Prevalence and Determinants of Immediate and Long-Term PTSD Consequences of Coronavirus-Related (CoV-1 and CoV-2) Pandemics among Healthcare Professionals: A Systematic Review and Meta-Analysis

Background: The COVID-19 pandemic continues to rise. In order to control the COVID-19 pandemic, healthcare professionals have been subjected to increased exposure to work stress. In this systematic review, we aimed at investigating the prevalence and determinants of immediate and long-term post-traumatic stress disorder (PTSD) effects on healthcare professionals by the COVID-19 (SARS CoV-2) and SARS-2003 (SARS CoV-1) pandemics. Methods: This systematic review was conducted according to the recommendations of the Protocols for Systemic Review and Meta-Analysis (PRISMA) statement. Only studies reporting the prevalence of PTSD (frequency, percentage) and related risk factors (adjusted odds ratio (OR)) in healthcare professionals (HCPs) during the SARS CoV-2 and SARS CoV-1 pandemics were included. The following databases were screened: Medline, Embase, PsychINFO, and Health Psychosocial Instrument (HaPI). Results: Six of eight studies reported PTSD symptoms among healthcare professionals during the COVID-19 pandemic in China (three), Singapore (one), India (one), and the United States of America (USA) (two), while two studies reported symptoms during the SARS-2003 pandemic in China (one) and Singapore (one). Sample sizes ranged from 263 to 5062 with a combined total of 10,074 participants. All of the studies self-reported the level of exposure to coronaviruses (CoV-1 and CoV-2) and severity of PTSD. Seven studies reported the prevalence of immediate PTSD and determinants, while one study reported delayed-onset PTSD (3 years after CoV-1 pandemic). Determinants of immediate PTSD were reported for the CoV-2 pandemic, while those for long-term PTSD were reported for the CoV-1 pandemic. Conclusions: A comprehensive understanding of the prevalence and determinants of immediate or long-term pandemic PTSD for healthcare workers can improve prevention, diagnosis, and management. Rigorous research measuring the prevalence of PTSD and its associated risk factors (adjusted OR) for the CoV-2 pandemic are envisaged. Although strategies to resolve immediate PTSD are key, long-term PTSD must not be overlooked.


Introduction
Early in December 2019, a newly discovered infectious coronavirus (CoV-2) disease occurred in Wuhan, Hubei Province, China [1]. The appearance of coronavirus CoV-2 did not prove to be an easy matter. The rapid global spread of the disease led to the declaration of a pandemic on 11 March 2020 [1]. It was subsequently termed Coronavirus Disease 2019 (COVID-19) [2]. The primary reported manifestation of COVID-19 is severe acute respiratory distress syndrome (SARS), ultimately leading to death in the most severe cases [3]. It was named CoV-2 since CoV-1 appeared as a pandemic in 2003 and caused a similar effect of SARS [1]. CoV-2 and CoV-1 share similarities at several levels. Both viruses have a high degree of homology and they share similar clinical features and disease dynamics [4]. The disease progression follows a similar trend for CoV-2 and CoV-1, with SARS occurring approximately 8-20 days after the first symptoms [5]. While this has yet to be confirmed, the two outbreaks appear to have a different epidemic trajectory. Whereas the SARS outbreak was brought under control in a matter of 8 months in 2003, the new pandemic in 2019 continues after more than a year. Although shorter in duration, the CoV-1 pandemic caused a great deal of distress not only in patients but also in healthcare professionals (HCPs), being named a "mental health catastrophe" [6]. Although both CoV-1 and CoV-2 pandemics resulted from coronaviruses, their impacts on healthcare professionals were different. Several determinants of occupational and psychosocial distress have been reported in HCPs during and after the CoV-1 pandemic, suggesting the need to establish strategies to help physicians through measures including timely knowledge sharing, effective infection control practices, income protection during outbreaks, and attention to family risk management [7]. Given the considerably higher severity of the CoV-2 compared to the CoV-1 pandemic, lessons could be gained by evaluating the burdens of pandemics on HCP psychosocial distress, such as post-traumatic stress disorder (PTSD) [8]. Although the prevalence of PTSD in the CoV-2 pandemic can be predicted to be higher, common determinants of PTSD may be shared by both CoV-1 and CoV-2 pandemics.
The three hallmark features of a traumatic event as defined in DSM-5 (Diagnostic and Statistical Manual of Mental Disorders) are unpredictability, uncontrollability, and the threat of death or serious injury [9,10]. Pandemics, described as traumatic incidents, trigger a great deal of concern for HCPs and the health authorities. HCPs are passing through unprecedented challenges with the COVID-19 pandemic in many aspects. They are risking their own health and lives, threatened not only by exposure to the coronavirus but also by ever-increasing stress at the workplace, with amounts to a parallel the pandemic itself. Increased exposure to work-related stress has been associated with deleterious effects for mental health with higher rates of anxiety disorders. The recognition of early and long-term PTSD in healthcare professionals is becoming increasingly relevant for health policymakers to develop preventive measures to mitigate or avoid PTSD and related diseases. Dutheil et al. [1] described PTSD as the tsunami of the COVID-19 pandemic. PTSD is a severe mental health condition caused by an unusual traumatic life event beyond the normal range of human experience [11]. HCPs are under enormous pressure to manage this disaster, as they have to reorganize resources and the workforce to manage an unusual medical emergency. Being worried about their health, their families' health, contagion, and their colleagues' safety make them more prone to acute distress and potentially chronic PTSD [1,3]. Furthermore, the response by worldwide governments involving quarantine, social distance, lockdown measures, and media might contribute to PTSD [1,11,12]. We could, thus, overcome the pandemic, but then face a global public mental health crisis [13]. The pooled prevalence of PTSD symptoms among HCPs exposed to the COVID-19 pandemic ranged from 13% (95% confidence interval (CI): 11%, 16%) to 20.7% (95% CI: 13.2%, 31%) [14,15]. Previous systematic and meta-analysis reviews on the COVID-19 pandemic reported combined odds ratios (ORs) of risk factors for mental disorders (anxiety, depression, occupational stress, PTSD, and insomnia) among HCPs [16,17]. However, to our knowledge, no systematic reviews investigated risk factors associated with immediate and long-term PTSD related to the SARS (CoV-1 and CoV-2) pandemics.
In this systematic review, we aimed at investigating the scientific evidence on immediate and long-term PTSD effects on healthcare professionals due to the COVID-2019 (SARS CoV-2) and SARS-2003 (SARS CoV-1) pandemics. We systematically reviewed the literature for the prevalence of PTSD symptoms and associated risk factors (adjusted OR) due to exposure to coronavirus pandemics COVID-2019 (SARS CoV-2) and SARS-2003 (SARS CoV-1) among healthcare professionals (HCPs).

Methods
This systematic review was conducted according to the recommendations of the Protocols for Systemic Review and Meta-Analysis (PRISMA) statement [18]. We included studies that focused on HCPs during pandemics COVID-2019 (SARS CoV-2) and SARS-2003 (SARS CoV-1). We narrowed our search to studies that focused on PTSD. The literature search was done on the following databases: Medline, Embase, PsychINFO, and Health Psychosocial Instrument (HaPI). The search was limited to primary studies published in peer-reviewed journals and in the English language from 1 January 2003 to 15 November 2020. As no cohort or case-control studies were found, only cross-sectional studies were included that reported prevalence of PTSD (frequency, %) and associated risk factors (adjusted OR) using multivariate regression analysis. Qualitative interview-based studies, review articles, editorials, opinion or letter articles, and studies on students or trainees were excluded. The studies retrieved from the search were imported into Covidence ® . Covidence ® is an online systematic review software management tool which allows uploading search results, screening abstracts and full-text study reports, completing data collection, conducting risk of bias assessment, and resolving disagreements (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia. Available at www.covidence.org, accessed on 22 February 2021). Duplicated papers were removed. Three groups consisting of two residents (B.a.F. and M.a.D., M.a.A. and M.a.M., and A.a.A. and M.a.K.) shared the systematic reviewing process equally and independently screened the title and abstract of each article according to their relevance. A second screening process was undertaken which involved reviewing the full text for each article to assess its eligibility. Any disagreement between the reviewing groups was resolved by three senior supervisory members M.D., M.a.Q., and H.a.T. Figure 1 shows study selection process. Data extraction was done by three investigators using a structured form. This included study information (author name, year of population, country, and study design), population (study sample, total number of participants, age, data collection time period, and follow-up period), exposure (definition, measurement, and categorization by degree of exposure), outcome (psychosocial outcomes, definitions, ascertainment, and classification by severity), confounders, point of prevalence estimate for outcomes (frequency, %) exposure-outcome association measure (adjusted OR), and level of statistical significance (95% confidence intervals). As the studies included in this systematic review were all cross-sectional, the Joanna Briggs Institute Checklist for Analytical Approach for cross-sectional studies was used to assess their methodological quality [19].

Results
The electronic search identified 822 candidate studies, 706 of which remained after the elimination of duplicates ( Figure 1). A total of 599 publications were excluded, and 107 remained for title and abstract screening, of which 44 were excluded and 63 remained for full-text examination, of which 55 were excluded, as the goal was to include crosssectional and cohort studies that reported the prevalence of PTDS and carried out a regression analysis of possible risk factors. The final search resulted in a total of eight studies enrolled in the qualitative synthesis. The eight studies were of cross-sectional design and are summarized by characteristics in Table 1, quality in Table 2, and estimates (prevalence and odds ratio) in Table 3 [3,[20][21][22][23][24][25][26]. Six of the eight studies reported PTSD symptoms among healthcare professionals during the COVID-19 pandemic in Asia (China [20][21][22], Singapore [3], and India [3]) and North America (United States of America [23,24]), while two studies were undertaken during the SARS-2003 pandemic in Asia (China [25] and Singapore [26]).

Results
The electronic search identified 822 candidate studies, 706 of which remained after the elimination of duplicates ( Figure 1). A total of 599 publications were excluded, and 107 remained for title and abstract screening, of which 44 were excluded and 63 remained for full-text examination, of which 55 were excluded, as the goal was to include cross-sectional and cohort studies that reported the prevalence of PTDS and carried out a regression analysis of possible risk factors. The final search resulted in a total of eight studies enrolled in the qualitative synthesis. The eight studies were of cross-sectional design and are summarized by characteristics in Table 1, quality in Table 2, and estimates (prevalence and odds ratio) in Table 3 [3,[20][21][22][23][24][25][26]. Six of the eight studies reported PTSD symptoms among healthcare professionals during the COVID-19 pandemic in Asia (China [20][21][22], Singapore [3], and India [3]) and North America (United States of America [23,24]), while two studies were undertaken during the SARS-2003 pandemic in Asia (China [25] and Singapore [26]).  Age, gender, presence of comorbidities

SARS-2003 (SARS CoV-1)
Chan [26] Sample size ranged from 263 to 5062 with a combined total of 10,074 participants. All of the studies self-reported level of exposure to coronaviruses (CoV-1 or CoV-2) and severity of PTSD symptoms using paper or online surveys with a response rate above 80% [3,25], of 60-80% [20][21][22]26], and below 15% [23,24]. Three studies were conducted in a single hospital [21,25,26], three studies were conducted in multiple hospitals [3,20,22], one study was conducted in a single state [23], and two studies were conducted in multiple states or provinces [20,24]. Two studies were investigated merely nursing staff [22,23], one study had a majority of female nursing staff (93.6%) [23], one study recruited just otolaryngology physicians [24], and two studies focused on nurses and doctors [20,26]. Three studies included allied HCPs and administrates [3,21,25]. All studies reported prevalence of acute or immediate PTSD symptoms during or immediately after the pandemic except one study [25] which reported delayed-onset or long-term PTSD (3 years after SARS-2003 pandemic). Evidently, to date, there is no study reporting long-term PTSD after CoV-2 pandemic.
All studies specified and categorized level of exposure with slight variations except one study [3] which did not report how the level of exposure was categorized. All studies used the Impact Event Scale (IES, 22-items, [3,[20][21][22]25] and 15-items, [24,26]) with different cutoff scores to detect PTSD symptoms and their severity in the past 7 days except one study [25] which evaluated the past 4 weeks 3 years post SARS-2003 outbreak, whereas another study [23] used the six-item Post-Traumatic Checklist (PCL-6) in the past 4 weeks. All studies used logistic regression model to control and adjust for different risk factors and confounders.

Quality Assessment
Total quality scores for methodology were seven [20], six [3,21,[23][24][25], and five [22] out of eight on the Joanna Briggs Institute Checklist for Analytical Approach for cross-sectional studies. None of studies used objective measurements or clinical interviews for PTSD symptoms (Table 2).

Prevalence and Determinants of Acute PTSD Symptoms
Regarding SARS CoV-1, the prevalence estimate of acute PTSD symptoms for Singapore was 3% [26].

Determinants of Acute or Immediate PTSD
The following variables have been shown by multivariable logistics regressive research to correlate with a higher risk of acute PTSD symptoms ( Figure 2) [3] found a bidirectional positive association between physical symptoms and PTSD, OR 2.70 (95% CI: 1. 40, 5.24) and between PTSD and physical symptoms, OR 2.20 (95% CI: 1.12, 4.35).

Discussion
In this systematic review and meta-analysis, results were pooled from eight real-world observational studies that reported the prevalence of PTSD (frequency, %) and associated risk factors (adjusted OR) on healthcare professionals (HCPS) during pandemics SARS CoV-2 and SARS CoV-1. These studies mostly addressed clinically significant PTSD in the acute phase during and immediately following a pandemic, while only one study addressed long-term PTSD. As evidenced by the data collected in this review, PTSD symptoms showed high prevalence variations ranging from 3.4% [3] to 71.5% [20] among HCPs. Determinants of immediate PTSD were reported for the CoV-2 pandemic, while those for long-term PTDS were reported for the CoV-1 pandemic.
While very limited data on the course of clinically important PTSD were available, our findings can be considered broadly consistent with the results of two meaningful pandemic meta-analyses on HCPs' mental health and pandemics [27,28]. While Pappa et al. studied the prevalence of depression, anxiety, and insomnia among HCPs during the COVID-19 outbreak [27], Allan et al. investigated the prevalence of common and stress-related mental health disorders in healthcare workers based in pandemic-affected hospitals [28]. Compared to Allan et al.'s meta-analysis [28], we limited our results to the SARS studies reporting odds ratios, while others did not [29][30][31][32][33][34] or had too small a sample size for regression analysis [35,36].
A series of early intervention trials aimed at individuals who were seen to be at high risk of PTSD development were prompted by a diagnosis of acute stress disorder [37]. Delayed-onset PTSD is defined as PTSD that develops at least 6 months after exposure to trauma, with cases of PTSD reportedly commencing years after the trauma occurrence [38]. Importantly, diagnosis during the acute phase after trauma is not intended to predict subsequent PTSD, but rather to describe people with elevated distress in the initial month who may benefit from mental health services [39]. No linear association could be identified between the severity of acute PTSD and the severity of delayed PTSD onset [40].

Prevalence of PTSD
In China, the highest prevalence reported was 71.5% [20] for acute PTSD symptoms among HCPs, while the lowest prevalence was 25.1% [22]. A possible explanation for the variation is the data collection period, as Lai et al. collected data from 29 January to 3 February 2020, i.e., during the initiation and acceleration phase of the COVID-19 outbreak curve and complete lockdown of Wuhan city, Hubei province, while Nie et al.'s data collection period was 3-10 February 2020 in Guangdong province, which was considered less affected by COVID-19 compared to Wuhan city, Hubei province [41][42][43][44].
In the USA, Civantos et al. [24] reported twofold higher prevalence of acute PTSD symptoms of 60.2%, compared to Arnetz et al. [23] (26.5%). One possible explanation for this discrepancy is the different use of self-reported tools used to ascertain PTSD. Civantos et al. [24] used the 15-item revised impact event scale (IES-R15) to detect symptoms over past 7 days among nurses (estimated response rate: 4%) during the month of May 2020, while Arnetz et al. [23] used the six-items Post-Traumatic Checklist (PCL-6) to detect PTSD symptoms in the past 4 weeks among otolaryngology physician (estimated repose rate: 10.2%) during the month of April.
The lowest prevalence of PTSD symptoms was reported in Singapore at 3% and 4% [3,26] and India at 3.4% [3]. There is a high variability in the association of traumatic events with PTSD, as it is not necessary for everyone exposed to a potentially traumatic event to develop a disorder [45].
Wu et al. [25] reported a prevalence of 10% for persistent PTSD symptoms among HCPs 3 years post SARS-2003 pandemic. Similar findings were reported by Maunder et al. [30], suggesting that impact of SARS can persist 1 to 2 years after the outbreak among HCPs compared with colleagues in settings that did not treat SARS patients.

Determinants of PTDS
Included studies in this review investigated varied determinants or risk factors associated with high prevalence of PTSD symptoms among HCPs. Gender, age, experience (working years), and degree of exposure (level, load, and amount of close contact or care) variables commonly showed significant association with high PTSD symptoms among different studies in different countries on different continents. As PTSD determinants, personal characteristics, conditions, and energies are resources of both instrumental and symbolic usefulness. Personal characteristics such as a positive attitude, view of events as predictable, and the ability to deal with stress are ways to help protect against stress [46]. Personal conditions, such as job training and seniority, are resources that are valued and sought after by an individual [47]. Any provided condition may be a resource for one person but may be harmful for another [48]. Personal energy resources include time, money, and expertise. This type of resource is useful, since it can be used to obtain additional resources [48,49].

Strengths and Limitations of the Study
This review had several strengths. It included research on two pandemics in Asia and North America, including investigation of PTSD symptoms with further regression analyses of possible risk factors. It demonstrated and highlighted the immediate and delayed traumatic effects of two pandemics in various settings and subpopulations of HCPs. This systematic review, therefore, provides valuable information and may serve as a guide for policymakers to establish and provide guidance on immediate and long-term traumatic stress among HCPs.
This review had several limitations. The search was limited to English language only and only those published in peer-reviewed journals; thus, it could be prone to publication bias. The studies were all cross-sectional studies, with a short follow-up period; therefore, the studies could not prove causality. Since all studies were self-reported with different cutoff scales and different versions of the Impact Event Scale (IES), and since no study utilized standardized clinical interviews for diagnosis as a confirmation tool, it is unknown whether the presence of an actual disorder existed, which could have led to misclassification and recall biases. However, in such pandemic situations, this was not a feasible way to conduct studies as direct contact was restricted. Nevertheless, it could be possible to investigate delayed or persistent PTSD once the pandemic is over. Furthermore, the majority of studies were heterogeneous in their operational assessment of exposure and used population screening scales to assess PTSD determinants with different cutoff scores. Therefore, summarizing the prevalence of PTSD and its determinants among HCPs with a single-point estimate was difficult, and reported findings need to be interpreted with cautions.
When used alone, the OR, which remains a representation of the power of association between the risk factor and the onset of PTSD, offers little detail. Meta-analyses represent an accumulation of knowledge that can often lead to meanings that do not address the clinical validity issue and leave the decision to be made in confusion [50]. This approach must then be used with experience and moderation and, where possible, should be supplemented by additional research illustrating the clinical validity of the meanings acquired.

Conclusions
A comprehensive understanding of the prevalence and determinants of immediate or long-term PTSD related to pandemics for healthcare workers may enhance its prevention, diagnosis, and management. Working experience, occupation, protective conditions, concomitant chronic diseases, present physical symptoms, mental disorder, negative coping style, and family-related factors were identified as determinants of an immediate PTSD. Exposure to SARS, being in quarantine, high work exposure, and age were reported as determinants of a long-term PTSD. The risk of immediate or long-term PTSD can be decreased on the basis of identifiable risk factors.
Studies are yet to investigate the long-term consequences of PTSD after a CoV-2 pandemic over longer periods of time. Rigorous studies evaluating the prevalence of PTSD and its related risk factors (adjusted OR) for the CoV-2 pandemic are envisaged. While strategies to tackle immediate PTSD are key, strategies for long-term PTSD must not be ignored. Data Availability Statement: Data sharing not applicable. No new data were created or analyzed in this study. Data sharing is not applicable to this article.