Incidentmelding: what makes an incident report usable

One word, two obligations

In most Dutch organisations, incidentmelding covers two things that happen to share a word.

The first is a notification you owe someone outside the building. A serious workplace accident, one that leads to death, hospital admission or lasting injury, has to reach the Nederlandse Arbeidsinspectie without delay, and the situation is in principle left as it is until they have been. For establishments working with dangerous substances there is a second line to the competent authority, DCMR in the Rijnmond area, with its own thresholds and its own clock.

The second is the internal report. The near miss someone types in on a phone at the end of a shift. The unsafe situation that three people walked past before one of them said something. Nobody outside is waiting for it, and that is exactly why it is the harder of the two to keep alive.

They are not the same obligation and they do not need the same form. Mixing them is how a near miss ends up with sixteen mandatory fields, and how the one report that legally had to go out in hours sat in a queue over a weekend.

What has to leave the building, and how fast

Worth settling before you redesign anything:

• Which events trigger an external notification, written down in plain language rather than a reference to a legal article.
• Who is authorised to make that call, including on a Saturday night.
• How that person is reached, and what they do if they cannot reach the next person up.
• Where the internal record of that notification lives afterwards.

Most organisations have an answer to the first question and a vague one to the second. Ask three shift leaders who makes the call outside office hours. If you get three answers, that is your finding, and you did not need an incident to surface it.

What makes a report usable six months later

A report is not a record of something that happened. It is the beginning of a decision someone else will have to make, often much later, usually with less context than you have right now.

That is the whole test. Everything in the form either survives into that moment or it is decoration.

What survives:

• When and where, precise to the installation, the line or the work area, not just the site.
• What was going on: the task, the permit, the shift, the phase of the operation.
• What happened, in the reporter's own words, before anyone classifies it.
• What kept it from being worse, or what was supposed to and did not.
• Who was told at the time, and what was changed on the spot.

What usually does not survive: the category the reporter picked from a dropdown at eleven at night. Classification is work for someone trained to do it, done afterwards, and it can be corrected. An observation cannot be reconstructed. Ask people for the observation.

Why reporting dries up

It is rarely a culture problem first. It becomes one.

Time, login, fields, language, and the feeling that nothing happens. A four page form at the end of a shift is not a reporting system, it is a filter that selects for people with an hour to spare. Nobody decides to stop reporting. They just report the big ones, and then only the ones with a witness.

The part that is almost always missing is the return trip. Someone reports something and hears nothing back, so the next one does not get reported, so the data thins out, so the trend you needed is not there when an inspector asks about it. If you fix one thing this quarter, fix who tells the reporter what happened with their report, and by when.

Not every finding belongs on the safety team's desk

The other failure mode is the opposite of silence. Everything comes in, everything lands with the same two people, and the serious ones queue behind the broken soap dispenser.

Triage is a decision you make once, in advance, not case by case:

• Low potential, routine: the team that found it resolves it in its own work, visibly, with the closure recorded.
• Potential serious injury or fatality, or anything touching a safety critical barrier: safety experts and corporate resources, a proper investigation, a named owner.

The second group is small. Treating it as if it were large is how organisations end up with a backlog of open actions and no way to tell which ones matter.

Where this stops being a form problem

Plan and Do are the easy halves. Most organisations execute. The gap shows up in Check and Act: did the measure we took actually work, and how would we know?

Repetition is the symptom. When the same type of incident comes back two years later with a different name, the first report was usually fine. What failed was everything after it, and nobody noticed, because closing an action and verifying an effect look identical in a spreadsheet.

For companies under Seveso III this runs into a further problem, because an incident report is also an input to the veiligheidsbeheerssysteem, to management of change, and to the evidence an inspector asks for. Generic tooling handles the first half of that well and tends to leave the connections to people's memory, which is a point we worked through in why generic safety management software falls short for Seveso III companies.

Five things to do with the reports you already have

1. Count last year's reports by how they ended, not by how they came in. Resolved, resolved and verified, closed without action, still open.
2. Time your own form. Fill it in as a reporter would, on a phone, with gloves on if that is the reality.
3. Take one category and read the free text instead of the codes. The codes tell you what people clicked. The text tells you what they saw.
4. Write the feedback loop into the procedure: who informs the reporter, with what, within how many days.
5. Pick two incidents from eighteen months ago and check whether the measure is still in place and still working. That is your Check phase, tested.

None of this requires software. It requires someone to sit with the list for an afternoon and be honest about what they find, which is harder to arrange than a project budget.

We build inspection and audit software, custom forms and workflows, and corrective action tracking, and we are not finished with any of it. If you want to compare notes on where your reports are getting stuck, send me a message. If the five questions above already told you where, you have what you need.