How to Run a Safety Evidence Review Before Monthly Decisions

A monthly safety meeting can contain dozens of numbers and still fail to show whether a critical control worked. ISO 45001:2018 specifies a management-system approach, while HSE guidance emphasizes leadership and active monitoring rather than passive record keeping. This guide shows EHS managers how to turn that principle into a repeatable evidence review before decisions are made.

The deliverable is not another dashboard. It is a short decision record that states what changed, which exposure remains, who owns the next control action, and how the result will be verified within a defined period. Andreza Araújo’s work consistently connects safety leadership with the quality of decisions that operators can see and trust.

Use a 30-day review window, a 24-hour escalation rule for uncontrolled critical exposure, and a seven-day verification check for urgent actions. The numbers are operating choices, not universal statistics, so adapt them to the risk profile, legal duties, and response capacity of the site.

Step 1: Define the decision the review must support

A safety evidence review works when it answers one management decision, such as whether to authorize work, fund a control, change a schedule, or escalate an unresolved exposure. Write that decision before opening the dashboard. ISO 45001:2018 links monitoring and evaluation to the organization’s OH&S objectives, which means the review should test whether evidence is relevant to the decision rather than whether every available field is complete.

Ask the meeting owner to complete one sentence. “At this meeting, we need to decide whether to…” The sentence can concern a shutdown, contractor mobilization, staffing change, engineering control, or overdue corrective action. If the answer is “review performance,” narrow it further because performance is too broad to govern a safe action.

Verify the scope by naming the affected operation, exposure, time window, and decision authority. A common error is allowing the data owner to define the question after the report has been produced, which makes the meeting follow the spreadsheet instead of the risk.

Step 2: Separate activity data from control evidence

Activity data shows that a process occurred, while control evidence shows whether the exposure changed or remained within an accepted condition. The distinction matters because inspection completion, training attendance, and audit closure can increase without proving that a critical barrier is available at the point of work. HSE’s active monitoring approach supports this separation by asking leaders to examine what is happening, not only what was recorded.

Create two columns for every indicator. In the first, record the activity, such as 96 percent inspection completion. In the second, state the evidence that would prove control, such as a verified guard, a tested isolation, or a corrected access condition. Do not label the activity a leading indicator until the team can explain how it anticipates or changes risk.

Verify three samples from the field or source record before accepting the indicator. A common error is treating a high completion rate as a high control rate, especially when the same person enters, approves, and closes the evidence.

Step 3: Build a small evidence set around the exposure

A monthly review should use a small evidence set that can be tested, not a complete export of every safety activity. Choose one lagging signal, one leading signal, one field observation, one open high-consequence action, and one worker or supervisor perspective. This five-part set gives leaders enough context to see whether the number, the barrier, and the lived work condition agree.

Use the same 30-day window for comparable measures, then mark exceptions such as shutdowns, major maintenance, seasonal work, or a new contractor. When a metric has a different denominator or reporting rule, show that difference instead of combining it with a smoother series.

Verify each item against its source owner and date. A common error is adding more metrics when the first five disagree. Disagreement is not a reporting failure; it is the reason the review exists.

For a deeper model of evidence quality, use the three evidence layers in safety assurance and test whether your review contains records, observations, and decision proof.

Step 4: Test the critical control at the point of work

Critical-control evidence must answer whether the barrier was present, capable, and used when the exposure existed. A dashboard cannot prove that by itself. Select one control connected to the decision and inspect it in the field, through a current record, or through a credible verification interview with the person who performs the work.

Ask four questions. What failure does this control prevent? What condition proves it is ready? Who can identify a failed state? What happens when the control is unavailable? The answers should match the procedure, the physical condition, and the escalation route. If they do not, the review should treat the disagreement as an open risk.

Verify the control using a dated observation, a named verifier, and a clear result. A common error is accepting a signed checklist whose entries were copied from the previous shift, because the signature proves completion rather than barrier performance.

Step 5: Assign the decision to the role with control

Every unresolved finding needs an owner who can change the exposure, not only an administrator who can update the tracker. The review should distinguish the person who reports the issue, the person who decides the risk response, and the person who verifies the result. That separation makes accountability practical when production, maintenance, procurement, and EHS interests compete.

Write the owner’s role, the decision due date, the temporary safeguard, and the evidence required for closure. If the decision depends on budget or engineering, name the approving authority rather than leaving the action with EHS. Andreza Araújo’s leadership writing in Make The Difference: Be a Leader in Health & Safety treats visible responsibility as an operating behavior, not a title on an organization chart.

Verify that the owner accepts the action and can describe the first move. A common error is assigning “management” as the owner, because an undefined group cannot be challenged when the date passes.

Step 6: Set an escalation rule before the meeting

Escalation should be triggered by a defined condition, not by how persuasive a person sounds in the meeting. Set the trigger before reviewing the result. Examples include a critical barrier unavailable, a high-consequence action overdue, repeated failure across two verification checks, or a work condition that differs from the approved method.

Use three response states. Continue with routine monitoring when the control is verified. Continue with a temporary safeguard and a dated repair decision when the exposure is bounded. Stop or restrict the work when the critical barrier is unavailable and no credible temporary control exists. The state must identify who can authorize the next move.

A 24-hour escalation rule is useful for uncontrolled critical exposure, but it does not replace an immediate stop-work decision when people face imminent serious harm. Verify the rule with the shift supervisor and the decision owner. A common error is writing an escalation path that starts after the next monthly meeting.

Step 7: Record the decision in four traceable fields

A useful decision record can fit on one page when it captures the evidence, the judgment, the owner, and the verification date. The record should preserve why the decision was made, which uncertainty remains, and what would change the decision. This is more valuable than a long narrative that cannot be searched or compared next month.

Use four fields. Record the exposure and evidence, the decision and its boundary, the accountable role and due date, and the verification method with its planned date. Link the record to the source data without copying a full dashboard into the minutes.

Verify traceability by asking another manager to reconstruct the decision from the record in less than five minutes. A common error is recording only “action agreed,” which hides the condition that made the action necessary.

If your organization needs a fuller governance rhythm, the safety decision log guide provides a complementary structure for recurring decisions.

Step 8: Verify whether the decision changed the work

The review is incomplete until the organization tests whether the decision changed the exposure, the barrier, or the escalation route. Verification should occur at a defined interval, such as seven days for urgent control repairs and 30 days for a recurring management change. The interval should reflect the consequence of failure and the speed at which conditions can drift.

Return to the same evidence source, then add one independent check from the work. Interview the supervisor, inspect the changed condition, review the next relevant permit, or compare the decision with the current shift plan. If the action was completed but the exposure remains, reopen the decision rather than protecting the closure rate.

Verify the result with a before-and-after statement that names the changed condition. A common error is closing the review because a meeting occurred, even though the decision never reached the people who control the task.

Andreza Araújo’s approach to safety culture keeps compliance connected to observable choices. That is the standard for this final step, because a decision that cannot be seen in the work has not yet become a control.

What to take into the next monthly meeting

Bring a decision-ready evidence set rather than a longer presentation. The meeting should leave with one clear sentence for each material exposure, including the control condition, the accountable role, the escalation state, and the date on which the result will be tested.

• Define one decision before opening the dashboard.
• Separate activity completion from control evidence.
• Use five evidence types within the same 30-day window.
• Test one critical control at the point of work.
• Assign the action to the role that can change the exposure.
• Set escalation states before discussing performance.
• Record the decision in four traceable fields.
• Verify the changed condition within seven or 30 days, according to consequence.

If the review identifies an uncontrolled critical exposure, do not wait for the reporting cycle. Restrict or stop the affected work through the site’s emergency and escalation process, then document the decision and the evidence that supports it.

ISO 45001 specifies a structure for an OH&S management system, while HSE explains how leadership and active monitoring support effective control. The ILO reports on the importance of preventing occupational injuries and ill health. These sources do not replace site judgment, but they reinforce why evidence must reach the decision that governs the work.

A safety evidence review earns its place when it makes the next decision clearer than the last one. Andreza Araújo’s books and practical resources on safety leadership are available through Andreza Araújo’s store.