Healthcare professionals working during the COVID-19 pandemic experienced documented clinical distress, anxiety, depressive symptoms, and sleep disturbances [7] [8] , along with profound experiences of loss and grief [28] . Their mental health declined due to extended working hours, contamination risks, family separation and isolation, and the burden of making difficult moral decisions regarding medical resource allocation [9] [29] .

Another study [30] found that among healthcare workers regularly treating COVID-19 cases, 32.9% reported unusual anxiety and 27.9% experienced depression. Additionally, 24.5% and 72.8% of participants reported mild and moderate stress levels, respectively. Key risk factors included insufficient infection control training and pre-existing stress-inducing medical conditions. Additionally, healthcare workers were seven times more likely to experience severe COVID-19 compared to non-essential workers [31] .

Healthcare professionals faced unprecedented challenges in treating patients while maintaining high levels of care. Stress in these situations increases the risk of physical and mental health disorders [32] . Similar patterns of elevated stress, anxiety, depression, and post-traumatic stress disorder symptoms were observed during previous epidemics, including H1N1 influenza and SARS.

Studies of frontline health workers in Bangladesh using the Patient Health Questionnaire-4 (PHQ-4) revealed anxiety prevalence at 17.6%, depression at 15.5%, PTSD at 7.6%, and insomnia at 5.9% [33] . Comparative research has consistently shown higher rates of psychological concerns, particularly insomnia, among healthcare workers versus non-healthcare workers [34] .

When the crisis began, the United States healthcare system was inadequately prepared to protect and support both medical and non-medical hospital staff, including essential frontline non-medical workers (ENMWs). During the COVID-19 pandemic, non-medical essential workers were found to be more at risk for mental health issues due to a lack of protective equipment, training, and organizational support [10] [35] , higher mortality rates [36] , and other functional impairments [37] [38] . These issues were compounded by workers reporting fear of contracting the virus, lack of organizational support, work disruption, and financial insecurity [16] .

Despite ranking among the lowest-paid hospital staff, ENMWs such as custodians require equivalent protection and support as frontline medical workers [10] . Within organizational hierarchies, custodians are often considered “invisible” workers [39] . This invisibility stems from their association with “dirty work” or tasks that carry social stigma or degradation [40] . Other positions considered essential non-medical workers include food transporters, supply chain workers, transportation workers, security personnel, and maintenance workers. These lower-ranked workers often came from marginalized backgrounds, experienced higher rates of COVID-19 [41] and lacked access to safety and health measures [31] .

Supervisor support reflects the quality of supervisor-subordinate relationships, and the degree of support employees receive [42] [43] . Enhanced workplace support from supervisors correlates with improved mental health outcomes among healthcare workers [44] . Notably, medical staff who felt valued at work experienced 40% less burnout compared to those feeling undervalued [45] . Fear of COVID-19 transmission and concerns about personal and family health became pervasive among frontline healthcare workers [45] [46] . This fear significantly impacted mental health, manifesting in elevated rates of post-traumatic stress symptoms (11% - 73.4%), depression (27.5% - 50.7%), anxiety (45%), and insomnia (34% - 36.1%) [47] .

COVID-19’s impact on healthcare workers’ physical and psychological well-being has been substantial [48] . Among essential medical staff, 34.3% of inpatient physicians and 53.4% of inpatient nurses expressed intent to leave their positions within two years. Interestingly, essential non-medical workers demonstrated higher job acceptance rates (62.7%) compared to nurses (60.4%) and physicians (50%) [10] .

Essential non-medical workers merit equal consideration to their medical counterparts, as they face similar risks when entering patient rooms. Their contributions are vital—without essential non-medical workers cleaning and sanitizing COVID-19-infected rooms, medical professionals cannot perform their duties effectively [49] .

The final sample consisted of N = 339 participants from a Southern California hospital. The majority of participants were male (n = 237, 69.9%), aged 36 - 50 years (n = 142, 41.9%), and single (n = 195, 57.5%). The sample was fairly evenly split between medical personnel (n = 158, 46.6%) and non-medical hospital employees (n = 181, 53.4%). Most participants worked full-time (n = 290, 85.5%), with smaller proportions working part-time (n = 33, 9.7%), per diem (n = 11, 3.2%), or temporary positions (n = 5, 1.5%). Regarding COVID-19 concerns, 38.6% (n = 131) reported fear of infection, 33.3% (n = 113) reported no fear, and 28.0% (n = 95) expressed neutral feelings. Notably, a substantial majority (n = 259, 76.4%) reported feeling unsupported by management during the pandemic, with only 23.6% (n = 80) indicating adequate support (see Table 1 ).

Note. Participant (N = 339) demographic data in aggregate, including gender, age, marital status, occupation, employment status, fearfulness of contracting COVID-19, and feeling supported by management. Percentages may not total 100 due to rounding.

The demographic data revealed several notable patterns. The high percentage of respondents reporting no managerial support contextualizes findings related to stress and job satisfaction. The predominantly male sample and older age distribution limit generalizability, with findings potentially more applicable to experienced workers. The high proportion of full-time employees suggests results may not reflect the experiences of part-time, per diem, or temporary workers. Additionally, control variables were omitted to maintain survey brevity and maximize response rates. Questions about pre-existing mental health conditions could have raised ethical concerns, while socioeconomic questions might have decreased participation due to their sensitive nature.

The first research question and corresponding hypotheses examined if or to what extent perceived supervisor support and fear of infection combined and separately predicted depression scores on the Depression, Anxiety, and Stress Scale (DASS) [14] .

A linear multiple regression was used to analyze the first research question. Perceived supervisor support and fear of infection were the predictor variables and the DASS depression scores were the criterion variable. The analysis aimed to determine how well the predictor variables, combined, explained the variance in the depression scores.

The model summary was statistically significant, F(2, 336) = 18.82, p < 0.001, where the two predictor variables, perceived supervisor support and fear of infection, explained 10.1% (R² = 0.101) of the variance of the criterion variable, depression scores from the DASS, with an adjusted R² = 0.095, indicating a slight adjustment for the number of predictors in the model. Overall, the model explains that the predictor variables, when combined, significantly predicted a meaningful and modest proportion of the variance in the dependent variable (See Table 2 ).

Note. a. Predictors: (Constant), I fear I might become infected with COVID19, my direct supervisor has provided me with adequate support during the COVID-19 pandemic.

Both the model summary and the ANOVA (below) results indicate that supervisor support and fear of becoming infected with COVID-19 significantly predict depression scores of the respondents. The model only explains 10.1% of the variance, the combined impact of both variables. The significant F-value and the low p-value (<0.001) in the ANOVA table suggest that the relationship between the predictor variables and the criterion variable is meaningful (See Table 3 ).

Note: a. Dependent Variable: Total score for depression on the DASS. b. Predictors: (Constant), I fear I might become infected with COVID-19, my direct supervisor has provided me with adequate support during the COVID-19 pandemic.

After conducting multiple regression for combining the predictor variables, a hierarchical regression analysis was conducted to determine how much variance was accounted for by the fear of contracting COVID and supervisor support of the depression scores from the DASS, separately. The first hierarchical regression was performed while holding the fear of infection constant and examining only supervisor support (See Table 4 ).

Note: a. Predictors: (Constant), Support.

The analysis revealed a weak but statistically significant relationship between supervisor support and depression scores from the DASS, F(1, 337) = 7.095, p < 0.008. The R value (0.144) indicates a weak positive correlation, and the R-squared value (0.021) indicates that approximately 2.1% of the variance in depression scores was explained by the level of supervisor support during the pandemic. In contrast, the adjusted R-squared result indicates that only 1.8% of the variance in depression scores was explained by the level of supervisor support. Despite the small variance, the ANOVA analysis confirmed the statistically significant finding that supervisor support does significantly contribute to the model in predicting depression scores (See Table 5 ).

Note: a. Dependent Variable: Fear. b. Predictors: (Constant), Support.

To complete the analysis for the first research question, the second hierarchical regression was performed while holding supervisor support constant and examining only fear of infection. The results revealed another weak but statistically significant relationship between supervisor support and depression scores from the DASS, F(1, 337) = 11.342, p < 0.001. The R value (0.180) indicates a weak positive correlation, and the R-squared value (0.021) indicates that approximately 3.3% of the variance in depression scores was explained by the fear of infection during the pandemic. In contrast, the adjusted R-squared result indicates that only 3.0% of the variance in depression scores was explained by the fear of infection (See Table 6 ).

Note: a. Predictors: (Constant), Fear.

The ANOVA from the second hierarchical regression again confirmed a statistically significant finding (p < 0.001), which explains only a small portion of the variance in depression scores when accounting only for fear of infection. While the F-statistic (11.342) shows that the fear of being infected variable has an impact on the overall model, there is a large portion of the variance that cannot be explained by the fear of infection predictor alone (See Table 7 ).

Note: a. Dependent Variable: Depression score b. Predictors: (Constant), Fear.

The second research question and corresponding hypotheses examined if stress, depression, and anxiety scores from the DASS, combined and separately, predicted fear of infection.

The model summary was statistically significant but weak relationship, F(3, 335) = 4.927, p < 0.002, where the three predictor variables, stress, anxiety, and depression scores on the DASS explained 4.2% (R² = 0.042) of the variance of the criterion variable, depression scores from the DASS, with an adjusted R² = 0.034, indicating a slight adjustment for the number of predictors in the model and the complexity of the model. Overall, the model indicates that depression, anxiety, and stress scores together have a significant relationship with the dependent variable, albeit a weak relationship (See Table 8 ).

Note: a. Predictors: (Constant), Depression score, Total score for anxiety on the DASS, Total score for stress on the DASS

Despite the small variance, as seen in the ANOVA table, this analysis confirmed the statistically significant finding that when the predictors are combined, it does significantly contribute to the model in predicting fear of infection (See Table 9 ). The overall variance is still rather low (4.2%) according to the model summary, but this model explains more variance than the previous model, where the predictor variable was supervisor support alone (4.2%).

Note: a. Dependent Variable: Fear. b. Predictors: (Constant), Depression score, Total score for anxiety on the DASS, Total score for stress on the DASS.

To complete the analysis for the second research question, three hierarchical regressions were performed, where each of the predictor variables was, in turn, held constant to determine which (if any) of the DASS scores on anxiety, depression, or stress predicted fear of infection. The first hierarchical regression was performed to determine if anxiety scores while holding depression and stress constant were a significant predictor of fear of infection.

The model summary was statistically significant but had a weak relationship, F(1, 337) = 9.603, p < 0.002, where the anxiety scores on the DASS explained 2.8% (R² = 0.028) of the variance of the criterion variable, anxiety scores from the DASS, with an adjusted R² = 0.025, indicating a slight adjustment for the number of predictors in the model and the complexity of the model. Overall, the model indicates that anxiety scores have a statistically significant relationship with the dependent variable (See Table 10 ).

Note: a. Predictors: (Constant), Total score for anxiety on the DASS. b. Predictors: (Constant), Depression score, Total score for stress on the DASS.

The corresponding ANOVA table for the first model indicates a significant finding. Taken together, the F-value (9.603) and the p-value (p = 0.002) both indicate that anxiety is a statistically significant predictor of fear of infection. This is a notable portion of variance from fear of being infected (see Table 11 ).

Note: a. Dependent Variable: Fear. b. Predictors: (Constant), Total score for anxiety on the DASS, Depression score, Total score for stress on the DASS.

The second hierarchical regression was performed to determine if depression scores were a significant predictor of fear of infection while holding anxiety and stress constant. The model summary was statistically significant, F(1, 337) = 11.342, p < 0.001, where the depression scores on the DASS explained 3.3% (R² = 0.033) of the variance of the criterion variable, depression scores from the DASS alone, with an adjusted R² = 0.030, indicating depression scores have a statistically significant relationship with the dependent variable (See Table 12 ). It should be noted here that the relatively low R Square value indicates that depression accounts only for a small part of the overall variance in fear of infection.

Note: a. Predictors: (Constant), Depression score, b. Predictors: (Constant), Total score for anxiety on the DASS, Total score for stress on the DASS.

The corresponding ANOVA table for the first model indicates a significant finding. Taken together, the F-value (11.342) and the p-value (p < 0.001) both indicate that depression is a statistically significant predictor of fear of infection. This is a notable portion of variance from fear of being infected (see Table 13 ).

Note: a. Dependent Variable: Fear, b. Predictors: (Constant), Depression score, c. Predictors: (Constant), Total score for anxiety on the DASS, Total score for stress on the DASS.

The third hierarchical regression was performed to determine if stress scores while holding anxiety and depression constant were a significant predictor of fear of infection. The model summary was statistically significant, F(1, 337) = 14.034, p < 0.001, where the stress scores on the DASS explained 4.0% (R² = 0.040) of the variance of the criterion variable, stress scores from the DASS alone, with an adjusted R² = 0.037, indicating depression scores have a statistically significant relationship with the dependent variable (See Table 14 ).

Note: a. Predictors: (Constant), Total score for stress on the DASS, b. Predictors: (Constant), Total score for stress on the DASS, Total score for anxiety on the DASS, Depression score.

The corresponding ANOVA table for the model indicates a significant finding. Taken together, the F-value (14.034) and the p-value (p < 0.001) both indicate that stress is a statistically significant predictor of fear of infection (see Table 15 ).

Note: a. Dependent Variable: Fear, b. Predictors: (Constant), Total score for stress on the DASS, c. Predictors: (Constant), Total score for stress on the DASS, Total score for anxiety on the DASS, Depression score.

This study presents both strengths and limitations. Key strengths include the methodological design, theoretical framework, and practical relevance, which collectively provided a thorough understanding and enabled the examination of bidirectional relationships (i.e., depression as both a predictor and an outcome). The theoretical foundation is built upon existing literature using the Job Demands-Resources (JD-R) Theory [11] . Furthermore, the study’s practical relevance and findings illuminate critical issues affecting healthcare workers both during and beyond pandemic situations. Healthcare administrators can potentially utilize these findings to implement workplace support and intervention strategies, identify at-risk workers, and develop organizational policies and support systems.

Regarding limitations, the cross-sectional design restricts our ability to establish causality [60] [61] and captures only a snapshot in time. Additional limitations include potential response variance due to self-report measures, social desirability bias, possible incomplete relationship capture through multiple regression analysis, and lack of control for pre-existing mental health conditions. The demographic composition of the sample presented several limitations. The predominantly male sample, older age distribution, and high proportion of full-time workers limited our ability to understand the experiences of women, younger employees, and those in part-time, per diem, or temporary positions.

The study’s generalizability is also constrained by its single-site sampling at one Southern California hospital. This study utilized personnel from a single hospital in California. This choice offered practical advantages: easier administrative access, lower resource requirements, and enhanced response rates due to one author’s prior employment and maintained relationship with the site. Methodologically, using one site eliminated hospital policy variations as confounding variables and enabled deeper contextual understanding. However, this approach limited our ability to account for regional differences among healthcare centers.

Research on hospital management highlights the crucial role of effective leadership in supporting both professional and non-professional staff. Evidence suggests that human resource (HR) functions often prioritize non-professional staff, leaving a gap in the management and support of core professionals [62] . Leadership practices in hospitals are frequently integrated into clinical work and shared across organizational boundaries, with their configurations varying depending on the specific healthcare setting [63] .

During the COVID-19 pandemic, nurses emphasized the importance of visible and accessible leadership, transparent communication, resilient supply chains, equitable policies, and psychological support from trusted sources [64] . The successful implementation of Total Quality Management (TQM) in healthcare depends on several key practices, including top-management commitment, teamwork, process management, customer focus, resource management, organizational culture, continuous improvement, and training [65] . These findings underline the necessity of integrated and context-specific leadership approaches to effectively support hospital staff, especially during crisis periods.

Socio-emotional support, particularly direct support from supervisors, has been shown to have a significant positive impact on the affective commitment of nurses and nurse aides [66] . Factors influencing human resource management (HRM) in hospital settings involve both quantitative and qualitative performance dimensions [67] . Future research should also focus on the integration of nursing assistants to mitigate staffing shortages and the design of HRM strategies tailored to the specific demands of healthcare providers [67] .

Future research should examine differences in supervisory styles and support levels across departments (e.g., emergency, ICU, administration). Our study did not account for variations in management structures, which could reveal important contextual differences within healthcare settings. For example, emergency and intensive care units face distinct high-stress situations and unpredictable stressors. The low percentage (23.6%) of workers reporting management support might reflect departmental variations in hierarchical structure, as clinical departments typically have multiple management layers while administrative units have more traditional hierarchies.

To address healthcare workers’ mental health and improve workplace support systems, several recommendations are proposed: conducting longitudinal studies to examine mental health trajectories throughout crisis periods, implementing comparative studies evaluating different types of supervisor support, investigating interactions between employees’ personal and organizational resources during crises, and replicating this study using more robust tools to measure fear of infection, such as the Workplace Safety Questionnaire (WSQ) [68] [69] . A longitudinal study may prove useful in that it would allow researchers to track measurements at multiple points to see how relationships between variables evolve, and would provide the ability for researchers to determine if the reciprocal /bi-directional relationships amplify over time, dimmish as workers become more adept at dealing with a crisis such as COVID-19, or if external circumstance of the pandemic (or any other such emergency) may result in fluctuations in the data; any of these things are not possible to evaluate without temporal data.

Future research should extend beyond quantitative methods to explore healthcare workers’ perceptions of mental health support systems, workplace environments, and professional development in response to the COVID-19 pandemic. Collecting rich qualitative data about workers’ lived experiences during the pandemic could offer valuable insights into their views on the effectiveness of supervisor support and its impact on mental health [70] .

Supervisors play a pivotal role in enhancing the performance and well-being of non-essential medical staff within hospital settings. The implementation of structured supervision models has been shown to improve the skills, confidence, and teamwork of non-physician clinicians, resulting in increased surgical capacity and a reduction in unnecessary referrals [71] . Acknowledging the significant contributions of non-clinical staff to the healthcare environment is also essential, as their involvement profoundly influences institutional culture and patient care outcomes [72] .

Hospital management plays a critical role in supporting non-essential medical staff, particularly during periods of crisis. Research highlights the importance of implementing policies that ensure all staff, including non-clinical workers, are provided with the necessary information, equipment, and support to maintain their safety and well-being during emergencies such as the COVID-19 pandemic [73] . This delegation allows medical professionals to focus on patient care while non-medical staff handle ancillary functions. Overall, strategic management approaches aimed at supporting non-essential staff can significantly improve hospital operations and foster a healthier, more resilient workforce [73] .

Healthcare administrators should consider these findings when addressing barriers to support service implementation, policy development, and supervisor support in workplace dynamics. Practical recommendations include implementing crisis-specific supervisor training, establishing standardized mental health screenings, mandating regular mental health check-ins for both employees and supervisors, tracking intervention effectiveness, monitoring staff well-being, and assessing the cost-benefit outcomes of implemented programs.

The overarching focus should remain on improving employee well-being, regardless of the specific research or practical recommendations implemented. Additionally, addressing peer support needs during epidemics or pandemics and exploring different mental health support models could provide valuable insights for both scholars and professionals in the field.