Psychological burden of the COVID-19 pandemic and its associated factors among frontline doctors of Bangladesh: a cross-sectional study [version 3; peer review: 2 approved]

Background: Frontline doctors are the most vulnerable and high-risk population to get the novel coronavirus disease 2019 (COVID-19) infection. Hence, we aimed to evaluate the anxiety, depression, sleep disturbance and fear of COVID-19 among frontline doctors of Bangladesh during the pandemic, and the associated factors for these psychological symptoms. Methods: In total, 370 frontline doctors who were involved in the treatment of suspected or confirmed COVID-19 patients during the pandemic took part in an online cross-sectional study. Recruitment was completed using convenience sampling and the data were collected after the start of community transmission of COVID-19 in the country. Anxiety and depression, sleep disturbance, and fear of COVID-19 were assessed by the Patient Health Questionnaire-4, twoitem version of the Sleep Condition Indicator, and the Fear of Coronavirus-19 scale, respectively. Socio-demographic information, health service-related information, co-morbidity, and smoking history were collected for evaluating risk factors. The proportion of psychological symptoms were presented using descriptive statistics and the associated factors were identified using multinomial logistic regression analysis. Results: Of the doctors, 36.5% had anxiety, 38.4% had depression, 18.6% had insomnia, and 31.9% had fear of COVID-19. In multinomial logistic regression, inadequate resources in the workplace were found as the single most significant predictor for all psychological outcomes: anxiety and/or depression (severe, OR 3.0, p=0.01; moderate, OR 5.3, p=0.000; mild, OR 2.3, p=0.003), sleep disturbance (moderate, OR 1.9, p=0.02), and fear of COVID-19 (severe, OR 1.9, p=0.03; moderate, OR 1.8, p=0.03). Open Peer Review


Introduction
Novel coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), was first recognized in December 2019 in Wuhan City in central China 1,2 . The World Health Organization declared the COVID-19 outbreak as a global pandemic on March 11, 2020 3 . Bangladesh confirmed its first COVID-19 outbreak on March 08, 2020, when the Institute of Epidemiology, Disease Control and Research (IEDCR) reported the first three confirmed cases 4 . As of July 31, 2020, IEDCR confirmed 234,889 COVID-19 cases in Bangladesh, including 3083 related deaths with a Case Fatality Rate of 1.31% 5 .
The COVID-19 pandemic has caused various challenges in Bangladesh's healthcare system. One of the biggest challenges is the spread of COVID-19 infections among frontline doctors 6 . Up to July 29, 2020, about 2453 doctors have been infected 7 , and 69 doctors have died 8 because of COVID-19 infection in Bangladesh. The mortality rate due to COVID-19 among doctors in Bangladesh is about 4%, which is the highest in the world among doctors 9 , and this rate is also higher than that of Bangladesh's national mortality rate for COVID-19 5 .
In addition to the surge of COVID-19 infection, the pandemic has caused mental health problems to rise among doctors in Bangladesh. Mental health problems during pandemics are common, and evidence has shown that the severe acute respiratory syndrome (SARS), Middle East respiratory syndrome (MERS), and H1N1 pandemics also impacted the mental health condition of healthcare workers 10-12 . A study showed that frontline healthcare workers feel tremendous mental pressure during a pandemic because of the diminution of personal protection equipment, extensive media reportage, lack of treatment resources, increasing pattern of cases, death tolls, tremendous workload and social stigmatization 13 . Recently, studies from Singapore, India, Greece and China have reported mental health issues of healthcare workers during the current rapidly evolving situation [14][15][16][17] . Besides all of these country-wise evidence, a case study of Bangladesh also reported an incident of suicide at a hospital due to fear of COVID-19. It was alleged that the suicide was committed because the victim was not treated by the health care professionals as they suspected the person was infected with COVID-19 18 . Again, another cross-sectional study reported COVID-19 suicidal behavior among the health professionals, and no comparable difference elucidated when compared with the general population 19 . However, these two Bangladeshi studies reported suicidal behavior and/or fear of infection that prompted us to evaluate the other most commonly studied symptoms like anxiety, depression, and sleep disturbance, along with fear of COVID-19 in frontline doctors.
Bangladesh is a lower-middle-income country where doctors have to provide services in an overburdened, understaffed, and insufficiently equipped setting due to massive shortage and disproportionate distribution of skilled health workers, which causes unusual mental stress 20 . Despite the challenges in their workplaces, during the COVID-19 pandemic Bangladeshi doctors have shown their competency and professionalism in providing the best care to the country's people. As part of their responsibility, they have to expose themselves to the risk of COVID-19 infection for the benefit of the mass population 21 . It is speculated that the risk of infection and professional stress has gradually worsened the mental health condition of doctors in Bangladesh as they are facing stigmatization, fear of spreading the infection to family members and fear of being isolated. Currently, there is no evidence in support of this assumption. Therefore, we conducted this study to evaluate the psychological burden among Bangladeshi frontline doctors during the COVID-19 pandemic. To assess psychological symptoms, we quantified the magnitude of anxiety, depression, sleep disturbance and fear of COVID-19. Besides, we explored the associated factors influencing the psychological outcome. The findings of the study could be used to identify potential gaps in practice that would need interventions.

Study design and participants
We conducted an online cross-sectional study among doctors working at different clinical settings to treat patients, either suspected or confirmed COVID-19 cases, during the pandemic. Participants' recruitment was completed by convenience sampling. Doctors from the professional and personal networks of the researchers were initially contacted through Facebook messenger and email. Doctors who showed interest were invited to participate in the study through an online questionnaire. We excluded doctors who did not complete intern training after graduation or were not involved in direct patient care.
A total of 370 frontline doctors took part in the study over two months from 1 st April to 30 th May 2020. Sample size was determined using the prevalence of anxiety, depression, insomnia, and distress among China's healthcare workers during the COVID-19 pandemic 13 . The highest sample number was taken using the prevalence of depression (31.8%) in the aforementioned study, i.e. 370 respondents.

Data collection
We circulated the questionnaire through online among the interested participants after the Government of Bangladesh confirmed community transmission in Bangladesh on March 28, 2020 22 . Data were collected by an online self-administered semi-structured questionnaire using the Google survey platform 23 . The questionnaire link was sent to participants electronically through Facebook and email.
Questionnaire content. The online questionnaire collected data on sociodemographic factors (age, gender, marital status, education, occupation), health service-related factors (the type of service, working place, professional designation, service level of health system, number of days of service provided, shifting duty or not, resource of working place), psychological parameters (anxiety, depression, sleep disturbance, fear), co-morbid conditions (diabetes, hypertension, asthma, chronic obstructive pulmonary disease, heart disease, chronic kidney disease, thyroid disorder), high-risk behavior as defined by tobacco use, and the living area of the physician where at least one COVID-19 case had been confirmed by the local authority.
The questionnaire was pre-tested before the final administration to detect any inconsistency and biases. To pre-test, 10 men and 10 women frontline doctors were selected randomly using the inclusion criteria (MBBS degree with completed intern training) and the questionnaire was sent to them through an online platform (Facebook messenger and email). The objective and importance of pre-testing were added with the questionnaire as an explanatory note. The researchers also informed that participation of the respondents was voluntary and they have the right to withdraw themselves at any time or refuse to answer any question. The collected responses were analyzed and interpreted based on the following: trends in responses; fundamental flaws with the design or format; attractiveness; comprehension; acceptance; and relevance.
Sleep disturbance was assessed via a two-item version of the Sleep Condition Indicator (SCI-02), an ultra-short clinical rating scale, which can be used to rapidly screen for insomnia in routine clinical practice 25 . Each item was scored on a 5-point scale as 0, 1, 2, 3, 4. By adding the item scores, the SCI total score was obtained, ranging from 0 to 8. A higher score means better sleep. This tool showed an acceptable level of Cronbach's α and the Spearman-Brown correlation at the point of 0.74. Again, the test-retest reliability (r) and intraclass correlation coefficient (ICC) in a sample repeating the test from 12 hours up to 7 days were r = 0.68 and ICC = 0.68, respectively 25 . To quantify the magnitude of severity, we categorized the sleep disturbance using percentiles of the SCI-02 score as follows: good sleep condition (score ≥75th percentile, score ≥7), moderate sleep condition (score ≥25th percentile and <75th percentile, score 3-6) and insomnia (score <25th percentile, score 0-2). Here, the cut-off value of insomnia was kept the same as DSM-5 threshold criteria 26 .
The Fear of Coronavirus-19 Scale (FCV-19S) was used to measure one's fear of COVID-19 27 . The FCV-19S consists of 7 items. Participants were asked to rate their agreement with each statement on a 5-point scale from '1 -strongly disagree' to '5 -strongly agree'. A higher score indicated greater fear. Recently, this instrument was validated among the Bangladeshi population 28 . Currently, the FCV-19S has no classification of severity, and hence, we developed a severity scale using percentiles of FCV-19S score as follows: mild (score ≤25th percentile, score ≤17), moderate (score >25th percentile and <75th percentile, score 18 to 23) and severe (score ≥75th percentile, score ≥24).

Statistical analysis
The data were entered in a pre-designed Microsoft Office Excel format, which was imported later into the software Statistical Package for Social Science version 20.0 for Windows (SPSS, Inc. Chicago. IL.USA). All the estimates of precision were presented at a 95% confidence interval (CI). Descriptive analysis included mean, standard deviation (SD), frequencies, and percentages. Background information (sociodemographic and professional) and the magnitude of psychological outcomes were presented using frequencies and percentages. The score of the instruments was presented using the mean with SD.
The associated factors of psychological outcomes were determined using multinomial logistic regression analysis. To find the factors that influenced the psychological outcomes, first, we run univariate analysis. Variables that showed p ≤0.25 in the univariate analysis were examined as an independent variable in the logistic regression 29,30 . We calculated odds ratios (OR) and 95% CI for each independent variable for multiple logistic regression analysis. In the regression table, factors that had OR >1 were presented for each outcome variable. We ensured no multicollinearity presence using the variance inflation factor (VIF) to run the regression analysis. The statistical tests were considered significant (2-sided) at a level of p ≤0.05.

Ethical approval
The Ethical Review Committee of Bangladesh University of Health Sciences approved the study (identification number: BUHS/ERC/20/16).
An information and consent form (Extended data 23 ) to take part in the study and for the publication of the participant's anonymized information was provided prior to the questionnaire. Completion of the questionnaire implied consent.

Results
As shown in Figure 1, 1000 individuals were contacted initially and 370 were included in the study after exclusion.

Demographic characteristics and health status of the doctors
The mean (SD) age of the doctors was 30.5 (4.4) years. Most of them were men (60.3%) and married (66.8%). A total of 69.5% had been living in areas that were affected by the COVID-19 outbreak. About a quarter of participants (24.8%) had been suffering from at least one chronic disease; the proportion of more-than-one chronic diseases was 4.3%. The most commonly reported chronic disease was chronic bronchial asthma (15.9%). Table 1 presents the detailed demographic and health-related characteristics of the study participants.
Professional and work-related characteristics of the doctors More than half of the total doctors (56.5%) had a Bachelor's (MBBS) degree, which is the entry-level degree for medical doctors in Bangladesh, and 19.7% had post-graduation degrees. The rest were post-graduate students (23.8%). The majority was employed in the private sector (55.4%), followed by the government sector (30.3%). Most of the doctors' primary working settings were a hospital (54.3%), and most of them worked at tertiary level healthcare settings (32.2%). The majority of the doctors had shifting duties (69.5%) and worked in a low resource setting (70.5%). On average, they worked five days a week during the pandemic (Table 2).

Psychological burden of COVID-19 outbreak among the doctors
The detailed result of psychological status is presented in Table 3. The mean (SD) score of PHQ4, GAD-2 score and PHQ-2 score were 4.5 (2.9), 2.3 (1.8) and 2.2 (1.6), respectively. Considering the total score of PHQ4, about 73% of doctors had anxiety and/or depression, of which the majority were affected by mild anxiety and/or depression (39.2%). Separately, the first two (GAD-2) and successive two (PHQ-2) items of PHQ-4 identified that 36.5% of the doctors had anxiety, and 38.4% had depression. Here, the mean (SD) score of SCI-2 and FCV-19S were 5 (2.4) and 20.3 (6.1), respectively. Moreover, in the SCI-2 score, 18.6% of the doctors were found to be insomniac. Furthermore, the FCV-19S identified that 31.9% and 37.6% of the physicians had a severe and moderate level of fear regarding the COVID-19 pandemic, respectively.

Discussion
The study aimed to assess the psychological burden of frontline doctors in Bangladesh during the COVID-19 pandemic, and factors that predict their psychological status. The study identified that anxiety, depression, insomnia, and fear related to the COVID-19 outbreak are common among frontline doctors of Bangladesh during this unprecedented time. The paucity of resources for providing care to patients in workplaces was found as the single most common predictor for poor psychological status. In addition, having shifting duty, living in a COVID-19 affected area, and the presence of asthma predicted poor quality of sleep among the frontline doctors.
A considerable proportion of frontline doctors in Bangladesh has experienced psychological symptoms due to the COVID-19 pandemic. The burden of psychological symptoms is higher than the burden of symptoms among healthcare workers of China, Singapore and India during the COVID-19 pandemic 14,15,17 . A meta-analysis study from China has presented the pooled prevalence of depression (22.8%), anxiety (23.2%), and insomnia (38.9%) 17 . Compared to the pooled prevalence of symptoms in China, the current study has shown a higher proportion of depression and anxiety, but a lower proportion of insomnia among Bangladeshi doctors. Furthermore, the prevalence of anxiety and depression were reported as 14.4% and 9%, respectively, in Singapore 15 and 17.1% and 12.4%, respectively, in India 14 , which are also lower than the magnitude of anxiety and depression observed among Bangladeshi doctors in this study. The burden of psychological symptoms in the current study is also higher than the burden of psychological symptoms among China's general population during the pandemic 31 . Similarly, the burden of depression in our study is also higher than the depression reported by another study among the general population of Bangladesh 32 . Moreover, a comparison with mental health symptoms (anxiety 77.4%, depression 74.2%, and sleep problems 52.3%) among health workers during SARS pandemics in Taiwan shows a lower burden of psychological symptoms in Bangladesh during COVID-19 pandemics 10 . It is noteworthy that there are variations in the methods of measuring psychological symptoms across the studies.
Many underlying factors for mental health problems among frontline health workers during the pandemic situation have been reported in the literature, including gender, age, living in a rural area, poor social support, poor self-efficacy, profession, place of work, disruption of routine clinical practice, fear of potential destabilization of health services, the sense of loss of control, having organic disease, and being at risk of contact with a patient with COVID-19 13,16,31,33-35 . Among all the reported   causes, COVID-19 can be an independent risk factor for healthcare workers' poor mental health 33 . In Bangladesh, the burden of COVID-19 is among the top 20 countries in the world. Along with the general population, frontline health workers have also been overwhelmed by the surge of infection. It has been reported that doctors in Bangladesh have been experiencing the highest infection and mortality in the world due to the virus 9 . Experts have suggested that lack of infection control measures, monitoring, proper management at hospitals, inappropriate use and disposal of safety gear, and lack of training for dealing with patients with COVID-19 are contributing to the highest infection and mortality of the doctors 9 . It is also believed that COVID-19 infection and its underlying causes contribute to the doctors' poor mental health condition.
The current study identified several factors that contribute to the burden of psychological symptoms among Bangladeshi doctors -the paucity of resources in the workplace is the most significant. Limited resources in the workplace include materials, trained workforce or any other things that are required to provide services. The current study has confirmed the association of inadequate resources with the poor psychological status of doctors. Inadequate resources such as masks, sanitizer, and personal protective equipment (PPE) in workplaces increase the chance of getting COVID-19 infection and can cause profound psychological pressure on frontline doctors. The lack of resources in workplaces in Bangladesh has been widely reported in news media 36 . The news media has reported inadequate and inappropriate PPE as a cause of widespread COVID-19 infection among healthcare professionals in Bangladesh 36 . However, Bangladesh is not the only country that faced a shortage of resources during the pandemic. The shortage of such resources has also been reported in many other countries because of the distorted supply chain across the world 37 . Lack of resources is also considered as a cause of poor psychological status among healthcare workers in many countries during the pandemic 16,38 . Experts have recognized sufficient resources as an essential factor for healthcare professionals to be resilient during an unprecedented time 39 .
The lack of skilled and trained workforce in hospitals is another underlying cause of the high burden of psychological symptoms among frontline doctors in Bangladesh. Amid the workforce shortage, frontline doctors have to do long shifting duties for a certain period and then stay in quarantine for 14 days before they return to work. This atypical work schedule for doctors has been introduced to reduce the frequency of exposure to COVID-19 virus in workplaces. However, it is believed that the long shifting duties and being isolated during quarantine may have triggered mental health problems among doctors. The current study has found that those who did shifting duties were more likely to have sleep problems linked with poor mental health. Although identifying links between the quarantine period and poor mental health was not a scope of the current study, other studies have confirmed the link between quarantine period and mental health during this pandemic 40 .
The current study is the first study in Bangladesh that provides the burden and associated factors for doctors' poor mental health outcome during the COVID-19 pandemic. There are some limitations in the study. As it is a cross-sectional study, causal relation could not be established. Thus, the study presents the factors linked with the psychological outcomes as associated factors. Moreover, the study is an online-based questionnaire. Therefore, the possibility of selection bias cannot be ruled out. Again, a small sample size limited the generalization of the study findings. The participants of the study were mainly young doctors. This happened because the younger population is more exposed to online platforms than the elderly. However, a recent review has shown that younger doctors are more affected by psychological symptoms than elder doctors 34 . Thus, the study has reflected evidence of the high-risk group of doctors for a psychological problem.

Open Peer Review
Reviewer Report 24 November 2020 Is not it better to report psychological burdens as "probable depression" instead of "depression", because the scales are short version (although their cutoff scores had higher specificity and sensitivity than the original one; for example, PHQ-2 and PHQ-9)? Similar suggestion for fear of COVID. Besides, the authors may like to perform ANOVA and linear regression -not mandatory if they refer how the cutoff scores were categorized. ○ Discussion is focused on the results. Discussing prevalence rate across countries is perfect, even the authors compared with the review article prevalence rate, which is appreciated. By the way, may I suggest to compare the Bangladeshi general people mental health problems rates, this may help the reader to be informed about how this study findings differ from the general people. Please refer Bangladeshi studies assessing the pandemic related psychological burdens (ref: https://doi.org/10.31234/osf.io/q4k5b).

○
The current study is the first study in Bangladesh that provides the burden and associated factors for doctors' poor mental health outcome during the COVID-19 pandemic. -I request to avoid the first word, because there is a published paper on suicidal behavior of the cohort.

○
Overall observations, the authors did a great job addressing mental health problems of the vulnerable cohort to virus infection, which may have influence in policy level. And the paper was written in a good flow, and also provided some of recommendations. Best of luck.

Is the work clearly and accurately presented and does it cite the current literature? Yes
Is the study design appropriate and is the work technically sound? Yes

If applicable, is the statistical analysis and its interpretation appropriate? Partly
Are all the source data underlying the results available to ensure full reproducibility? Yes We mentioned the study you suggested to discuss.

Comment 6
More information regarding sampling is needed. Please add scale reliability (e.g., Cronbach's alpha) as the scales are not validated in Bangla.

Response 6
We applied a convenience sampling technique that was clearly mentioned. Moreover, a flow chart was used to show the pool of samples from where the subjects were recruited based on inclusion criteria. Among the tools applied, only FCV-19S validated among the Bangladeshi population and we mentioned it in the manuscript. For PHQ-4 and SCI-02, as per your suggestion, we added the Cronbach's alpha as appropriate.

Comment 7
The authors already mentioned, there is no cutoff score for Fear of COVID scale. However, they categorized it. It would be better if they provide how the category was consideredmedian?
To convince the readers, in the revised version, we have elaborated the method of categorization adding the percentiles with each cut-off value. This method of categorization also applied for another health-related tool also (https://doi.org/10.1111/jdi.13331).

Comment 8
It is requested to report the mean and SD of fear of COVID (along with percentage; as it was done for depression and anxiety) as the scale has no cutoff score yet.

Response 8
We have added.

Comment 9
Is not it better to report psychological burdens as "probable depression" instead of "depression", because the scales are short version (although their cutoff scores had higher specificity and sensitivity than the original one; for example, PHQ-2 and PHQ-9)? Similar suggestion for fear of COVID. Besides, the authors may like to perform ANOVA and linear regression -not mandatory if they refer how the cutoff scores were categorized.

Response 9
As the generic version of PHQ-2 avoided the term "probable" to represent depression and showed an acceptable level of reliability, we used 'depression' to align with the main paper (A 4-item measure of depression and anxiety: Validation and standardization of the Patient Health Questionnaire-4 (PHQ-4) in the general population, https://doi.org/10.1016/j.jad.2009.06.019).
Regarding fear of COVID-19, we presented the severity using the scientific method as described above. As this severity scale has not yet developed and validated, hence we pretested it (as part of the whole questionnaire) in a small sample of frontline doctors as mentioned in the section entitled 'Questionnaire content'. We used multinomial logistic regression based on the categorization of the tools as explained in the revised version and also stated above to respond to the reviewer inquiry.

Comment 10
Discussion is focused on the results. Discussing prevalence rate across countries is perfect, even the authors compared with the review article prevalence rate, which is appreciated. By the way, may I suggest to compare the Bangladeshi general people mental health problems rates, this may help the reader to be informed about how this study findings differ from the general people. Please refer Bangladeshi studies assessing the pandemic related psychological burdens (ref: https://doi.org/10.31234/osf.io/q4k5b).

Response 10
Thank you for your valuable input. As the suggested study is a pre-print version, we avoided it and cited a nation-wide study that reported psychological symptoms among the general population of Bangladesh (https://doi.org/10.1016/j.jad.2020.10.036). We compared the prevalence of depression as appropriate.

Comment 11
The current study is the first study in Bangladesh that provides the burden and associated factors for doctors' poor mental health outcome during the COVID-19 pandemic. -I request to avoid the first word, because there is a published paper on suicidal behavior of the cohort.

Response 11
As per our knowledge, this study is absolutely first as it reported the most commonly reported psychological symptoms that were not previously reported by any study of Bangladesh among frontline doctors. We respect the comment of the reviewer and want to mention that the previous study assessed suicidal behavior, not the 'anxiety, depression, fear, and sleep disturbance'. Moreover, this study solely conducted among frontline doctors (not intern doctors) who were involved in treating the confirmed or suspected cases of COVID-19.

(D) Overall Comments
Overall observations, the authors did a great job addressing mental health problems of the vulnerable cohort to virus infection, which may have influence in policy level. And the paper was written in a good flow, and also provided some of recommendations. Best of luck.

Response
Thanks for the overall comments.
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias • You can publish traditional articles, null/negative results, case reports, data notes and more • The peer review process is transparent and collaborative • Your article is indexed in PubMed after passing peer review • Dedicated customer support at every stage • For pre-submission enquiries, contact research@f1000.com