Your incident reports doubled. Is that bad news?

A rising line can mean more harm, more honesty or a change in how you count. Before you colour it red, find out which.

Imagine the Monday morning meeting. The dashboard goes up on the screen. Incident reports have doubled.

Someone asks what went wrong.

It sounds like the obvious question. But suppose the business has just made reporting easier. Contractors can finally submit concerns. Supervisors have started thanking people for raising problems. Things that were once mentioned in passing are now making it into the record.

The workplace might be getting worse. It might be becoming more willing to tell the truth. Both could be happening at once.

The line on the chart cannot settle that argument.

An incident report count measures what reached the system. It does not, on its own, measure everything that happened.

That distinction can change which team gets praised, which manager gets challenged and which warning gets missed.

The reporting paradox

There is a name worth giving this problem: the reporting paradox. Improve your ability to hear about problems and, for a while, the numbers may look worse.

Consider a deliberately simplified example. In one month, 100 reportable events occur and 20 reach the reporting system. In the next, 80 occur and 40 are recorded.

Actual events have fallen by 20%. Recorded events have doubled.

Now imagine a different workplace. Events rise from 100 to 200, with one in five recorded in both months. Its dashboard also goes from 20 reports to 40.

The same headline. Two very different stories.

Illustrative arithmetic, not observed company data or a forecast. “Reportable” here means within a fixed internal reporting definition, not a statutory reporting category. Real organisations usually do not know the full underlying event count.

This is why you cannot simply reverse the traffic lights and declare rising reports a success. The missing information is precisely what you need to establish.

This is more than a hypothetical problem

NHS England makes the distinction explicit in its patient safety reporting statistics: a low recording rate may reflect under-recording, while a high rate may reflect greater openness. Neither is a reliable verdict on safety by itself. That is a warning about interpreting healthcare data, but the measurement problem is familiar wherever people must choose to raise a concern. NHS England, patient safety event data.

There is also a substantial gap in workplace injury reporting. In its 2024/25 accident statistics, Britain’s Health and Safety Executive records 59,219 reported non-fatal injuries to employees, and states that such injuries are substantially under-reported by employers, with current levels of reporting estimated at around half. That estimate concerns a specific statutory injury category. It is not a reporting rate for every accident, hazard or near miss. HSE, kind of accident statistics.

The uncomfortable implication is that a quiet system may be reassuring you about its own blind spots.

What did the green number cost?

Picture a supervisor who is congratulated every month for keeping incident numbers down. Then picture the first person who threatens that record by submitting a report.

Nobody needs to say “keep quiet” for the incentives to become awkward.

OSHA’s worker participation guidance warns that incentives must not discourage injury and illness reporting, because hazards can otherwise remain undetected. It also recommends regular feedback so workers can see that their concerns are being addressed. OSHA, worker participation.

This creates a difficult management test. When someone raises a problem, does the response make the next report more likely or less likely?

The same question belongs beside a falling complaints total or a shrinking defect log. Have the underlying problems reduced? Or has the route into the record become harder, less trusted or less worthwhile?

A low number deserves curiosity before congratulations.

Five questions to ask before judging the trend

1. What has actually increased?

Separate injuries from near misses, hazards and general observations. A surge in newly captured concerns tells a different story from an increase in serious harm.

Look at potential consequences as well as actual outcomes. A heavy object falling into an empty walkway may cause no injury. The absence of injury does not make the failed control reassuring.

Read a sample of the reports behind the total. Are they describing different problems, or repeated warnings about the same unresolved one?

2. Has the opportunity for incidents changed?

More working hours, deliveries, sites or completed jobs can mean more opportunities for things to go wrong. Compare like with like, using an activity measure that fits the work. The US Bureau of Labor Statistics, for example, calculates workplace injury and illness incidence rates using hours worked, and recommends comparisons with similar operations. BLS, computing incidence rates.

Even the denominator needs checking. NHS England corrected its October to December 2025 release after two months of ambulance contact data were omitted, making recording rates appear too high. The missing activity data changed the apparent story. NHS England, correction dated 7 April 2026.

An adjusted rate is useful. It still cannot correct for problems that never get reported, or for a shift towards more hazardous work.

3. Has the reporting system changed?

Mark changes on the timeline: a simpler form, a new reporting category, contractor access, training, a campaign or a backlog being entered.

Check whether the chart groups reports by the date an event happened or the date someone submitted it. A catch-up exercise can make this month’s submissions spike without this month’s events doing the same.

Look for duplicate records and changes in classification too. A new system can improve visibility and disrupt comparability at the same time.

4. Who is speaking, and who is missing?

Are reports coming from a broader range of people, shifts and locations? Or is the entire increase being driven by two enthusiastic reporters?

Ask people what happens when they report something. Include those who rarely use the process. An empty column for the night shift is a reason to investigate, not evidence that nothing happens at night.

More participation supports the explanation that visibility has improved. It does not prove that underlying risk has fallen.

5. What changed after the report?

How quickly was the concern assessed? Was someone responsible for the response? Were urgent risks controlled? Did the person who reported it hear back? Did the problem recur?

OSHA includes response time and timely corrective action among the measures organisations can track alongside injury outcomes. Its leading indicators guide also warns that unresponsive management can put people off reporting further hazards. OSHA, leading indicators guide.

Keep response speed in proportion to risk. A quick acknowledgement is useful; a quick administrative closure is no substitute for checking that the remedy worked.

Better reporting is a hypothesis to test

There are encouraging combinations: wider participation, shorter reporting delays, useful new information and evidence that problems are being addressed. Together, they strengthen the case that the organisation is seeing more clearly.

There are worrying combinations too: more serious events, repeated failures, overdue responses and the same hazard appearing again after it was supposedly resolved.

Mixed signals need investigation. A rise in near-miss reporting must not be used to explain away worsening injuries. Equally, a fall in injuries over a short period cannot establish that every important risk is under control.

The aim is to understand the change well enough to act, rather than pick whichever interpretation makes the meeting easier.

The number that should make you uneasy

Suppose the reports keep coming. The same damaged barrier. The same confusing handover. The same defect on another shift.

At that point, the story has moved on from willingness to report. People are telling you. The question is whether the organisation is learning.

HSE draws a distinction between identifying lessons and embedding them in how an organisation works. It highlights the failure to implement earlier recommendations as a feature of organisations that do not learn. HSE, learning organisations.

So when the line rises, resist the instant verdict. Find out what changed in the work, what changed in the reporting and what happened next.

And when the line falls, ask an equally demanding question:

If something went wrong here tomorrow, how confident are we that somebody would tell us?