Long-term psychological outcomes among frontline healthcare workers following the COVID-19 pandemic: A systematic review

Abstract

BACKGROUND

Studies of short-term mental health (MH) outcomes among frontline (FL) healthcare workers (HCWs) during the pandemic have been extensively published and have reached dramatically variable conclusions. Long-term MH outcomes in this population have received less study. A synthesis of available data on long-term MH outcomes among FL HCWs associated with the coronavirus disease 2019 (COVID-19) pandemic is needed to inform efforts for screening, mobilizing psychiatry resources, and preparing for the long-term MH consequences of future pandemics.

AIM

To determine the prevalence of MH sequelae of FL HCWs 1-3 years after the COVID-19 pandemic.

METHODS

All studies published between March 1, 2023, and December 1, 2025, that reported on the post-pandemic MH of FL HCWs affected by the COVID-19 pandemic, such as insomnia, stress, burnout, anxiety, and depression, were identified using PubMed, PsycINFO, Scopus, EMBASE, Cochrane Database of Systematic Reviews, Google Scholar, MEDLINE, and Web of Science. Articles identified in the search were screened using the approaches recommended by the PRISMA 2020.

RESULTS

Of the 1819 articles, 1679 were excluded after review of the title and abstract, 140 were reviewed at length, and 19 were included. The available data (entirely based on symptom screening measures) suggest that FL HCWs are at risk for stress-related, anxiety, depressive, and sleep disorders at least 1 year after the end of the pandemic.

CONCLUSION

These results suggest that enhanced access to psychiatry services is warranted in HCW populations exposed to pandemics.

Key Words: COVID-19; Mental health; Front-line healthcare workers; Long-term psychological outcomes; Psychiatry services

Core Tip: Healthcare workers (HCW) are at risk for stress-related, anxiety, depressive, and sleep disorders at least 1 year after the end of the pandemic. These results suggest that enhanced access to psychiatry services is warranted in HCW populations exposed to pandemics.

INTRODUCTION

Postdisaster psychopathology has been found in up to 50% of disaster survivors, especially among those highly exposed to severe events such as the Oklahoma city bombing and the September 11 terrorist attacks[1,2]. Much of this postdisaster psychopathology has been found to persist in disaster survivors years and even decades later[1,2]. The coronavirus disease 2019 (COVID-19) pandemic was distinct from other disasters, most of which represented catastrophic one-time events. According to the World Health Organization, the COVID-19 pandemic began in March 2020 and ended in May 2023, with the highest rates of hospital admissions for COVID-19 occurring between March 2020 and March 2022[3]. Studies of short-term psychological outcomes among frontline (FL) healthcare workers (HCWs) during the pandemic have been extensively published and have reached dramatically variable conclusions[4]. Long-term psychological outcomes in this population have received less study. A synthesis of available data on long-term psychological outcomes among FL HCWs associated with the COVID-19 pandemic is needed to inform efforts for screening, mobilizing psychiatry resources, and preparing for the long-term psychological consequences of future pandemics. The objective of this systematic review is to determine the prevalence of psychological sequelae of FL HCWs 1-3 years after the COVID-19 pandemic.

MATERIALS AND METHODS

All studies published between March 1, 2023, and December 1, 2025, that reported on the post-pandemic mental health (MH) of FL HCWs affected by the COVID-19 pandemic, such as insomnia, stress, burnout, anxiety, and depression, were identified using PubMed, PsycINFO, Scopus, EMBASE, Cochrane Database of Systematic Reviews, Google Scholar, MEDLINE, and Web of Science. Institutional review board, ethics committee approval, and informed consent were not needed because data were obtained from existing published literature. Computer-based search terms and inclusion and exclusion criteria were based on prior systematic reviews and meta-analyses conducted to evaluate psychological outcomes among FL HCWs during the pandemic (Table 1 and Supplementary Table 1)[4,5]. Studies published in peer-reviewed journals that reported data on FL HCWs and used a validated method including psychiatric screening tools or diagnostic psychiatric interviews to assess post-pandemic psychological outcomes of FL HCWs were included. Only studies that collected data primarily after March 1, 2023, were included. March 1, 2023 was chosen because this date was 1 year following the dramatic decline in hospitalizations from COVID-19 when the majority of healthcare facilities in the United States and around the world returned to normal operations and several months after China lifted its social blockade and epidemic control measures on January 8, 2023[3,6]. Articles identified in the search were screened using the approaches recommended by the PRISMA 2022[7].

Table 1 Inclusion and exclusion criteria.

Inclusion	Exclusion
Published in English	All languages other than English
Data collected on or after March 1, 2023	Data collected before March 1, 2023
All or part of the analyses targeted FL HCWs caring for patients with COVID-19: Physicians, nurses, advanced practice providers	Reports limited to the context of another pandemic (e.g., Middle East Respiratory Syndrome, influenza) or event (e.g., earthquake or another natural event) not related to the COVID-19 pandemic
Reporting quantitative data (e.g., prevalence rates) on psychological problems experienced by FL HCWs after the COVID-19 pandemic	Reports limited to other professionals not working in the health system (i.e., police officers, firefighters, teachers) or healthcare students
Reporting quantitative data (e.g., prevalence rates) on psychological problems (i.e., depressive symptoms, anxiety symptoms, insomnia, posttraumatic stress disorder symptoms) and other psychological data, including burnout symptoms, sleep problems, psychological stress and distress)	Reports limited to HCWs who did not perform hospital-based work as a member of the primary healthcare team on hospitalized COVID-19 patients (i.e., dentists, orthopedic surgeons)
	Reports limited to types of qualitative data other than prevalence rates or results of interventional studies
	Reports limited to data on emotions (e.g., fear) or behaviors (e.g., physical activity, smoking)
	Reports limited to surges of COVID-19 infection occurring locally at time of study (i.e., several provinces in China had surges in early 2023; studies reporting data collection during these surges were excluded)

FL HCW: Frontline healthcare workers; COVID-19: Coronavirus disease 2019.

This study’s first author independently extracted the following information from each article into a systematic form for recording details of study design, country, survey period, specialty, sample size, average age, screening instrument used, screening instrument results, and reported prevalence of insomnia, stress, anxiety, and depression. Multiple studies of the same population and multiple identical publications were not duplicated in the review. When FL HCWs were not specified as a survey question or an inclusion criterion, work in the intensive care unit (ICU) during the pandemic was considered a proxy for FL HCWs. Screening tools to evaluate for burnout or job turnover intention were counted as measures of distress because burnout is a type of distress and turnover intention is highly associated with burnout[8].

Statistical analysis

The quality of non-randomized studies was assessed by the modified version of the Newcastle-Ottowa Scale (NOS)[9]. Studies with 7-9 stars were deemed good quality, 4-6 stars fair quality, and 0-3 stars poor quality[9]. I² statistic was calculated to assess heterogeneity. Odds ratios (OR) and forest plots using random effects model were generated for comparisons using Graphpad Prism, Boston, MA, United States.

RESULTS

Searches of PubMed, PsycINFO, Scopus, EMBASE, Cochrane Database of Systematic Reviews, Google Scholar, MEDLINE, and Web of Science were completed on December 2, 2025, generating 1819 records. Of the 1819 articles, 1679 were excluded after review of the title and abstract, 140 were reviewed at length, and 19 were included (Figure 1). The 19 articles contained 317133 total participants (Table 2)[10-28]. The following geographic areas are represented: Asia (12 from China, 1 from Taiwan, 1 from Japan), Europe (1 from France and Belgium, 1 from Spain, 1 from Albania), the United States (1 article), and Iran (1 article). Two studies utilized different surveys of the same population and were both included, as their surveys did not overlap. All of the studies in these articles screened for psychiatric effects using a brief self-report symptom screening survey.

Figure 1 Findings from systematic review. FL HCW: Frontline healthcare workers; COVID-19: Coronavirus disease 2019.

Table 2 Study characteristics, n (%)/mean ± SD.

Ref.	Country	Survey year	Occupation	Comparison	Number of participants	Age (mean ± SD or most common age group)	Men	Depression screen	Anxiety screen	PTSD screen	Sleep screen	Distress screen
Azoulay et al[10]	Europe	2024	ICU HCWs	None	850	39 ± 7	276 (32)	Yes	Yes	No	No	Yes
Cheng et al[11]	Taiwan	2023	Nurse practitioners	None	98	35-44	3 (3.1)	Yes	Yes	No	No	Yes
Yang et al[12]	China	2023	Nurses	None	2,065	26-35	370 (17.9)	Yes	Yes	No	Yes	No
Qian et al[13]	China	2023	HCWs	FL vs NFL	1419	26-35	343 (24.2)	No	No	No	No	Yes
Mohr et al[14]	United States	2018-2023	HCWs	Year; FL vs NFL	293164	46.3 ± 12.1	83258 (28.4)	No	No	No	No	Yes
Liu et al[15]	China	2023	HCWs	None	2070	25-34	469 (22.7)	Yes	Yes	Yes	Yes	No
Wen et al[16]	China	2022, 2023	Nurses, physicians	Year; ICU	2292	31-40	389 (17.0)	No	No	Yes	No	No
Narita et al[17]	Japan	2023	HCWs	FL vs NFL	660	42.7 ± 10.9	214 (32.4)	No	No	Yes	No	No
Liu et al[18]	China	2023	Nurses	Year	784	NR	366 (49.6)	Yes	Yes	No	Yes	No
Giménez-Díez et al[19]	Spain	2023	Nurses	Year; FL vs NFL	1045	42.7 ± 10.9	128 (12.2)	Yes	Yes	Yes	Yes	No
Xu et al[20]	China	2023	Nurses, physicians	None	2094	< 35	517 (24.7)	Yes	Yes	No	No	Yes
Xie et al[21]	China	2023	Nurses	FL vs NFL	253	18-45	22 (8.7)	Yes	Yes	No	Yes	No
Wang et al, 2023[22]	China	2023	Nurses	None	2210	26-35	436 (19.7)	Yes	Yes	No	No	No
Zhang et al[23]	China	2023	Nurses	FL vs NFL	2513	25-44	98 (3.9)	No	No	Yes	No	Yes
Wang et al, 2024[24]	China	2023	Nurses	None	2210	26-35	436 (19.7)	No	No	No	No	Yes
Ding et al[25]	China	2023	HCWs	FL vs NFL	1412	18-34	609 (43.1)	Yes	Yes	No	Yes	No
Yang et al[26]	China	2022, 2023	Nurses	Year	816	NR	381 (49.9)	Yes	Yes	No	Yes	No
Çerçizaj et al[27]	Albania	2024	Nurses	FL vs NFL	288	38.6 ± 10.4	52 (17.9)	Yes	Yes	No	No	Yes
Hoseini-Marvast et al[28]	Iran	2023	Nurses	FL vs NFL	890	34.89 ± 7.84	107 (12.0)	Yes	Yes	No	No	No

FL HCW: Frontline healthcare workers; ICU: Intensive care unit; NFL: Non-frontline; NR: Not report; PTSD: Posttraumatic stress disorder.

Fifteen articles were cross-sectional, and 4 were longitudinal. Fourteen studies were rated fair or good quality by the NOS and 1 study was poor quality (Supplementary Table 2). Various populations of HCWs were examined in these studies, with 10 evaluating nurses, 4 all HCWs, 3 physicians and nurses, 1 HCWs in the ICU, and 1 nurse practitioners. Twelve articles had a comparison group (7 comparing FL HCWs vs non-FL (NFL) HCWs only, 3 comparing both FL HCWs and NFL HCWs and 2023 vs pre-2023, and 2 comparing 2023 vs pre-2023 only)[10-28].

Thirteen studies screened for depression

Total 5 Self-Rating Depression Scale (SDS), 3 Patient Health Questionnaire-9 (PHQ-9), 2 Depression Anxiety Stress Scale-21 (DASS-21), 1 Hospital Anxiety and Depression Scale (HADS), 1 Beck Depression Inventory, 1 Center for Epidemiologic Studies Depression Scale (Table 3)[10-12,15,18-22,25-28]. Cutoff values were inconsistent across studies for the SDS and PHQ-9. Two studies reported only a combined variable for multiple psychological screening tools used in the study and did not report the results of the depression screening tool independently. The prevalence of positive depression screens ranged from 12%-65% in the included studies. Studies reporting an OR for a positive depression screen were pooled in a random effects model. Despite the differences in surveys and cutoff values used between studies, studies were pooled for analysis because heterogeneity may exist between the optimal cutoff value for a screening survey in different populations. Pooled risk ratio indicated a greater risk of a positive depression screen in FL HCWs compared to NFL HCWs (OR = 1.64, 95%CI: 1.12-2.16) (Figure 2). I² statistic is 0.0 (Q = 0.2, DF = 1, P = 0.65) for studies including a depression screen that provided positive screen prevalence data for FL HCWs and NFL HCWs.

Figure 2 Forest plot. A: For depression screen; B: For anxiety screen.

Table 3 Depression data, n (%)/mean ± SD.

Ref.	Depression screening tool	Positive depression screen definition	Depression prevalence	Depression survey score	Results of comparison
Azoulay et al[10]	HADS	7 of 21 on the depression subscale	219 (26)	Mean NR; median 4 (IQR: 2-7)	NCG
Cheng et al[11]	DASS-21	> 9	NR	8.2 ± 6.9	NCG
Yang et al[12]	SDS	≥ 40	NR	53.6 ± 13.1	NCG
Hoseini-Marvast et al[28]	Beck-2	Minimal 0-7, mild 8-15, moderate 16-25, severe 26-63	845 (50)	20.8 ± 13.7	Not significant for comparison between FL vs NFL (data NR)
Çerçizaj et al[27]	DASS-21	Mild 10-13, moderate 14-20, severe 21-27, extremely severe ≥ 28	Mild 16 (5.6); moderate 11 (3.8); severe/extremely severe 9 (3.1)	NR	OR 2.67 (95%CI: 1.2-6.1) for FL vs NFL
Liu et al[15]	PHQ-9	≥ 5	865 (41.8) (95%CI: 39.7-43.9)	NR	NCG
Yang et al[26]	SDS	≥ 40	NR	55.4 ± 12.8	Mean 62.46 ± 4.69 in 2022 vs 55.4 ± 12.8 in 2023, P < 0.001
Ding et al[25]	PHQ-9	≥ 10	231 (16.4)	NR	OR 1.66 (95%CI: 1.47-1.9) P < 0.001 for FL vs NFL
Liu et al[18]	SDS	≥ 73 for severe depression	36 (4.9)	55.6 ± 12.6	Prevalence 36 (4.9) in 2023 vs 13 (1.7) in 2022, P < 0.001
Giménez-Díez et al[19]	PHQ-9	Pooled symptom profile	191 (73.9)	8.9 ± 5.9	OR = 1.38 (95%CI: 1.00-1.90), P = 0.048 FL vs NFL
Xu et al[20]	CES-D	Pooled symptom profile	1300 (62.1) low, 605 (28.9) moderate and 189 (9) high	NR	NCG
Xie et al[21]	SDS	≥ 50	137 (64.9)	NR	OR = 1.44 (95%CI: 0.39-1.25), P = 0.017 FL vs NFL
Wang et al[22]	SDS	≥ 50	NR	54.3 ± 9.4	NCG

NCG: Bo comparison group; NR: Not reported; NS: Not significant; SDS: Self-Rating Depression Scale; PHQ-9: Patient Health Questionnaire-9; DASS-21: Depression Anxiety Stress Scale-21; HADS: Hospital Anxiety and Depression Scale; Beck-2: Beck depression inventory 2; CES-D: Center for Epidemiologic Studies Depression Scale; FL: Frontline; NFL: Non frontline; OR: Odds ratio; NR: Not report; IQR: Interquartile range.

Thirteen studies screened for Generalized Anxiety Disorder

Total 5 Generalized Anxiety Disorder-7 (GAD-7), 4 Zung Self-Rating Anxiety Scale (SAS), 2 DASS-21, 1 HADS, and 1 Beck Anxiety Inventory (Table 4)[10-12,15,18-22,25-28]. Cutoff values were inconsistent across studies for the SAS and GAD-7. Two studies reported only a combined variable for multiple psychological screening tools used in the study and did not report the results of the anxiety screening tool independently. The prevalence of positive anxiety screens ranged from 8%-80%. Pooled risk ratio revealed an increased risk of a positive anxiety screen among FL HCWs compared to NFL HCWs (OR = 1.90, 95%CI: 1.15-2.66). I² statistic is 91.4 (Q = 23.3, DF = 2, P < 0.001) was calculated for studies including an anxiety screen that provided positive screen prevalence data for FL HCWs and NFL HCWs.

Table 4 Anxiety data, n (%)/mean ± SD.

Ref.	Anxiety screening tool	Positive anxiety screen cutoff	Estimated anxiety prevalence	Anxiety survey score	Results of comparison
Azoulay et al[10]	HADS	≥ 7	492 (58)	median score 7 (IQR 5-10)	NCG
Cheng et al[11]	DASS-21	> 7	NR	6.92 ± 6.32	NCG
Yang et al[12]	SAS	≥ 50	NR	49.62 ± 11.9	NCG
Liu et al[12]	GAD-7	≥ 5	638 (30.8) (95%CI: 28.8-32.8)	NR	NCG
Liu et al[18]	SAS	NR	NR	51.5 ± 12.2	Mean SAS 69.6 ± 5.04 in 2022 vs in 51.5 ± 12.2 in 2023, P < 0.001
Giménez-Díez et al[19]	GAD-7	Dichotomous MH variable (no/mild vs moderate/severe); GAD-7 cutoff not provided	839 (80.3)	9.4 ± 5.7	OR for FL vs NFL 1.38 (95%CI: 1.00-1.90)
Xu et al[20]	GAD-7	Psychological symptom profile created: Low, moderate, and high level	1300 (62.1) low, 605 (28.9) moderate, and 189 (9) high	NR	NCG
Xie et al[21]	GAD-7	≥ 5	120 (56.9)	NR	OR = 0.67 (95%CI: 0.36-1.24), P = 0.20
Wang et al[22], 2023	SAS	≥ 36	1438 (65.1)	NR	NCG
Ding et al[25]	GAD-7	≥ 10	254 (18)	NR	OR for FL vs NFL 2.047 (95%CI: 1.68-2.24), P < 0.001
Yang et al[26]	SAS	≥ 50	61 (8.0)	51.4 ± 12.2	426 (52.3) in 2022 vs 61 (8.0) in 2023, P < 0.001
Çerçizaj et al[27]	DASS-21	Normal 0-7, mild 8-9, moderate 10-14, severe 15-19, extremely severe 20+	Mild 9 (3.1), moderate 25 (8.7), severe 3 (1), extremely severe 9 (3.1)	NR	OR for FL vs NFL 4.17 (95%CI: 1.6-10.6), P = 0.003
Hoseini-Marvast et al[28]	Beck anxiety inventory	Low 0-13, mild 14-19, moderate 20-28, severe 29-63	579 (65)	22 ± 13.4	NS FL vs NFL (data NR)

NCG: No comparison group; NR: Not reported; NS: Not significant; FL: Frontline; NFL: Non frontline; GAD-7: Generalized Anxiety Disorder-7; SAS: Zung Self-rating Anxiety Scale; OR: Odds ratio.

Five studies screened for posttraumatic stress disorder

Total 1 Impact of Events Scale (IES), 1 IES-Revised (IES-R), 1 COVID-19-specific IES-R-Revised, 1 posttraumatic stress disorder (PTSD) checklist for Diagnostic and Statistical Manual of Mental Disorders, 5^th Edition, and 1 Posttraumatic Diagnostic Scale. Of importance, none of the available studies assessed trauma exposure history, a necessary requirement for assessment of posttraumatic symptoms and disorders. Therefore the results of PTSD screens in the included studies were determined to be methodologically invalid and were not pooled for analysis.

Seven studies screened for sleep problems

Total 4 Insomnia Severity Index (ISI) and 3 Pittsburgh Sleep Quality Index (PSQI). Cutoff values were inconsistent across studies using the ISI and PSQI. Prevalence of positive sleep-related disorder screens ranged from 20.5%-79.4% (Table 5)[12,15,18,19,21,25,26]. Only one study compared FL HCWs vs NFL HCWs for sleep disorders and found an increased risk of a positive sleep disorder screen in FL HCWs compared to NFL HCWs (OR = 1.09, 95%CI: 1.02-1.16, P = 0.01). I² statistic was not calculated as only 1 study compared FL HCWs to NFL HCWs.

Table 5 Sleep disorder data, n (%).

Ref.	Sleep screening tool	Positive sleep screen definition	Sleep prevalence in FL 2023	mean ± SD	Results of comparison
Yang et al[12]	PSQI	≥ 8	1088 (52.70)	10.11 ± 4.2	NCG
Liu et al[15]	ISI	≥ 8	782 (34.3) (95%CI: 32.2-36.3)	NR	NCG
Liu et al[18]	PSQI	≥ 8	NR	9.4 ± 4.7	10.7 ± 2.5 in 2022 and 9.4 ± 4.7 in 2023, P < 0.001
Giménez-Díez et al[19]	ISI	NR	698 (66.80)	10.8 ± 6.4	NR
Xie et al[21]	PSQI	≥ 7	201 (79.4)	NR	NR
Ding et al[25]	ISI	≥ 15	289 (20.5)	NR	OR for FL vs NFL 1.09 (95%CI: 1.02-1.16), P = 0.01
Yang et al[26]	PSQI	≥ 8	NR	9.3 ± 4.7	10.7 ± 2.5 in 2022 vs 9.3 ± 4.7 in 2023, P < 0.001

PSQI: Pittsburgh Sleep Quality Index; ISI: Insomnia severity index; NCG: No comparison group; NR: Not reported; NS: Not significant; FL: Frontline; NFL: Non frontline; OR: Odds ratio.

Eight studies screened for distress or burnout

Total 3 Maslach Burnout Inventory (MBI), 2 DASS-21, 1 visual analog distress scale, 1 IES-R, 1 Turnover Intention Scale. Various cutoff values were used for the MBI depending on the subset of the MBI administered. Distress was as low as 1% in 1 study to as high as 75% (Table 6)[10,11,13,14,20,23,24,27]. A forest plot and I² statistic could not be generated because only one study compared prevalence of burnout between FL HCWs and NFL HCWs. However, studies comparing FL HCWs vs NFL HCWs revealed higher turnover intention scores and higher work stress in FL HCWs.

Table 6 Distress and turnover intention table, n (%)/mean ± SD.

Ref.	Distress screening tool	Distress screen definition	Distress prevalence in FL 2023	mean ± SD	Results of comparison
Azoulay et al[10]	VAS and intention to leave yes/no	NR	Intention to leave 484 (57); VAS NR	NR	NCG
Cheng et al[11]	DASS-21 and Stress Scale of Caring for Patients with Highly-Infectious Disease	> 14	NR	DASS-21 stress score 1324 ± 8.14; HSS 48.89 ± 18.5	NCG
Qian et al[13]	IES-R; change in work stress and rest time yes/no	NR	NR	IES-R score 173 ± 12.7	NS increase in work stress and decrease in rest time reported in FL vs NFL
Mohr et al[14]	2 items from MBI and coronavirus disease 2019 professional stress	Burnout ≥ once per week; coronavirus disease 2019 professional stress positive if answered high or extreme stress	95278 (32.50)		ICU had highest work stress at the start of the pandemic (48.6%) with the greater decline (12.9%) from 2020-2023 than other work areas
Xu et al[20]	MBI	Psychological symptom profile created: Low, moderate, and high level	1300 (62.1) low level, 605 (28.9) moderate level, and 189 (9) high level	NR	NCG
Zhang et al[23]	Turnover Intention Scale	< 6 very low; 6-12 low; 13-18 high; > 19 very high	High turnover intention 804 (32) and very high 302 (12)	NR	FL 2.09 ± 0.88 vs NFL 1.85 ± 1.85, P < 0.01
Wang et al[24]	MBI-HSS	> 26 in EE, > 9 in D, and < 33 in PA	1666 (75.38)	NR	NCG
Çerçizaj et al[27]	DASS-21	Normal 0-14, mild 15-18, moderate 19-25, severe 26-33, extremely severe ≥ 34	Mild 3 (1.0), Moderate 0 (0), severe or extremely severe 0 (0)	NR	NR

VAS: Visual analog scale; IES-R: Impact of Event Scale-Revised; MBI: Maslach Burnout Inventory; HSS: Health Services Scale; DASS-21: Depression Anxiety Stress Scale-21; NCG: No comparison group; NR: Not reported; NS: Not significant; FL: Frontline; NFL: Non frontline; OR: Odds ratio; ICU: Intensive care unit; EE: Emotional Exhaustion; D: Depersonalization; PA: Personal accomplishment.

DISCUSSION

This systematic review found that FL HCWs during the COVID-19 pandemic have a higher prevalence of positive screening surveys for depression, anxiety, and sleep disorder more than 1 year after the end of the pandemic than NFL HCWs. Positive screening surveys for distress remained prevalent in FL HCWs but available data were insufficient to compare distress between FL HCWs and NFL HCWs.

The screening survey results found in this systematic review of FL HCWs after COVID-19 support prior findings on long-term MH of survivors of other disasters, including the Oklahoma City bombing and the September 11 terrorist attacks, with considerable persistence of psychopathology up to 7 and even 25 years later[1,2]. Notably, many disaster MH studies have used diagnostic psychiatric interviews, but the available data on post-COVID MH outcomes among FL HCWs were all measured with brief self-report symptom screening surveys; none used diagnostic psychiatric interviews. While screening surveys can indicate risk for psychiatric illness, they do not assess full diagnostic criteria for psychiatric disorders. Conclusions from the study therefore suggest a higher prevalence of positive 1-year post-pandemic depression and anxiety screening surveys among FL HCWs compared to NFL HCWs, but the prevalence of actual psychiatric illness in this population remains unclear because of the absence of studies utilizing diagnostic interviews. Cross-sectional studies using diagnostic psychiatric interviews at relevant points in time are needed to define long-term prevalence of psychiatric disorders among FL HCWs, and the findings can be used for planning acute and long-term psychological responses to HCWs in future pandemics and other disasters. Because individuals with positive psychological screening surveys will require further evaluation with diagnostic psychiatric interviews, the findings of this study suggest that increased access to psychiatry services for purposes of full clinical diagnostic evaluation and treatment selected based on the results will be warranted for FL HCWs for at least 1 year after the end of a pandemic.

This review also noted substantial limitations of the evaluation of PTSD symptoms in published studies of this population. According to the Diagnostic and Statistical Manual of Mental Disorders, 5^th Edition, Text Revision[29], working with COVID-19 patients as a FL HCW is not a qualifying trauma exposure for the diagnosis of PTSD on 2 accounts: (1) Naturally-occurring medical illness is classified a stressor not meeting the definition of trauma; and (2) Working with patients does not constitute indirect exposure to a loved one’s trauma. Thus, FL HCWs cannot qualify for a diagnosis of PTSD on the basis of their workplace exposure to COVID-19 patients alone. Because none of the included studies assessed trauma exposure history, conclusions about PTSD following the pandemic cannot be drawn, despite a reported prevalence of a positive PTSD screen ranging from 21.3%-58.4% in the included studies (data not shown). Future research is needed to evaluate whether COVID-19 FL HCWs should be reconsidered for a qualifying trauma in future revisions of the Diagnostic and Statistical Manual of Mental Disorders, which would require comparison of symptoms among FL HCWs to individuals with traditional trauma exposures. This systematic review did not identify any studies that addressed such comparisons.

Limitations of this systematic review and meta-analysis include the use of different screening surveys and variable cutoff values for the surveys, which may lead to bias in pooled results. Substantial heterogeneity existed in anxiety screening survey results. Further, the majority of the studies were conducted in China, which may decrease generalizability to a global population, as cultural, healthcare-system, and occupational differences may exist between nations, Longitudinal studies of MH of FL HCWs outside of China are needed.

CONCLUSION

In conclusion, diagnostic psychiatric interviews are needed to define the long-term prevalence of psychiatric illness in FL HCWs, but the available data (entirely based on symptom screening measures) suggest that FL HCWs have an increased prevalence of positive screening surveys for anxiety, depressive, and sleep disorders at least 1 year after the end of the pandemic compared to NFL HCWs. These results suggest that enhanced access to psychiatry services is warranted in FL HCWs exposed to pandemics.