When Good Safety Numbers Hide Great Risks

Perfect safety stats can mask hidden dangers. Learn why "zero-injury" goals often silence workers and how to build a culture of true prevention.

• By Huynh Tran An Binh
• Feb 20, 2026

A safety dashboard can tell a reassuring story: recordables down, lost-time cases down, rates trending in the right direction. Leaders relax. Teams get recognized for another “injury-free” month. Yet on the floor, people still talk about close calls, sore backs, small cuts handled quietly, and hazards everyone knows, but nobody wants to put in writing.

If that sounds familiar, you may not have a data problem. You may have a trust problem. Safety metrics don’t just measure performance—they shape it. When the metric becomes the goal, people learn how to protect the numbers rather than protect each other. The result is a culture where the scoreboard looks great, but risk remains the same (or worsens) because the system has trained people to keep bad news out of the official record.

Why “good numbers” can mislead

Most organizations rely heavily on lagging indicators such as Total Recordable Incident Rate (TRIR), Days Away, Restricted or Transferred (DART) rate, lost-time injury rates, and severity. They are useful for tracking outcomes and benchmarking. But they share a fundamental weakness: they depend on events being reported and classified correctly.

Research on the U.S. Survey of Occupational Injuries and Illnesses (SOII) has documented concerns about undercounting in employer-reported injury and illness data and has explored why cases may be missed or misclassified. In comparisons between employer logs and other sources, mismatches were often attributed to recordkeeping errors, data transfer issues, and misunderstandings of reporting rules, rather than to the absence of harm [1].

In other words: “good numbers” can be produced by the way a system is designed.

A five-question trust check

Before you redesign anything, run a quick diagnostic. Ask a cross-section of workers (different shifts, crews, and languages) these five questions and compare answers. The goal is not to “score” people—it is to find where fear, friction, or confusion is blocking the flow of information.

• Can you submit a near-miss or hazard report in under two minutes, using the tools you have?
• Do you believe you can report an injury and still be treated fairly—no blame, no hassle, no pay or job impact?
• When someone reports, do they get acknowledged and see what changed?
• What happens at the first point of contact: curiosity and support, or interrogation and disbelief?
• Are rewards, bonuses, or praise tied to “no incidents,” or to prevention and learning activities?

If you see big gaps between leadership and frontline answers, start there. A reporting system cannot be “fixed” with new metrics if workers do not trust what happens after they speak up.

Three common ways metrics “lie.”

1) Underreporting driven by fear or perceived consequences.
When workers believe reporting will lead to blame, discipline, stigma, or hassle, reporting drops—especially for less visible conditions and near misses. In one workplace case study, fewer than 5% of workers had officially reported a work-related injury or illness in the prior year, while over 85% reported work-related symptoms; 50% had persistent problems; and 30% reported lost time or work restriction. Workers cited reasons such as fear of reprisals and doubts about management's willingness to respond. [2]

2) Misclassification (often from confusion, not malice).
Recordkeeping is complex. “First aid vs. medical treatment,” restricted work definitions, and work-relatedness judgments can lead to inconsistent classification. SOII undercount research highlights how incomplete or inaccurate logs and misunderstandings of recordkeeping requirements can contribute to missing cases. [1]

3) Incentives that reward silence instead of learning.
Programs that celebrate “zero injuries” can unintentionally punish truth-telling. Survey research among union carpenter apprentices found that injury reporting was 50% less prevalent when workers were disciplined for reporting injuries. [3] That is exactly how a metric becomes a target: people learn that reporting carries a downside.

OSHA’s recordkeeping rule requires employers to establish a reasonable procedure for employees to report work-related injuries and illnesses and prohibits discrimination against employees for reporting. A procedure is not reasonable if it would deter or discourage a reasonable employee from reporting. [4]

What a trustworthy reporting culture looks like

A trustworthy reporting culture is not one in which injuries never occur. It’s one where:

• Reporting is easy, fast, and accessible
• Workers can speak up without fear of retaliation or shame
• The organization responds quickly and visibly
• The system learns—so hazards actually get fixed

OSHA’s Recommended Practices for Safety and Health Programs treat worker participation as a core element and encourage employers to remove barriers such as fear of retaliation and disincentives that suppress reporting. The guidance also suggests multiple reporting channels (including anonymous reporting when needed) and closing the feedback loop so workers can see what changed. [5]

The playbook: 7 actions you can implement now

1) Fix incentives: reward prevention and learning, not “no news.”
Remove incentives tied directly to low recordables or “zero injuries.” Replace them with leading measures that reflect prevention and learning: hazard reports, investigation quality, timely closure of corrective actions, and verification that controls are effective. OSHA cautions that incentive programs should be designed to avoid discouraging reporting. [5]

2) Make reporting frictionless.
If reporting takes more than a couple of minutes, your system is asking workers to choose between safety and productivity. Offer multiple channels (supervisor, QR link, short form, hotline). Make it usable across language and literacy levels.

Mini metric example: focus on flow, not just volume.

Instead of counting “how many reports,” measure whether the system moves information into action. Track reports per 100 employees, median time to acknowledge, percent assigned an owner within 48 hours, percent closed with verification, and repeat-hazard rate. If you want one simple signal: the share of reports that results in a documented control change.

A case-control study in manufacturing, comparing web-based versus paper-based near-miss reporting, found a 78.04% lower recordable accident rate and a 4.15% greater decrease over three years at the facility using the more accessible web-based approach. [6]

3) Close the loop—fast and predictably.
Silence after a report kills future reporting. Set a simple response standard: acknowledgement within 24–48 hours, assignment to an owner, visible status (open/in progress/verified/closed), and a short “what changed because of your report.” OSHA’s leading indicators guidance includes examples such as tracking time from worker report to management acknowledgement and the share of hazards corrected within a defined timeframe. [7]

4) Train supervisors on how to receive a report.
Most reporting breakdowns happen at the first human contact. Train frontline leaders to respond with three messages: “thank you,” “you did the right thing,” and “here’s what happens next.” Also train leaders to escalate medical concerns and avoid minimizing or re-labeling reports.

5) Audit the reporting pipeline, not just the injury rate.
Periodically compare first-aid logs and clinic referrals vs. recordables, restricted duty assignments vs. logs, and a sample of classifications for consistency. SOII undercount research shows that errors and misunderstandings can drive missing cases—meaning “clean data” may reflect process weaknesses rather than true risk reduction. [1]

6) Add “psychological safety checks” as a leading indicator.
Ask workers quarterly whether they feel safe reporting. Keep it short: “I can report a near miss without negative consequences.” “Management acts when we raise concerns.” “I know how to report and what will happen after.” If trust scores drop, injury rates may still look great—until they don’t.

7) Use a balanced scorecard: lagging + leading + learning.
Lagging indicators still matter—but they should be the rearview mirror, not the steering wheel. OSHA recommends using leading indicators tied to core program elements and using trends to drive continuous improvement. [7] A practical set might include TRIR/DART, near-miss/hazard reports per 100 employees, percentage of corrective actions verified as effective, median time to close hazards, and a worker trust pulse score.

How to measure if it’s working

A healthier reporting culture often looks “worse” before it looks better, because more truth enters the system. Watch for an initial rise in near-miss and hazard reporting, faster acknowledgement and closure times, fewer repeat hazards, improved trust scores, and—over time—fewer serious exposures and high-severity events. OSHA’s leading-indicator approach emphasizes evaluating trends, identifying gaps in program elements, and adjusting resources, training, and controls to improve outcomes. [7]

One caution: don’t chase report counts. If the system floods you with low-quality reports, workers will conclude that speaking up creates paperwork, not improvement. Coach for clarity, allow photos when appropriate, and publish a short “learning summary,” so crews can see patterns and fixes. The best trend is fewer repeat hazards and more verified controls that prevent high-severity events.

The bottom line

Safety metrics should be a window into risk—not a mask. If your numbers are “too good,” the goal is not to accuse people of dishonesty. The goal is to redesign the system so honesty is easy, safe, and useful—because the most trustworthy metric is simple: workers believe reporting will help, and they act like it.