Impact of COVID-19 on the mental health of frontline and non-frontline healthcare workers in Sri Lanka.

Introduction
Healthcare workers (HCWs) are at risk of mental health problems during a pandemic. Being stationed at the frontline or not may have implications on their mental health.


Objectives
The aims of this study were to assess depression, anxiety and stress among HCWs, to explore differences between frontline and non-frontline workers, and to investigate associated factors.


Methods
In this cross-sectional study, frontline and non-frontline HCWs were recruited from a COVID-19 screening hospital in Sri Lanka. Mental health impact was assessed using Depression, Anxiety and Stress Scale (DASS-21). Sociodemographic data and perceptions of social and occupational circumstances were gathered. Categorical variables were analyzed using Chi square and logistic regression. Odds ratios were calculated for the effect of different perceptions on psychological morbidity.


Results
A total of 467 HCWs participated, comprising 244 (52.2%) frontline and 223 (47.8%) non-frontline workers, with female preponderance (n=341, 77%). Prevalence of depression, anxiety and stress among HCWs were 19.5%, 20.6%, 11.8%, respectively. Non-frontline group showed a higher prevalence of depression (27% vs. 11%, p<0.001), anxiety (27% vs. 14%, p=0.001) and stress (15% vs. 8%, p=0.026). Being married, having children, living with family and higher income were associated with better psychological outcomes. Perceived lack of personal protective equipment, inadequate support from hospital authorities, greater discrimination, and lack of training to cope with the situation predicted poor mental health outcomes, and non-frontline HCWs were more likely to hold such perceptions.


Conclusion
Addressing factors leading to negative psychological outcomes in HCWs should be a key concern during this pandemic.


Methods
In this cross-sectional study, frontline and nonfrontline HCWs were recruited from a COVID-19 screening hospital in Sri Lanka. Mental health impact was assessed using Depression, Anxiety and Stress Scale . Sociodemographic data and perceptions of social and occupational circumstances were gathered. Categorical variables were analyzed using Chi square and logistic regression. Odds ratios were calculated for the effect of different perceptions on psychological morbidity.
Conclusion Addressing factors leading to negative psychological outcomes in HCWs should be a key concern during this pandemic.

Background
Since its initial identification from China in early January 2020, COVID-19 has been taking its unrelenting toll on myriad aspects of human lives worldwide [1]. The ever-rising numbers of cases and deaths announced repeatedly on media, often paired with cues of danger and despair, and the ensuing fear of contracting the disease, have understandably placed the human population at risk of mental health sequelae [2]. An increase in psychological morbidity in this background has already been reported from various places [2][3][4]. Since the first case in Sri Lanka was confirmed on 27 January 2020, the island has witnessed a steady rise in the number of cases, exceeding 2000 cases in June 2020, while this study was underway.
Healthcare workers (HCWs), whose services were brought to the limelight during this pandemic, are at a higher risk of psychological distress, as they interact with patients potentially having COVID-19 infection. Being a frontline (FL) HCW has been hypothesized to be a risk factor for adverse mental health outcomes, and a few studies from China [5] and Italy [6] have provided evidence of this. Even though not many HCWs in Sri Lanka had contracted COVID-19 at the time of this study, the staggering numbers of HCWs infected with COVID-19 as reported from other countries [7] may have led FL workers in Sri Lanka to vicariously experience this danger. Vol. 66, No. 1, March 2021

Original article
Prolonged curfews and social distancing policies have imposed constraints on coping strategies usually adopted by people to alleviate stress. Religious activities, social gatherings, family outings, entertainment events, physical exercise and sport events have been all curtailed. Maladaptive coping strategies such as substance use were altered as well.
Investigating the psychological impact of the current pandemic on HCWs was important for many reasons. Good mental well-being is crucial for optimal occupational and social functioning of HCWs. Psychological consequences of a pandemic may persist even one year after the crisis among HCWs [10]. Therefore, appropriate psychological support must be provided to HCWs. To this end, the World Health Organization has issued instructions to healthcare leaders worldwide to ensure access to mental health services for HCWs [11]. However, in order to plan such strategies, it is important to gauge the prevailing nature and gravity of mental health issues among HCWs in the local setting. The objective of this study was to assess the psychological impact of COVID-19 in terms of depression, anxiety and stress among FL and NFL HCWs at a tertiary care hospital in Sri Lanka, and to investigate associated factors.

Method Study design and setting
This cross-sectional study was conducted between June-August 2020 at North Colombo Teaching Hospital (NCTH), a screening centre for COVID-19 in the Western Province of Sri Lanka, deemed at 'high risk' for COVID-19 transmission.

Study participants
All categories of HCWs -doctors of all grades, nurses of all grades, ancillary staff (pharmacists, medical laboratory technologists, radiographers), and supporting staff were included.

Participant selection
Convenient sampling was used to recruit both FL and NFL participants. FL staff included HCWs who dealt with potentially COVID-19-infected patients being screened at the outpatient department, emergency treatment unit, fever corner, isolation wards, medical wards, pediatric wards, high dependency units and intensive care units. Employees of laboratories within the hospital who ran RT-PCR diagnostic tests for COVID-19 and ambulance drivers who transported suspected and confirmed patients were also considered as FL. All other HCWs employed at the hospital were considered NFL workers.

Sample size
Sample size was calculated using a standard equation used when proportions (prevalence rates) are compared between two groups (FL and NFL) [12].
An α of 0.05, a power of 80%, and proportion values (p 1 = 0.58; p 2 = 0.45) based on a prior study in China [5], were used for the calculation. This yielded a sample size of 225 for each group, for a total of 450 subjects. A final sample of 500 was deemed appropriate, considering possible non-responders.

Measures
Mental health status was assessed using Depression, Anxiety and Stress Scale (DASS-21). This 21-item, selfadministered scale includes three 7-item subscales assessing depression, anxiety and stress separately, as experienced during the past week. Responses are provided on a 4-point Likert scale. Subscale scores are generated by summing the item scores in each subscale, and multiplying by two. DASS-21 has been translated and validated into Sinhala and Tamil [13,14]. Validity statistics of the Sinhala version, such as concurrent, criterion and construct validity, as well as reliability measures were comparable to the original English version. DASS-21 provides cut-offs to determine the severity of symptoms ('normal', 'mild', 'moderate', 'severe' and 'extremely severe') in each subscale.
A socio-demographic questionnaire was administered, alongside DASS-21. Perceptions about social and occupational circumstances and coping strategies were also assessed, using a 5-point Likert scale. This sociodemographic questionnaire was developed and refined by an expert panel that consisted of two psychiatrists and a physician. Previous literature was reviewed and local socio-cultural and administrative factors were considered in developing this questionnaire. The panel discussed and ensured its content and face validity.

Procedure
Ethics approval was obtained from the Institutional Ethics Review Committee. The permission of the Director of NCTH was also obtained. A list of FL and NFL units was prepared, and these units were visited in a planned and coordinated manner by the data collectors to invite potential participants. In order to ensure inclusiveness, different shifts were covered and staff leave was factored in, so that almost all members of staff in these identified units were invited to participate. Informed written consent was obtained. Participants filled the sociodemographic questionnaire and the DASS-21, on their own, in the language of their choice. Filling both questionnaires took approximately 15-20 minutes. In some instances, the participants from a certain unit were given a specified period to fill the questionnaires (up to 3 days), and the Original article filled questionnaires were collected at a later time; the respective units were contacted before visiting them to collect the filled forms. All hospital policies and guidelines for prevention of COVID-19 transmission, such as social distancing, wearing masks/visors, and hand hygiene, were adhered to, when interacting with participants.

Ethical issues
Ethical issues that may arise due to concerns of COVID-19 transmission were considered; however, we were able to minimize these risks by adhering to guidelines for prevention of COVID-19 transmission within the hospital. Considering the scientific importance of mental health research in the context of the COVID-19 pandemic and the potential impact on healthcare policy, benefits were deemed to outweigh risks.

Data analysis
IBM SPSS Version 21 was used for data analysis. Descriptive statistics were used to describe population characteristics. Prevalence rates were calculated based on DASS-21 cut-offs. Factors associated with binary outcomes (e.g. presence of depression) were assessed using Chi square test. Sociodemographic differences between FL and NFL groups were assessed using Chi square test; factors which were significant were fitted onto a logistic regression model, as confounders in the association between position (FL/NFL) and DASS-21 outcomes. To describe the effect of social and occupational perceptions on DASS-21 outcomes, the 5-point Likert scale was recoded into a binary variable; the affirmative responses, i.e. 'Agree' and 'Strongly agree', were combined into one category, and the negative responses i.e. 'Disagree' and 'Strongly disagree' were combined similarly; the neutral response was excluded from the analysis. Odds ratios were calculated to show the association between different perceptions and DASS-21 outcomes. Mann-Whitney U test compared the responses given by FL and NFL groups on the Likert scale.

Sociodemographic factors associated with depression, anxiety and stress
The associations between socio-demographic factors and DASS-21 outcomes of depression, anxiety and stress are shown in Table 4. Having children (p=0.015), higher income (p=0.012) and living with family (p=0.002) were protective against depression. Being married (p=0.010), having children (p=0.009) and living with family (p=0.001) were associated with less anxiety. Being married (p=0.001) and having children (p=0.002) were protective for stress. Note: AOR = Adjusted odds ratio; ref = reference category a Only the socio-demographic variables which were found to be significantly different between frontline and non-frontline participants (see Table 1), were included in the logistic regression model as covariates.

Prevalence of depression, anxiety and stress
b Cut-offs for depression, anxiety and stress on DASS-21 subscales were >9, >7 and >14, respectively. *significant at 95% confidence

Perceptions of work environment, social circumstances and coping strategies, and their relationship with depression, anxiety and stress
The participants' responses regarding their social circumstances, support received at the hospital, and coping strategies used, are tabulated against DASS-21 outcomes, in Table 5. The risk of DASS-21 outcomes if one agrees with each statement is expressed using odds ratios. Workplace-related perceptions such as the fear of contracting COVID-19, inadequacy of PPE, lack of administrative support, lack of psychological support and constraints on delivering patient care were significantly associated with higher psychological morbidity. Perceptions related to personal circumstances including the fear of transmitting the infection to family members, discrimination experienced due to working in a COVID-19 screening hospital, and the difficulty in balancing professional and domestic lives were also associated with greater mental health burden. Among coping strategies, seeking help from family and friends, and spending time productively appeared to be protective. An increase in alcohol and other substance consumption was associated with depression, anxiety and stress. Practicing religion or taking up new hobbies were not associated with outcomes.
The differences in responses given by FL and NFL groups revealed that the NFL group was more likely to feel they were not provided adequate PPE, they were vulnerable to contract COVID-19 in spite of PPE, and they had not been provided adequate administrative and psychological support. NFL workers felt that they had not been trained to cope with workplace changes. Also, NFL workers had experienced greater discrimination due to their employment at a hospital, and were more worried about loss of employment or income. NFL workers were less likely to cope using the help of family and friends.

Discussion
This study reports for the first time, higher depression, anxiety and stress among NFL HCWs in comparison to FL HCWs, during the COVID-19 pandemic, although both groups had higher psychological morbidity when compared to the general population. Having children, living with family, being married and higher income appeared to be protective. Perceived lack of PPE, administrative and psychological support, and discrimination were associated with psychological outcomes.
The study assessed HCWs from a COVID-19 screening centre located in a high-risk area. As the hospital was not a treating centre, FL HCWs who were at risk of being exposed to COVID-19 prior to testing, did not continue to treat patients who became positive. The NFL HCWs only ran the risk of unknowingly being exposed to COVID-19 patients. However, the criteria used to define the FL/NFL status of a HCW in this study may not accurately reflect their actual exposure to COVID-19. Nevertheless, we assumed that FL HCWs (as per present criteria) are at greater perceived risk of exposure to COVID-19 than their NFL counterparts. These two groups were dissimilar in perceptions about their social and work-related circumstances, which may have contributed to the disparity in psychological outcomes. For instance, the lack of PPE for NFL workers may have been a real concern as FL HCWs were given priority. Similarly, NFL workers may have been stigmatised unfairly by their communities for working in the hospital, despite having little risk of exposure to COVID-19. Adverse psychological effects of stigma during the pandemic have been similarly reported from Italy [15].
Consistent with previous observations [5,16], our study found perceived inadequacy of PPE to increase the risk of mental health disturbance. This highlights the need to ensure PPE for both FL and NFL HCWs. Also, NFL workers felt they had not been adequately trained to handle workplace changes caused by the pandemic. Such a lack of psychological preparedness would have increased their risk of mental health issues. These findings underscore the need to direct more attention to the mental health needs of NFL HCWs.
The prevalence of psychological morbidity in all HCWs is higher than the general population of Sri Lanka. A previous study [17] among non-HCWs in the country showed depression and anxiety disorder prevalence of 6.6% and 9.1%, respectively.
These prevalence rates among Sri Lankan HCWs seem to fall on the lower end of the range of findings from other countries. A review of studies on HCWs' mental health during this pandemic has reported a prevalence of 20-40% for depression and 30-70% for anxiety [9]. It should be borne in mind that these prevalence rates were derived using different instruments and cut-offs. A study from Singapore [18], not included in the foregoing review, used the same instrument and cut-offs as the current study, and found a relatively lower prevalence of depression (9%), anxiety (14.5%) and stress (6.6%) among HCWs.
Being a FL HCW was shown to elevate the risk of psychological sequelae in China [5,16,19] and Italy [6,20]. However, an absence of such a risk difference has also been reported from China [21]. Comparable research from other parts of the world is scarce. Among Chinese HCWs [5], females exhibited greater levels of depression and anxiety. The present findings did not show a gender disparity. Female overrepresentation in the sample limits our ability to draw inferences about male HCWs. However, a preponderance of females in HCW samples is seen in other studies as well [18,22].
Higher mental health burden has been reported among nurses compared to other professional categories in China [5,23] and Japan [22], whereas in Singapore, non-medical professionals demonstrated greater levels of depression and anxiety [18]. However, no significant variation across professions was observed in the current study.
In line with established etiological understandings and empirical evidence [19], social support, in the form of being married or living with family, was shown to be protective against psychological problems. Having children also reduced the risk of depression, anxiety and stress; however, this could be the result of a confounding effect by marital status.
The fear of contracting COVID-19 and infecting family members being significant predictors of mental health issues in previous reports [5,9] was replicated in the present analysis. Those who felt it a challenge to balance their work and domestic life, and those who had been compelled to live away from their families demonstrated a higher prevalence of psychological issues. Coping with the help of family and friends was protective against mental health problems. A Chinese cohort similarly reported seeking support from family and friends as the most salient coping method [23]. These observations illustrate the impact of family-related concerns on HCWs' mental health during a pandemic. However, the directionality of this association cannot be verified from present findings, as those who are suffering from depression are in turn more likely to hold such negative cognitions.
The maladaptive nature of alcohol and other substance use is illustrated by the higher prevalence of depression, anxiety and stress among those who experienced an increase in this behaviour.
The high response rate of over 90%, which can be attributed to the convenience sampling and the different strategies used by the research team to ensure good response rate, is a strength of this study. However, nonprobability convenience sampling would have introduced a sampling error. The study findings cannot be generalized to COVID-19 treating hospitals where FL HCWs may have constant exposure to confirmed cases. As the sample is Original article from a single centre, these context-specific issues are a limitation. Furthermore, the findings are from self-reported measures and cross sectional in nature and therefore do not establish clear morbidity or causality. Although mental health problems were assessed using a validated instrument, the sociodemographic questionnaire, which assessed social and occupational circumstances and coping methods, was not a formally validated tool, and therefore, may limit the validity of some inferences based on responses to this questionnaire.

Author contributions
All authors contributed to the conceptualization and design of the study. KB, DK, CAA and DSK contributed to the acquisition of data. AB and SW conducted the data analysis. AB, SW, STD and AH contributed to data interpretation and writing the manuscript. All authors read and approved the final manuscript.

Conflicts of interest
The authors declare that they have no conflicts of interest.