The short answer: A falling TRIR or DART rate tells you what already happened, usually to a few people over a short window, and it can keep dropping while the conditions for a serious event build underneath it. Leading indicators, near-miss reporting, how fast you close corrective actions, inspection completion, training currency, and participation, tell you whether the work that prevents injuries is actually getting done. Our view: a program judged only on its recordable rate is grading itself on last year's test.
What do TRIR and DART actually measure?
TRIR and DART are lagging indicators. They count injuries that have already happened, and they do it consistently enough to compare one year against the next.
The recordable rate you already report is a rear-view mirror. DART narrows the view to the more serious cases: the ones that cost someone days away, restricted duty, or a move to another job. Both are useful for what they are, and a sustained rise in either is a real warning.
The trouble starts when one number becomes the scoreboard for the whole program. A recordable rate is an accurate record of harm already done. It is not a forecast of harm still forming.
Why can a falling injury rate hide rising risk?
Because a recordable rate is a small number over a short period, and small numbers move for reasons that have nothing to do with safety. The metric that rewards low counts also quietly rewards not recording.
Three things pull it off the truth. Noise: a smaller site can swing on a single case, so a drop often means luck, not progress. Incentive: when bonuses or contracts ride on the rate, pressure lands on whether a case gets logged, not on whether the hazard gets fixed. Scope: everyday injury frequency says little about the rare event that kills someone.
The Chemical Safety Board found that BP judged its safety largely by personal injury rates while the process hazards that caused the fatal 2005 Texas City explosion went unmanaged.
A falling injury rate can sit right beside a rising catastrophic risk, because the two are not the same measurement.
Which leading indicators predict what is coming?
Leading indicators measure the prevention work that happens before an injury. They move when the work moves, which is what makes them worth watching.
The ones that carry weight are familiar: how many people take part, how many near misses get reported and how fast you respond, what inspections turn up, who is current on training, and how quickly corrective actions close.
Two are easy to misread. A rising near-miss count usually means trust, not danger. And corrective-action closure time is the quiet one: an open action is a hazard you already know about and have not yet removed.
| Indicator | Type | What it tells you |
|---|---|---|
| TRIR / recordable rate | Lagging | Injuries already logged over the period |
| DART rate | Lagging | The serious cases: days away, restricted duty, or transfer |
| Near-miss reporting rate | Leading | Hazards surfaced before harm, and how much workers trust the system |
| Corrective-action closure time | Leading | Whether the hazards you found are actually getting removed |
| Inspection completion | Leading | Whether the hazard-finding work you planned is getting done |
| Training currency | Leading | Share of workers current on the training their job needs |
| Participation | Leading | How engaged people are in reporting and walkthroughs |
Should you drop lagging indicators?
No. Lagging indicators measure the thing that matters most, real harm to real people, and they do it in a standardized, comparable way that leading indicators rarely match.
Keep them. They benchmark across sites and years, they feed insurance and prequalification, and a rising DART is a genuine alarm. Leading indicators can be gamed too: an inspection checked off without rigor teaches nothing. The problem was never the recordable rate. It was the weight we put on it.
How do you build a balanced set?
Pair every lagging number with the leading indicators that would have moved first. Keep the set small enough that people actually watch it.
- Show TRIR and DART as a trend, not a single figure. A bare number hides the swings that look like improvement, so a trend line lets leaders react to real movement instead of noise.
- Read the near-miss reporting rate as health, not failure. A program that discourages reporting goes quiet before it goes wrong, so a rising count means hazards are surfacing while they are still cheap to fix.
- Put a clock on corrective actions. An open action is a known risk still on the floor, so tracking median days from found to closed shrinks the backlog that precedes many serious events.
- Track inspection completion and training currency as coverage. The gaps are where uncontrolled hazards and untrained exposure live, so measuring the share done lets you see the hole before someone falls through it.
The Long View: measure the work, not just the wreckage
A falling recordable rate is worth having and worth reporting. It describes last year, and the long decline is real: injury and illness rates have dropped to 2.4 per 100 workers in 2023, down from 10.9 in 1972.
But 5,283 people still died from work injuries that year, a reminder that everyday injuries and fatal events run on different tracks. The numbers that tell you about next quarter are the ones measuring whether hazards get found and fixed, whether people are trained, and whether they still feel safe enough to report. Watch those, and keep the recordable rate honest company.
