[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"article-body-the-safety-metric-that-measures-nothing":3},"\nThe most consequential number in workplace safety has no statistical\nrelationship to the outcomes it is supposed to prevent. That is the\nfinding of peer-reviewed research on TRIR, the Total Recordable\nIncident Rate, the number contractors win bids on and boards read as\nthe state of safety.\n\nHallowell and colleagues examined the metric's statistical behaviour\nin [The Statistical Invalidity of TRIR as a Measure of Safety\nPerformance (Professional Safety, 2021)](https://www.csra.colorado.edu/tyrannyoftrir).\nThere is no statistical relationship between a company's TRIR and\nits risk of serious injury or fatality, and at the size of most real\nworkforces, year-to-year movement is dominated by random variation.\nThe decimal-point changes that drive celebration and alarm are\nmostly noise.\n\n## The national data says the same thing\n\n[US Bureau of Labor Statistics injury and illness data](https://www.bls.gov/news.release/osh.nr0.htm)\nshows recordable injury rates falling by nearly half since 2006.\nOver the same period [the fatal injury rate](https://www.bls.gov/news.release/cfoi.nr0.htm)\nfell by roughly a fifth. If recordables were a faithful proxy for\ncatastrophic risk, the two curves should have moved together. They\ndid not.\n\nThe global picture is harsher. The International Labour Organization\nestimates around 2.93 million work-related deaths per year, the\nlarge majority from occupational disease, harm that appears in no\nincident log. And IOGP data on oil and gas fatalities shows 81% were\ncontractors, the segment the reporting systems cover least.\n\n## Why a real number can measure nothing\n\nEvery recordable is a real event. The rate still fails, for three\nreasons.\n\nFirst, serious events are too rare to count. A rate built on total\nrecordables is dominated by frequent, minor events, which arise from\ndifferent causes than the catastrophic ones.\n\nSecond, the metric invites its own corruption. A number tied to\ncontracts and bonuses attracts classification pressure (e.g. the\ncase managed into first aid, or the report discouraged). It can\nimprove while nothing about the hazard changes.\n\nThird, it reports after the harm. Even a statistically sound count\nof injuries arrives too late to prevent them.\n\nA number with those three properties is not a safety instrument. It\nis an accounting convention.\n\n## What to steer by instead\n\nThe data to do better usually already sits in the safety function's\nown records (e.g. observations, permits to work and near-misses).\nAssembled, it supports different instruments.\n\nCount the warning conditions, not just the injuries. The dangerous\nsituations and failed safeguards that research links to serious\nevents can be counted directly.\n\nPredict where the next incident is likely, with odds per site and\nper activity tested against history, so the stated odds have\nactually matched what happened.\n\nAnd wire the warnings to people. Predictions go to named owners with\nagreed trigger points that schedule an intervention.\n\nNone of this retires the recordable count; regulators require it.\nThe shift is in what the function steers by.\n\nPredicted incidents with tested odds are what\n[the Prophesee Compliance Suite](/solutions/compliance/ehs)\nproduces from the records a safety function already keeps. Built\nwith and proven inside global enterprises. See your own sites' odds.\n[Start here](/contact).\n",1786984937746]