A descriptive cross-sectional study design was adopted in this study. Doctors (consultants and medical residents) and nurses at the Lagos State University Teaching Hospital (LASUTH), Ikeja Lagos State, were surveyed. Lagos State University Teaching Hospital is a tertiary institution owned by the Lagos State Government with several departments such as paediatrics and surgery. Data were collected between June and July 2021 via a self-administered questionnaire. Doctors and nurses that worked in the facility as of March 2020 were eligible to participate; interns and house officers were excluded as they were not working in the study location as of the beginning of the pandemic.

The sample size was calculated with the Cochrane formula (Z²[pq]/d²),¹⁸ where Z was 1.96 at a 95.0% confidence interval (CI), and ‘p’ was the proportion of anxiety among healthcare workers in a previous study conducted in Ghana (27.8%),¹⁵ q = 1–p, and ‘d’ was the level of precision at 0.05.

The calculated sample size was 308, with an additional 10% included to correct for non-response, yielding a sample size of 340 HCPs. According to the hospital records, a total of 1452 doctors and nurses (511 doctors, 941 nurses) worked in the 14 departments (such as psychiatry internal medicine, etc.) that make up LASUTH. By proportionate sampling, the HCPs to be selected from each department were determined. Systematic random sampling was employed from the sampling frame of HCPs. Doctors and nurses were selected for all departments. In all, 120 doctors and 220 nurses (a total of 340 HCPs) were recruited for this study.

Sociodemographic characteristics such as age, gender, marital status and occupation were obtained and work-related details such as years of employment and history of anxiety. The Patient Health Questionnaire 9 (PHQ-9), Generalised Anxiety Disorder 7 (GAD-7) and Perceived Stress Scale (PSS) were used to measure depression, anxiety and stress, respectively, while the coping strategy subscale of the COVID-19 questionnaire was used to assess how the HCPs coped with the infection. Other variables include access and adequacy of personal protective equipment (PPE), exposure to COVID-19, the fear of transmitting COVID-19 to colleagues and family and the history of depression, anxiety and comorbidity were sought from the respondents.

Depression was measured with the PHQ-9, which has been previously used to assess depression among healthcare workers and the general population.^12,19,20 The nine-item tool has a four-point Likert scale (not at all = 0 to nearly every day = 3). The scale contains questions such as, ‘feeling down, depressed, or hopeless’ and ‘poor appetite or overeating’. The score range of the PHQ-9 is 0–27, HCPs that score ≥ 10 were considered to be depressed, while those that score below 10 were considered not to be depressed.^21,22 The PHQ-9 has a Cronbach alpha reliability of 0.84–0.89.^20,22 In this present study, the Cronbach alpha reliability of the PHQ-9 was 0.868, which is considered very good according to,²³ and the area under the curve (AUC) was 0.795 (95% CI: 0.713–0.877), which is considered good.²⁴

Anxiety was measured with the GAD-7, which is a seven-item screening measure with a four-point Likert scale (not at all = 0 to nearly every day = 3). The scale contains questions such as ‘feeling nervous, anxious, or on edge’ and ‘feeling afraid, as if something awful might happen’. The GAD-7 score ranges between 0 and 21. Healthcare professionals that score ≥ 10 were considered to be anxious, while those that score less than 10 were considered not to be anxious.^21,22 GAD-7 has been previously used to assess anxiety in several studies with a Cronbach alpha reliability range of 0.80–0.92.^12,22,25 The reliability of the GAD-7 in this present study was very good²³ at 0.876, and the AUC was fair²⁴ with 0.689 (95% CI: 0.566–0.813).

The PSS was used to measure stress among the HCPs. The 10-item tool is measured on a five-point Likert scale (from never to very often) and has a score range of 0–40. Four positively worded items (4, 5, 7, 8) were reversed (0 = 4, 1 = 3, 2 = 2, 3 = 1 and 4 = 0). The scale contains questions such as ‘in the past month, how often have you felt that you were unable to control the important things in your life’ and ‘in the past month, how often have you felt nervous and stressed’. Healthcare professionals that score ≥ 14 were considered to be stressed.¹² The Cronbach alpha reported by an earlier study was 0.82,¹² while the Cronbach alpha was low²⁶ in this present study with 0.575. The AUC for PSS was good²⁴ with 0.718 (95% CI: 0.661–0.774).

The coping strategies subscale of the COVID-19 questionnaire developed by Zhang et al. was adapted to assess coping strategy with a reliability of 0.77.¹⁶ The reliability of the tool was 0.713 in this present study. Twelve statements were used to assess coping strategies, such as ‘I actively learn about COVID-19 (symptoms, route of transmission)?’ and ‘I keep busy to refrain from thinking about the epidemic?’. The items were placed on a five-point Likert scale (1 = Never to 5 = Always). However, our concern was the individual items as they are relevant to this present study.

The SPSS version 26 was used to analyse the data. Simple frequency and means (standard deviation) were used to describe the variables. For binary analysis, Pearson’s chi-square or Fisher’s texts, where appropriate, were used. Spearman’s correlation was used to test the association between the dependent variables because of skewness. Binary logistic regression was used at multivariate analysis to control for confounders. Before then, multicollinearity between the independent variable was tested with the variance inflation factor (VIF). No evidence of collinearity was found as all the VIF were < 2.0. Variables with a p-value of ≤ 0.1 at bivariate analysis were considered for multivariate analysis, while regression analysis was computed at a 95% CI. However, some potential confounders were forced into a model, if necessary, for statistical relevance as suggested by Bursac et al.²⁷ Hosmer-Lemeshow test was used to assess the fitness of the models. Nagelkerke R² was deployed to measure the level of variance explained by the independent variables. Later, AUC was used to assess the predictability power of the dependent variables.

Ethical approval was obtained from the Health Research Ethics Committee (HREC) of the Lagos University Teaching Hospital (HREC Assigned NO: ADM/DSCT/HREC/APP/4331 and LASUTH with reference number LREC/00/10/1598). Written informed consent was sought from the participants after an explicit explanation of the study purpose. Confidentiality was maintained, and participants were allowed to withdraw from the study at any time. Participation was purely voluntary; non-participation was without consequence.

Of the 340 questionnaires distributed, 338 were fit for data analysis giving a response rate of 99.4%. The majority (45.9%) of the health workers were between the ages of 30 years and 39 years, with a mean ± standard deviation of 34.03 ± 8.03 years. The majority (72.5%) were females, and two-thirds (64.5%) were nurses. More than half (58.6%) were married and 41.4% were not married. The majority of the HCPs were Christians (75.7%), and one-fifth lived alone. The mean length of working in the hospital was 6.53 ± 5.96 years, and the majority (58.3%) had worked up to five years in the facility, while only 18% had worked for more than a decade. While 9 in 10 of the HCPs had access to PPE, 61.5% classified the PPE as inadequate. The average working days for the HCPs was 3.66 ± 1.32 during the peak of the pandemic, while the majority (55.9%) worked between three and four days weekly. The majority (89.1%) had no comorbidity, and more than one-fifth (23.1%) have a history of depression or anxiety (Table 1).

Seven in 10 of the HCPs had contact with suspected cases, while more than half (53.3%) had contact with a confirmed case of COVID-19. One in six (16.9%) of the HCPs contracted COVID-19. Many (79.9%) had a colleague infected with the virus and more than one-third (35.2%) of these colleagues died from COVID-19 infection. One in five (20.4%) of the HCPs had a family member that contracted COVID-19, while 5.6% of the HCPs lost a family member to COVID-19. A large proportion (77.5%) of the HCPs are willing to work as frontline staff (Table 2).

The prevalence of depression, anxiety and stress was 9.8%, 5.0% and 62.4%, respectively. Several coping strategies were employed by the HCPs to cope with stress and anxiety. For instance, 45.9% and 45.9%, respectively, learnt about COVID-19 always to cope with the pandemic, while 11% stopped watching TV to avoid COVID-19 news. To cope with the pandemic, 59.2% constantly wore face masks and routinely washed their hands. Ten per cent of employees constantly communicated their worries and needs to their supervisors, while 9.5% took adjuvant medicine, and 4.4% let their feelings out through tears. Six per cent sought psychological support from colleagues, 11.5% always sought spiritual guidance regularly, and 26.6% engaged in chatting with family and friends.

We found a significant positive association between the three dependent variables: depression and anxiety (rs = 0.683, p = < 0.001), depression and perceived stress (rs = 0.460, p = < 0.001) and anxiety and perceived stress (rs = 0.521, p = < 0.001).

Independent variables with a p-value of ≤ 0.1 at bivariate analysis were considered for multivariate analysis. Of the seven variables associated with depression (Table 3 and Table 4), five variables were identified as factors associated with depression among the HCPs when the regression model was fitted (Table 5). Healthcare professionals that were at least 40 years were less likely to be depressed compared with those between 20 years and 29 years (odds ratio: 0.167; 95% CI: 0.043–0.649). Nurses were 1.19 more likely (95% CI: 1.194–13.103) to be depressed than doctors; while HCPs that came in contact with specimen were 2.50 times (95% CI: 1.033–6.070) more likely to be depressed. Healthcare professionals whose colleague died of COVID-19 were 2.91 times (95% CI: 1.210–7.004) more likely to be depressed. Also, those whose family members were infected with COVID-19 were 2.66 times (95% CI: 1.128–6.265) more likely to be depressed. The independent variables explained 22.3% of the variance.

The two variables associated with anxiety at bivariate analysis retained their significant power at regression analysis. Nurses, compared with doctors, were 6.82 times more likely (95% CI: 1.360–34.157) to experience anxiety. Healthcare professionals that were not afraid of contracting COVID-19 were 3.12 times (95% CI: 1.022–9.510), more likely to experience anxiety (Table 6). The independent variables explained 11.9% of the variance.

Age, occupation, a history of depression or anxiety and COVID-19 infection were identified as factors associated with stress among the HCPs. Healthcare professionals that were ≥ 40 years were less likely (95% CI: 0.147–0.981) to be stressed compared with their younger counterparts. This implies that younger HCPs (20–29 years) were 2.64 times more likely to be stressed (reversed reference). Nurses were more likely to be stressed (OR: 2.35; 95% CI: 1.388–3.974) than doctors. Those with a history of anxiety or depression were almost inevitably bound to be stressed compared to those without such history. Stress became more significant, up to 3.33 times more probable (95% CI: 1.599–6.935) if the HCPs had been infected with COVID-19 (Table 7). The independent variables explained 17.7% of the variance.