Underreporting Risk: 9 Distortions That Make Safety Metrics Look Clean

Underreporting risk is the possibility that injuries, illnesses, near misses or weak control signals are not entering the safety measurement system. It matters because a low injury rate can reflect real prevention, but it can also reflect fear, incentives, classification pressure, poor access to reporting, or leadership routines that punish bad news.

OSHA recordkeeping rules and BLS data-quality research both point to a hard truth: reported injury numbers depend on worker voice, classification discipline and organizational pressure, not only on actual harm. This article explains 9 distortions that make safety metrics look clean while risk stays active in the operation.

Why does underreporting risk distort safety performance?

Underreporting risk distorts safety performance because the dashboard receives a filtered version of reality, while leaders treat the filtered number as if it were the work itself. OSHA's recordkeeping requirements call for work-related injuries and illnesses to be recorded under defined criteria, but the rule cannot force a worker to speak, a supervisor to escalate, or an executive committee to distrust a number that looks convenient.

As Andreza Araujo argues in Muito Alem do Zero, glossed for English readers as Far Beyond Zero, lagging indicators look in the rearview mirror and show consequence rather than cause. The problem becomes sharper when the rearview mirror is partly covered. A low TRIR can mean the operation improved, but it can also mean pain became invisible, first aid was stretched, near misses were rebranded, or people learned that reporting creates trouble.

The practical correction is not to abandon injury rates. It is to test whether those rates deserve trust. Pair the reported outcome with worker voice, medical referral patterns, near-miss quality, exposure volume and field verification. The article on green safety dashboards expands this point for executive reviews where clean color can become false confidence.

1. Incentives reward silence instead of prevention

Incentives distort reporting when bonuses, recognition, contractor scores or team awards depend on low injury counts rather than on verified control strength. The OSHA 2016 anti-retaliation provisions under 29 CFR 1904.35 reinforced that employees must be able to report injuries and illnesses without retaliation, which is why any incentive tied to the absence of cases deserves careful review.

The distortion is subtle because leaders rarely say, "hide the injury." They say, "protect the month," "do not break the streak," or "we are close to the bonus." Across 25+ years leading EHS at multinationals, Andreza Araujo identifies this as one of the easiest ways to convert a metric into a loyalty test. The worker is no longer deciding whether to report pain. The worker is deciding whether to disappoint the crew.

Replace injury-free rewards with control-based rewards. Recognize verified hazard removal, quality near-miss reporting, strong stop-work decisions, and corrective actions sustained for 30 days. The comparison in safety KPIs, bonuses and control checks shows why incentive design can either protect the signal or corrupt it.

2. Supervisors filter weak signals before they reach EHS

Supervisor filtering happens when first-line leaders solve, soften or relabel events before the formal reporting system sees them. A crew lead may treat a strain as soreness, a vehicle contact as property damage only, or a dropped object as housekeeping, even when each event should trigger a wider risk conversation.

This distortion often comes from role conflict. Supervisors are held accountable for production, overtime, team morale and injury performance at the same time, and the cleanest number is the one that creates the least friction this week. In Safety Culture: From Theory to Practice, Andreza Araujo argues that culture appears in repeated decisions, not in slogans. A repeated decision to soften bad news becomes culture even when the safety policy says the opposite.

EHS managers should audit the first 24 hours after weak signals. Select 10 informal events from shift logs, clinic notes, maintenance records or supervisor handovers, then compare them with the formal reporting system. If the informal trail is richer than the official database, the dashboard is not measuring risk. It is measuring what survived the filter.

3. First-aid boundaries become metric management

First-aid boundaries distort metrics when the organization treats classification as a performance strategy rather than as a technical decision. OSHA recordkeeping distinguishes first aid from medical treatment, restricted work and days away, but a classification rule can be misused when the main goal is to avoid a recordable case.

The trap is not the existence of first aid. The trap is pressure around the boundary. A nurse, supervisor or manager may be pushed toward the least visible classification because a recordable case affects the dashboard, the contract review or the manager's reputation. 1 classification decision can change the monthly rate, even though the person's pain and the failed control remain exactly the same.

Build an independent classification review for borderline cases. The review should include the medical basis, task exposure, witness account and control failure, not only the recordability outcome. A clean classification process protects both compliance and learning because it separates the legal threshold from the prevention question.

4. Near misses rise or fall for the wrong reason

Near-miss trends distort underreporting risk because both increases and decreases can be misread. A rise may show deteriorating controls, but it may also show better trust. A fall may show safer work, but it may also show that workers stopped believing the report leads to action.

The useful question is not whether near misses are high or low. The useful question is whether the content is getting sharper. Strong near-miss reports name energy, exposure, barrier failure, decision point and recovery action. Weak reports name vague conditions and produce generic reminders. A dashboard that counts reports without reading quality can celebrate volume while missing serious precursors.

For a monthly review, sample 20 near misses and score them for specificity. Look for task, exposure, control, immediate action and owner. If most reports lack those elements, the issue may be reporting quality rather than reporting volume. The guide on auditing leading indicator quality gives a practical method for testing whether an indicator is close enough to the hazard.

5. Contractor data arrives polished

Contractor data arrives polished when vendors know that low injury rates protect prequalification, renewal, price position or access to future work. The client may receive a neat monthly number while subcontractor crews carry pressure to keep cases away from the record.

This distortion matters because contractor exposure can sit outside the cultural radar of the host company. A site may have excellent employee reporting and weak contractor reporting at the same gate. In more than 250 cultural transformation projects, Andreza Araujo observes that contractor safety maturity depends less on the packet submitted before mobilization and more on what the host organization verifies during execution.

Require contractor reporting checks that compare injury data with clinic visits, replacement labor, stoppages, first-aid logs, permit exceptions and supervisor notes. The goal is not to accuse the contractor. The goal is to test whether the client's governance system can see the work it outsourced.

6. Clean departments receive less field verification

Clean departments receive less field verification when leaders assume that low injury numbers mean low risk. The assumption feels efficient, but it moves attention away from the very areas where underreporting may be hiding.

The pattern is predictable. A department with repeated reports gets visits, coaching and questions. A department with no reports gets praise and autonomy. If silence is caused by fear, peer pressure or weak access to reporting, leadership has just rewarded the condition that hides the signal. As Andreza Araujo writes in Sorte ou Capacidade, glossed as Luck or Capability, good numbers do not prove capability when the system has not been tested.

Build a reverse audit into the calendar. Each month, select at least 1 area with unusually clean numbers and test the field reality through observation, worker interviews and control assurance. The article on control assurance through field evidence explains why paper checks and field proof cannot be treated as the same signal.

7. Workers report hazards but not harm

Workers may report hazards while withholding harm when they believe condition reporting is safer than injury reporting. This creates a dashboard that looks participative because hazard cards are flowing, while pain, symptoms and treatment needs remain hidden.

The distortion can fool mature companies because hazard reporting volume feels like trust. Yet the worker who reports a missing guard but hides a shoulder strain is giving the organization only half of the truth. The missing guard may get fixed, but the body signal that reveals task overload, awkward reach or repetitive exposure disappears from the trend.

Separate the channels but connect the analysis. Hazard reports, discomfort reports, first-aid notes and near misses should be reviewed together for repeated task patterns. If one palletizing task produces hazard cards, informal complaints and stretching requests, the organization has a work-design issue even before a recordable injury appears.

8. Metric hygiene hides classification drift

Metric hygiene hides underreporting risk when definitions, owners, update dates and exception rules are weak. A safety database can look precise while different sites classify similar cases differently, close reports with missing fields, or change coding practices after leadership pressure.

Classification drift is dangerous because it looks like improvement. If one plant starts coding minor injuries as discomfort, another plant delays entries until month-end, and a third plant merges repeated first-aid visits into a single event, the regional dashboard becomes a comparison of habits rather than risk. 4 hygiene checks matter first: definition, owner, source record and exception rule.

Audit metric hygiene before debating performance. Select a sample of cases, trace each case back to the source document, then verify whether the same rule would have been applied on another shift or site. The article on metric hygiene data defects gives EHS managers a direct checklist for this layer.

9. Executive reviews ask for explanation, not contradiction

Executive reviews distort reporting when leaders ask managers to explain bad numbers but do not ask anyone to contradict good numbers. The meeting becomes a defense of exceptions rather than a search for hidden exposure.

During her PepsiCo South America tenure, where the accident ratio fell 50% in 6 months, Andreza Araujo learned that indicators improve when leaders change the rhythm of management. The decisive shift is not a prettier chart. It is a review habit that asks what the number might be missing, who has not spoken, and which control has been verified in the field.

For the next review, choose 3 clean indicators and assign a contradiction owner for each. That person must bring one field sample, one worker voice sample and one data-quality challenge. If no contradiction appears, the team can trust the number more. If contradiction appears, the team has found the work before it becomes a recordable event.

How can leaders detect underreporting risk before harm appears?

Leaders can detect underreporting risk before harm appears by testing clean metrics against independent traces of work, including medical visits, near-miss quality, contractor exposure, hazard reports, supervisor logs and field verification. The method should be repeated monthly because the reporting climate changes when incentives, deadlines or leadership pressure change.

Use a simple underreporting review with 5 lenses. First, compare injury rates with exposure volume. Second, check whether near-miss reports contain enough detail to guide prevention. Third, review borderline medical classifications. Fourth, interview workers in areas with unusually clean numbers. Fifth, test whether contractor and employee data are equally visible. This is metric assurance, not suspicion.

Each month without underreporting review allows the organization to confuse silence with control, especially in areas where high-consequence work continues without a strong worker voice channel.

Reported safety versus visible safety

Reported safety is the safety picture created by forms, databases and official classifications, while visible safety is the picture leaders build by comparing those records with the work. The gap between the 2 is where underreporting risk lives.

The organization needs both views. Reported safety supports compliance, trend analysis and external reporting. Visible safety protects decision quality because leaders can see whether the official system is still connected to reality.

What should change in the next monthly safety review?

The next monthly safety review should include an underreporting-risk test before leaders accept clean performance as proof of control. A practical review can be done in 30 minutes if the team selects a small sample, asks for contradiction, and records one decision about metric trust.

Start with the 9 distortions in this article. Check incentives, supervisor filtering, first-aid boundaries, near-miss quality, contractor data, clean departments, hazard-versus-harm reporting, metric hygiene and executive questioning. If 2 or more distortions appear, the dashboard should be treated as provisional until field evidence confirms the signal.

Underreporting risk does not mean every clean number is false. It means clean numbers must earn trust. If your leadership team needs to redesign safety metrics so they protect life rather than reputation, Andreza Araujo supports safety culture diagnostics, executive metric reviews and control-verification routines. Schedule a conversation with Andreza Araujo.