Effects of Region, Epidemic Stage, and Demographic Characteristics on Sleep Quality and Mental Disturbances among Health Care Workers during COVID-19 Outbreak

Objective To investigate the prevalence of sleep quality and mental disturbances of medical staff and identify the effects of region, epidemic stage, and demographic characteristics during the COVID-19 outbreak in China. Method Self-administered questionnaire were sent to health care workers (HCWs) in China from 30 Jan to 2 March, 2020. The Pittsburgh Sleep Quality Index, the Patient Health Questionare-9, the Generalized Anxiety Disorder-7 and the Impact of Event Scale were used to assess sleep quality, depression symptoms, anxiety symptoms and Post-traumatic stress disorder (PTSD) of HCWs, respectively. The inuencing factors of psychological and sleep disturbances were identied by univariate analysis and multiple regression. Result The incidence of people getting depression, anxiety or PTSD symptoms were 11.6%,13.3%, 14.3%, respectively. HCWs in Hubei province experienced signicantly poorer sleep quality (t=5.034, P<0.001). The predictors of sleep quality among HCWs were COVID-19 stage 2 (β=.135, p=.014) and 3 (β=.184, p=.001), female (β=.141, p=.003) and older age (β=.160, p=.001).Not working in Hubei province (β=-.264, P<0.001) showed to be a protective factor of sleep quality. The risk factors of depression symptoms were not working in Hubei province (OR=4.318, P<0.001), administrative and logistic staff and others (OR=3.538, p=.011), and higher PSQI score (OR=1.282, P<0.001). Having children (under-age: OR=.292, p=.001, grown-up: OR=.293, p=.042) was identied as a protective factor of having depression symptoms. Poor sleep quality showed to be the risk factor of anxiety and PTSD symptoms as well (both p<.001). Furthermore, administrative and logistic staff and others (OR=3.399, p=.006) were found to be the risk factor of PTSD symptoms among HCWs. 2and 3 of the outbreak. HCWs in Hubei had poorer sleep quality but lighter depression condition. gender, age, occupation and status of having children were associated with sleep and mental health. Mental health programs should be considered for HCWs especially those with specic characteristics. Key words COVID-19,


Introduction
Several pneumonia cases caused by unknown etiology were rstly detected in Wuhan, Hubei Province of China at the end of 2019 and the World Health Organization (WHO) China Country O ce was timely informed and responded to the outbreak of Novel Coronavirus (COVID-19) together with Regional O ce for the Western Paci c and headquarters on January 5, 2020 (1). The outbreak of COVID-19 was declared from being a Public Health Emergency of International Concern (PHEIC) to a global pandemic on 11 March by WHO Director General after the epidemic has widely spread to the rest of the world (2).Fever, cough, fatigue, diarrhea and vomiting are the major symptoms of COVID-19. At the same time, the sign of having di culty in breathing has been found in a few cases. The in ltration in both lungs was shown on chest radiographs of the infected (3)(4)(5).
Although COVID-19 is similar to SARS-CoV and MERS-CoV as an acute respiratory disease, it is different (5). Early cases suggest that the COVID-19 may not be as severe as SARS-CoV and MERS-CoV, but it is more infectious according to rapidly increasing incidence and evidence of human-to-human transmission (6).
Health care workers (HCWs) are those at risks confronting outbreaks and pathogens unknown to date (7).There are multiple ways causing health care workers increasing risks of attracting infectious diseases (8). Previous Research found that workers in the medical services had a higher rate of being infected by patient contacts and had a higher prevalence of work-related morbidity (9). Working in health care industry can endanger health, using SARS as a serious reminder (10). Among all SARS patients, about one fth of them are HCWs, and HCWs comprised more than 40% of patients in Canada and Singapore (11).
Early research concerning the outbreaks such as SARS and Ebola virus disease found that workers in healthcare system have faced with more severe emotional stress during the epidemic (12)(13)(14).
Data from several studies have shown that natural disasters such as earthquakes and pandemics caused severe mental disorders, among which attracted most attention are posttraumatic stress, major depression and sleep disturbances (18)(19)(20). During the epidemic, medical staff usually have a high level of anxiety because of the fear of transmitting infectious diseases to family, friends and colleagues [21].
PTSD has enormous effects on nursing care providers by changing their attitudes toward professionalism which may further result in increased physical and mental distress, decreased job satisfaction, increased turnover rate and reduced e ciency. Furthermore, it would also negatively affect the nursing care quality (22)(23)(24)(25). Depression anxiety and sleep problems could also cause adverse consequences to HCWs and the work settings. Therefore, it is necessary to investigate the risk factors of mental and sleep health among HCWs during COVID-19 and speci c solutions should be made targeting those who are vulnerable to the outbreak.
As the epidemic progresses to different stages, the sleep and psychological conditions of HCWs may change. No previous study has investigated the impact of the stage of epidemic on sleep quality and mental health of HCWs during the COVID-19 outbreak.
In this study, we assessed the mental health and sleep quality of HCWs. We sought to analyse the predictive effect of region, epidemic stage and sociodemographic variables on sleep quality, and together with sleep quality on anxiety, depression and PTSD symptoms during the epidemic.
As the COVID-19 continues and will continue to spread over the next few months, this research may help nd those working in health care industry who are more prone to suffer from emotional and sleep problems, Which may provide a basis for further intervention.

Study design and Participants
The study was a quantitative survey for HCWs in China including doctors, nurses, medical technicians, administrative, logistic staff and others working in hospital. The time span of the study was between 30 January and 2 March, 2020. All health care workers voluntarily participated, completed a questionnaire anonymously online. The study was approved by the Research Ethics Commission of Naval Medical University,.

Measures
Sociodemographic variables were collected, such as gender, age, years of working, education level, marital status, occupation, if one is the only child in his family and child status. COVID-19 time were normalized into three stages: stage1 (30 Jan to 17 Feb), stage2 (18 Feb to 24 Feb) and stage3 (25 Feb to 2 March). The reason why the stages have been de ned is that on 18 Feb, for the rst time, the number of new cases was less than the number of new cured cases. The number of con rmed cases rose to the rst peak in China between 24 Feb to 28 Feb (26). So we chose 25 Feb as the second cut-off point.
The Pittsburgh Sleep Quality Index (PSQI) was used in this study to test HCWs' sleep quality. It is a selfadministered scale including 19-items which consists of 7 dimensions including subjective sleep quality, sleep latency, sleep duration, sleep e ciency, sleep disturbances, use of sleeping medications and daytime dysfunction. The PSQI global score ranges from 0 to 21 with higher score indicating worse sleep quality (27).
The Patient Health Questionare-9 (PHQ-9) was used to detect depression among HCWs in this study. Each item of the scale were scored from 0 to 3, with total PHQ-9 score ranging from 0 to 27. A cut-off point of 10 was used in this study with high speci city (88%) and sensitivity(88%). Scores ≥ 10 are recognized as the presence of major depression symptoms (28).
The Generalized Anxiety Disorder-7 (GAD-7) was administered to participants to test anxiety scores. The total GAD-7 score ranges from 0 to 21. Those whose score ≥ 8 were diagnosed to have anxiety symptoms. The sensitivity (77%) and speci city (82%) are appropriate (29).
The Impact of Event Scale (IES-R) scale was used to assess subjective stress caused by traumatic event.
It includes 22 items and consists of 3 subscales which are intrusive, avoidance and hyperarousal. The scale ranges from 0 to 88. IES-R total score 33 are identi ed to have psychological stress (30).

Statistical analysis
Depression, anxiety, and PTSD scales were converted to dichotomous variables because the scale values were not normally distributed.
The group comparisons of categorical variables were conducted by Chi-square test and continuous variables by T-test.
Simple linear regression was used for univariate analysis of PSQI total score. For univariate analysis of depression syndrome, anxiety syndrome and PTSD syndrome, Chi-square test and Fisher's Exact test were used for categorical variables and simple logistic regression were used for interval variables.
For categorical variables, the count and frequency were presented and the internal variables were summarized as mean ± standard deviation.
Thirdly, variables with p 0.2 in univariate analysis were subjected to multiple regression analysis. multiple linear regression with stepwise method was performed for predicting risk factors of PSQI score. Multiple logistic regression analysis with forward: conditional procedure was conducted for detecting risk factors of Depression syndrome, Anxiety syndrome and PTSD syndrome.
Two-sided P value < 0.05 was identi ed as statistically signi cant. All statistical analyses were performed using SPSS 26.0 (Statistical Package for the Social Sciences) for Windows (SPSS, Chicago, IL).

Result
Baseline information 421 respondents completed the questionnaire. There are 59 (14%) participants existing missing values over sleep e ciency. Expectation Maximization (EM) interpolation method was used to ll in the missing values. Among all the HCW's (421) participating in the study, 177 (42%) of them were working in Hubei province, which is the epicenter of the ongoing outbreak in China. Most of the respondents in our study were female (291, 69.1%), and nurses (203, 48.2%). We divided the time span of the study into 3 stages, and we have 163 (38.7%) respondents in stage 1, 133 (31.6%) respondents in stage 2 and 125 (29.7%) in stage 3. The mean age of the participants were 35.48 (SD = 8.543) years old. The average years of working of participants were 14.05 (SD = 9.687) years. The mean PSQI score of the subjects were 5.82 (SD = 3.315). The data showed that the number of people with depression, anxiety or PTSD symptoms was 49 (11.6%), 56 (13.3%), 60 (14.3%), respectively (Table 1).

Sleep Quality And Mental Health Across Different Covid-19epidemic Stages
The sleep quality of HCWs was signi cantly worse in COVID-19 epidemic stage 2 (β = .135, p = .014) and stage 3 (β = .184, p = .001). While depression, anxiety and PTSD symptoms had no signi cant differences  The PSQI score was signi cantly associated with Hubei province(p < .001), gender (p = .005), occupation (p = .014), COVID-19 time (p = .046), age (p = .048) and years of working (p = .039). Nurses had signi cantly higher PSQI scores than that of doctors (p = .003, Table 2). Table 3, the results showed that variables of being working in Hubei province, occupation, education level, marital status and child status were signi cantly associated with depression symptoms (all p < 0.05). Variables of occupation (p = .008), years of working and child status were signi cantly associated with PTSD symptom(all p < 0.05). The PSQI score were highly signi cantly associated with depression, anxiety and PTSD symptoms (all p .001).

Risk Factors Associated With Psqi, Depression, Anxiety And Ptsd
We put all variables identi ed in univariate analysis with p value smaller than 0.2 into multiple regression analysis.
For PSQI, the risk factors were listed as follows: COVID-19 stage 2 (p = 0.014) and stage3 (p = 0.001), older age (p = 0.001) and female (p = 0.003). The protective factor was revealed to be not working in Hubei province (beta=-0.264, p < .001). The adjusted R 2 of the model was 0.109( Table 2).   49 (11.6%), 56 (13.3%) and 60 (14.3%) of HCWs were shown to have depression, anxiety and PTSD symptoms in this study. The rates are lower comparing with the results of previous studies on SARS which presented that 18 to 57% of the medical staff endured serious emotional distress during and shortly after the epidemic (12,13,(15)(16)(17). This may be due to the experience of ghting SARS in 2003 and timely and effective psychological assistance to medical staff during this pandemic.

As shown in
Few studies have explored the impact of the epidemic stage on sleep quality and mental health of HCWs.
This study found that comparing withCOVID-19epidemic stage 1, the sleep quality at stage 2 and 3 were signi cantly poorer, which might probably be owing to the rapidly rising number of con rmed cases and severe cases over time. This result was contrary to that of Chen (2006) who found that compared with the mean score of PSQI of medical staff before caring for patients, the mean score of PSQI was signi cantly decreased 2 weeks and 1 month after the initiation of SARS prevention program (31). And the study of Chen also showed that the anxiety and depression level of nursing staff had also been alleviated after 2 weeks of the SARS caring program (31). However, the ndings of the current study indicating that the depression and anxiety level of HCWs had no signi cant difference across 3 stages.
The difference might be because that the epidemic stage 1 of our study was the early stage of the prevention program instead of the period before a prevention program began. High-risk HCWs were more risky of being fatigue and having poor sleep quality during the outbreak of SARS (32). Hubei Province is the center of the epidemic where con rmed cases and severe cases are signi cantly beyond those in other provinces. Therefore, the workload and work intensity of medical personnel in Hubei Province will be much higher than those in other provinces. Research found that higher workload of nurses were related to poorer sleep quality which could explain the result of the study (33). In addition, female and older age were two well established risk factors for sleep disturbances in our study and previous studies (34).
This study found that HCWs working in or outside Hubei province had equal level of prevalence to have psychological stress and anxiety symptoms. Previous studies also indicated that similar psychological morbidity and perceived stress were found between high-risk and low-risk HCWs (12,32).COVID-19 has a long incubation period and the virus carriers are undetectable and could transmit virus during the latent period. Although non-Hubei provinces are not as severe as Hubei province, the number of con rmed cases was rising rapidly as well. The HCWs outside Hubei provinces did not recognize themselves as exempt from the danger. Moreover, medical staff in Hubei has higher vigilance and con dence as the attention of the whole country was focused on the epidemic in Hubei. Thus medical staff in Hubei had a morale and sense of responsibility to conquer this challenge, which is bene cial for their mental health maintenance. These reasons may help explain the unusual result that HCWs not working in Hubei province had higher risk of depression.
It has not been described before that the status of having children is associated with depression of HCWs. This study showed that having under-age or grown-up children was a protective factor of depression symptom among HCWs. Previous research found that people with older age (46.65 ± 13.82) and coexisting conditions (28.8%) were more likely to be affected by COVID-19 (35). Children were not the main affected group of this outbreak. And the postpone of the reopening of schools and lockdown of the cities made children stayed in home under supervision. These might explain part of the results but still acquire further studies.
Surprisingly, our study suggested that administrative and logistic staff and others working in health care industry had higher prevalence of being depression or PTSD compared with that of doctors. This nding is contrary to previous studies which have suggested that being a nursing professional and providing direct care with SARS patients were the risk factors of high level of job-related stress (12,36). But the results of our study accords with earlier observations. Attack rate among HCWs during SARS varied by occupations. The attack rate of HCWs in Vietnam, 2003, were 16%, 35%, 2% for doctors, nurses and administrative staff, respectively. Those with highest attack rate was named as "other staff with patient contact" which accounted for 53% (37). A study in Hongkong during SARS outbreak reported the attack rate as follows: nurses (1.21%), medical staff and technicians (0.29%), and others including assistants and cleaners (2.72%) (38).The studies above have indicated that the logistic and administrative staff and others was the most vulnerable groups that withstand the highest attack of SARS virus. This might be one of the reasons why administrative and logistic staff in our study endured with higher prevalence of depression and PTSD. Another reason could be that the logistic staff and others working in hospital might not be as psychologically prepared as doctors and nurses. Furthermore, situations of low exposure can also be risky of getting affected (37) as there were asymptomatic carriers. Administrative and logistic staff and others like cleaners are staff that are easily neglected in a outbreak. We should be more concerned about their mental health.
Our data showed that the risk of having depression, anxiety or PTSD symptoms tended to increase with enhancing PSQI score among HCWs. Previous study found that Service members and veterans who had higher PTSD, depression and anxiety scores were those who had relatively bad or very bad sleep quality (39). A study among South Korean workers found that the variance in weekday and weekend sleep duration are highly associated with PHQ-9 score which indicated depression (40). A recent research suggested that poor sleep quality and social support lead to anxiety, stress and self-e cacy problems among HCWs in China dealing with COVID-19 during early stages of the outbreak (41), which is consistent with our study. Several studies found that insomnia was often accompanied by mental disorders, especially anxiety and depression (42,43). In fact, longitudinal studies have found that risk of depression and anxiety increased when insomnia existed (44)(45)(46)(47) which would even cause the relapses (48). A considerable number of longitudinal studies (49-52) and a meta-analysis (53) suggested that people with sleep disturbances, compared with those without sleep di culties, have over twice times risk of getting depression symptoms over the next one to three years of follow-up. Insomniac patients with depression have higher prevalence to remain depressed despite standard therapy (54). In this case, sleep disturbances are likely to be the risk factor and consequence of anxiety and depression. In this case, we could deal with mental health problems by coping with sleep disturbances which is less-stigmatizing.
The study has several limitations. Firstly, The study is a cross-sectional study, which can not investigate the causal relationship. Secondly, there is self-report bias because all results were from self-reported