Mental Health at Work: 4 Decisions That Stop Support From Becoming a Referral Loop

A manager notices that a usually reliable employee has started missing handovers, working through breaks, and avoiding difficult conversations. The manager recommends the employee assistance program, sends a wellbeing article, and closes the conversation with a promise to check in later. The support sounds caring. The work system remains untouched.

Mental-health support at work becomes a safety issue when the organization treats a work signal as a private condition only the individual must solve. A referral can be appropriate, especially when clinical care is needed, but it cannot carry the whole response when workload, staffing, conflict, scheduling, or decision pressure continues to create the strain.

Why manager support can become a referral loop

A referral loop occurs when a manager repeatedly directs a worker toward support services without changing the work conditions, decision rights, or follow-up process connected to the concern.

The loop usually begins with good intent. The manager wants to respect privacy, avoid diagnosis, and connect the person with qualified help. Those are sensible boundaries. The failure appears when those boundaries become an excuse to avoid examining the job itself.

A manager does not need to diagnose anxiety, depression, or burnout to ask whether the task is feasible, whether staffing matches demand, whether conflict is being managed, or whether a high-risk decision is being left with one exhausted person. Work design is not a clinical diagnosis. It is an operational responsibility.

Andreza Araujo’s work across more than 250 cultural transformation projects supports this distinction. In Safety Culture: From Theory to Practice, the practical question is not whether an organization has stated concern for people, but whether its routines make safer decisions possible when pressure rises.

Decision 1: Separate the person’s privacy from the work exposure

Protect personal medical information while still investigating the work factors that may be increasing strain or weakening safe decision-making.

Managers often make one of two errors. They ask for private details that they are not qualified or authorized to collect, or they avoid every question that might reveal a work design problem. Both responses leave the employee alone with the operational consequences.

The better approach is to ask about the work without asking for a diagnosis. What has changed in the workload? Which deadlines are colliding? Where is the employee unable to take a break, obtain help, or transfer a decision? Which interactions are creating repeated conflict? What task becomes unsafe when attention, recovery, or staffing is reduced?

The answers should be recorded as work conditions and control questions, not as labels about the individual. If the concern involves driving, isolation, machinery, hazardous energy, work at height, or another serious exposure, the manager should verify whether the current arrangement still protects the work. Privacy does not remove the duty to manage the exposure.

Decision 2: Decide what changes before the next shift

Choose an interim work adjustment before the next high-consequence task, even when the longer-term cause has not yet been established.

A support conversation is incomplete when the employee returns to the same overloaded shift, the same unresolved conflict, and the same decision queue without an interim plan. The manager does not need a perfect explanation before reducing immediate pressure.

An interim change might involve reassigning a high-consequence task, adding an independent check, changing the handover sequence, removing nonessential work, setting a protected break, or bringing a second decision-maker into an escalating conflict. The choice should match the exposure rather than follow a generic wellbeing script.

That decision also needs an owner and a review point. Otherwise, a temporary adjustment becomes an informal favor that disappears when production pressure returns. A visible work change tells the employee that the concern has entered management practice, while a referral alone may communicate that the problem belongs somewhere else.

In projects led by Andreza Araujo, the value of a safety intervention is tested through changed decisions and follow-up, not through the presence of a campaign. Her documented PepsiCo South America result, a 50% reduction in accident ratio in six months, is cited here as a leadership case in operating discipline, not as evidence that one mental-health action produces a fixed outcome.

Decision 3: Match the route to the signal

Use the right response route for the signal, because a clinical referral, a work redesign, and a safety escalation solve different problems.

A manager check-in can reveal several kinds of concern at once. The person may need confidential clinical support. The work may contain an excessive demand or unresolved conflict. A safety-critical task may need an immediate control. A policy or legal issue may require HR, occupational health, or another formal route. Treating all signals as one category creates delay.

The manager should therefore make the route explicit. A personal health concern can be offered confidential support without turning the manager into a clinician. A workload concern should enter a work planning conversation. A conflict concern should have a named process and owner. A safety exposure should be controlled before the discussion is considered complete.

This separation improves trust because the employee does not have to disclose more private information to make the work safer. It also improves accountability because each route has a different owner, evidence standard, and follow-up expectation.

Decision 4: Make follow-up a control, not a courtesy

Set a follow-up that checks both the person’s ability to work safely and the condition that prompted the concern.

“Let me know if you need anything” sounds supportive, yet it transfers the next move back to the person who may already be struggling to initiate difficult conversations. A manager should agree on when they will reconnect, what work condition will be reviewed, and what change will trigger escalation.

The follow-up should not ask the employee to prove recovery or disclose medical information. It should ask whether the agreed work adjustment happened, whether the workload remains feasible, whether the conflict route is functioning, and whether any safety-critical task still depends on a person whose capacity is being stretched.

Edgar Schein’s work on organizational culture helps explain why this matters. People learn what an organization truly values from repeated management behavior. If a concern leads to a referral and no operational response, workers learn that support is separate from production decisions. If the work is adjusted and the decision is reviewed, support becomes part of how the operation is controlled.

What leaders should stop calling a mental-health solution

A mental-health solution is incomplete when it offers personal resilience advice while leaving preventable workload, conflict, staffing, or role-design problems in place.

Resilience resources can help people cope, and employee assistance programs can provide valuable confidential care. The trap is presenting those resources as proof that the organization has addressed the risk. A meditation link does not resolve a shift that cannot be staffed safely. A manager training module does not resolve a role with contradictory priorities. A referral does not resolve retaliation fear or a conflict that nobody owns.

The same mistake appears in safety when training is treated as a substitute for control. James Reason’s model of organizational accidents remains useful here because it directs attention toward conditions that allow several defenses to fail together. Mental-health signals can be one indication that the defenses around work design, supervision, recovery, and escalation are under pressure.

Leaders should ask whether the offered support changes the conditions of work, changes the decision path, or only changes the information sent to the employee. The first two are management controls. The third may be helpful, but it is not sufficient.

How EHS, HR, and line leaders can share ownership

Shared ownership works when each function keeps a clear boundary while coordinating the work condition, support route, and safety consequence.

Line leaders see the work and can change priorities, staffing, sequencing, and immediate controls. HR can manage employment processes, conflict routes, accommodations, and policy obligations. Occupational health or qualified clinical providers can assess health needs within their professional boundaries. EHS can connect work strain to exposure, control verification, and serious-risk decisions.

Coordination fails when every function assumes another one owns the whole case. The manager says HR has it. HR says the employee assistance program has it. EHS receives only an incident after the decision quality has already degraded. A simple ownership record can prevent this drift by naming the immediate work protection, the confidential support route, the next review, and the person accountable for each action.

The record should contain only the information needed for the work decision. It should not become a parallel medical file. The purpose is to keep a safe work response visible without turning a human concern into a bureaucratic object.

The executive test is whether the work changed

Executives can test the quality of mental-health support by asking what changed in work design after a concern was raised, not only how many resources were made available.

A useful review examines whether repeated overload is visible, whether managers can escalate workload without being treated as weak, whether high-consequence tasks have backup decision capacity, and whether conflict or harassment signals have an accountable route. It also checks whether people receive feedback after raising a concern and whether the same condition returns under the next production peak.

These questions move the discussion from program activity to operating evidence. They do not require leaders to collect private diagnoses. They require leaders to examine decisions, conditions, and controls, which are within the organization’s responsibility.

Across 25+ years of multinational EHS work, Andreza Araujo has consistently linked culture to what leaders reinforce under pressure. Her book Antifragile Leadership frames leadership as the capacity to improve through disruption rather than merely appear calm during it. For mental-health support, that means treating strain as information about the system when the work pattern points beyond an individual case.

A practical reset for the next manager conversation

The next manager conversation should end with a protected work decision, a suitable support route, a named owner, and a review that checks whether the original condition changed.

Start by listening without diagnosing. Name the work condition that appears to be creating strain or weakening safe performance. Decide what needs protection before the next shift. Offer confidential support when it fits, while keeping the work response active. Record who owns each route and when the manager will return to the conversation.

This approach respects privacy without making privacy an excuse for inaction. It gives the employee more than a resource list, because it changes the conditions in which the employee must make decisions. It gives leaders more than a wellbeing metric, because it shows whether management responded to the work that produced the signal.

Mental-health support becomes a safety control when it changes the work, not when it only redirects the worker. The decisive question is simple to ask and difficult to avoid: after the concern was raised, what became safer, more feasible, or more speakable?