Fix the Work, Not the Worker: Managing Psychosocial Risk as an Occupational Safety Duty
Download the article here:
Listen to a review of this article:
Abstract: Workplace mental health has long been managed as a matter of individual coping, addressed through employee assistance programs, resilience training, and wellbeing apps. A growing body of evidence and a shift in regulation now point in a different direction. Psychosocial hazards such as excessive workload, low job control, poor recognition, unclear roles, and weak communication are increasingly treated as occupational safety risks that employers must identify, assess, and control at the source. This article reviews the conceptual foundations of psychosocial risk management, the prevalence and consequences of work-related psychological strain, and the evidence on organizational-level interventions. It examines five evidence-based responses: systematic psychosocial risk assessment, workload and working-time redesign, job control and supervisor support, protected recovery and communication norms, and safety-style governance with capable managers. Drawing on examples from public administration, information technology, long-term care, telecommunications, automotive manufacturing, health services, and defence, it concludes with guidance for building durable organizational capability for psychologically healthy work.
On 13 October 2026, the European Agency for Safety and Health at Work launches its Healthy Workplaces Campaign 2026–2028 under the banner "Together for mental health at work," with the European Week for Safety and Health following from 19 to 24 October (European Agency for Safety and Health at Work, 2026). The agency frames the campaign around preventing psychosocial risks, not around encouraging workers to take better care of themselves. That framing matters. It signals that a safety agency, whose institutional history lies in guarding machinery, controlling chemical exposures, and preventing falls, now regards the way work is organized as a hazard in its own right.
The numbers behind the campaign are sobering. In EU-OSHA's OSH Pulse 2025 survey of more than 28,000 workers, 29% reported stress, depression, or anxiety caused or made worse by their work in the previous 12 months, and 44% said they were exposed to severe time pressure or work overload (Belli & De Keulenaer, 2025). Globally, the World Health Organization estimates that around 12 billion working days are lost every year to depression and anxiety, at a cost of roughly US$1 trillion in lost productivity (World Health Organization, 2026).
Most employers already do something about mental health. They offer counselling lines, mindfulness sessions, and awareness weeks. The trouble is that these offerings tend to sit downstream of the problem. They help people cope with conditions that the organization continues to produce. A large cross-sectional study of 46,336 workers in 233 UK organizations found that participants in common individual-level interventions, such as resilience training, mindfulness, and wellbeing apps, appeared no better off on several wellbeing measures than colleagues who did not participate (Fleming, 2024). That finding does not prove such programs are useless for everyone, but it should unsettle any leadership team that treats them as its mental health strategy.
Regulators have reached a similar conclusion through a different route. In Victoria, Australia, the Occupational Health and Safety (Psychological Health) Regulations 2025 came into force on 1 December 2025, requiring employers to identify and control psychosocial hazards in the same way they would physical ones (Premier of Victoria, 2025). New South Wales amended its work health and safety regulations on the same theme in October 2022 (NSW Resources Regulator, 2022). In Britain, the Health and Safety Executive has long told employers that they have a legal duty to assess the risk of work-related stress and act on the findings (Health and Safety Executive, n.d.). And in late 2025 an Australian court convicted the Department of Defence after a technician's death, in what Comcare described as the first penalty of its kind for a Commonwealth employer (Comcare, 2025).
This article makes the case that leaders should treat mental health at work as an occupational safety issue first, and an individual wellness issue second. It sets out what the evidence says about psychosocial hazards, why they matter for performance as well as health, and what organizations can do about them. The central argument is simple to state and harder to practise: if the work is making people unwell, fix the work.
The Psychosocial Risk Landscape
Defining Psychosocial Hazards and Risk Management
A psychosocial hazard is a feature of how work is designed, organized, and managed, or of the social relationships at work, that has the potential to cause psychological or physical harm. Victoria's new regulations define these hazards as factors arising from work design, systems of work, the management of work, the carrying out of work, or workplace interactions that may cause a negative psychological response creating a risk to health or safety (Norton Rose Fulbright, 2025). Psychosocial risk is the likelihood and severity of harm that follows from exposure to such hazards. Psychosocial risk management is the systematic process of identifying those hazards, assessing the risk they pose, controlling them, and reviewing whether controls are working.
This vocabulary deliberately borrows from conventional safety practice. The international standard ISO 45003:2021 provides guidance for managing psychosocial risk within an occupational health and safety management system built on ISO 45001, and it applies to organizations of any size or sector (International Organization for Standardization, 2021). The UK HSE Management Standards organize the problem around six areas of work design: demands, control, support, relationships, role, and change (Health and Safety Executive, n.d.). The WHO guidelines on mental health at work, released with a companion WHO and International Labour Organization policy brief in September 2022, put organizational interventions that address risks such as heavy workloads and negative behaviours at the front of their recommendations (World Health Organization, 2022).
Three well-established theories explain why these working conditions matter. Karasek's (1979) demand-control model holds that strain arises when high job demands are combined with low decision latitude, the worker's ability to influence how and when the work gets done. Siegrist's (1996) effort-reward imbalance model argues that strain follows when high effort is met with low reward, where reward includes pay, esteem, and career security as well as recognition. The job demands-resources (JD-R) model integrates and extends these ideas, proposing that every job has demands that drain energy and resources, such as autonomy, support, and feedback, that buffer those demands and foster engagement (Bakker & Demerouti, 2007). When demands chronically outrun resources, a health-impairment process unfolds that leads toward exhaustion.
That process has a name. WHO's ICD-11 describes burnout as a syndrome resulting from chronic workplace stress that has not been successfully managed, characterized by exhaustion, mental distance or cynicism toward one's job, and reduced professional efficacy (World Health Organization, 2019). WHO is explicit that burnout is an occupational phenomenon rather than a medical condition, and that it refers specifically to the work context. Maslach and Leiter (2016) have long argued that burnout reflects a mismatch between people and their work environment across areas such as workload, control, reward, community, fairness, and values. The definition carries a quiet but important message for leaders: burnout is described as a failure of stress management at the workplace, not a failure of personal character.
Finally, psychosocial safety climate refers to shared perceptions about the policies, practices, and procedures an organization uses to protect workers' psychological health. Dollard and Bakker (2010) positioned it as an upstream factor that shapes the demands and resources people encounter. In other words, the way senior leaders prioritize psychological health tends to show up later in workloads, support, and wellbeing.
Prevalence, Drivers, and Distribution
The OSH Pulse 2025 survey provides the most current cross-national picture. Ipsos interviewed 28,220 workers in the 27 EU member states, Iceland, and Norway between 31 March and 14 April 2025, with a separate Swiss sample (Belli & De Keulenaer, 2025). Figure 1 summarizes how often EU workers report exposure to six psychosocial risk factors.
Severe time pressure or work overload tops the list at 44%. About one in three workers (34%) report a lack of rewards, including pay, career opportunities, and recognition, for the effort they put into their jobs, and 29% report poor communication or cooperation within their organization, up three percentage points from 26% in 2022. Smaller but still substantial shares report a lack of autonomy or influence over work pace and processes (17%), violence or verbal abuse from customers, patients, or pupils (16%), and harassment or bullying (8%) (Belli & De Keulenaer, 2025). EU-OSHA's public summary of these results led with the 29% of workers reporting work-related stress, depression, or anxiety and the one in three who feel their efforts go unrewarded (European Agency for Safety and Health at Work, 2025).
The distribution is uneven. More than half of workers in Greece (57%), Luxembourg (54%), France and the Netherlands (both 52%), and Finland (51%) reported severe time pressure, and poor communication was reported by 42% of Finnish and 40% of Swiss workers (Belli & De Keulenaer, 2025). Work-related stress, depression, or anxiety was more common among workers aged 25 to 39 (32%) than among those aged 16 to 24 (23%) and was higher among more highly educated workers (Belli & De Keulenaer, 2025). Fatigue was reported by 40% of workers in education and in health and social care.
Other sources point the same way. In Great Britain, an estimated 964,000 workers suffered from work-related stress, depression, or anxiety in 2024/25, accounting for 52% of all work-related ill health and 62% of working days lost to work-related ill health; HSE's labour force data identify workload pressures and lack of managerial support as the main causes (Health and Safety Executive, 2025). Safe Work Australia (2025) reports that mental health conditions made up 12% of serious workers' compensation claims in 2023–24, up 161% over a decade, with median time lost almost five times longer than for other injuries and diseases.
Employer and consultancy surveys, which use different samples and definitions, offer corroborating texture. DHR Global's (2025) survey of 1,500 knowledge workers across North America, Europe, and Asia found that 83% reported at least some degree of burnout, with overwhelming workloads (48%) and long hours (40%) the leading drivers, and lack of reward or recognition cited by 32%. The 2026 NAMI-Ipsos poll of 2,153 US full-time employees found that 53% felt burned out because of their job in the past year and that 38% felt their mental health suffered because of work demands (National Alliance on Mental Illness, 2026). Gallup's (2026) global data for 2025 show that 40% of employees experienced a lot of stress during the previous day, with managers reporting more daily stress (45%) than individual contributors (39%).
The pattern across these sources is consistent. The most commonly reported drivers are not exotic. They are too much work, too little time, too little say, and too little recognition, delivered through communication that leaves people guessing. These are design and management problems, which means they are, at least in principle, controllable.
Organizational and Individual Consequences
Organizational Performance Impacts
Psychosocial risk is expensive, and much of the expense is hidden. The WHO estimate of 12 billion lost working days and about US$1 trillion in lost productivity each year captures only depression and anxiety (World Health Organization, 2026). At the national level, Deloitte (2024) estimated that poor mental health among employees costs UK employers £51 billion a year, with presenteeism, working while unwell at reduced capacity, the largest single component at around £24 billion. Deloitte's review of published studies also suggested an average return of about £4.70 for each £1 spent on employee mental health and wellbeing, though such return-on-investment figures depend heavily on the studies included and should be read as indicative rather than precise.
Absence data tell a similar story. Because work-related stress, depression, or anxiety accounts for most working days lost to occupational ill health in Britain, at an average of 22.9 days per case in 2024/25 (Health and Safety Executive, 2025), even modest reductions in incidence translate into recovered capacity. In Australia, the much longer duration of mental health claims compared with physical injuries makes them disproportionately costly for insurers and employers alike (Safe Work Australia, 2025). Victoria's government noted that only 42% of workers with a mental injury returned to work within six months, compared with 75% of those with physical injuries (Premier of Victoria, 2025).
The costs extend beyond absence. DHR Global (2025) reported that 52% of surveyed workers said burnout dragged down their engagement, and that overall engagement in its sample fell sharply year on year. Gallup (2026) found that only 20% of employees worldwide were engaged in 2025. While engagement and burnout are distinct constructs, the JD-R model predicts that the same resource deficits that drive exhaustion also suppress motivation, so organizations that allow demands to outrun resources tend to pay twice: once in ill health and once in discretionary effort (Bakker & Demerouti, 2007).
There is also legal and reputational exposure. Court Services Victoria was convicted and fined A$379,157 in 2023 after admitting it had failed to identify and assess risks to employees' psychological health at the Coroners Court (WorkSafe Victoria, 2023). The Department of Defence was convicted and fined A$188,000 in December 2025, with an adverse publicity order (Comcare, 2025). Fines of this size are small relative to large organizational budgets, but the precedents matter. They show that regulators will prosecute psychosocial failures and that courts will treat poor work design and management as safety breaches.
Individual Wellbeing and Stakeholder Impacts
For individuals, the consequences range from fatigue to serious illness. In OSH Pulse 2025, 37% of EU workers reported overall fatigue caused or made worse by work, and 29% reported stress, depression, or anxiety, up from 27% in 2022 (Belli & De Keulenaer, 2025). A major Lancet review concluded that exposure to certain adverse working conditions is associated with a higher risk of depressive disorders, while noting that causal inference remains debated and requires better exposure measurement and theory (Rugulies et al., 2023). The prudent reading for practitioners is that the association is consistent enough to act on, especially given that the hazards involved are under managerial control.
Stigma compounds the harm. Workers in the EU are evenly split on whether disclosing a mental health condition would damage their careers, with 48% agreeing and 48% disagreeing, although 58% say they would feel comfortable speaking to their manager about their mental health (Belli & De Keulenaer, 2025). In the United States, employees in the NAMI-Ipsos poll were more comfortable discussing mental health with a friend at work (83%) than with a manager (57%), HR (39%), or senior leadership (30%) (National Alliance on Mental Illness, 2026). When people fear the consequences of speaking up, problems surface later and in more severe form.
The most painful consequences are the ones regulators now investigate. WorkSafe Victoria's account of the Coroners Court case describes exposure to traumatic material, high workloads, role conflict, poor workplace relationships, and inappropriate behaviours over almost three years, during which several workers took leave for anxiety and stress and a senior solicitor died by suicide after being diagnosed with work-related major depressive disorder (WorkSafe Victoria, 2023). The Defence case involved a 34-year-old RAAF technician who died after a series of performance-related work plans that supervisors had not been trained to recognize as a psychosocial hazard (Comcare, 2025). These cases are not typical outcomes of stressful work, and they should not be presented as such. But they illustrate why safety regulators insist that psychological harm be managed with the same seriousness as physical harm.
Effects also spill into families and communities. Caregivers in the NAMI-Ipsos poll reported higher burnout than non-caregivers (61% versus 49%) (National Alliance on Mental Illness, 2026), and Deloitte (2024) estimated that working parents' concerns about their children's mental health cost UK employers around £8 billion a year. The boundary between work strain and home strain is porous, which is one reason that interventions targeting schedule control and recovery time have shown promise.
Evidence-Based Organizational Responses
The research base on workplace mental health interventions has a well-documented imbalance. Interventions disproportionately target individual workers and illness rather than working conditions (Rugulies et al., 2023), even though systematic reviews have long found that organizationally focused approaches benefit both individuals and organizations, while individually focused approaches tend to help individuals without showing organizational-level benefit (LaMontagne et al., 2007). Table 1 summarizes five responses that rebalance the portfolio toward prevention at the source, ordered roughly from the system that finds hazards to the governance that sustains control.
Table 1: Evidence-based organizational responses to psychosocial risk
Response | Evidence Base | Key Practices | Illustrative Examples |
Systematic psychosocial risk assessment | HSE Management Standards; ISO 45003; reviews of organizational interventions (LaMontagne et al., 2007; Montano et al., 2014); Stress Check evaluations | Group-level hazard surveys, worker consultation, action plans, review cycles | Japan's Stress Check Program; 40 organizations implementing Canada's National Standard |
Workload and working-time redesign | Demand-control and JD-R theory; national and multi-employer working-time trials | Staffing to demand, removing low-value work, reduced-hours trials with process redesign | Reykjavík City Council and Icelandic government trials; UK four-day week pilot |
Job control and supervisor support | Randomized STAR trials (Moen et al., 2016; Kossek et al., 2019) | Schedule control, results-focused work, supervisor support training | STAR in a large US IT firm; STAR in 30 nursing facilities |
Protected recovery and clear communication | Effort-reward and recovery logic; collective bargaining and statutory disconnection rights | Right-to-disconnect agreements, after-hours email limits, communication norms | Orange (France); Daimler; Volkswagen |
Safety-style governance and manager capability | WHO (2022) guidelines; psychosocial safety climate research; enforcement cases | Board-level accountability, manager training, safe performance processes, structured peer support | Court Services Victoria and Department of Defence enforcement; NHS Schwartz Rounds |
The hierarchy of controls, familiar from physical safety, offers a useful way to see how these responses fit together. Figure 2 adapts it to psychosocial hazards. Primary prevention eliminates or reduces hazards at the source through work design. Secondary measures help workers and managers detect and respond to strain early. Tertiary measures support recovery and return to work once harm has occurred. All three levels have a place, but the regulatory logic in Victoria and New South Wales, and the evidence summarized here, place the burden of proof on primary controls.
Systematic Psychosocial Risk Assessment
The foundation of any safety system is a reliable way to find hazards. For psychosocial risk, that typically means collecting group-level data on working conditions, interpreting it with the people who do the work, and turning it into specific changes. The HSE Management Standards approach structures this around its six work-design areas and reminds employers that they have a legal duty to assess and act on stress risks, with written assessments required where five or more people are employed (Health and Safety Executive, n.d.). ISO 45003 places the same activity inside a management system with planning, implementation, and review cycles (International Organization for Standardization, 2021).
The evidence on organizational-level interventions that follow from such assessments is encouraging but mixed. Montano et al. (2014) reviewed 39 organizational-level intervention studies and found that about half reported significant health effects, with success rates appearing higher for more comprehensive interventions that addressed material, organizational, and working-time conditions together, though that comparison did not reach statistical significance. Nielsen and Randall (2013) argued that much of this variability reflects implementation rather than intervention theory, and that evaluations must examine how and why changes succeed or fail, including whether managers support them and whether employees actually experience the intended change. Risk assessment, in other words, is only as good as the action and follow-through it produces.
Effective approaches:
Measure conditions, not just symptoms: Survey the demands, control, support, relationships, role clarity, and change management that drive strain, rather than relying solely on wellbeing scores.
Analyze at team level: Aggregate data to work units large enough to protect confidentiality but small enough to reveal specific hazards and accountable managers.
Consult the people exposed: Use structured worker participation to interpret findings and design controls, consistent with the consultation duties built into safety law.
Prioritize by risk: Rank hazards by severity, frequency, and duration of exposure, as NSW guidance requires employers to consider (NSW Resources Regulator, 2022).
Close the loop: Set dates, owners, and review points for each control and re-measure to confirm change.
Japan offers the largest natural experiment in mandated psychosocial monitoring. Since December 2015, an amendment to the Industrial Safety and Health Law has required workplaces with 50 or more employees to run a Stress Check at least once a year (the 57-item Brief Job Stress Questionnaire is recommended but not mandatory), to return results confidentially to each worker, to offer physician interviews to highly stressed workers, and to analyze results for groups to improve the psychosocial work environment (Kawakami & Tsutsumi, 2016). The program's architects were candid that evidence for the individual screening and feedback components was weak and that the work environment component was the more promising, though at launch it lacked detailed procedures.
A subsequent one-year cohort study of 2,492 workers supported that judgement. Workers who both completed the stress survey and experienced work environment improvements showed significantly greater reductions in psychological distress than those who had neither, but the effect was small, and the overall group-by-time interaction was not significant (Imamura et al., 2018). Only 65 of the workers in the study fell in that combined group, a hint, though not proof, that survey results often did not translate into workplace change. Japan is now extending the obligation to smaller workplaces, with implementation expected by around 2028 (Vanguard Lawyers Tokyo, 2026). The lesson for employers elsewhere is that measurement mandates do little on their own. The value lies in what happens after the data are analyzed.
The Mental Health Commission of Canada's three-year Case Study Research Project, which followed 40 organizations implementing Canada's National Standard for psychological health and safety, including Bell Canada, Rogers Communications, Via Rail, several hospitals, and provincial governments, reinforces the point (Mental Health Commission of Canada, 2017). Participating organizations raised their average compliance with the Standard's core elements from 55% to 72%. Yet the most commonly reported actions were respectful workplace policies and education (78%), employee and family assistance programs (70%), and mental health awareness (66%). The nine promising practices the researchers distilled included considering psychological health and safety during organizational change and measuring results regularly, both of which point back toward work design.
A common pitfall is to treat the survey as the intervention. Organizations launch an annual pulse, publish an engagement score, and then wait for next year's numbers. Without a documented link between each identified hazard and a specific control, risk assessment becomes a ritual that raises expectations and then deflates them, which can itself damage trust.
Workload and Working-Time Redesign
If the single most commonly reported psychosocial hazard is too much work for the time available, then the most direct control is to change the ratio. That can mean staffing to actual demand, removing low-value tasks, sequencing priorities more honestly, or reducing working hours while redesigning processes so that output is maintained. The demand-control and JD-R models predict that lowering chronic demands, or raising the resources available to meet them, should reduce strain (Bakker & Demerouti, 2007; Karasek, 1979).
Working-time experiments provide some of the most visible evidence, with the important caveat that most are not randomized and rely on self-reported outcomes from volunteering organizations. They are best read as demonstrations that workload can be reengineered without loss of service, rather than as proof of causal health effects.
Effective approaches:
Audit demand against capacity: Compare workload, staffing, and deadlines at team level and treat persistent overload as a hazard requiring control, not a performance issue.
Remove before adding: Identify meetings, reports, approvals, and duplicated processes that can be stopped before asking anyone to work faster.
Redesign, then reduce: Where hours are cut, pair the change with process redesign so the reduction is sustainable rather than compressed.
Pilot with measurement: Run time-limited trials with clear baseline and follow-up measures of strain, service, and output.
Protect high-risk roles: Pay special attention to roles with emotional demands or exposure to aggression, where workload and emotional load interact.
Reykjavík City Council and Iceland's national government ran two large trials of a 35- to 36-hour week without pay cuts between 2015 and 2019, covering about 2,500 workers, or over 1% of Iceland's working population (Haraldsson & Kellam, 2021). According to the report by Autonomy and the Icelandic association Alda, worker wellbeing improved across several indicators, including perceived stress and burnout, while service provision and productivity remained the same or improved in most workplaces, and the trials were revenue neutral for the public bodies involved. Subsequent collective agreements mean that 86% of Iceland's workforce now either works shorter hours or has the right to do so. Critically, the trials were not simply about cutting hours. The report describes workplaces shortening or replacing meetings, cutting out unnecessary tasks, and reorganizing shift arrangements (Haraldsson & Kellam, 2021).
In the private sector, the UK four-day week pilot ran from June to December 2022 across 61 organizations and about 2,900 employees, who reduced working hours by 20% with no reduction in pay (University of Cambridge, 2023). Researchers from the University of Cambridge, Boston College, and the think tank Autonomy reported that 71% of employees had lower levels of burnout and 39% were less stressed than at the start of the trial, while sick days fell by 65% and staff departures by 57% compared with the same period of the previous year. Of the 61 companies, 56 intended to continue, with 18 making the change permanent. The participating firms volunteered, so selection effects are likely, but the scale of the reported change in burnout is difficult to dismiss.
A common pitfall is to compress rather than redesign. Offering a shorter week while expecting the same volume of meetings, messages, and deliverables simply raises intensity, which can worsen the hazard it was meant to control. Workload controls succeed when the work itself changes.
Job Control and Supervisor Support
Two of the most robust predictions in occupational health psychology are that greater control over how and when work is done buffers high demands, and that supportive supervisors reduce strain (Bakker & Demerouti, 2007; Karasek, 1979). The Work, Family, and Health Network tested these ideas with unusual rigour through STAR (Support. Transform. Achieve. Results.), an organizational intervention combining training for all employees, additional training for supervisors, and practical changes that increased employees' control over when and where they worked and encouraged supervisors to show support for workers' personal and family lives.
The STAR trials matter because they were group-randomized field experiments, a rare design in organizational research. They also produced honest, nuanced results that practitioners can learn from, including null findings.
Effective approaches:
Shift the default to control: Give teams discretion over scheduling, location, and methods wherever operational constraints allow, rather than treating flexibility as an individual accommodation.
Train supervisors in support: Teach managers to recognize and respond to non-work demands and signs of strain, and to discuss workload openly.
Focus on results, not visibility: Replace presence-based norms with clear expectations about outputs and priorities.
Redesign together: Use facilitated team sessions to identify practices that create unnecessary strain and agree on alternatives.
Watch context: Anticipate that restructuring or uncertainty can swamp the benefits of control-enhancing changes.
At a large US information technology company, 56 work groups were randomized to STAR or usual practice (Moen et al., 2016). Among the 455 employees and managers in the main analysis, STAR reduced burnout (effect size 0.36) and increased job satisfaction (0.27), with smaller reductions in perceived stress and psychological distress (both 0.18). Benefits for psychological distress were concentrated among women. Notably, STAR had no effects for a later group of employees who began after a merger was announced, which suggests that organizational upheaval can overwhelm even well-designed interventions.
In long-term care, the network adapted STAR for 30 nursing facilities, randomizing 15 to the intervention and 15 to control, with 420 and 511 employees respectively measured over 18 months (Kossek et al., 2019). The intervention did not produce overall effects on stress or psychological distress. It was, however, more effective for employees who also cared for family members outside work, the so-called double-duty and triple-duty caregivers. This is an important corrective to overconfident claims. Increasing control is harder in shift-based, low-wage care work where staffing ratios constrain flexibility, and benefits may concentrate among workers with the greatest competing demands.
A common pitfall is to grant flexibility on paper while leaving the underlying workload untouched. Control over when work happens offers little relief if the volume of work fills every available hour, and supervisors who have not been equipped to manage by results often revert to monitoring presence.
Protected Recovery and Clear Communication
Recovery is the process through which people restore the resources that work depletes. When work follows people home through constant connectivity, recovery is interrupted, and the effort-reward balance tips further toward effort (Siegrist, 1996). At the same time, OSH Pulse 2025 shows poor communication or cooperation rising as a reported hazard, to 29% of EU workers (Belli & De Keulenaer, 2025). Communication practices therefore sit at the intersection of two hazards: they can create overload and intrusion, and when they fail they can create ambiguity and conflict.
France moved early to address the first problem. From 1 January 2017, French employers with recognized union representatives for mandatory annual negotiations have been required to negotiate on employees' right to disconnect from digital devices and the processes regulating their use, and to adopt an internal policy if no agreement is reached (Rivez-Domont, 2017). The legal duty is procedural, which means the substance depends on what organizations actually agree.
Effective approaches:
Set explicit norms: Agree when messages may be sent and when responses are expected, with clear exceptions for genuine emergencies.
Use technology as a control: Delay or hold messages outside working hours rather than relying on individual restraint.
Protect leave: Make it possible for people to take holidays without returning to an overflowing inbox.
Clarify channels and priorities: Reduce ambiguity by stating which channels carry which kinds of request and who sets priorities when demands conflict.
Negotiate with workers: Develop rules jointly with employee representatives so they reflect real work patterns.
Orange, the French telecommunications group, signed an agreement on digital transformation in September 2016 with three unions representing 58.5% of its workforce (Mind RH, 2016). The 122-page agreement, covering 96,000 employees in France, recognized a right to disconnect alongside provisions on digital training and the use of employees' personal data, following a year of negotiation and the rejection of an earlier draft. The agreement's commitments on disconnection were framed as broad principles, which illustrates both the value of collective agreements in legitimizing boundaries and the risk that principles alone change little without operational follow-through.
German automakers adopted more mechanical controls. Volkswagen restricted its email servers so that employees covered by union-negotiated contracts in Germany stopped receiving messages on company devices 30 minutes after their shift ended and resumed 30 minutes before the next one began, with senior managers excluded (Siu, 2011). Daimler went further for holidays, offering its German employees a "Mail on Holiday" option that automatically deleted incoming email during vacation and told senders the message had not been received, after a trial the previous year (Silverberg, 2014). Both measures treated after-hours connectivity as a design problem rather than a matter of individual discipline, which is exactly the logic of the hierarchy of controls. Neither has, to our knowledge, been subject to a published evaluation of mental health outcomes, so their value lies mainly as worked examples of engineering controls applied to communication.
A common pitfall is to ban after-hours email while leaving the workload that generates it unchanged. People then work late offline, or front-load messages that land the next morning. Communication rules are most effective when they accompany realistic deadlines and staffing.
Safety-Style Governance and Manager Capability
The final response connects the others. Psychosocial risk will be managed seriously only when it is governed like other safety risks: with senior accountability, defined duties, competent people, and assurance that controls are working. WHO's 2022 guidelines recommended, for the first time, that managers be trained to prevent stressful work environments and respond to workers in distress, alongside organizational interventions that address risks such as heavy workloads and negative behaviours (World Health Organization, 2022). Psychosocial safety climate research suggests that the visible priority leaders give to psychological health shapes the working conditions that follow (Dollard & Bakker, 2010).
Recognition belongs here too. A third of EU workers report insufficient reward for their effort (Belli & De Keulenaer, 2025), and lack of reward or recognition was cited as a burnout driver by 32% of DHR Global's (2025) respondents. Effort-reward imbalance theory implies that recognition is not a soft extra but a protective resource whose absence creates risk (Siegrist, 1996). Fair performance processes, transparent promotion criteria, and routine acknowledgement of effort are therefore controls, not perks.
Effective approaches:
Assign senior accountability: Put psychosocial risk on the same governance agenda as physical safety, with named executive ownership and regular reporting.
Train managers as risk controllers: Equip managers to recognize psychosocial hazards in workload allocation, change, conflict, and performance management.
Make performance processes safe: Treat performance improvement plans, investigations, and restructurings as high-risk activities that require trained handling and welfare checks.
Build recognition into routines: Ensure effort is acknowledged and rewards are distributed fairly and transparently.
Provide structured peer reflection: Offer forums where staff in emotionally demanding roles can process the human side of their work.
Court Services Victoria's conviction illustrates what happens when governance is absent. Between December 2015 and September 2018, workers at the Coroners Court were exposed to traumatic material, high workloads, role conflict, and poor workplace relationships, and complaints of bullying and intimidation went unaddressed (WorkSafe Victoria, 2023). The organization admitted it had not adequately identified or assessed risks to employees' psychological health. WorkSafe Victoria's executive director of health and safety emphasized that a positive culture and appropriate controls depend on leadership from the top.
The Department of Defence case shows how an ordinary management process can become a hazard. Comcare (2025) found that supervisors had not been trained on a draft work plan procedure being used as a performance management tool, nor to recognize such plans as a psychosocial hazard, and that they did not refer a visibly distressed technician for support, place him on leave, or pause the process. The conviction under federal work health and safety law was the first of its kind for a Commonwealth employer. For HR leaders, the implication is direct: performance management is a work process with risk, and managers who run it need training and safeguards.
On the supportive side, NHS organizations in England have adopted Schwartz Center Rounds, regular forums in which staff from all disciplines discuss the emotional and social aspects of care. A national mixed-methods evaluation found that poor psychological wellbeing among attenders fell from 25% to 12%, compared with a fall from 37% to 34% among non-attenders, although the authors noted reliance on self-report and fewer regular attenders than intended (Maben et al., 2018). Rounds are a secondary rather than primary control. They do not reduce workload or staffing pressure, but they do strengthen the social support resource that the JD-R model identifies as protective.
A common pitfall is to delegate psychosocial risk entirely to HR or the wellbeing team. When accountability sits outside the line, managers continue to allocate work, run restructurings, and conduct performance processes without regard to psychological risk, and wellbeing staff are left to treat the results.
Building Long-Term Psychosocial Safety Capability
Integrating Psychosocial Risk into Safety Management Systems
The first pillar of durable capability is integration. Psychosocial hazards should appear in the same risk registers, audits, incident investigations, and board reports as physical hazards. ISO 45003 provides a template for doing this within an existing ISO 45001 system (International Organization for Standardization, 2021), and Victoria's regulations now make the integration mandatory for employers in that state (Premier of Victoria, 2025). Practically, integration means that a change proposal, such as a restructuring, a new technology, or a staffing reduction, triggers a psychosocial risk assessment just as a new machine would trigger a physical one.
Integration also means investigating psychological injuries with the same rigour as physical ones. When a worker's claim for a mental injury is accepted, the question should not stop at how the person is supported back to work. It should extend to what about the work contributed and whether others remain exposed. Japan's experience shows that collecting data without systematic action delivers little (Imamura et al., 2018). The system needs to link findings to controls and controls to review.
Finally, integration requires leading as well as lagging indicators. Claims and absence data arrive late. Measures of workload, role clarity, recognition, and communication quality, gathered at team level and reviewed regularly, give earlier warning and allow leaders to act before harm occurs.
Developing Managers as Work Designers
The second pillar is manager capability. Most of the hazards described in this article, including workload allocation, role clarity, communication, recognition, and the handling of change and performance concerns, are produced or mitigated through everyday managerial decisions. WHO's recommendation that managers be trained to prevent stressful work environments reflects this reality (World Health Organization, 2022). Yet only about three in ten US managers in the NAMI-Ipsos poll had been trained to talk about mental health with their teams (National Alliance on Mental Illness, 2026).
Training should go beyond mental health literacy, which helps managers recognize distress, to work design literacy, which helps them recognize and change the conditions that cause it. The STAR trials show the value of training supervisors to support employees' lives outside work and to manage by results (Moen et al., 2016). The Defence case shows the cost of failing to train supervisors in high-risk processes (Comcare, 2025). Managers also need relief themselves; Gallup (2026) found that managers report more daily stress than individual contributors. Spans of control, administrative burden, and the competing demands placed on middle managers are part of the psychosocial system.
Organizations should therefore assess managers' own workloads and decision rights, give them authority to adjust team demands, and hold them accountable for psychosocial outcomes in the same way they are held accountable for physical safety and budget performance.
Sustaining Worker Voice and Learning
The third pillar is participation. Safety law has long recognized that the people exposed to hazards are often best placed to identify them, and psychosocial risk is no exception. In OSH Pulse 2025, 45% of EU workers reported that they were consulted about stressful aspects of work, while 53% reported access to awareness raising or training on coping with stress, and only 35% reported other measures such as changes to work organization (Belli & De Keulenaer, 2025). Figure 3 shows this gap between individual-oriented provision and structural change.
Closing that gap requires durable channels for worker voice, from health and safety committees and union agreements, as at Orange, to facilitated team redesign sessions, as in STAR. Nielsen and Randall (2013) emphasize that how employees experience an intervention, including whether they believe it is genuine and whether they participated in shaping it, strongly influences whether it works.
Learning also depends on evaluation. Many well-intentioned initiatives, including after-hours email controls, have never been evaluated for mental health outcomes. Organizations that pilot changes with baseline and follow-up measures, share results honestly, including null results like those in the long-term care STAR trial, and adjust course will build a body of local evidence that is more useful than any vendor claim.
Conclusion
The case for treating mental health at work as an occupational safety issue rests on three converging lines of evidence. First, the hazards are common and well understood: excessive workload, low control, insufficient recognition, and poor communication are reported by large shares of workers across Europe and beyond, and decades of research link them to strain, burnout, and depression. Second, the most common organizational response, offering individual coping resources, does not address those hazards and, at least in recent large-scale evidence, shows little measurable benefit on its own. Third, regulators are converting expectations into enforceable duties, and courts are beginning to penalize failures.
None of this means that counselling, employee assistance programs, or mental health awareness should be abandoned. Tertiary and secondary supports remain essential for people who are already struggling. The argument is about sequence and proportion. Organizations should start by asking what about the work is making people unwell, act on the answer, and then provide support for the residual risk. The evidence from Japan, Iceland, the UK four-day week pilot, the STAR trials, and the enforcement cases in Australia all point to the same conclusion: changing the conditions of work is harder than offering a wellbeing app, but it is where the leverage lies.
The EU-OSHA campaign that begins this month gives leaders a timely prompt. The question it implicitly poses is whether an organization would accept a physical hazard that harmed nearly three in ten of its workers. If not, it should not accept the psychosocial equivalent.
Start with hazards, not symptoms: Assess workload, control, support, relationships, role clarity, recognition, and change management at team level, and link every identified hazard to a specific control.
Apply the hierarchy of controls: Prioritize work redesign that removes or reduces hazards before relying on training, coping resources, or counselling.
Treat overload as a design problem: Audit demand against capacity, remove low-value work, and pilot working-time changes with process redesign and honest measurement.
Build control and support into jobs: Give teams discretion over how and when work is done, and train supervisors to support people and manage by results.
Engineer recovery and clarity: Use collective agreements and technical controls to protect off-hours and leave, and set clear norms for communication channels and priorities.
Govern it like safety: Assign executive accountability, train managers to handle high-risk processes such as performance management, and report psychosocial indicators to the board.
Keep workers in the room: Consult the people exposed, evaluate what you change, and share what you learn.
Research Infographic
References
Bakker, A. B., & Demerouti, E. (2007). The job demands-resources model: State of the art. Journal of Managerial Psychology, 22(3), 309–328.
Belli, S., & De Keulenaer, F. (2025). OSH Pulse 2025: Occupational safety and health in the era of climate and digital change. European Agency for Safety and Health at Work; Publications Office of the European Union.
Comcare. (2025, December 19). Defence convicted after RAAF worker's death [News release].
Deloitte. (2024, May 17). Poor mental health costs UK employers £51 billion a year for employees [Press release].
DHR Global. (2025). Workforce trends report 2026.
Dollard, M. F., & Bakker, A. B. (2010). Psychosocial safety climate as a precursor to conducive work environments, psychological health problems, and employee engagement. Journal of Occupational and Organizational Psychology, 83(3), 579–599.
European Agency for Safety and Health at Work. (2025, October 10). World Mental Health Day: 29% of EU workers suffer stress, depression or anxiety [Highlight].
European Agency for Safety and Health at Work. (2026). Healthy Workplaces Campaign 2026–2028: Together for mental health at work.
Fleming, W. J. (2024). Employee well-being outcomes from individual-level mental health interventions: Cross-sectional evidence from the United Kingdom. Industrial Relations Journal, 55(2), 162–182.
Gallup. (2026). State of the global workplace: 2026 report.
Haraldsson, G. D., & Kellam, J. (2021). Going public: Iceland's journey to a shorter working week. Autonomy and Alda.
Health and Safety Executive. (2025). Work-related stress, anxiety or depression statistics in Great Britain, 2025.
Health and Safety Executive. (n.d.). What are the Management Standards?
Imamura, K., Asai, Y., Watanabe, K., Tsutsumi, A., Shimazu, A., Inoue, A., Hiro, H., Odagiri, Y., Yoshikawa, T., Yoshikawa, E., & Kawakami, N. (2018). Effect of the National Stress Check Program on mental health among workers in Japan: A 1-year retrospective cohort study. Journal of Occupational Health, 60(4), 298–306.
International Organization for Standardization. (2021). Occupational health and safety management: Psychological health and safety at work: Guidelines for managing psychosocial risks (ISO Standard No. 45003:2021).
Karasek, R. A. (1979). Job demands, job decision latitude, and mental strain: Implications for job redesign. Administrative Science Quarterly, 24(2), 285–308.
Kawakami, N., & Tsutsumi, A. (2016). The Stress Check Program: A new national policy for monitoring and screening psychosocial stress in the workplace in Japan. Journal of Occupational Health, 58(1), 1–6.
Kossek, E. E., Thompson, R. J., Lawson, K. M., Bodner, T., Perrigino, M. B., Hammer, L. B., Buxton, O. M., Almeida, D. M., Moen, P., Hurtado, D. A., Wipfli, B., Berkman, L. F., & Bray, J. W. (2019). Caring for the elderly at work and home: Can a randomized organizational intervention improve psychological health? Journal of Occupational Health Psychology, 24(1), 36–54.
LaMontagne, A. D., Keegel, T., Louie, A. M., Ostry, A., & Landsbergis, P. A. (2007). A systematic review of the job-stress intervention evaluation literature, 1990–2005. International Journal of Occupational and Environmental Health, 13(3), 268–280.
Maben, J., Taylor, C., Dawson, J., Leamy, M., McCarthy, I., Reynolds, E., Ross, S., Shuldham, C., Bennett, L., & Foot, C. (2018). A realist informed mixed-methods evaluation of Schwartz Center Rounds® in England. Health Services and Delivery Research, 6(37).
Maslach, C., & Leiter, M. P. (2016). Understanding the burnout experience: Recent research and its implications for psychiatry. World Psychiatry, 15(2), 103–111.
Mental Health Commission of Canada. (2017, March 16). Backgrounder: Case Study Research Project final report.
Mind RH. (2016, September 29). France: Orange signs its first agreement on digital transformation.
Moen, P., Kelly, E. L., Fan, W., Lee, S.-R., Almeida, D., Kossek, E. E., & Buxton, O. M. (2016). Does a flexibility/support organizational initiative improve high-tech employees' well-being? Evidence from the Work, Family, and Health Network. American Sociological Review, 81(1), 134–164.
Montano, D., Hoven, H., & Siegrist, J. (2014). Effects of organisational-level interventions at work on employees' health: A systematic review. BMC Public Health, 14, Article 135.
National Alliance on Mental Illness. (2026). 2026 NAMI-Ipsos workplace mental health poll.
Nielsen, K., & Randall, R. (2013). Opening the black box: Presenting a model for evaluating organizational-level interventions. European Journal of Work and Organizational Psychology, 22(5), 601–617.
Norton Rose Fulbright. (2025, December). Victoria's new psychological health regulations are now in effect.
NSW Resources Regulator. (2022, September 29). Managing psychosocial risks and hazards at work.
Premier of Victoria. (2025, December 1). New regulations start today to protect workers [Media release].
Rivez-Domont, E. (2017, March). France: Employees and digital devices outside working hours. In European labour & employment update. Jones Day.
Rugulies, R., Aust, B., Greiner, B. A., Arensman, E., Kawakami, N., LaMontagne, A. D., & Madsen, I. E. H. (2023). Work-related causes of mental health conditions and interventions for their improvement in workplaces. The Lancet, 402(10410), 1368–1381.
Safe Work Australia. (2025). Key work health and safety statistics Australia 2025 now available [News release].
Siegrist, J. (1996). Adverse health effects of high-effort/low-reward conditions. Journal of Occupational Health Psychology, 1(1), 27–41.
Silverberg, M. (2014, August 14). The latest perk: Force employees to enjoy vacation by deleting their email. Government Executive (originally published in Quartz).
Siu, J. (2011, December 23). Volkswagen to turn off BlackBerry email servers for off-shift employees. AutoGuide.
University of Cambridge. (2023, February). New results from world's largest trial of four-day working week [News release].
Vanguard Lawyers Tokyo. (2026, April 29). Changes to Industrial Safety and Health Act. Legal 500.
WorkSafe Victoria. (2023, October 19). Court body fined almost $380,000 for deadly work culture [News release].
World Health Organization. (2019, May 28). Burn-out an "occupational phenomenon": International Classification of Diseases [News release].
World Health Organization. (2022, September 28). WHO and ILO call for new measures to tackle mental health issues at work [News release].
World Health Organization. (2026). Mental health at work [Fact sheet].
Jonathan H. Westover, PhD, Chief Research Officer (Nexus Institute for Work and AI); Co-Founder & Chief Workforce and Learning Officer (Future State University); Founder & CEO (Human Capital Innovations); Professor of Organizational Leadership & Change (UVU). Read Jonathan Westover's executive profile here.
Suggested Citation: Westover, J. H. (2026). Fix the Work, Not the Worker: Managing Psychosocial Risk as an Occupational Safety Duty. Human Capital Leadership Review, 40(1). doi.org/10.70175/hclreview.2020.40.1.2
