How a Global EHS Portfolio Turned Mental-Health Concerns Into Safer Work Decisions

A mental-health concern rarely arrives in an EHS dashboard as a clean clinical label. It appears as repeated overtime, a supervisor who stops asking questions, a skilled operator who is present but no longer fully fit for the task, or a return-to-work plan that treats attendance as the only outcome. Across more than 250 companies and 30+ countries, Andreza Araujo's work shows why these signals become safety issues when leaders leave them inside HR or outside the operating system.

This case study follows the decision pattern that turns early strain into a manageable work-design and safety question. It does not diagnose employees, replace occupational-health care, or claim that one intervention solves every mental-health concern. It shows how EHS, HR, and line leadership can stop waiting for absence before they act.

Initial scenario: the concern was visible but nobody owned the decision

The recurring pattern was not a lack of concern. Managers noticed fatigue, withdrawal, conflict, presenteeism, and repeated requests for schedule changes, yet each signal was sent to a different destination. HR handled the employee conversation, occupational health handled clinical assessment, and EHS handled physical exposure. The work itself, where the risk was being created or intensified, remained largely unchanged.

That separation created a dangerous delay. A person could receive support and still return to a task whose workload, night rotation, isolation, production pressure, or supervisory relationship had not been reviewed. The presenteeism pathways that turn hidden strain into safety exposure make this distinction clear. Attendance is not proof that the work is safe to perform.

In the portfolio pattern, the first measurable change was therefore not a reduction in a medical diagnosis. It was a change in the decision record. Leaders began to document what signal had been observed, which work condition could be contributing, who had authority to adjust it, and when the adjustment would be checked.

Decision: treat mental-health signals as work decisions without turning managers into clinicians

The pivotal decision was to separate three questions that organizations often collapse. The first question was whether the person needed confidential clinical or occupational-health support. The second was whether the work created a foreseeable exposure that required a management response. The third was whether the person could perform a specific task safely under current conditions.

Managers were not asked to diagnose depression, anxiety, or any other condition. They were asked to manage observable work conditions and escalate concerns through the correct route. This boundary protected privacy while making accountability visible. It also prevented the EAP from becoming a place where every organizational problem disappeared without a change to workload, staffing, scheduling, or supervision.

The decision model followed ISO 45003:2021 guidance on psychosocial risks by examining work design, organizational conditions, and management practices rather than treating individual resilience as the main control. The related work-design tests for psychosocial risk provide a practical starting point for this review.

Execution: build one short pathway from signal to action

The execution began with a short review that could happen inside an ordinary operating rhythm. A supervisor recorded the observable concern without speculation, HR or occupational health confirmed the appropriate support route, and EHS tested whether the work design introduced a material safety exposure. The operational owner then decided what could change immediately and what required a deeper review.

The pathway was deliberately small because an elaborate process would have made supervisors wait for a specialist. It asked five questions.

• What changed in the person's work pattern or behavior?
• Which task, schedule, workload, or relationship may be increasing exposure?
• What temporary adjustment protects the person and the operation now?
• Which professional or confidential route must handle the personal concern?
• What evidence will show that the work condition has improved?

Each question had a different owner. That distinction mattered. The manager owned the work arrangement, HR owned the employment process, occupational health owned clinical boundaries, and EHS owned the safety-risk review. When the concern crossed those boundaries, the case was escalated rather than passed from one function to another.

Execution: connect support to work design

The second execution move was to test the work itself. Teams reviewed roster changes, excessive overtime, unclear priorities, inadequate recovery time, poor handoffs, isolation, and conflicting instructions. A concern that had first appeared to be an individual performance issue sometimes revealed a design problem that affected an entire shift.

This was where the mental-health process became relevant to occupational safety. If a worker was repeatedly asked to recover equipment alone at the end of a long shift, the response could not stop at a wellbeing conversation. The team had to examine staffing, authorization, fatigue exposure, escalation rules, and the physical task. If a new manager created uncertainty through contradictory priorities, the response had to include decision rights and workload clarity.

The difference between a burnout check, a workload review, and a return-to-work plan helps leaders choose the intervention that matches the signal. A support service can be necessary, yet still insufficient when the work continues to generate the same strain.

Measured result: the portfolio changed what leaders could see

The verified result from Andreza Araujo's work is the scale and repeatability of the operating pattern, not a fabricated clinical success rate. Her background includes 25+ years in multinational EHS leadership, work across 30+ countries, and more than 250 companies served worldwide. The lesson from that reach is that mental-health risk becomes more governable when the organization measures decisions and work conditions rather than waiting for absence or a crisis.

This is a management result because it changes the organization's field of view. The question moves from “Is this employee coping?” to “What condition is the operation creating, and which leader can change it?” That shift is consistent with James Reason's distinction between active errors and latent conditions. A personal struggle may be real, while the surrounding system still determines whether exposure is increased or reduced.

Measured result: the safety conversation became more specific

Once the pathway existed, leaders could discuss mental health without using vague promises. They could ask whether the next shift had enough recovery time, whether a supervisor had authority to remove a conflicting demand, whether a return-to-work restriction matched the actual task, and whether a temporary accommodation had an expiry and review point.

Specificity also reduced stigma. The conversation did not require an employee to disclose a diagnosis in a team meeting. It required the organization to examine the conditions under which work was performed. That approach protects dignity and produces better safety evidence, because the record contains a changeable exposure rather than a label attached to a person.

Generalizable lessons from the case

First, early signals matter only when they lead to a decision. A survey score, absence pattern, or supervisor concern has limited value if nobody owns the next change.

Second, confidentiality and operational accountability are compatible. Personal information can remain restricted while the organization still records the work condition, the temporary protection, and the review date.

Third, a manager's role is not to provide therapy. It is to notice a change, protect the task, start the correct referral, and remove work conditions that are within managerial control.

Fourth, EHS should not wait for a serious incident before reviewing psychosocial contributors. Fatigue, workload, isolation, and role conflict can affect attention, coordination, escalation, and willingness to report a concern long before an injury appears.

Fifth, the strongest indicator is not the number of wellbeing messages sent. It is whether the organization can show that a material work condition changed and that the change was checked under ordinary pressure.

What to apply in your operation

Start with one high-risk population, such as night-shift maintenance, process control, emergency response, or a team returning from a major change. Do not launch a broad campaign before you know how a real concern will move from observation to decision.

Define the minimum record. It should include the observable signal, the task or condition under review, the immediate protection, the accountable owner, the confidential referral route, and the date for checking whether the work changed. The 30-day escalation protocol for workplace mental health can help structure the first version.

Then test one case in the field. Ask the supervisor to describe the work without diagnosing the person. Ask EHS to identify the exposure. Ask HR or occupational health to confirm the appropriate professional route. Finally, ask the operational leader to make the decision that only the operation can make, such as changing the roster, removing conflicting priorities, adding supervision, or redesigning the task.

Andreza Araujo's experience across 250+ companies points to a practical conclusion. Mental-health support becomes part of safety management when leaders connect human signals to work conditions, decision rights, and proof that the exposure changed. The organization does not need to become a clinic. It needs to become accountable for the work it designs and permits.