and severe depression (30-63) (Beck et al., 1988).
Brief Resilient Coping Scale (BRCS), consisting of 4 items, was used to assess resilience. This scale demonstrated good construct and criterion validity, good test-retest reliability, and adequate internal consistency (Sinclair & Wallston, 2004). Respondents were asked to rate each item from 1 ("does not describe me at all") to 5 ("it describes me very well"). The scores were divided using tertiles, into low, medium, and high resilience groups.
One additional question was introduced in the questionnaire. It was a closed-ended, multiple-choice question, regarding the dominant thought content related to anxiety and fear. The subjects could choose one or more items as an answer to the question, "What are your fears or anxiety mostly related to?" The answers offered are shown in the results section.

| Statistical analysis
Analysis of the collected data was performed in SPSS version 20 (IBM Corp; RRID:SCR_002865). As the Kolmogorov-Smirnov test confirmed that the scores of the used tools were not distributed normally, the nonparametric Mann-Whitney U test and Kruskal-Wallis test were used to compare values between groups. Chisquare test was used to compare the differences in the relative frequency of different score categories, and Z test was used for pairwise comparison. We used binary logistic regression analysis to determine the potential impact of working with COVID-19 patients on the probability of increase in stress, anxiety, and depression levels in medical personnel. Medical workers were classified into two categories for each dimension-0 if they were in the groups with low stress, minimal anxiety, or minimal depression, and 1 if they were in any of the other categories. Three logistic regressions were performed, and the impact factor was working with COVID-19 patients (frontline or second-line medical workers), and the dependent variables were indicators of stress, anxiety, and depression. The significance value of 0.05 or less was considered the significance threshold.
Hierarchical regression analysis was performed to determine the possible role of resilience as a mediator between occupation (medical personnel or other professions) and stress, anxiety, and depression. In the first model, a binary variable, occupation, was used as a predictor. In the second model, resilience was also added as a predictor, and we investigated potential changes in the significance of the first predictor variable. The process was repeated three times-for prediction of stress, anxiety, and depression, respectively.

| Levels of stress and distribution of stress scores
To investigate levels of stress in these groups, we calculated mean scores and standard deviations. As shown in Table 2, the chi-square test was used to analyze the relative frequency of different score groups and significant differences were found (χ 2 = 12.495, p = .014). Z test was used for pairwise comparison. The group of frontline medical personnel had a significantly lower percentage of respondents in "low stress" group compared to both second-line personnel and group of other professions.

| Levels of anxiety and distribution of anxiety scores
The Cronbach α for the GAD-7 in our sample was 0.919. The mean level of anxiety ± SD in the group of medical personnel was 7.18 ± 5.94 and in people of other professions 6.34 ± 5.52. Mean levels ± SD in the subgroups of medical personnel were as follows: frontline doctors, 7.37 ± 5.68; second-line doctors, 5.31 ± 4.93; frontline nurses, 9.58 ± 6.57; and second-line nurses, 8.05 ± 6.16.
Using the chi-square test, statistically significant differences were found between relative frequencies of score groups (χ 2 = 24.831, p < .001), and pairwise comparison was done using the Z test. Frontline medical personnel had a significantly lower percentage of respondents in the "minimal anxiety" group and a significantly higher percentage of respondents in the "severe anxiety" group compared to both second-line medical personnel and the group of respondents of other professions.

| Levels of depression and distribution of depression scores
For the BDI-IA, Cronbach's α in our sample was 0.882. Depression score, expressed as mean ± SD, in the group of medical personnel was 7.84 ± 7.57 and in the group of other professions was 8.20 ± 7.68.
Mean scores ± SD in subgroups of medical personnel were as follows: Note:: Symbols in superscript ( a and b ) represent the results of the Z test; the groups with the same symbol ( a and a or b and b ) do not significantly differ; groups marked with different symbols ( a and b ) differ significantly. The groups marked with both symbols ( a,b ) do not significantly differ neither from group marked with a nor from the group marked with b .
A statistically significant difference was not found between frontline medical personnel, second-line medical personnel, and respondents of other professions in the relative frequency of score groups, as analyzed by chi-square test (χ 2 = 6.667, p = .353) ( Table 2).

| Resilient coping style
In our sample, Cronbach's α for the BRCS was 0.775. Mean BRCS scores ± SD in the groups of our respondents were as follows: 13.81 ± 3.24 in the group of medical personnel and 14.32 ± 3.09 in the group of other professions. In the subgroups of medical personnel, mean scores ± SD were as follows: frontline doctors,

| Differences between frontline and second-line medical personnel
As shown in Table 3, statistically significant differences in levels of stress (p = .006), anxiety (p = .001), and depression (p = .006) were found between the frontline and second-line medical personnel.
Frontline medical personnel had significantly higher levels of stress, anxiety, and depression than second-line medical personnel.

Three binary logistic regressions were performed. Our results
suggest, as shown in Table 4, that working as frontline personnel is associated with an increase in the probability of exhibiting elevated stress (p = .017) and anxiety (p = .001) levels, but not depression (p = .094) levels, although the significance is close to the 0.05 threshold. In frontline medical personnel, the probability of exhibiting anxiety symptoms in the range of mild, moderate, and severe scores is increased by 100% (exp{B} = 2.003).

| Differences between medical personnel and other professions
Medical personnel, frontline and second-line combined, had significantly higher (p = .009) mean levels of anxiety ± SD (7.18 ± 5.94) than respondents of other professions (6.34 ± 5.52), as determined by Mann-Whitney U test. Also, there was a significant difference in BRCS scores between these two groups (p = .005).
The differences in stress levels and depression were not found to be statistically significant. Note: Symbols in superscript (a, b and c) represent the results of the Z test; the pairs with the same symbol (a and a or b and b or c and c) do not significantly differ; pairs marked with different symbols (e.g., a and b) differ significantly.
Abbreviation: df, degree of freedom.

| D ISCUSS I ON
Our results showed that levels of stress, anxiety, and depression were significantly higher in frontline than in second-line medical personnel and that the probability of developing more severe anxiety symptoms doubles in frontline medical personnel. These differences could possibly be explained by the distress caused by unfamiliarity with the pathogen, known characteristics of the disease itself, direct contact with confirmed patients, longer work hours, work burden, exposure to much death and dying, having to make difficult decisions regarding patient treatment (Holmes et al., 2020), and with differences in dominant anxiety-producing thought content, such as fear of infecting loved ones. It should be taken into account that there is a possibility that these symptoms could persist even after the end of the current pandemic, as it was the case during the SARS and Ebola epidemics (Tam et al., 2004). Delayed onset of post-traumatic stress disorder may also occur after an acute phase of a pandemic (Mak et al., 2009).
Not only frontline medical workers exhibit higher levels of anxiety, although they are the most affected. All medical personnel differ significantly in the level of anxiety from the group of other professions, which could be explained by the increased risk of working with undiagnosed or asymptomatic COVID-19 patients.
Our findings show that all medical workers have significantly lower levels of resilience compared to the group of other professions. As a previous study has shown that medical workers with low resilience are more likely to develop anxiety compared to medical workers with high resilience (Mosheva et al., 2020), this could also be a factor contributing to higher anxiety levels in medical personnel.
Since one's ability to cope with adverse life situations significantly affects investigated mental health outcomes (World Health Organization, 2005), we also examined the role of resilience as a mediator variable between the occupation and levels of stress, anxiety, and depression. It was shown that resilience partially explains the effect of occupation on anxiety, but not on stress and depression.
However, all three models are significantly improved when resilience is added as a predictor variable, in addition to the occupation. This result may prove useful to future studies.
There are several limitations to our study. Firstly, given the limited time frame, we used the cross-sectional design, and conducting a longitudinal study would be necessary to determine a causal link and long-term effects of the COVID-19 pandemic on mental health.
Secondly, the snowball sampling method was based on the nonrandom selection of the sample and may indicate selection bias. Finally, assessment of mental health was based on an online survey and self-report measures, which have disadvantages compared with an in-person interview.

| CON CLUS ION
Our study has shown, though with limitations, that levels of anxiety, depression, and stress during the COVID-19 pandemic are higher in medical personnel working with COVID-19 patients than in secondline medical personnel, and that the probability of developing more severe symptoms of anxiety doubles in frontline medical personnel.
In addition, it was shown that all medical personnel had higher levels of anxiety and stress than respondents of other professions.
It is the authors' opinion that, in accordance with our findings, measures should be taken to reduce the psychological burden on the frontline workers in situations such as the COVID-19 pandemic.
Furthermore, it is essential to monitor the mental health of medical personnel even after this pandemic is over in order to make an early diagnosis of any possible long-term conditions, and provide them with adequate treatment.

ACK N OWLED G M ENT
We thank Professor Ivana Binic (University of Nis, Serbia) for her patient guidance and useful critiques of this research and Dr. Dimitrije Pancic (University of Nis, Serbia) for encouragement and support throughout the process of doing this research. We would also like to express our gratitude to Associate Professor Milan Latas (University of Belgrade, Serbia) for moral support and advice and to Ivona Ilic for her generous help.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to declare.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/brb3.1881.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.