TRIR counts injuries. It doesn't reveal where the next serious one will come from.

TRIR is arithmetic: recordable cases, scaled by hours worked. A sprain that leads to restricted duty and a life-altering amputation each add one to the numerator. The formula was designed to compare how often injuries occur, and it does that job well. It was never designed to say which of those injuries came close to killing someone.

That limitation matters most when the rate is improving. Minor and moderate injuries are common, so their counts move the rate. Severe events are rare, so they barely register statistically. A company can cut its rate for several years by getting better at the everyday injuries while the handful of high-energy tasks that produce serious harm stay exactly as dangerous as before. A year without a serious event is also weak evidence that exposure is low, because small numbers are noisy.

ASTM E2920-26 addresses the classification side of the problem. It defines Level 1 and Level 2 categories for injuries and illnesses and sets common criteria for deciding whether a case is work-related. The labels matter less than the agreement behind them. When dozens of sites each apply their own judgment, a corporate roll-up reflects local habits as much as real conditions. Shared definitions make severity comparable across sites, business units and years, and they remove work-relatedness as a hidden source of variation.

The practical shift is in the questions leaders can ask. Instead of “how many recordables did we have,” a team can ask which tasks, energy sources and locations keep producing the most serious classified outcomes, and then point root cause analysis, controls and audits there. That only holds at scale if classification happens in the workflow where incidents are recorded, not in a spreadsheet after the fact. Benchmark Gensuite’s Incident Management application is built around that principle. TRIR stays in the picture for trending and reporting; severity becomes the lens that sets priorities.