A near miss is a warning, not a safety score

A near-miss report tells a safety team where to investigate; its count alone does not tell management how safe the workplace is. A rise could reflect more hazardous events, better reporting, a change in work or several of these together. Treat the number as a prompt to examine exposure and controls, not as a league table.

For US employers, OSHA's incident-investigation guidance encourages investigation of injuries and close calls to identify underlying causes and corrective action. It favours participation by workers and managers and looking beyond individual blame. That guidance should not be confused with a requirement to submit every near miss to OSHA; assess applicable recordkeeping and reporting rules separately, including the relevant state-plan context.

Imagine a team introduces an easier reporting form and records more close calls the following month. That is a hypothetical scenario, not a documented case. The increase cannot establish that hazards rose or that the new form improved safety. The form changed how observations reach the register, so the measurement process changed alongside the reported result.

The reverse problem is equally important: a quiet register may mean few events, little work, a poorly understood reporting route or reluctance to report. An executive dashboard cannot distinguish these explanations without further evidence. Compare periods only after checking activity, exposure, definitions, site coverage and the reporting process.

Avoid targets that reward silence or require a fixed number of reports from each person. A quota can encourage trivial entries while a low-count target can discourage useful ones. The management question is whether important warnings reach someone able to act and whether that action reduces the specific exposure.

An initial account should capture the task, location, time, equipment or material involved, what happened, the potential exposure and any immediate protective action. Separate observed facts from possible explanations. An observer need not diagnose a root cause to provide a valuable warning.

Route an imminent or continuing danger through the site's emergency and safety arrangements immediately. Completing a form must not delay making the area safe through authorised action. Preserve relevant evidence without asking staff to recreate a dangerous event for a photograph or investigation.

Use a proportionate triage process. Consider the credible potential consequence and whether the exposure remains, rather than sorting only by the injury that happened to occur. A competent person should decide the investigation depth; this article does not supply a universal numerical risk threshold or replace task-specific safety expertise.

Keep a distinction between immediate correction and prevention of recurrence. Clearing an obstruction may address today's access problem; it does not explain why the route became obstructed or whether the same condition occurs on another shift. Record the contributing conditions, supporting evidence and uncertainties before selecting the action.

For each action, name an owner, completion date and a way to check effectiveness. A new briefing can be recorded as delivered, but delivery alone does not demonstrate that the relevant control works during the task. Choose evidence appropriate to the failure mechanism, such as an authorised observation of the revised work arrangement or inspection of an engineered safeguard.

The existing guide to corrective-action effectiveness explains why completion and effectiveness need separate evidence. Keep workers informed about the outcome, including why an action was selected, without exposing personal or confidential details unnecessarily.

Fictional dashboard example; these are not safety benchmarks. One site records four near misses in 10,000 work-hours during Period A and eight in 20,000 work-hours during Period B. It also introduces an easier reporting form between the periods.

The raw count doubled. The illustrative rate did not: 8 ÷ 20,000 × 10,000 = 4. Neither comparison establishes that safety improved, worsened or stayed unchanged. The reporting-method change affects comparability, and work-hours may not represent the particular exposure behind each event.

Suppose three Period B entries describe repeated obstruction of the same access route while a previously assigned action remains open. That specific recurrence is a better investigation prompt than the aggregate count. The management record should identify the mechanism, current protective action, accountable owner and evidence required before closure.

This rate is an analytical example, not an OSHA reporting metric or a target for workers. Keep severity potential, exposure, reporting changes and control evidence visible. Do not reward a lower number or infer a safe workplace from a quiet register.

Review unresolved high-potential events, recurring failure mechanisms, overdue actions and the evidence supporting closure. Retain report counts as context, together with exposure and reporting changes. Do not compress these different questions into an invented universal safety score.

Ask three practical questions about one recent event: what protected people this time, what could make the next event worse, and what verified change addresses that mechanism? The answers should come from the investigation, not from a generic training statement. Where the evidence is incomplete, leave the conclusion qualified and the investigation open.

A useful reporting system makes warnings actionable and gives employees a credible response. It cannot prove safety simply because the chart moves in a preferred direction.

AI assistance was used in research and drafting. No client engagement, firsthand test or independently verified outcome is claimed.

This article is general information. Verify current requirements and seek qualified advice for your specific circumstances.

Spotted an issue? Contact our editorial desk.