Digitalization and Employee Health and Well-Being During COVID-19

Abstract

Employees were required to adopt new working methods within a very short time frame during the COVID-19 period through digitalization. While digitalization has been largely perceived as an enabler during the pandemic, its impact on employee health and well-being remains complex and underexplored, particularly in the public sector, where employees have less discretion to adapt digital tools. This study examines how rapid workplace digitalization during COVID-19 affected employee health and well-being in the public sector. Drawing on the job demands–resources (JD-R) framework, we focus on three specific forms of digital work—digital meetings, digital clearance, and digital training—selected because they represent distinct theoretical pathways through which digitalization affects well-being, such as digital meetings and digital training can increase job demands that can deplete employee energy and increase stress, whereas digital clearance operates as a job resource that reduces bureaucratic hurdles and enhances autonomy. To test these ideas, this study uses data from the 2020 Australian Public Service Commission Census (n = 108,085), and applies ordinal and multinomial generalized structural equation modeling (GSEM) to assess the effects of three new ways of working—digital meetings, digital clearance, and digital training—on employees’ health and well-being, as well as the mediating roles of organizational support. The results demonstrate that while digital clearance is positively associated with employee health and well-being, digital meetings and digital training are negatively associated. Organizational support mediates these relationships, underscoring its importance in mitigating adverse effects. These findings highlight the mixed consequences of digitalization for public employees’ health and well-being and point to the need for supportive organizational strategies in times of crisis. As a practical implication, this study suggests that public sector organizations should prioritize employee mental health in teleworking policies, adopt employee-centered digital transformation strategies that provide adequate resources and training support, and implement digital clearance processes that enhance employee well-being, particularly during a crisis.

1. Introduction

COVID-19 has significantly impacted public and private sector employees, much like it has affected millions of citizens worldwide since December 2019. Despite the availability of vaccines, the emergence of new variants led to continued infections. Governments, therefore, have been particularly concerned about the safety of their employees, even as they strived to maintain many public services. The “new normal of work” (NNW) included the adoption of telework facilitated by digitalization, which played a crucial role in ensuring the safety and well-being of employees (Raghavan et al., 2021; Moller et al., 2024; Lunde et al., 2022, for a systematic review). Digitalization also enhanced the productivity of employees and ensured service continuity during the crisis (Ha et al., 2024; Aristovnik et al., 2021), thus providing stability during the pandemic.

While digitalization has been largely perceived as an enabler during the COVID-19 pandemic, its impact on employee health and well-being remains complex and underexplored, particularly in the public sector. Prior research in the private sector has established that digital technologies and remote work have a positive effect, such as enhancing flexibility; they also have negative effects, such as increasing stress, overload, and blurred work–life boundaries (Beyea et al., 2025; Howarth et al., 2018; Johnson et al., 2020; Standaert et al., 2023). For example, a meta-analysis identified a major source of strain as videoconferencing fatigue, finding that it is significantly driven by psychological factors such as the feeling of “hindered movement” (Beyea et al., 2025). While increased virtual meeting participation is associated with negative well-being indicators such as workload, stress, and fatigue, it is also positively associated with a greater sense of work influence (Standaert et al., 2023). More broadly, technology is considered a “double-edged sword” impacting mental health at work (Johnson et al., 2020), although tailored digital health interventions show promise for improving well-being, especially by reducing depression and anxiety in employees with higher levels of psychological distress (Howarth et al., 2018).

However, we know little about how these dynamics unfold in public organizations, where digitalization, driven by the critical goal of delivering public services and constrained by distinct managerial and governmental policies, has a much broader impact than in the private sector. Prior studies thus demonstrate that digitalization can both support and strain employees, but the evidence comes largely from the private sector and treats digitalization and digital technologies in relatively broad terms, leaving limited knowledge on specific digital practices and their implications for public employees’ well-being. Unlike private firms, public sector agencies emphasize accountability, transparency, and equity over profit and productivity (Rainey et al., 2021), so public sector employees often have less discretion to adapt digital tools to their needs. As a result, the effects of workplace digitalization on well-being in the public sector may manifest differently.

This article aims to investigate the impact of digitalization during COVID-19 on employee health and well-being in the public sector by analyzing this research question: How did the three specific facets of workplace digitalization—digital meetings (DMs), digital clearance (DC), and digital training (DT)—affect the health and well-being of public sector employees during the COVID-19 pandemic? Drawing on large-scale data from the Australian Public Service Commission’s (APSC) 2020 State of the Service Employee Census (n = 108,085), this study assesses not only the direct relationships between digitalization and well-being but also the mediating role of organizational health and well-being support (shortly organizational support) in these associations.

By focusing on a national public sector context, this study contributes to both the public administration and organizational health literature by showing how digitalization can yield mixed outcomes for employees depending on organizational support and managerial practices. The empirical findings from the generalized structural equation models (GSEMs) with mediation effects, which is an empirical methodology increasingly used in the public administration area (i.e., Moller et al., 2024), demonstrate that digital meetings and digital training negatively affected employees’ health and well-being, whereas digital clearance had a positive impact. Organizational support partially or fully mediates relationships, underscoring the importance of supportive management practices in mitigating the adverse effects of rapid digital adoption. These insights provide both theoretical and policy implications for designing employee-centered digital transformation strategies in public organizations.

The rest of the article is organized as follows: The following section reviews the literature on COVID-19, digitalization, and employee well-being. Section 3 discusses the data and methodology. Section 4 presents the findings, followed by discussion and policy implications. This article ends with conclusions.

2. COVID-19 and Employee Health and Well-Being

Employees’ health and well-being are a multidimensional concept comprising psychological, physical, and social dimensions and are defined as the overall quality of an employee’s experience and functioning at work (Steijn & Giauque, 2021; Grant et al., 2007). Psychological health and well-being focus on the subjective experiences of individuals, and physical health and well-being are related to both objective and subjective aspects of bodily health and include work-related stress and sick leave. Lastly, social well-being is related to interactions that occur between employees, between employees and supervisors, and between employees and their leaders (Steijn & Giauque, 2021; Grant et al., 2007). Based on the job demands–resources (JD-R) model, Bakker et al. (2014) has identified two pathways linking job demands and resources to employee health and well-being: (1) job demands if not matched by adequate resources can lead to health problems like burnout, since they deplete energy, and (2) job resources can also provide motivation and engagement, and thereby contribute to the fulfilment of basic psychological, physical and social needs. As such, external changes in the environment that significantly affect existing structures and processes in public sector organizations are also likely to have an impact on employee health and well-being through changes in the job demands and resources available to them.

COVID-19 created many unexpected changes, including work task changes and team reorganizations for public sector employees, which had adverse effects on their psychological health and well-being (Ervasti et al., 2022; Ingusci et al., 2021). Remote work and increased use of digital tools created a “new normal” for employees in the public sector (Raghavan et al., 2021), thereby altering both job demands and resources. While remote work had a positive impact on the physical and psychological health and well-being of employees in the public sector (Parent-Lamarche & Boulet, 2021), it had a negative impact on employees’ social health and well-being in terms of workplace relationships and work–life balance (Juchnowicz & Kinowska, 2021). Although these effects may vary in scale across different countries, similar patterns have also been observed in other countries, especially in the healthcare sector (Kabasakal et al., 2021). While the flexibility offered by remote work has helped employees by increasing their productivity (Varotsis, 2022; Guler et al., 2021), the pandemic negatively affected the employees’ right to disconnect and work–life balance (Todisco et al., 2023).

Young adults and women in particular were at greater risk of mental health issues in the initial weeks of the pandemic (Ortiz-Lozano et al., 2022; O’Connor et al., 2021). Employees with children and families faced social disruption in the form of “financial insecurity, caregiving burden, and confinement-related stress (e.g., crowding, changes to structure, and routine)” (Ortiz-Lozano et al., 2022; Prime et al., 2020, p. 1). Work from home (WFH) negatively affected the physical and mental health of employees, especially due to factors like decreased physical activity, increased junk food intake, lack of communication with coworkers, having a toddler at home, and higher workload and increased working hours (Ortiz-Lozano et al., 2022; Xiao et al., 2021). Moreover, stressors like employees’ overall perception of safety, quarantine, social exclusion, and job insecurity (Hamouche, 2023), in addition to the psychological distress resulting from social distancing (Doberstein & Charbonneau, 2022; Sibley et al., 2020), may have contributed to long-term mental health effects for employees.

As a result, it is important for organizations to develop human resource policies that take a more “flexible, reflective, holistic, and person-centered approach” (Berry et al., 2024; p. 1) while keeping in mind employees’ individual characteristics like work experience and gender (Ríos Villacorta et al., 2024). Scholars also highlight the need to support employees with dependent family members through professional organizations that work for child and family well-being and mental health (Prime et al., 2020). And lastly, it is essential for organizations to effectively mitigate fears associated with job loss and resulting distress for employees, some of whom may be disproportionately affected due to the pandemic (Bilotta et al., 2021; Hamouche, 2023).

3. Workplace Digitalization, COVID-19, and Employees’ Health and Well-Being

Digitalization refers to a process in which social life and organizations are “restructured around digital communication and media infrastructures” (Kreiss & Brennen, 2016, p. 1). According to Bloomberg (2018), digitalization transforms work processes, increases process efficiency, improves data transparency, and requires employees to learn new digital skills. As a result, many public organizations in central and local government implement digital technologies to serve citizens more effectively and efficiently (Di Loreto et al., 2025). In the public sector, digitalization has been clearly accelerated during the pandemic, largely due to the benefits it offered the sector in dealing with the pandemic (Puddister & Small, 2020). Its benefits led to massive adoption of digital applications at all levels of governance, in developed as well as developing contexts, across public sector functions (Puddister & Small, 2020).

However, some scholars warned that the accelerated massive digitalization adoptions during the pandemic might bring out some negative effects such as digital sclerosis (Andersen et al., 2020) or digital anxiety and training stress (Pfaffinger et al., 2020). For example, Andersen et al. (2020, p. 8) refer to digital sclerosis as “the stiffening of work, usually caused by a replacement of the normal human work with digital work” and explain it with early warnings or symptoms such that the rapid pace of digitalization during the pandemic might significantly reduce the bargaining power or discretion of public employees, or extend the workplace in time and space while seeking responsibility or availability for 24 h during seven days. A case of reduced bargaining power or discretion can be found in German workers exposed to robotization at their workplace (Dauth et al., 2017). Specifically, the extended workplace in time and space can be explained by the fact that many organizations pushed for greater accountability of employees, who were expected to be present beyond the typical 9:00 a.m. to 5:00 p.m. work schedule through smartphones or digital platforms.

Additionally, public employees had to adopt digital tools within very short time frames and use them as part of their work processes during COVID-19 (Pfaffinger et al., 2020). Similarly, employees were also expected to be able to quickly learn and participate in training. Many times, employees were left to their own devices without adequate support or time to apply new technological processes. Therefore, this pandemic situation and the organizational expectations might cause digitalization anxiety to public employees. Moreover, many employees faced personal barriers to adoption, including a lack of technological affinity, inability to quickly adapt to fast-changing processes (including digital modes of communication), and a perceived loss of agency as changes outpace employees’ ability to adapt (Pfaffinger et al., 2020). In particular, Bregenzer and Jimenez (2021) have identified four risk factors of digital work being related to higher stress in the workplace (distributed teamwork, mobile work, constant availability, and inefficient technical support). Thus, due to digital sclerosis, digital anxiety, or higher training stress, some factors of workplace digitalization during the pandemic might weaken public employees’ mental health and well-being or negatively impact the work–life balance (Beckel & Fisher, 2022; O’Connor et al., 2021; Xiao et al., 2021; Prime et al., 2020). Taken together, these studies demonstrate that COVID-19 disrupted employee health and well-being through multiple mechanisms, yet they rarely disentangle how specific work practices or organizational responses systematically shape these outcomes. In particular, most of these studies treat remote or digital work as a broad condition rather than examining concrete practices that may increase demands or provide resources for employees.

During the COVID-19 pandemic, digitalization or digital transformation of the workplace has accelerated new ways of working (NWWs), such as flexibility in working relations, time- and location-independent work, and access to organizational knowledge (Mele et al., 2021; Duque et al., 2020). The job demands–resources (JD-R) framework (Bakker & Demerouti, 2007) suggests that digitalization can simultaneously introduce new job demands (stressors that deplete energy) and job resources (factors that support motivation and well-being). The balance between these determines employee health and well-being outcomes. In this article, we focus on three NWWs, which are digital meeting (DM), digital clearance (DC), and digital training (DT), in public organizations and their relationship with employee well-being and health during the COVID-19 pandemic.

3.1. Digital Meetings

Digital meetings (DMs) exemplify how digitalization can increase job demands. While virtual platforms (Zoom, Teams, Skype) facilitate collaboration and flexibility, they also impose constant connectivity, information overload, and reduced nonverbal cues, which heighten cognitive load and interpersonal strain (Potter et al., 2022; Bennett, 2017). For example, Potter et al. (2022) argue that even though digital communications via digital meetings facilitate flexibility, collaboration, and access to resources for employees, they require employees’ constant accessibility and connectivity, provide information overload, and create interpersonal conflict and issues by less face-to-face interaction. These demands align with JD-R’s “hindrance demands,” which undermine well-being by fostering stress, fatigue, and blurred work–life boundaries (McEwan, 2016; Beyea et al., 2025). Although digital meetings may provide some resources (e.g., influence over work processes; Standaert et al., 2023), the net effect during COVID-19 was likely negative, as demands outweighed resources. In sum, digital meetings bring constant connectivity and information overload, draining employees’ energy through increased screen time and potential cognitive fatigue—classic job demands in the JD-R theory that lead to exhaustion. Thus, the relationship between digital meetings and employees’ health and well-being can be hypothesized as follows:

H1. 

Digital meetings have a negative relationship with employees’ health and well-being.

3.2. Digital Clearance

Digital clearance (DC) refers to streamlining processes of work, such as removing one or more steps in the approval process that could make employees’ jobs and lives easier and less stressful (Puddister & Small, 2020; Dolan et al., 2022). According to Dolan et al. (2022), human resource (HR) management in the digital era, due to the integration of social, mobile, analytics, and cloud (SMAC) technologies, can be innovated or reinvented in how HR operations are provided. So, digital clearance (DC)—such as e-filing, streamlined approvals, and real-time access—represents a job resource within the JD-R framework. By reducing bureaucratic hurdles and administrative workload, digital clearance enhances efficiency, autonomy, and work–life balance (Dolan et al., 2022). These resources foster motivation and buffer against stress, consistent with JD-R’s resource pathway that promotes engagement and well-being (Bakker & Demerouti, 2017). In the public sector, where procedural rigidity often constrains discretion, digital clearance can be particularly valuable in alleviating strain and supporting employee health. In sum, digital clearance cuts red tape and streamlines approvals, giving employees more time and control over their work, so based on JD-R, it can increase motivation and reduce stress. Thus, the relationship between digital clearance and employee well-being is hypothesized as follows:

H2. 

Digital clearance has a positive relationship with employees’ health and well-being.

3.3. Digital Training

Digital training (DT) refers to programs that enable employees’ ability to use digital tools effectively and responsibly in professional contexts (Budai et al., 2023), or structured learning initiatives aimed at enhancing the digital literacy and competencies of public sector employees to support digital transformation efforts (Lopes et al., 2023). According to Zhong et al. (2021), digital (skills) training is one of the HR issues due to the pandemic affecting employees’ health and well-being. In other words, digital training (DT) illustrates how digitalization can create additional job demands. While training aims to build digital competencies (Budai et al., 2023; Lopes et al., 2023), its implementation during COVID-19 often increased workload, required rapid adaptation, and generated digital stress when support was insufficient (Bordi et al., 2018; Salanova et al., 2014). In JD-R terms, inadequate training resources (e.g., time, infrastructure, managerial support) transform training into a hindrance demand, depleting energy and undermining well-being (Day et al., 2010; Parkin et al., 2023). In sum, digital training can increase employees’ workload and reduce autonomy, may create confusion and stress, and thus increase demands that overwhelm rather than empower employees. Thus, digital training may exacerbate anxiety and reduce health outcomes unless accompanied by strong organizational support:

H3. 

Digital training has a negative relationship with employees’ health and well-being.

3.4. Organizational Support as a Mediator

Organizational support for employee health and well-being plays a central role in shaping how digitalization affects outcomes. Within the job demands–resources (JD-R) framework, organizational support can be conceptualized as a resource-providing mechanism that channels the effects of digital work practices into employee well-being (Bakker & Demerouti, 2007). Unlike leadership behaviors, HR practices, or organizational culture—which represent broader contextual or structural features—organizational support refers specifically to the provision of tangible and intangible resources (e.g., guidance, budgets, training opportunities, psychosocial support) that directly influence employees’ ability to cope with new demands (Eisenberger et al., 1986; Rhoades & Eisenberger, 2002).

Treating organizational support as a mediator rather than a moderator is theoretically justified. A moderator would imply that support merely buffers or weakens the relationship between digitalization and well-being (e.g., reducing strain from digital meetings). However, empirical evidence suggests that organizational support is not only a buffer but also a mechanism through which digitalization exerts its effects. For instance, health-promoting leadership and supportive HR practices often translate into organizational support that provides employees with participatory opportunities, technical assistance, and psychosocial resources (Jiménez et al., 2017). In this sense, digitalization initiatives (e.g., implementing digital meetings, clearance systems, or training programs) require organizational support to be effective. Without adequate support, these initiatives may generate stress and anxiety; with support, they can enhance well-being (Bregenzer & Jimenez, 2021; Bregenzer et al., 2019).

Thus, organizational support functions as a mediating pathway: digitalization practices influence the level of support employees perceive, which in turn shapes health and well-being outcomes. This aligns with JD-R’s resource mechanism, where organizational support transforms potentially demanding digital practices into manageable or even beneficial experiences (Bakker & Demerouti, 2017). In sum, organizational support provides clear guidance, time allowances, empowerment, and mental health resources that turn draining digital demands into manageable work based on the insights from JD-R. Thus, the mediating effects of organizational support can be hypothesized as follows:

H4. 

Organizational support mediates the relationships between each digitalization factor (DM, DC, DT) and employees’ health and well-being.

More specifically, the mediation effects can be hypothesized as follows:

H4a. 

Digital meeting (DM) has a positive relationship with organizational support.

H4b. 

Digital clearance (DC) has a positive relationship with organizational support.

H4c. 

Digital training (DT) has a positive relationship with organizational support.

H4d. 

Organizational support has a positive relationship with employees’ health and well-being.

To sum up, these theoretical and empirical models with hypotheses are displayed in Figure 1.

4. Data and Methodology

This section will provide information about the data, followed by the variables used in this study (dependent (or criterion) variable, independent (or predictor) variable, mediator, and control variables), and estimation strategy.

4.1. Data

The data for this study come from the Australian Public Service Commission’s (APSC) 2020 State of the Service Employee Census, which were collected between October and November 2020 (APSC, 2021). This is the only year the APSC asked questions about COVID-19 and its impacts. A total of 108,085 valid responses were received, representing a response rate of about 70 percent of the APS workforce. The survey employed a census approach rather than a random sample, conducted online via a secure platform managed by the Australian Public Service Commission (APSC). Participation was voluntary and confidential, with all employees across APS departments and agencies invited through internal communication channels such as email and intranet. As a result, the dataset provides a comprehensive and representative view of employees’ experiences across the Australian public sector. We handled missing variables as follows: firstly, we deleted respondents who did not answer key questions, such as changes in general health and well-being since COVID-19 (27 February 2020), or questions about actions implemented in the employee’s workgroup during COVID-19, even though response options included different policies along with “other” and “there were no new actions.” Consequently, 5044 observations were removed due to nonresponse. Secondly, 1433 respondents who did not answer questions related to organizational support for health and well-being were also deleted. Similarly, 928 observations were excluded because they did not answer questions regarding job satisfaction, resources, workload, age, or job level. Therefore, the final sample size was reduced to 100,680, indicating that over 93% of the survey data were usable. We found no significant differences in mean values before and after removing these missing data.

4.2. Variables

This study’s main dependent variable (DV) or criterion variable is employees’ health and well-being, which is measured by a questionnaire, “Has there been a change in your general health and well-being since COVID-19 (27 February 2020)?”, whose response is categorized in 5 levels from 1 = very negative change through 3 = no change to 5 = very positive change. Specifically, approximately 3%, 28.8%, 47.1%, 17.1%, and 4% of public employees reported that their health and well-being since the onset of COVID-19 have very negatively changed, negatively changed, not changed, positively changed, and very positively changed, respectively. Additionally, considering the complex model with 5 levels of DV with a mediator, the dependent variable with 5 levels (DV1) is reduced to another dependent variable with 3 levels (DV2), by changing the values from “1. = very negative change” and “2. = negative change” into “1 = negative change”, and from “4. = positive change” and “5.= very positive change” into “3 = positive change”, while from “3. = no change” into “2. = no change”. The descriptive statistics of the DVs and all other variables used in this study are reported in

Table 1. Descriptive statistics.

Variable	Mean	Std. Dev.	Min	Max	VIF
Health and Well-being (DV1)	2.90	0.85	1.00	5.00
Health and Well-being (DV2) *	1.89	0.72	1.00	3.00
Digital Meeting (DM)	0.86	0.34	0.00	1.00	1.07
Digital Clearance (DC)	0.17	0.37	0.00	1.00	1.05
Digital Training (DT)	0.34	0.47	0.00	1.00	1.14
Organizational Support (HC:PCA)	0.00	2.07	−7.34	3.14	1.56
HC1	3.76	0.90	1.00	5.00
HC2	3.77	0.90	1.00	5.00
HC3	3.74	0.93	1.00	5.00
HC4	3.58	1.04	1.00	5.00
HC5	4.15	0.91	1.00	5.00
HC6	3.89	0.90	1.00	5.00
Job Satisfaction	3.82	0.92	1.00	5.00	1.51
Resources	3.59	1.01	1.00	5.00	1.36
Workload	3.68	0.93	1.00	5.00	1.1
Agency Size **					1.02
1_Small	0.04	0.20	0.00	1.00
2_Med	0.09	0.29	0.00	1.00
3_Large	0.87	0.34	0.00	1.00
Female ***	0.62	0.49	0.00	1.00	1.03
Age					1.02
<40	0.38	0.49	0.00	1.00
40–54	0.44	0.50	0.00	1.00
>=55	0.17	0.38	0.00	1.00
Level of Job					1.11
1_APS 1–6	0.67	0.47	0.00	1.00
2_EL 1–2	0.31	0.46	0.00	1.00
3_SES	0.02	0.16	0.00	1.00

Note: N = 100,680. * There is only one (same) dependent variable. However, to simplify the dependent, we reduced the five-level category into a three-level category. ** Large agency (>1000), medium (251~1000), small (<251). *** For gender, N = 97.490.

, and the correlation matrix among the variables is displayed in

Table 2. Correlation matrix.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)
Health and Well-being (DV1) (1)	1
Health and Well-being (DV2) (2)	0.958 *	1
Digital Meeting (DM) (3)	0.018 *	0.012 *	1
Digital Clearance (DC) (4)	0.054 *	0.053 *	0.032 *	1
Digital Training (DT) (5)	0.071 *	0.064 *	0.174 *	0.159 *	1
Organizational Support (HC:PCA) (6)	0.243 *	0.221 *	0.143 *	0.093 *	0.271 *	1
Job Satisfaction (7)	0.222 *	0.203 *	0.113 *	0.079 *	0.166 *	0.529 *	1
Resources (8)	0.164 *	0.154 *	0.048 *	0.095 *	0.135 *	0.431 *	0.427 *	1
Workload (9)	−0.051 *	−0.055 *	0.077 *	−0.052 *	−0.010 *	−0.080 *	0.026 *	−0.170 *	1
Agency Size (10)	0.003	0.003	−0.076 *	0.059 *	−0.069 *	−0.050 *	−0.017 *	0.003	−0.032 *	1
Female (11)	−0.005	−0.010 *	−0.012 *	0.065 *	0.092 *	0.030 *	0.048 *	0.040 *	−0.001	0.004	1
Age (12)	0.017 *	0.020 *	−0.024 *	0.014 *	0.045 *	0.012 *	−0.007 *	−0.027 *	0.036 *	0.038 *	−0.020 *	1
Level of Job (13)	−0.024 *	−0.027 *	0.156 *	−0.080 *	0.022 *	0.060 *	0.095 *	−0.038 *	0.214 *	−0.082 *	−0.104 *	0.083 *	1

* Pairwise correlation significance (p < 0.05).

.

The main independent variables (IVs) or predictor variables are the digitalization of the workplace. In particular, with COVID-19, because most employees were not able to go to their offices as frequently as before, or even in some cases not able to go there for a while, organizations and workgroups started implementing more frequent use of digital platforms for meetings. Additionally, many, if not most, of the organizations have streamlined clearance processes by removing one or more steps in the approval process or moving to a digital-only clearance process. Finally, many organizations have implemented increased access to online well-being platforms and well-being training. Thus, due to the digital transformation of the workplace during the pandemic, the main independent variables (IVs) are three new ways of working (NWWs). As shown in Appendix A, the first NWW, digital meeting (DM), is measured in a binary by a questionnaire, “My work group implemented more frequent use of digital platforms for meetings during COVID-19”. Table 1 indicates that 86% of employees recognized the implementation of digital meetings in the workplace. The other NWWs, digital clearance (DC) or digital training (DT), are also measured in a binary by “My workgroup streamlined clearance processes (e.g., removing one or more steps in the approval process)” or “My workgroup increased access to online wellbeing platforms and wellbeing training during COVID-19,” respectively. Table 1 shows that 17% and 34% of employees answered positively.

The mediating variable between the 3 NWWs or digital factors and employees’ health and well-being is organizational support of health and well-being, which is measured by 6 questions. These questions are asking about employees’ assessments on the organizational supports or efforts to improve employees’ health and well-being: (1) “I am satisfied with the policies/practices in place to help me manage my health and wellbeing”, (2) “My agency does a good job of communicating what it can offer me in terms of health and wellbeing”, (3) “My agency does a good job of promoting health and wellbeing”, (4) “I think my agency cares about my health and wellbeing”, (5) “I believe my immediate supervisor cares about my health and wellbeing”, and (6) “ I am satisfied with my agency’s efforts to maintain a safe environment at work”, which are measured by a 5-point Likert scale from 1 = strongly disagree to 5 = strongly agree. These questions consistently measure the concept of organizational support reliably since the scale reliability coefficient is 0.92 from Cronbach’s alpha, which is reported in

Table A1. Operationalization variables.

Dependent variable	(from 1 = very negative change to 5 = very positive change)
Employee Health & Well-being	Has there been a change in your general health and wellbeing since COVID-19 (27 February 2020)?
Independent variable	(Binary: 1 = Yes, 0 = No)
Digital Meeting (DM)	My work group implemented more frequent use of digital platforms for meetings during COVID-19
Digital Clearance (DC)	My workgroup streamlined clearance processes(e.g., removing one or more steps in the approval process)
Digital Training (DT)	My workgroup increased access to online wellbeing platforms and wellbeing training during COVID-19
Mediator variable	(From 1 = strongly disagree to 5 = strongly agree)
Organizational Support of Health & Well-being	(1) I am satisfied with the policies/practices in place to help me manage my health and wellbeing
	(2) My agency does a good job of communicating what it can offer me in terms of health and wellbeing.
	(3) My agency does a good job of promoting health and wellbeing
	(4) I think my agency cares about my health and wellbeing.
	(5) I believe my immediate supervisor cares about my health and wellbeing.
	(6) I am satisfied with my agency’s efforts to maintain a safe environment at work.
	Scale Reliability coefficient from the Cronbach’s alpha = 0.9159
Control variables
Job Satisfaction	Overall, I am satisfied with my job. (From 1 = strongly disagree to 5 = strongly agree)
Resources	My workgroup has the tools and resources we need to perform well (From 1 = strongly disagree to 5 = strongly agree)
Workload	What best describes your current workload? (1) Below capacity—not enough work. (2) Slightly below capacity—available for more work. (3) At capacity—about the right amount of work to do. (4) Slightly above capacity—lots of work to do. (5) Well above capacity—too much work
Agency Size	By number of Employees: 1 = Small (less than 251 employees), 2 = Medium (251–1000), 3 = Large (Over 1000).
Gender	Respondent’s gender: Male = 0; Female = 1
Age	Respondent’s age: 1= Under 40 years old; 2 = Between 40 and 54 years old; 3 = 55 years or older
Level of Job Classification	1 = APS 1–6; 2 = Executive Level (EL) 1–2; 3 = Senior Executive Service (SES).

. The Scree plot in Appendix A Figure A1 shows that there is only one principal component whose eigenvalue is bigger than 1, implying that there exists only 1 unit root. So, we extracted by principal component analysis (PCA) an index or the one principal component from the 6 questionnaires, and we called the index the construct of organizational support of health and well-being, which is reported in Table 1.

Finally, we control for employee job satisfaction, resources, and workload because these variables might affect employee health and well-being. Employee job satisfaction is known to have a greater impact on the psychological health of employees than physical health (Faragher et al., 2005) and is also known to have an impact on employees’ overall health, happiness, subjective well-being, and self-esteem (Satuf et al., 2018). Owing to the increased workplace demands on employees in the public sector during COVID-19, the importance of resources to fulfill these demands and mitigate the impact of the pandemic was significantly increased (Demerouti & Bakker, 2023; Ha et al., 2024). As such, drawing from the job demands–resources (JD-R) model, Roczniewska et al. (2022) have identified the beneficial role of organizational resources on employee health and well-being and on employee productivity (Ha et al., 2024). Demerouti and Bakker (2023) extended this impact beyond organizational factors to other life domains by integrating crisis management as part of the JD-R model, and identified that organizational strategies in response to crisis events like COVID-19 can impact the resources and workload of employees, and thereby affect their health and well-being.

With respect to the third variable, Ingusci et al. (2021) identified that increased workload during the COVID-19 pandemic led to behavioral stress among employees. This effect was mitigated through access to suitable resources at the organizational level, which likely supported employees who were struggling with remote working conditions and the increased use of digital tools as part of their work. Interestingly, the existing literature has identified that the above three variables tend to interact differently with employee health and well-being. For instance, while employees who have higher job satisfaction tend to be satisfied with their health, those with higher workloads may report less satisfaction with their health (Subramaniam et al., 2021; Horii & Sakurai, 2020). Additionally, because the size of the agency, gender, age, and job level may also affect employees’ health and well-being, we also control these demographic and agency-level variables. According to Table 1, most of the respondents work in large agencies (87%) and in the level of job APS 1–6, (67%). In addition, 62% of respondents are female. Finally, the age distribution is as follows: 38% of the respondents are below 40 years old, 44% are between 40 and 54 age, and 17% are at or over 55 years old. All variables and their descriptive statistics are displayed in Table 1. In addition, the operationalizations of the control variables are reported in Table A1.

4.3. Estimation Strategy

The dependent variable, employee health and well-being, has a 5-point Likert scale. According to Long and Freese (2006) and Wooldridge (2010), when the dependent variable has an ordinal scale with unequal distances, ordinary least squares (OLS) models may cause incorrect interpretations and inconsistency due to the nonlinearity of the variable. Therefore, ordered logit or multinomial logit models are recommended, based on the parallel regression assumption (Long, 1997). Since the Brant test results show that the parallel regression assumption has been violated (p-value < 0.05), multinomial logit models (MNLMs) are preferable. Nevertheless, as this test is sensitive to sample size (>100,000) and small categories (for the dependent variable, “very negative” respondents are less than 3%), these test results should be taken cautiously (Long & Freese, 2006). In addition, since the dependent variable is ordinal and the mediation effects are hypothesized to see whether the relationships between digital factors and employees’ health and well-being are mediated by organizational support, we will apply a generalized structural equations model (GSEM) with a categorical DV. To see whether a variable is a moderator or a mediator, Baron and Kenny (1986)’s 4 steps approach is usually applied. For more details, refer to MacKinnon et al. (2007) or Hayes (2013). A GSEM is an advantageous method for exploring the mediating effect on categorical variables (Bi et al., 2021). Thus, in the following section, we will report the findings from both an ordinal GSEM (

Table 3. Results from ordinal GSEM.

	Health & Well-Being (5 Levels)					Health & Well-Being (3 Levels)
	Coef.	S.E.	P > |Z|	[95% CI]		Coef.	S.E.	P > |Z|	[95% CI]
Health & Well-being (N)
Digital Meeting (DM)	−0.102 ***	0.018	0.000	−0.137	−0.066	−0.107 ***	0.018	0.000	−0.143	−0.072
Digital Clearance (DC)	0.114 ***	0.017	0.000	0.081	0.147	0.120 ***	0.017	0.000	0.088	0.153
Digital Training (DT)	−0.010	0.014	0.462	−0.036	0.017	−0.004	0.014	0.770	−0.031	0.023
Organizational Support (HC:PCA)	0.164 ***	0.004	0.000	0.156	0.171	0.146 ***	0.004	0.000	0.139	0.154
Job Satisfaction	0.284 ***	0.008	0.000	0.268	0.301	0.260 ***	0.008	0.000	0.244	0.277
Resources	0.071 ***	0.007	0.000	0.057	0.085	0.071 ***	0.007	0.000	0.057	0.085
Workload	−0.072 ***	0.007	0.000	−0.086	−0.059	−0.073 ***	0.007	0.000	−0.086	−0.059
Agency Size	0.033 *	0.013	0.010	0.008	0.058	0.029 *	0.013	0.025	0.004	0.055
Female	−0.108 ***	0.013	0.000	−0.132	−0.083	−0.114 ***	0.013	0.000	−0.139	−0.089
Age	0.074 ***	0.008	0.000	0.058	0.091	0.073 ***	0.008	0.000	0.057	0.090
Level of Job	−0.145 ***	0.012	0.000	−0.168	−0.121	−0.146 ***	0.012	0.000	−0.170	−0.123
Organizational Support (HC:PCA) (n)
Digital Meeting (DM)	0.592 ***	0.018	0.000	0.556	0.628	0.592 ***	0.018	0.000	0.556	0.628
Digital Clearance (DC)	0.279 ***	0.017	0.000	0.246	0.313	0.279 ***	0.017	0.000	0.246	0.313
Digital Training (DT)	1.071 ***	0.013	0.000	1.044	1.097	1.071 ***	0.013	0.000	1.044	1.097
_cons	−0.924 ***	0.017	0.000	−0.958	−0.891	−0.924 ***	0.017	0.000	−0.958	−0.891
cut1	−2.707	0.064		−2.832	−2.582	0.054	0.062		−0.067	0.176
cut2	0.174	0.062		0.053	0.294	2.259	0.062		2.136	2.381
cut3	2.384	0.062		2.263	2.506
cut4	4.291	0.064		4.166	4.416
var(e.HC_pca)	3.907	0.017		3.873	3.941	3.907	0.017		3.873	3.941

Note: N = 97,490, n = 100,680; * p < 0.05, *** p < 0.001.

) and a multinomial GSEM (

Table 4. Results from multinominal GSEM.

	Health & Well-Being (3 Levels)
	Coef.	S.E.	P > |Z|	[95% CI]
1. Health & Well-being (N)	(base outcome)
2. Health & Well-being (N)
Digital Meeting (DM)	−0.300 ***	0.023	0.000	−0.346	−0.255
Digital Clearance (DC)	−0.004	0.021	0.843	−0.046	0.038
Digital Training (DT)	−0.231 ***	0.017	0.000	−0.265	−0.197
Organizational Support (HC:PCA)	0.143 ***	0.005	0.000	0.134	0.152
Job Satisfaction	0.262 ***	0.010	0.000	0.242	0.281
Resources	0.079 ***	0.009	0.000	0.062	0.096
Workload	−0.150 ***	0.008	0.000	−0.167	−0.134
Agency Size	0.105 ***	0.016	0.000	0.074	0.137
Female	−0.163 ***	0.016	0.000	−0.194	−0.132
Age	0.271 ***	0.011	0.000	0.250	0.292
Level of Job	−0.081 ***	0.015	0.000	−0.110	−0.051
_cons	−0.514 ***	0.076	0.000	−0.662	−0.365
3. Health & Well-being (N)
Digital Meeting (DM)	−0.097 **	0.030	0.001	−0.156	−0.039
Digital Clearance (DC)	0.193 ***	0.025	0.000	0.144	0.241
Digital Training (DT)	0.033	0.020	0.106	−0.007	0.073
Organizational Support (HC:PCA)	0.204 ***	0.006	0.000	0.193	0.216
Job Satisfaction	0.360 ***	0.013	0.000	0.334	0.386
Resources	0.098 ***	0.011	0.000	0.077	0.120
Workload	−0.074 ***	0.011	0.000	−0.095	−0.053
Agency Size	0.020	0.019	0.291	−0.017	0.058
Female	−0.146 ***	0.019	0.000	−0.184	−0.108
Age	0.065 ***	0.013	0.000	0.039	0.091
Level of Job	−0.232 ***	0.019	0.000	−0.268	−0.195
_cons	−1.590 ***	0.095	0.000	−1.775	−1.405
Organizational Support (HC:PCA) (n)
Digital Meeting (DM)	0.592 ***	0.018	0.000	0.556	0.628
Digital Clearance (DC)	0.279 ***	0.017	0.000	0.246	0.313
Digital Training (DT)	1.071 ***	0.013	0.000	1.044	1.097
_cons	−0.924 ***	0.017	0.000	−0.958	−0.891
Var (e.HC_pca)	3.907	0.017		3.873	3.941

Note: N = 97,490, n = 100,680; ** p < 0.01, *** p < 0.001.

and

Table 5. Direct and indirect effects of multinomial GSEM.

	2. Health & Well-Being		3. Health & Well-Being
	Direct Effect	Indirect Effect+	Direct Effect	Indirect Effect+
Digital Meeting (DM)	−0.300 ***	0.085 ***	−0.097 **	0.121 ***
Digital Clearance (DC)	−0.004	0.040 ***	0.193 ***	0.057 ***
Digital Training (DT)	−0.231 ***	0.153 ***	0.033	0.218 ***

Note: + results from Bootstrap repetition 1000; ** p < 0.01, *** p < 0.001

). In particular, to avoid structural complications and make concise interpretations in a multinomial GSM, we have reduced the DV’s categories by three levels.

We conducted several model specification tests. Firstly, the variance inflation factor (VIF) scores are calculated and reported in Table 1 for the multicollinearity test. The highest VIF score is 1.56 (organizational support from principal component analysis), and the mean VIF score is 1.18, suggesting that multicollinearity is not a problem since all VIFs are less than 10 (Hair et al., 2014). Secondly, common method bias (CMB) may be a concern because all variables were self-reported from the same source (Chang et al., 2010; Podsakoff et al., 2003; Spector et al., 2019). Several procedural remedies were implemented to mitigate this risk, including pretesting the survey, assuring the confidentiality and anonymity of respondents (Podsakoff et al., 2003; Spector et al., 2019; Williams & McGonagle, 2016), and employing different measurement formats for dependent, independent, and control variables (Vogel & Hattke, 2018). In this study, the dependent variable was measured on a five-point ordinal scale (ranging from very negative change to very positive change), whereas the independent variables were binary (yes vs. no), reducing the likelihood of inflated correlations due to common method variance. Additionally, we conducted Harman’s single-factor test, which indicated that a single factor accounted for 46.6% of the variance—below the conventional 50% threshold. While this test provides some evidence against severe CMB, we acknowledge that it is limited by contemporary methodological standards (for example, unmeasured latent method factor (ULMF), Williams & McGonagle, 2016) and does not fully rule out endogeneity or reverse causality (Podsakoff et al., 2012). We therefore emphasize that our findings should be interpreted as associations rather than definitive causal relationships.

5. Findings

To examine the relationships between three digital factors and employees’ health and well-being, and the mediation effect of organizational support on the relationships, we applied the generalized structural equations model (GSEM) since the employees’ health and well-being (DV) is categorical. The GSEM is a generalized SEM that is a multivariate statistical analysis framework that allows the simultaneous estimation of a system of equations when the dependent variable (DV) is ordinal or multivariate (StataCorp, 2025; Cain, 2021) and has been used in analyzing mediation effects in social science studies (Moller et al., 2024; Bi et al., 2021; Maron et al., 2019). The models in this article were fitted by using the maximum likelihood method, assuming ‘logit’ link functions and an ‘ordinal’ or ‘multinomial’ family for the categorical DV or ‘identity’ link function and ‘Gaussian’ family for the continuous DV (StataCorp, 2025; Cain, 2021). The GSEMs were estimated using Stata16.

5.1. Ordinal Generalized Structural Equation Modeling (GSEM)

The ordinal GSEM shows that digital meeting (DM) and digital clearance (DC) have a significant negative and positive effect on employees’ health and well-being (5 levels) at the 5 percent significance level, respectively, holding other variables constant (B = −0.102, P = 0.000; B = 0.114, P = 0.000). However, digital training (DT) has no significant effect on employees’ health and well-being (B = −0.010, P = 0.462). As expected, digital meeting (DM), digital clearance (DC), and digital training (DT) have significant positive effects on organizational support, implying that more digitalization requires more resources and better organizational support (B = 0.592, P = 0.000; B = 0.279, P = 0.000, B = 1.071, P = 0.000, respectively). Also, organizational support has a significantly positive relationship with employees’ perception of health and well-being (B = 0.164, P = 0.000). These results are similar when the DV has three levels, as seen in Table 3.

Regarding control variables, what matters most for employees’ health and well-being is employee job satisfaction (B = 0.284, P = 0.000). Employees who have higher satisfaction with their jobs tend to report highly regarding their health and well-being. Additionally, employees who possess more resources, are older employees, and are in a lower level of job (front-line employees) report that they have higher health and well-being, compared to the employees who have less resources, are younger, and are in a higher level of job (manager groups). Furthermore, employees in larger agencies report better health and well-being, implying that larger agencies have more resources and better management systems. However, employees who have a higher workload report less health and well-being and female employees feel less health and well-being, keeping other variables constant. A reason would be that their workload may increase, particularly for those who have family responsibilities (Ortiz-Lozano et al., 2022). These findings are summarized in Table 3 and shown in Figure 2 and Figure 3.

5.2. Multinomial Generalized Structural Equation Modeling (GSEM)

Since the parallel regression assumption is not satisfied by the Brant test, the multinomial GSEM might be better suited to the data. Considering the complex model with a 5-level DV with a mediator, the DV is reduced to three levels (negative change, no change, and positive change) and named DV2. As shown in

Table A2. Distribution of dependent variables.

DV1 (Original)	Freq.	Percent	DV2 (Collapsed)	Freq.	Percent
1. very negative	2787	2.85	1. negative	30,730	31.44
2. negative	27,943	28.59
3. no change	46,254	47.33	2. no change	46,254	47.33
4. positive	16,851	17.24	3. positive	20,742	21.22
5. very positive	3891	3.98
Total	97,726	100.00	Total	97,726	100.00

, the original five-point dependent variable exhibited highly uneven distributions, with “no change” accounting for nearly half of all responses (47.33%) and “very negative” and “very positive” categories comprising less than 4% each. Given the sparse counts in extreme categories, collapsing into three groups (negative, no change, positive) ensured sufficient cell sizes for estimation in the GSEM framework while retaining substantive interpretability of health outcomes. This practice aligns with methodological recommendations for collapsing underutilized ordinal categories to improve model fit and stability (Quan & Wang, 2025). Nonetheless, concerns that collapsing may reduce some nuance should be acknowledged (Roszkowska, 2025). The findings are reported in Table 4, where the base outcome is 1 (health and well-being [negative change]). First of all, the digital factors’ direct effects on employees’ health and well-being are as follows. Employees who report no change in their health and well-being are more likely than those who are perceived negatively to be adversely affected by digital meetings (DMs) and digital training (DT). These effects are statistically significant at the 5 percent level, controlling for other factors (B = −0.300, P = 0.000 for DM; B = −0.231, P = 0.000 for DT). However, digital clearance (DC) does not affect employees’ health and well-being (B = −0.004, P = 0.843). Meanwhile, employees who report better health and well-being are more likely than those who report worse health and well-being to perceive significantly that digital clearance (DC) improves their health and well-being, ceteris paribus (B = 0.193, P = 0.000). They feel that digital meeting (DM) still affects their health and well-being negatively (B = −0.097, P = 0.000), but digital training (DT) has no significant effect (B = 0.033, P = 0.000). The negative or null effects of digital training (DT) on employees’ health and well-being suggest that its implementation during COVID-19 often increased workload, demanded rapid adaptation, and generated digital stress. This aligns with previous findings indicating that inadequate training resources (e.g., time, infrastructure, managerial support) can transform training into a hindrance demand, depleting energy and undermining well-being (Day et al., 2010; Parkin et al., 2023).

Next, for the mediation effect of organizational support on the relationship between digital factors and employees’ health and well-being, Table 4 also reports the results. The three digital factors, such as digital meetings (DMs), digital clearance (DC), and digital training (DT), have positive relationships with organizational support (B = 0.592, P = 0.000; B = 0.279, P = 0.000; B = 1.071, P = 0.000, respectively). We can see that digital training (DT) is the most demanding factor. Finally, organizational support has a positive relationship with employees’ perception of health and well-being. More specifically, those who report better health changes are more likely than those who report worse health changes to say that organizational support has a positive relationship with their health and well-being. This effect’s size (B = 0.204, P = 0.000) is much bigger than that of the one (B = 0.143, P = 0.000) from those who report no change. For the control variables, the results are similar to those in the ordinal GSEM. Figure 4 shows these results well.

5.3. Post Hoc Tests

To examine the mediating role of organizational support in the relationship between digital factors and employees’ health and well-being, we tested indirect effects using the Bootstrap method with 1000 repetitions, as summarized in

Table A3. Indirect effects from Bootstrap method (repetition = 1000).

IV	DV: Health & Well-Being	Coef.	Bootstrap se	z	P > |z|	[95% CI]
Digital Meeting (DM)	Level 2	0.0848	0.0040	21.29	0.000	0.0770	0.0926
	Level 3	0.1209	0.0054	22.59	0.000	0.1104	0.1313
Digital Clearance (DC)	Level 2	0.0400	0.0028	14.39	0.000	0.0346	0.0455
	Level 3	0.0570	0.0038	14.89	0.000	0.0495	0.0646
Digital Training (DT)	Level 2	0.1533	0.0053	29.02	0.000	0.1429	0.1637
	Level 3	0.2186	0.0068	32.07	0.000	0.2053	0.2320

Note: n = 100,680.

and Table 5. For example, the indirect effect of digital meetings (DMs) on employees’ health and well-being (no change: 2. Health & Well-being) through organizational support is 0.085 (DM × HC: PCA = 0.592 × 0.143 = 0.085), which is statistically significant based on the Bootstrap method. Digital training (DT) shows a similar indirect effect for employees reporting no change, while digital clearance (DC) demonstrates an indirect-only mediation pattern (B = 0.04, P = 0.000), as its direct effect is no longer statistically significant (B = −0.004, P = 0.843). For those reporting better health change (3. Health & Well-being), digital training (DT) also exhibits an indirect-only mediation pathway (B = 0.218, P = 0.000). We caution against the use of terms such as “full mediation” or “perfect mediation,” as these can be sensitive to sample size and statistical power and may oversimplify interpretation. Instead, we adopt more measured terminology (e.g., indirect only mediation or evidence consistent with mediation) to ensure methodological transparency (see Hayes, 2013, pp. 119–121). In sum, organizational support consistently demonstrated statistically significant indirect associations consistent with a mediation pathway between digital factors and employees’ health and well-being.

6. Discussion

This study analyzed the relationships between three new ways of working (NWWs) due to workplace digitalization and employees’ health and well-being and the statistical indirect associations consistent with a mediation pathway (so called, briefly mediation role) of organizational support in their relationships. The results from the multinomial as well as ordinal GSEM on the categorical DV with a mediator showed that digital meeting (DM) and digital training (DT) decreased employees’ health and well-being during the COVID-19 pandemic period, but digital clearance (DC) increased employees’ health and well-being. The negative association between digital meetings (DMs) and employee well-being likely reflects the phenomenon of digital fatigue, whereby the rapid and compulsory shift to video-based communication during COVID-19 intensified screen overload, reduced physical movement, and eroded the boundaries between work and personal life (Beyea et al., 2025; Standaert et al., 2023). Similarly, the adverse effect of digital training (DT) may be attributed to the additional cognitive burden it imposed on employees who were simultaneously adapting to remote working conditions, suggesting that poorly timed or inadequately supported digital upskilling can deplete employee resources. In contrast, the positive association with digital clearance (DC) points to its role as a genuine job resource, reducing bureaucratic friction and enhancing employees’ sense of autonomy and control over their work, consistent with the job demands–resources framework (Bakker et al., 2014).

Furthermore, the organizational support showed statistical indirect associations consistent with a mediation pathway in their relationships. Specifically, each digital factor (DM, DC, and DT) required more organizational support, which in turn increased employees’ health and well-being. Among the mediation effects, digital meeting (DM) has direct and indirect effects on employees’ health and well-being through organizational support. However, digital clearance (DC) and digital training (DT) have no direct effects in the category of DV. In other words, for DC, the organizational support showed an indirect-only mediation effect on employees’ health and well-being in the second category, and for DT, the organizational support showed an indirect-only mediation effect on employees’ health and well-being in the third category (refer to Table 5). These findings advance current knowledge by demonstrating that the adverse effects of workplace digitalization on employee well-being are not inevitable, but can be meaningfully mitigated through deliberate organizational support, thereby extending the JD-R framework to the specific context of rapid digital transformation in the public sector. From a practical standpoint, public sector organizations should prioritize targeted support policies—such as structured guidelines to limit unnecessary digital meetings, adequately resourced and well-timed digital training programs, and streamlined digital clearance processes—that collectively enhance employees’ capacity to adapt to digital work environments while protecting their health and well-being.

These results explain that all three NWWs had a significant effect on employee health and well-being since they were new at the time and required significant adjustments to be made on behalf of employees to adjust to them. Moreover, digital meeting (DM) and digital training (DT) both broke away from employees’ traditional working methods, which included face-to-face communication within a work environment that offered more avenues to connect with their colleagues (Mele et al., 2021). Employees also had to use these tools within the confines of their home as they juggled family responsibilities and a lack of a proper work desk, reliable internet connection, and other resources.

In particular, women employees empirically reported lower health and well-being in our dataset. While the survey does not directly measure caregiving burdens or home working conditions, these patterns are consistent with prior research suggesting that gendered differences in work–family responsibilities, emotional labor, and access to resources during the pandemic may have contributed to intensified work–life conflict and digital fatigue (Ortiz-Lozano et al., 2022; Fauville et al., 2023). We interpret these findings as theoretically informed explanations rather than direct evidence. These gender differences also raise broader concerns about the organizational dynamics of visibility and career progression under remote work conditions. Although the APSC 2020 State of the Service Employee Census does not include items directly measuring the influence of remote work on career advancement or managerial perceptions—particularly for women—it does capture related dimensions such as perceptions of supervisor support, fair access to development, and leadership inclusivity during COVID-19. Prior research suggests that remote work can exacerbate preexisting gender disparities by reducing informal visibility and access to influential networks (Chung & Van der Lippe, 2020). Furthermore, career advancement in many organizations, including the public sector, often depends on being seen and recognized by supervisors and peers (Pfeffer, 1992). Thus, women who combine telework with caregiving responsibilities may be less visible and consequently report lower health and well-being. Future research should build on these findings by conducting intersectional subgroup analyses—examining gender alongside caregiving status and job level—to more fully theorize how digital transformation interacts with organizational support and career trajectories.

Conversely, older employees in Australian public organizations tend to have higher job security and benefits, and they reported higher job satisfaction (Rice et al., 2022). For example, the Australian government “introduced financial and taxation incentives to encourage workers to remain at work beyond 60 years of age” and “older workers [in Australia] may find public services a more attractive and accepting employer” (Colley, 2014, p. 1037). In addition to job security, they may also benefit from organizational resources, the JD-R model (Bakker et al., 2014). Furthermore, while younger employees may have more family responsibilities, such as having younger children, particularly in school age, older employees may have fewer family responsibilities. So, the closing of schools and other institutions due to COVID-19 may have less impact on older people, although they may have less ability to use technology (e.g., cybersecurity) and, at the same time, greater responsibilities at work and in terms of organizational performance.

The results also indicate that public sector organizations in Australia play a vital role in enabling employees to better adapt to the NWW by developing employee-centric policies, providing adequate resources, and supporting employees as they adapt to digitalization as part of their daily work. In the case of digital clearance (DC), organizations actively reduce the complexity of processes, thus saving employees precious time and allowing them to focus on more important tasks. This way, organizational support played a vital role in having a positive effect on employees’ health and well-being. In the case of digital meeting (DM) and digital training (DT), however, the role of organizational support was partial, since employees had to make active attempts to learn and use technology within a very short span of time, despite limited resources, and in a way that replaced their usual ways of working from the office.

Crises like COVID-19 necessitate that public sector organizations think differently when they deploy digitalization. They need to consider the circumstances in which employees at different levels, especially women and younger employees with higher responsibilities, are trying to adapt to NWWs, as they juggle healthcare risks with family responsibilities (Champagne et al., 2023; Ortiz-Lozano et al., 2022). NWWs should be designed and deployed in a way that reduces stress and enhances the productivity of employees, rather than affecting their health and well-being negatively. To enable this, organizations must carefully reevaluate their teleworking and digital transformation policies such that employees’ health, both physical and mental, is at the core of the efforts to implement NWWs (Beckel & Fisher, 2022; Boulet & Parent-Lamarche, 2022). Employees should be treated as co-producers of solutions as organizations strive to develop and/or adopt new digital tools as part of working methods. Regular feedback across departments and levels of the organization should be a key component of efforts to adapt to the changing times of a crisis. An employee-first approach will help leadership to deliver the best possible outcomes for the organization without negatively affecting employee health and well-being.

6.1. Policy Recommendations

Based on this study’s empirical findings and our extensive literature review (both academic and practical pieces), this study offers three policy recommendations. The order of the recommendations does not reflect their importance.

1.

Mental health should be a key part of teleworking policies

Organizational support of health and well-being is a vital factor that can mitigate the mental health risks associated with NWWs. However, it is important to understand the nature of telework that employees are engaged in to best address their concerns. For instance, employees working more than eight hours in telework may have a higher risk of experiencing depressive symptoms (Beckel & Fisher, 2022). As such, organizations should provide sufficient organizational support and mindfulness for such employees to mitigate their emotional exhaustion (Karatepe, 2015; Hülsheger et al., 2013). Johnson et al. (2020) also recommend the use of digital detox initiatives, such as one practiced by Boston Consulting Group for a mandatory smartphone-free night every week. Organizations should thus make mental health a centerpiece of teleworking policies, which, in turn, can have a positive impact on employee well-being and improve their productivity in the long run (Beckel & Fisher, 2022; Johnson et al., 2020). For example, communication can be a key factor that can include employees in decision-making and reduce their sense of uncertainty and fear associated with job loss (Hamouche, 2023). This includes providing employees with the choice to telework versus choosing to work from the office or hybrid work. For instance, Beckel and Fisher (2022) suggest that employees who choose to telework may be better prepared to handle the stress that comes with it.

2.

Telework policies should prioritize work–life balance (WLB)

Work–life balance was greatly affected during COVID-19 and, as a result, is a crucial factor that public sector organizations should take into account. A healthy work environment is especially perceived as important in the post-COVID “new normal” as public sector organizations are moving to hybrid models of work (Gaspar et al., 2024; Raghavan et al., 2021). As a result, organizations should make it a priority to integrate the work–life balance of employees as a key aspect of their telework policies (Ríos Villacorta et al., 2024), and there should be a review of the same at regular intervals (Johnson et al., 2020). Moreover, there is a need to ensure that women employees and employees with disabilities are adequately supported through this transition (Champagne et al., 2023). One way to achieve this is to train them through co-production activities with private or academic institutions, for instance, supporting employees through organizations that provide counseling and support for child and family well-being and mental health (Prime et al., 2020), and through flexible skill development programs that employees can access at their own convenience (Johnson et al., 2020; ILO, 2020). Beckel and Fisher (2022) propose that organizations should delegate telework to employees who would benefit most from it, provide them with the same benefits as regular employees (especially when it comes to family and health support policies) (Johnson et al., 2020) and mandate supervisors to manage employees based on their workload, work hours, and teleworking behaviors (Kim et al., 2021).

3.

A digital-first approach to prepare for future pandemics

Our findings highlight the positive association of digital clearance with employees’ health and well-being, particularly when supported by organizational resources. Building on this evidence, organizations should adopt a digital-first approach by ensuring adequate technological infrastructure, including hardware (computers, phones, headphones, web cameras, broadband connection) and software (digital applications, premium subscriptions), so employees can perform effectively in remote environments (Beckel & Fisher, 2022). To further support well-being, organizations should subsidize ergonomic tools to reduce health risks associated with prolonged remote work (Beckel & Fisher, 2022). Policymakers can complement these efforts by developing accessible neighborhood-based co-working spaces designed to remain safe during health emergencies, thereby promoting equitable digital working conditions across geographic contexts. These spaces would support employees who lack adequate home offices, reduce commuting burdens, and ensure organizational continuity during crises. Furthermore, to avoid the decrease in socializing opportunities in telework, organizations can choose to provide them with as many face-to-face interactions as possible, potentially through online web conferencing platforms like Zoom, Microsoft Teams, etc., and communication applications like Slack and Microsoft Yammer, which can help employees to engage in meaningful interactions with colleagues (Beckel & Fisher, 2022; Johnson et al., 2020). Such tools can enable supervisors to work closely with telework employees to conduct results-based management and trust-building efforts to support them (Kim et al., 2021). Finally, effective leadership is critical to sustaining digital transitions; organizations should invest in digital leadership training for middle managers to strengthen their capacity to support employees’ expectations and needs in telework arrangements (Melo & Demo, 2024; Mourão et al., 2021).

6.2. Limitations and Future Research

This study has some limitations. Firstly, we focus only on the three new ways of working (NWWs), such as digital meeting (DM), digital clearance (DC), and digital training (DT), in public organizations. According to Aroles et al. (2021), however, NWWs can also be related to socio-cultural aspects of respondents, such as flexibility in working conditions and working hours (Mele et al., 2021; Duque et al., 2020) or digital leadership in fostering a more adaptable and digitally competent public sector workforce (Solopi & Qutieshat, 2023). We could not analyze these variables due to data limitations, but future studies, particularly qualitative studies, can analyze other and more in-depth analyses of other NWWs.

Secondly, the dataset lacks variables distinguishing between urban and rural respondents, preventing a systematic analysis of geographic variation. This limitation may obscure important contextual differences, such as while employees working in urban areas (e.g., in Sydney, Melbourne, or Canberra) may have faced greater digital overload and blurred work–life boundaries, employees working in rural areas may have experienced weaker connectivity and limited technical support. Thus, future studies may aim to include geographic identifiers or link survey responses with administrative data to capture such locational dynamics.

Thirdly, we applied the GSEM with cross-sectional data so that we could recognize the short-term effect of digital factors on employees’ health and well-being. To see any long-term effects, longitudinal or panel data can be preferable because a GSEM with a growth model can be applied, which can demonstrate longer-term impacts (Perrot et al., 2024).

Fourthly, we utilized the APS 2020 cross-sectional data to see any mediation effects between the new ways of working and employees’ health and well-being during the COVID-19 pandemic. However, as Kline (2015) criticizes, cross-sectional data have a limitation on arguing mediation effects since it is hard to establish time precedence among the independent variable (IV), mediator (ME), and dependent variable (DV). In other words, we assume that the effect of the IV on the ME occurs before the effect of the ME on the DV without any theoretical rationale or empirical evidence. So, to establish time precedence, empirical mediation models with panel data are needed.

Fifthly, although CMB is less of a concern due to the procedural and statistical remedies, it cannot be eliminated because this study uses self-reported cross-sectional data. Thus, future studies may focus on longitudinal as well as multiple datasets to reduce CMB. Furthermore, while controlling for agency size and job level, due to data limitations, we could not analyze how digital work experiences might vary across specific departments (e.g., IT vs. policy) or job roles, suggesting a need for more granular analyses in different public organizations and roles. As a result, future studies may collect more data on agency types and job roles.

Finally, both the dependent variable and the main independent variables were measured using a single questionnaire due to the survey limitations. Although single-item measures can yield acceptable validity and reliability, particularly under space or time limitations (Milton et al., 2013; Matthews et al., 2022), they necessarily lack the nuance of multi-item assessments and may not fully capture distinct dimensions of psychological, physical, and social well-being. Moreover, because the dependent variable reflects perceived change in general health and well-being since COVID-19, it may be influenced by recall bias or general effects. Future research should therefore employ multi-item measures to provide a more comprehensive assessment of well-being across its different dimensions (Matthews et al., 2022; Song et al., 2023). In addition, the independent variables (digital meetings, clearance, training) were measured as binary indicators, which oversimplifies the intensity, frequency, and quality of digitalization experiences. Future studies should adopt more nuanced measures to better capture variation in employees’ digital practices.

With these limitations acknowledged, this article nonetheless contributes to the literature on digitalization and employees’ health and well-being by offering both theoretical and empirical insights based on APS data. Future research could extend these findings through comparative studies across countries to examine whether the well-being effects of digitalization vary by country and organization. Sectoral comparisons among public, private, and non-profit organizations can be valuable in terms of comparing how digitization impacts employee well-being. Finally, longitudinal analyses with panel data tracking the same employees over time would reveal whether COVID-19 digital practices represent temporary crisis responses or enduring new normals with lasting well-being consequences.

7. Conclusions

We observe in this study that new ways of working (NWWs) like digital meeting (DM), digital clearance (DC), and digital training (DT) in the public sector had a direct impact on employees’ health and well-being in the public sector, mediated by the organizational support of health and well-being. While digital clearance (DC) had a positive impact on employee health and well-being, mediated fully by the organizational support of health and well-being, digital meeting (DM) and digital training (DT) had a negative impact on the health and well-being of employees, partially mediated by organizational support. Additionally, employees who are in better health perceived NWWs as having a positive impact on their health overall. Organizational support plays a pivotal role in developing telework policies in public sector organizations, providing employees with adequate resources to do their work, and also to support them sufficiently through family well-being and mental health programs to mediate the negative effects associated with the increased adoption of telework and digitalization during the COVID-19 pandemic. To prepare for future crises, public sector organizations should reevaluate their telework and digital transformation policies to include mental health, work–life balance and a digital-first approach to implement telework more effectively, such that the health and well-being of employees are protected while enhancing their overall productivity and contribution to the organization.