Burnout, depersonalization, and anxiety contribute to post‐traumatic stress in frontline health workers at COVID‐19 patient care, a follow‐up study

Abstract Introduction We designed a follow‐up study of frontline health workers at COVID‐19 patient care, within the same working conditions, to assess the influence of their general characteristics and pre‐existing anxiety/depression/dissociative symptoms and resilience on the development of symptoms of post‐traumatic stress disorder (PTSD), while monitoring their quality of sleep, depersonalization/derealization symptoms, acute stress, state anxiety, and burnout. Methods In a Hospital reconfigured to address the surge of patients with COVID‐19, 204 frontline health workers accepted to participate. They completed validated questionnaires to assess mental health: before, during, and after the peak of inpatient admissions. After each evaluation, a psychiatrist reviewed the questionnaires, using the accepted criteria for each instrument. Correlations were assessed using multivariable and multivariate analyses, with a significance level of .05. Results Compared to men, women reporting pre‐existing anxiety were more prone to acute stress; and younger age was related to both pre‐existent common psychological symptoms and less resilience. Overall the evaluations, sleep quality was bad on the majority of participants, with an increase during the epidemic crisis, while persistent burnout had influence on state anxiety, acute stress, and symptoms of depersonalization/derealization. PTSD symptoms were related to pre‐existent anxiety/depression and dissociative symptoms, as well as to acute stress and acute anxiety, and negatively related to resilience. Conclusions Pre‐existent anxiety/depression, dissociative symptoms, and coexisting acute anxiety and acute stress contribute to PTSD symptoms. During an infectious outbreak, psychological screening could provide valuable information to prevent or mitigate against adverse psychological reactions by frontline healthcare workers caring for patients.


| INTRODUC TI ON
First responders to epidemics are vulnerable to psychological distress, with influence from the potential to be both victim and service provider (Brooks et al., 2016). The most severe disorder elicited by exceptionally stressful life events is post-traumatic stress disorder (PTSD) (APA, 2013), which has been related to depersonalization (altered perceptions of the self and the environment) (Spiegel, 1997).
On the contrary, the ability to adapt successfully in the face of adversity (resilience) may be protective to most people (Bonanno & Mancini, 2012).
The psychological impact of the COVID-19 pandemic on frontline health workers has been assessed around the world. However, the majority of studies available are cross-sectional surveys that were performed remotely. In Italy, a web-based cross-sectional study on 1,379 health workers showed that younger age and female gender were related to adverse reactions (Rossi et al., 2020), with symptoms of post-traumatic stress in 49.3% of respondents, severe depression in 24.7%, anxiety in 8.2%, insomnia in 19.8%, and high perceived stress in 21.9%. In Spain, an online survey was conducted on 1,422 health workers (1,228 women), displaying that emotional exhaustion and depersonalization contribute to psychological distress, while resilience and personal fulfillment can be protective factors (Luceño-Moreno et al., 2020), with symptoms of post-traumatic stress disorder in 56.6% of the respondents. In the United States of America, during a peak of inpatient admissions, a cross-sectional web survey conducted on 657 physicians and nurses showed that especially nurses and advanced practice providers were experiencing psychological distress (Shechter et al., 2020), with 57% positive screen for acute stress, 48% for depressive symptoms, and 33% for anxiety symptoms. A systematic review and meta-analysis to examine the pooled prevalence of depression, anxiety, and insomnia on 33,062 health workers, who participated in 13 cross-sectional studies (mainly in China), showed female gender and nurses exhibiting higher rates of affective symptoms, compared to male gender and physicians, respectively (Pappa et al., 2020), with a prevalence of symptoms of anxiety and depression higher than 20%, and a prevalence of sleeping difficulties of 38.9%. A systematic review of 117 studies (65% conducted in Asian countries), including 119, 189 participants, also supports that younger age and female gender are risk factors for psychological distress during an infectious outbreak (Serrano-Ripoll et al., 2020), with a pooled prevalence of 40% for acute stress (6,949 participants), 30% for anxiety (43,751 participants), 28% for burnout (1,168 participants), 24% for depression (61,463 participants), and 13% for post-traumatic stress disorder (24,540 participants).
During epidemics, risk factors for adverse psychological reactions can be related to both context and individual factors (Brooks et al., 2018;Kisely et al., 2020). During the COVID-19 pandemic, several context factors have been related to stress and burnout (Raudenská et al., 2020), including limited resources, threat of exposure to the virus, longer shifts, sleep disruption, and work-life balance with increased workload. However, the main individual factors related to psychological distress include age, gender, and occupation (Luceño-Moreno et al., 2020;Pappa et al., 2020).
We designed a follow-up study of health workers at the frontline of COVID-19 patient care, within the same working conditions, to assess the influence of their general characteristics and pre-existing anxiety/depression/dissociative symptoms and resilience on the development of symptoms of post-traumatic stress disorder (PTSD), while monitoring their quality of sleep, depersonalization/derealization symptoms (DD), acute stress, state anxiety, and burnout.

| Participants
The study protocol was approved by the institutional Research and gave answer to all the questionnaires at the 3 evaluations (Appendix S1). The 33 (13.9%) health workers who accepted to participate, but did not complete the study protocol, were 26-58 years old (mean age 34.3 years), 23 were women, and 10 were men, and they were working in clinical areas (n = 22), laboratory and imaging (n = 4), and support areas (n = 2). Among them, five workers were absent from the Hospital during the peak of inpatient admissions, due to sickleave (n = 3) or vacations (n = 2); three were transferred to another Hospital; one died (COVID-19); 18 responded incomplete questionnaires; and five workers drop out of the study before the peak of | 3 of 9 MIGUEL-PUGA Et AL. inpatient admissions, while one drop out after the peak of admissions (Appendix S1).

| Procedures
The study protocol included three evaluations: (1) The first evaluation was performed, while clinical spaces were reconfigured for COVID-19 and before clinical teams were reorganized, (2) the second one, during the peak of inpatient admissions, and (3) the last one, just before clinical spaces were opened again for patients with diseases other than COVID-19.
In the first evaluation, we performed a short assessment on demographics, contact with patients or biological samples of COVID-19, and a medical history. Then, a psychological screening was performed using the following self-administered questionnaires: Hospital Anxiety and Depression Scale (HADS) by Zigmond and Snaith (1983), Dissociative Experiences Scale (DES) by Bernstein and Putnam (1986) and Resilience scale by Connor et al. (2003).
The follow-up of the participants was performed by administering the next inventories at the 3 time points: Pittsburgh Sleep Quality Index by Buysse et al. (1989), Depersonalization/derealization inventory (DD) by Cox and Swinson (2002), Stanford Acute Stress Questionnaire by Cardena et al. (2000), the short-form of the State-Trait Anxiety Inventory (STAIsv) by Marteau and Bekker (1992) and the short version of the Burnout Measure by Malach-Pines (2005). At the end of the third evaluation, we also administered the Posttraumatic Stress Disorder Symptom Severity Scale-Revised by Echeburúa et al. (2016). After each of the three evaluations, a psychiatrist reviewed all the questionnaires, using the accepted criteria for each instrument.

| Analysis
Statistical analysis was performed using STATISTICA software (StatSoft Inc.). The significance level was set at 2-tailed α = 0.05.
We assessed data distribution using Kolmogorov-Smirnov test; since the scores of the inventories were not normally distributed, they are presented as medians with the first and third quartiles (Q1-Q3). To assess linear correlations among the inventory scores, we used Pearson's correlation coefficient. To assess variations through time on the proportion of participants with bad quality of sleep and those with burnout, we used the Cochran Q test. To assess variations through time on the scores of the DD inventory, the acute stress questionnaire and the STAIsv, considering influence from age, gender, and occupation, we used repeated measures multivariate analysis of covariance. To assess the influence of age, gender, and occupation on the scores of the HADS, DES, and Resilience scale, as well as their influence on the PTSD score, considering the inventories administered at each of the 3 time points of evaluation, we performed multivariable regression analyses, using a generalized linear model (with log transformation) and Wald test.

| Ethical approval
The study protocol was approved by the Local Institutional Research and Ethics Committees (R-2020-3601-042). The study protocol was conducted in accordance with the Declaration of Helsinki and its amendments. Table 1 shows the inventory scores of the 204 participants who completed the study protocol. Noteworthy, the 33 health workers who did not complete the study protocol showed similar scores than those observed on the 204 participants, with median total score on the HADS of 11 (Q1-Q3 = 4-17.5), on the DES of 3.5 (1.9-7.8), and on the Resilience scale of 76 (68-85). Table 2 shows the results of the regression analysis on the influence of gender, age, and occupation of the 204 participants on the score of the HADS, DES, and Resilience scale. There was no independent influence from gender on any of the three scales, while the age of the participants had influence on the scores of the three scales, with a negative relationship with HADS and DES and a positive relationship with Resilience; and occupation had influence on the DES, with higher scores on support workers than clinical workers. In addition, the interaction between gender and occupation had influence on the scores of the Resilience scale and DES among women, and just on the DES among men.

| Follow-up assessments
During the three evaluations, good correlations were observed among the scores of all the inventories, including those administered for psychological screening and those administered during follow-up (Appendix S2).
In the three evaluations, the quality of sleep was bad (score > 5) in the majority of participants, with good correlation among the scores obtained during each evaluation (Pearson's r from 0.58 to 0.61, p < .0001; Appendix S2). Repeated measures multivariate analyses on the scores of the sleep quality index, the STAI, the DD inventory, the acute stress TA B L E 1 Median and quartiles 1 & 3 of the inventory scores that were administered to 204 health workers, (1) before, (2) during, and (3) after the peak of inpatient admissions

| D ISCUSS I ON
The results of this follow-up study support that pre-existing anxiety/depression and dissociation symptoms may contribute to the development of PTSD symptoms in frontline health workers, during a peak of inpatients admissions at COVID-19 patient care, with protective influence from pre-existing resilience, while persistent burnout may contribute to depersonalization symptoms and acute stress.
In addition, before, during, and after the epidemic crisis, women with anxiety symptoms may be more prone to acute stress than men.
Burnout and stress among healthcare workers have been associated with negative impacts on the individual, patients and healthcare system (Walton et al., 2020;West et al., 2018). In this study, several factors may have contributed to a lower frequency of burnout and PTSD, compared to reports from a variety of countries (Dobson et al., 2020;Luceño-Moreno et al., 2020;Matsuo et al., 2020). In this study, the scores on the resilience scale were high, with an inverse relationship with all the scales of psychological distress. The study was performed in a Hospital reorganized to address the surge of patients with COVID-19, including balance between shifts and resting time, as well as task assignment according to professional profile, which F I G U R E 2 Mean and standard error of the mean of the score on the shortform of the State-Trait Anxiety Inventory (STAIsv) of 204 healthcare workers, according to burnout before and after the peak of inpatients admissions, computed at mean scores on the Resilience scale and the Dissociative Experiences Scale ( The main limitation of this study was the recruitment from a single hospital, which may not be representative of other hospitals.
However, this setting endorsed control for the working conditions and direct interaction with the participants. A second limitation of the study was the sample size, which allowed us to address just the more evident relationships among the study variables. In addition, the time frame of the study was not enough to allow assessment of long-term mental health outcomes. Another limitation was the reliance on self-report; since the study was designed to assess psychological reactions of busy workers, we had to rely on a selected

TA B L E 3
Results of the regression analysis on the evidence of post-traumatic stress disorder after the peak of inpatient admissions; including the general characteristics of the 204 participants, the scores on the psychological screening inventories, and the scores on the questionnaires administered (1) before, (2) during, and (3) after the peak of inpatient admissions set of self-administered questionnaires, which were reviewed by the same psychiatrist; besides, since the Hospital was reorganized for the surge of patients with COVID-19, no adequate controls were available to assess changes over time; however, the stability of the follow-up measurements supports that the findings were not just time related.

| CON CLUS IONS
In health workers at the frontline of COVID-19 patient care, within the same environment and working circumstances, the assessment of the influence of individual predisposing factors to adverse psychological reactions to peak inpatient admissions showed that: • Pre-existent anxiety/depression, dissociative symptoms, as well as coexisting acute anxiety and acute stress contribute to PTSD symptoms.
• Persistent burnout may contribute to state anxiety, acute stress, and depersonalization symptoms.
• Compared to men, women reporting pre-existent anxiety may be more prone to acute stress.
• The quality of sleep of healthcare workers may be bad, even before the epidemic crisis, and possibly contribute to stress.
• Younger age could be related to pre-existent common psychological symptoms and less resilience.
• Pre-existent resilience may have favorable influence both during and after an epidemic crisis.
During an infectious outbreak, in order to protect the mental health of vulnerable personnel and decrease the probability of latent human errors, psychological screening could provide valuable information to prevent or mitigate against adverse psychological reactions by frontline healthcare workers caring for patients.

CO N FLI C T O F I NTE R E S T
All authors report no disclosures relevant to the manuscript.

AUTH O R CO NTR I B UTI O N
JAMP and DCB executed the research project; were data curators; and reviewed/critiqued the data analysis and the manuscript. FJAG, LASH, TCM, EEP, JCAG, JIGD, OBSS, and LCVA executed the research; organized the data; and reviewed/critiqued manuscript preparation. KJR conceived, organized, and executed research project; designed and executed data analysis; and first drafted and reviewed/ critiqued the manuscript.

E TH I C A L A PPROVA L
The study protocol was approved by the Local Institutional Research and Ethics Committees. The study protocol was conducted in accordance with the Declaration of Helsinki and its amendments.

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/brb3.2007.