A spanner slips from a hand six metres up and rings on the concrete. Nobody was underneath. Nothing is broken, nobody is hurt, and most workplaces write nothing down. That event has a name: a near miss. It is an accident that did not happen, and it is the most useful thing a workplace can count.
After the harmaccidents, sick days
vsvs
Before the harmnear misses, checks
There are two ways to measure safety. One counts after the harm: accidents, days off sick, illnesses. The other counts before it: near misses, inspections done, guards checked, training given. After-the-harm numbers tell you about the past. Before-the-harm numbers tell you what is coming.
Dropped spanner
↓→
Before, or after?
A dropped spanner that hit nobody. Which kind of measure is it?
Before the harm.
Right. Nobody was hurt, so there is no harm to count. What there is, is a warning: the same spanner, the same height, and one day somebody below.
Not a measure at all, since nothing happened.
Not yet. Something happened: a heavy tool fell six metres onto a place where people stand. Nothing was hurt is the lucky part, not the whole story.
After the harm.
Not yet. After-the-harm measures count injuries and illness. There was none here, which is exactly what makes a near miss worth counting.
about 300Near misses
↓→
about 30Small injuries
↓→
1Serious injury
An old ratio, argued over. The shape is the point.
An old study, from 1931, claimed that behind every serious injury sit about 30 small ones and about 300 near misses. The exact numbers are argued over, and they differ from trade to trade. The shape is what lasts: the warnings are many and the disasters are few, so the warnings are where the information is.
Move the control to set how many near misses get reported. Picture 300 happening in a year, the old study's number, used here only as a shape.
Near misses that happened300 in a year
Near misses the firm saw0 in a year
0 in 100 reportedNobody reports. 300 near misses happened and the firm's record says zero. Its first warning will be an injury.
Near misses that happened300 in a year
Near misses the firm saw30 in a year
10 in 100 reported10 in 100 reported. 300 happened; the firm's record shows 30. The other 270 were warnings it never heard.
Near misses that happened300 in a year
Near misses the firm saw90 in a year
30 in 100 reported30 in 100 reported. 300 happened; the firm's record shows 90. The other 210 were warnings it never heard.
Near misses that happened300 in a year
Near misses the firm saw180 in a year
60 in 100 reported60 in 100 reported. 300 happened; the firm's record shows 180. The other 120 were warnings it never heard.
Near misses that happened300 in a year
Near misses the firm saw270 in a year
90 in 100 reported90 in 100 reported. 300 happened; the firm's record shows 270. The other 30 were warnings it never heard.
When 1 in 10 are reported, what does the firm's record show?
30 near misses, and 270 warnings it never heard.
Right. The record is honest about what it was told. The 270 it was not told about are the same spanner falling in rooms where nobody wrote it down.
30 near misses, which means 30 happened.
Not yet. Thirty is what was reported, not what happened. A near-miss count is a count of what people chose to say, and it is low when saying is unsafe.
300 near misses.
Not yet. Three hundred happened. The firm sees only the reported tenth, and its record says 30. The gap between the two is the part it is blind to.
Move the control.
Each card is a number a workplace could keep. Is it counted after the harm, or before it?
One card at a time. Tap the pile it belongs to.
Card 1 of 6
Days lost to back pain last year.
Fire doors found wedged open on the walk round.
A hand crushed in the press.
Hours of forklift training given this quarter.
A pallet that fell where someone had stood a minute earlier.
A case of deafness confirmed by the doctor.
Days lost to back pain last year. → After the harm
Not yet. After. The backs are already hurt; the number is a bill for harm that has happened.
Fire doors found wedged open on the walk round. → Before the harm
Not yet. Before. Nobody has been hurt by an open fire door yet. The count is a warning about the day there is a fire.
A hand crushed in the press. → After the harm
Not yet. After. An injury is the clearest after-the-harm number there is; it tells you what the press was capable of, too late.
Hours of forklift training given this quarter. → Before the harm
Not yet. Before. Training is something done to lower the chance of harm, and counting it says whether the plan is being done.
A pallet that fell where someone had stood a minute earlier. → Before the harm
Not yet. Before. A near miss: the harm did not happen, by a minute. It is a warning with a time stamp.
A case of deafness confirmed by the doctor. → After the harm
Not yet. After, and years after. The noise did its work long ago; the diagnosis is the harm finally being counted.
Afterthe past
vsvs
Beforewhat is coming
Right. If the card names harm that has already happened, it is after. If it names a warning, a check, or something done to prevent harm, it is before. A workplace that counts only the first kind learns only from its injuries.
Told offthe next one is hidden
vsvs
Thankedthe next one is reported
Near misses go unreported for a simple reason: reporting one feels like confessing. If the worker who dropped the spanner is told off, the next spanner is not reported. Workplaces that hear about near misses have made reporting safe, and they thank the reporter. Their near-miss count goes up, and that is good news.
Near misses reporteddoubled
vsvs
Accidentsfell
A site's near-miss count doubled in a year, and its accidents fell. What most likely happened?
People started reporting. More warnings were heard, so more were fixed before they became accidents.
Right. The count of near misses is a count of what people say. When saying becomes safe, the count rises and the warnings get acted on, and the accidents fall behind them.
The site became twice as dangerous.
Not yet. If the site were more dangerous, the accidents would have risen with the near misses. They fell. What doubled was the reporting, not the danger.
The two numbers have nothing to do with each other.
Not yet. They are the two ends of the same triangle. Near misses heard become near misses fixed, and fewer of them grow into the injuries at the top.
0 reportedSite A
vsvs
150 reportedSite B
Two sites, 200 workers each. Site A reported no near misses last year. Site B reported 150. Which is safer?
Site B, for certain. More reports always means safer.
Not yet. More reports usually means a site that listens, which is a good sign. It is not proof: you would still need to know whether the 150 warnings were acted on.
You cannot tell from these numbers. A zero can mean safe or silent, and B's 150 may be a site that listens.
Right. A near-miss count measures reporting as much as danger. To judge safety you would want the accident numbers and the walk-round findings beside it.
Site A. Zero near misses is the best possible score.
Not yet. Zero is the score of a perfectly safe site and of a site where nobody dares speak, and the number alone cannot tell you which. With 200 people, silence is the likelier story.
Lesson complete
Accidents are counted after the harm; near misses are the warnings before it, and a zero may mean silence.