Three key problems with that one safety metric
Total Recordable Incident Frequency (TRIF or TRIR) remains one of the most widely reported occupational health and safety metrics. It appears in annual reports, contractor prequalification questionnaires, registry platforms, and executive dashboards. It is required under the United States Occupational Safety and Health Administration (OSHA) recordkeeping framework and appears prominently in the Global Reporting Initiative (GRI) 403 framework for reporting on occupational safety.
Despite its widespread use, dissatisfaction with TRIF justifiably continues to grow.
The common complaints are familiar. TRIF is a lagging indicator. It is reactionary. It focuses on incidents rather than prevention. It can disproportionately impact small businesses.
These criticisms are valid and collectively reveal deeper mathematical and behavioural flaws within the metric.
A more structured critique of TRIF may be summarized into the following three problems:
- The Small Numerator Problem
- The Perverse Incentive Problem
- The One-Metric Problem
These three problems help clarifies why TRIF is often misleading, and why it should carry far less weight in decision-making.
1. The small numerator problem
TRIF = (# of recordable incidents × 200,000) ÷ total hours worked.
Thankfully, most organizations experience relatively few recordable injuries each year. While every incident matters, small sample sizes produce large variability in the data.
When sample sizes are small, the metric becomes hypersensitive to isolated and often random events. One injury can make performance appear catastrophic. Zero injuries can make it appear flawless. Neither tells us much about the underlying quality of a company’s management system, nor does it reflect incident severity.
This concern has been formalized in academic critiques of incident frequency data. Researchers including Dr. Marloes Nitert, Dr. Sidney Dekker, and Dr. Matthew Hallowell have questioned the statistical validity of using small injury counts for meaningful comparison or trend analysis.
Put simply: small sample sizes in the numerator produce volatile rates. Volatile rates produce far more noise than signal.
Compounding the issue, TRIF summarizes fundamentally different outcomes into a single count. A serious injury requiring weeks off work and a minor injury requiring one restricted day both register as “one.” When risk exposure and severity are ignored, there is no context in this small number of ‘recordables.’ And data is meaningless without context.
However, TRIF is routinely compared across companies, sectors, and years as though it were precise, and it is often heavily weighted in contractor risk scoring.
It is not precise. It cannot be.
Treating it as mathematically robust reflects a misunderstanding of statistical variability.
2. The perverse incentive problem
Because contractors across North America are routinely required to submit TRIF as part of prequalification or reporting processes—and because that one number may influence an overall risk score, site eligibility, or contract awards—it will inevitably drive behaviours associated with it.
This dynamic is not unique to safety. In every domain, performance indicators create incentives. As Peter Drucker famously observed, “What gets measured gets managed.” Tie compensation to quarterly earnings and accounting behaviour shifts. Tie school rankings to standardized test scores and instruction follows; sometimes, to the point of misconduct.
This does not require unethical individuals. It requires rational actors responding to incentives.
High-stakes consequences related to TRIF create the following unsurprising incentives:
- Reclassification of injury severity
- Aggressive interpretation of modified duties
- Selective inclusion or exclusion of contractor hours
- Reluctance to record borderline cases
When organizations are judged primarily on one thing, that one thing begins to matter more than the underlying system. When systems reward outcomes over process integrity, behaviour predictably follows.
3. The one-metric problem
As stated, a common critique is that TRIF is a lagging indicator. That, by itself, is not really a problem. Most commonly understood financial measures (e.g., revenue, net income, earnings per share) are lagging. Executives do not dismiss them simply because they look backward.
However, it would be a problem if an executive made a strategic decision based solely on any one of these. Instead, financial metrics are interpreted collectively. They reflect thousands of transactions. They are segmented, audited, and analyzed within integrated reporting systems.
The difference is context.
Similarly in personal health, we do not rely on Body Mass Index (BMI) alone. We consider resting heart rate, VO₂ max, strength measures, activity levels, caloric deficit or surplus, and more. One number can never tell the whole story.
TRIF, however, is often presented alone. It is rarely paired with exposure metrics, severity weighting, audit findings, or leading indicators. It stands as a single ratio expected to summarize organizational safety performance.
We would never assess corporate health based on one minor accounting line item. We could not tell whether a person was healthy or not based what was shown when they stepped on a scale.
Yet, procurement systems sometimes assign disproportionate weight to a single frequency rate.
A more constructive approach
Eliminating TRIF entirely is neither realistic nor necessary. Injury data has value. Workers deserve transparency. Outcomes (sort of) matter.
That said, we can choose what to emphasize. A more constructive approach measures the behaviours and system strengths we want to see.
These may include:
- Evidence of hazard identification and control effectiveness
- Demonstrated senior leadership commitment and worker participation
- Supervisor competency and involvement in the safety management system
- Audit findings (e.g., COR or ISO-aligned systems)
- Psychological health and safety measures
- Leading indicators tied to actual operational risk
Activity-based and system-based metrics address all three core problems. They rely on larger samples, incentivize desired behaviours, and allow for more meaningful comparisons.
Moving forward
The goal of occupational health and safety is continual improvement. If we continue treating TRIF as a primary representation of safety capability, we risk confusing measurement with meaning.
TRIF may remain part of internal reporting. But it should never be the driving force behind any consequential and data-driven decision.
References
Erkal, E. D. O., & Hallowell, M. R. (2023). Moving beyond TRIR: Measuring & monitoring safety performance with high-energy control assessments. Professional Safety, 68(5), 26–35.
Global Reporting Initiative. (2018). GRI 403: Occupational health and safety 2018. Global Reporting Initiative.
Hallowell, M., Quashne, M., Salas, R., MacLean, B., & Quinn, E. (2021). The statistical invalidity of TRIR as a measure of safety performance. Professional Safety, 66(4), 28–34.
Jacob, B. A., & Levitt, S. D. (2003). Rotten apples: An investigation of the prevalence and predictors of teacher cheating. The Quarterly Journal of Economics, 118(3), 843–877. https://doi.org/10.1162/00335530360698441
Martin, L. (2026, January 8). What directors get wrong about workplace safety. https://www.nacdonline.org/all-governance/governance-resources/directorship-magazine/online-exclusives/2026/january-2026/workplace-safety-metrics/
Nitert, M., & Dekker, S. (2019, June 26). When does a reduction in injury numbers become statistically significant? Safety Differently. https://safetydifferently.com/when-does-a-reduction-in-injury-numbers-become-statistically-significant/
Occupational Safety and Health Administration. (n.d.). 1904.7—General recording criteria. U.S. Department of Labor. https://www.osha.gov/laws-regs/regulations/standardnumber/1904/1904.7
ISN (ISNetworld). (2014). Incident classification flow chart [Flow chart]. ISN Software Corporation.
