Part 310 min

3.2 TRIR, DART, lost time and severity

Four rates, one formula, four different questions. TRIR asks how often people get hurt. DART asks how often it costs them time or duty. The lost time case rate asks how often they stop working altogether. The severity rate asks how long they are gone. Confusing any two of them produces a comparison that is out by a factor of two or three, and it happens constantly.

1 The four, side by side

The four rates and the 300A columns behind them
Rate	Counts	Columns	The question it answers
TRIR	Cases	G + H + I + J	How often does a recordable injury or illness happen here?
DART	Cases	H + I	How often does one cost time away, restriction or a transfer?
Lost time case rate	Cases	H	How often does someone stop working entirely?
Severity rate	Days	K (and sometimes L)	When it happens, how long are they gone?

Two structural facts fall out of the columns. A fatality is in TRIR and not in DART, because death is column G. And days are in none of the case-based rates, because K and L are day totals. A year with one catastrophic case and a year with one stitched cut produce the same TRIR contribution.

2 Reading the gaps

Because all four share a denominator, the ratios between them are stable enough to be diagnostic. Nationally, BLS put private industry in 2024 at a total recordable case rate of 2.3, a DART rate of 1.4 and a days-away rate of 0.8. So roughly six recordable cases in ten are DART cases, and a little over half of those involve actual days away.

What an unusual ratio suggests
Pattern	First thing to check
DART close to TRIR	Column J. The case that ended in stitches or a prescription with no lost time is the one most often missed.
DART far below TRIR	Restricted work capture. Partial shifts, informal “take it easy” arrangements and unwritten accommodations that never reach the recordkeeper.
TRIR flat, DART rising	Severity. The same number of cases is producing more time off and more restriction.
DART flat, severity rate rising	A small number of long cases. Look at the individual day counts before concluding anything about the year.

On the job: none of these patterns is a conclusion. Each one is a reason to open the 300 Log and read the case descriptions — which is why column F was worth writing properly.

3 Why OSHA watches DART

DART is a crude severity filter: it drops the cases that ended at the clinic with no time away. That makes it the number OSHA's programmed inspection planning for general industry is built on. The Site-Specific Targeting directive CPL 02-01-064, dated February 7, 2023 and now marked archived on OSHA's site, generates inspection lists in four ways from the Form 300A data establishments submit electronically:

Establishments with elevated DART rates, with separate thresholds for manufacturing (NAICS 31–33) and for non-manufacturing, because average rates differ so much between them.
Establishments with upward-trending rates — at or above twice the private sector national average in the base year, and still climbing.
A random sample of establishments with low DART rates, explicitly to verify data accuracy and quality control.
A random sample of non-responders that failed to submit the required data.

Construction is excluded from that plan; it is covered by other programs. And note the third bullet once more: under-recording is not a route off the list.

4 The light duty question, settled

Bringing an injured worker back on modified duty moves the case from column H to column I. DART counts both columns. The rate does not move by a decimal, and neither does TRIR, because the case was already recordable.

What a return-to-work program actually changes
Measure	Effect
TRIR	No change
DART	No change
Lost time case rate	Falls — column H empties into column I
Severity rate / days away	Falls, often sharply
Workers' compensation indemnity cost	Falls

Modified duty remains worth doing — people recover better at work, days lost fall and claim costs fall with them. Say that, and not the thing about the DART rate, because the 300A a customer asks to see will contradict you.

5 Leading indicators, and why rates are not enough

Every rate in this chapter is a lagging indicator: it counts harm that has already happened. At the scale most establishments operate, it also counts a small number of events, which makes it statistically noisy and slow to respond to anything you do.

The numbers that move first are the ones nobody is required to keep. Near miss reports filed. Inspection findings raised, and the share closed within their due date. Corrective actions verified as effective rather than merely marked done. Training gaps carried into the next month. Toolbox talks where somebody raised something.

Common mistake: setting a management target on TRIR alone. A target on a lagging indicator that a supervisor cannot influence directly is a target on the reporting behavior of that supervisor's crew, which is precisely the thing you need to keep clean.

6 A rate pack that survives questions

1
Quote all four rates together, for the same establishment and the same year.
2
Show the case count and the hours worked underneath them.
3
Show the three-year combined rate alongside the single year if you are under about 200,000 hours.
4
Name the universe: establishment or company, employees only or contractors included.
5
Put two leading indicators on the same page — near misses reported, and actions closed on time.

Key takeaways

TRIR = G+H+I+J, DART = H+I, lost time = H, severity counts days from K
A fatality raises TRIR and not DART; days raise neither case-based rate
Nationally in 2024: TRIR 2.3, DART 1.4, days away 0.8 for private industry
OSHA’s Site-Specific Targeting selects on DART — high, rising, and implausibly low
Light duty moves a case from H to I: it lowers days and cost, not DART or TRIR
Rates lag and are noisy at small scale — pair them with leading indicators you control

Free educational content — not OSHA-authorized training, no certificate or card issued. Follow your employer's program and the standards cited.