[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"article-body-the-safety-metric-that-measures-nothing":3},"\nThe most consequential number in workplace safety is the Total\nRecordable Incident Rate. Contractors win and lose bids on it.\nExecutives carry it in bonus scorecards. Boards receive it quarterly\nand read its movements as the state of safety.\n\nThen there is the research. Hallowell and colleagues, in [The\nStatistical Invalidity of TRIR as a Measure of Safety Performance\n(Professional Safety, 2021)](https://www.csra.colorado.edu/tyrannyoftrir),\nexamined the metric's statistical behaviour and reached a conclusion\nthe industry has still not absorbed: there is no statistical\nrelationship between a company's TRIR and its risk of serious injury\nor fatality, and at the sizes of most real workforces, year-to-year\nTRIR movement is dominated by random variation. The decimal-point\nchanges that drive celebration and alarm are, statistically, mostly\nnoise.\n\n*The industry's primary instrument does not measure the thing the\nindustry most needs to prevent.*\n\n## The macro data says the same thing\n\nZoom out and the national statistics rhyme with the statistical\ncritique. [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.\nAcross 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\nfaithful proxy for catastrophic risk, the two curves should have\nmoved together. They did not, and the divergence is the entire\nproblem in one chart: we improved the measured thing dramatically\nwhile the outcome that matters most improved far more slowly.\n\nThe global picture is harsher still. The International Labour\nOrganization estimates around 2.93 million work-related deaths per\nyear, with the large majority caused by occupational disease, harms\nthat accumulate over years and appear in no incident log at all. And\nin high-hazard industries the exposure concentrates where visibility\nis weakest: IOGP data on oil and gas fatalities shows 81% were\ncontractors, the workforce segment least covered by the reporting\nsystems producing the headline metric.\n\n## Why a real number can measure nothing\n\nTRIR is not fabricated. Every recordable is a real event. The failure\nis statistical and structural at once:\n\n- **Serious events are too rare to count.** Fatalities and\n  life-altering injuries are, thankfully, rare enough that a rate\n  built on total recordables is dominated by the frequent, minor\n  events, which research shows arise substantially from different\n  causal pathways than the catastrophic ones.\n- **The metric invites its own corruption.** A number tied to\n  contracts and bonuses attracts classification pressure: the case\n  managed into first aid, the report discouraged. The metric can\n  improve while nothing about the underlying hazard changes.\n- **It is a lagging count of harm.** Even if TRIR were statistically\n  sound, it reports after the fact. Prevention needs instruments\n  that move before the event.\n\n> A metric that is unrelated to the worst outcomes, gameable under\n> pressure, and only available after the harm is not a safety\n> instrument. It is an accounting convention.\n\n## What a predictive safety instrument looks like\n\nThe data to do better usually already exists inside the safety\nfunction's own records, in observations, permits to work, near-misses,\ninspection findings, contractor onboarding, maintenance backlogs.\nAssembled, that history supports a different set of instruments:\n\n- **Precursor tracking**: the presence of the specific high-energy\n  conditions and failed controls that research links to serious\n  events, counted directly rather than inferred from minor-injury\n  frequency.\n- **Predicted incidents**: site-level and activity-level probability\n  estimates, backtested against history and honest about their\n  calibration, so \"this site, next quarter\" carries tested odds\n  rather than last year's rate.\n- **Decision-linked leading indicators**: not dashboard decoration,\n  but predictions wired to named owners with thresholds, so a rising\n  probability triggers a scheduled intervention, and the outcome is\n  recorded against the prediction that prompted it.\n\nNone of this retires the recordable count; regulators require it and\ntrend context has value. The shift is in what the function steers by:\nrecording harm remains the floor, predicting and preventing it\nbecomes the instrument panel.\n\n*Count recordables because you must. Steer by predictions, because\nthe count cannot see what kills people.*\n\nPredicted incidents with published calibration are what the\nProphesee Compliance Suite's EHS module produces from the records a\nsafety function already keeps. Built with and proven inside global\nenterprises. See your own sites' odds. [Start here](/contact).\n",1786833838562]