Incident and near miss reporting, and the investigation that follows

Two things go wrong with incident reporting, and neither is the form.

The first is that the reporting net is drawn too narrowly, so most of the useful events never enter the system. The second is that investigations conclude somebody was careless, which is not a root cause, will not prevent a recurrence, and will not survive an audit.

Here is what has to be reported, what has to be investigated, and what a finished investigation actually looks like.

What a near miss is, and why it is worth more than it looks

A near miss is the same event as a recordable incident, minus the luck running out.

The load swings and nobody is standing in the arc. The ladder slides and the worker catches themselves on the rail. The excavator swings and the spotter happens to be looking. Nothing goes on an injury log, because nobody was hurt.

That is precisely what makes it valuable. The failure has already demonstrated itself and it cost you nothing. Every near miss is a free run of an experiment you did not have to pay for, and a company that collects them is being handed its own future incident list in advance.

Companies that report near misses well tend to look, on paper, like they have more safety events than their competitors. They usually have fewer injuries.

What has to be reported and investigated

This is where most programs are quietly incomplete.

The ACSA COR audit document list expects a reporting process covering four categories, not one:

• Incidents
• Near misses
• Occupational illness
• Work refusals

It also expects the process to address internal and external reporting, meaning to the workers compensation board and the regulator where that applies.

The investigation procedure is expected to cover the same list, and to state the purpose of investigating, identify root causes, follow up on corrective actions, and set out the responsibilities of senior management, management and supervisors.

Source: ACSA COR Audit Document List, elements 9.1, 9.4 and 9.6, youracsa.ca, reviewed August 2026. Audit tools are revised, so get the current version from your certifying partner.

What counts as recordable in the United States

If you work across the border, the recording question has a different and much more mechanical answer.

Under the US recordkeeping rule at 29 CFR 1904, a work related injury or illness is recordable if it results in death, days away from work, restricted work or transfer to another job, medical treatment beyond first aid, or loss of consciousness. Some diagnosed conditions, including a fractured or cracked bone and a punctured eardrum, are always recordable. Recordable cases go on the OSHA 300 log within seven calendar days.

A near miss does not go on the log, because no injury or illness occurred. That does not make it uninteresting. It makes it invisible to the metric, which is a different thing, and investigating serious close calls before they become recordable is ordinary good practice.

Worth connecting to the money: your recordable count is what your injury frequency rate is calculated from, and that rate is the number clients ask for during contractor prequalification, usually before they hire you rather than after.

The root cause problem

This is the section that matters most, and it is where most real investigation reports fail.

The audit expects completed investigation reports to identify root causes resulting from deficiencies in the health and safety system, and corrective actions designed to implement system changes to prevent reoccurrence.

Read that twice, because it rules out the conclusion most reports reach.

"The worker was not paying attention" is not a root cause. Neither is "failed to follow procedure", "used poor judgement" or "was in a hurry". Those describe what happened. The question the audit is asking is why the system allowed it, and it is a different question with a different answer.

The tell is what the corrective action looks like. If the entire corrective action is retrain the worker, the investigation has almost certainly stopped at the person. Retraining is sometimes part of the answer and it is very rarely the whole of it, because it leaves the system exactly as it was for the next person.

There is a practical test worth applying before you close a report: if this same worker left the company tomorrow, would your corrective action still prevent the next occurrence? If not, you have not finished.

This is also why the form matters more than it looks. The free investigation form below asks for contributing factors by category before it asks for a root cause, which makes it awkward to write "not paying attention" and move on. A form that goes straight from what happened to corrective actions will collect the shallow answer every time, because that is the easiest thing to write at the end of a bad day.

The corrective action is the part that gets audited

An investigation that identifies a cause and produces no closed action is worse than useless, because it is documentary proof you knew.

The audit asks for a corrective action log alongside the investigation reports, and expects completion in a timely manner with involvement from management at the appropriate level. In practice the most common finding across the whole of COR is an action that was found, recorded, assigned, and then nothing.

Put every action from every source, investigations, inspections, hazard assessments, worker reports, into one register with a name, a date and a verification column. One list, one owner, checked on a schedule. The corrective action tracker is free and does exactly this.

The verification column is the part people leave off, and it is the difference between the action being done and somebody saying it was done.

Why people stop reporting

The mechanism is simple and it is worth naming.

If reporting a near miss produces an investigation that looks like a search for who is at fault, people stop reporting near misses. Not because they do not care, but because they have watched what happens and drawn the obvious conclusion.

The result is a company whose near miss numbers fall while its actual risk does not move at all. A falling near miss count should be treated as suspicious until proven otherwise, because the two explanations are "we got safer" and "people stopped telling us", and the second is far more common.

The repair is the same as everywhere else in a safety program: respond visibly to something somebody reported, say publicly where it came from, and never let the first question be who did it. It is the same dynamic that governs whether crews fill out safety forms at all.

Turning it into something management reads

One requirement people miss sits in program administration rather than in investigations: the audit expects periodic health and safety reports, analysis of injury reports, and graphical reporting of statistics to show trends.

Which is a formal way of saying the events have to be added up and looked at. Individual reports filed and never aggregated tell you nothing about whether the same thing keeps happening in different clothes, and the pattern is usually where the real finding is.

If three separate reports over a year all trace back to work being scheduled without enough people, that pattern is the finding, and no single report contains it.

The free investigation form

This is the form for everything above. It is a report and an investigation form on one document, because the investigation you run is the same whether somebody was hurt or nearly was.

Print ready PDF and editable DOCX, no email required.

It is laid out to make stopping at the person awkward. After what happened and the immediate actions, it walks you through contributing factors by category, then root cause, then corrective actions, and then a section for communicating back to the crews, which is the step that decides whether anybody reports the next one. The sign off block carries the management involvement the audit expects to see.

The tracker is the register those corrective actions live in until they are closed and verified.

The full library has the rest. In the app, a report raised in the field creates the corrective action, routes it to an owner and holds the verification, which is the loop an auditor is actually testing.

Related

• Hazard assessments: formal and field level, because an incident usually means an assessment missed something.
• Safe work practices and safe job procedures, the documents a corrective action often ends up changing.
• Worker orientation, where a new hire is told how reporting works, which decides whether they ever use it.
• What actually happens in a COR audit, including how records get sampled.