Near miss underreporting: why do reports go missing and how can you spot it in your data?

Practitioners on Reddit keep returning to one scenario: the standup says to report every near miss, and then a supervisor says not to log it. This piece grew out of those conversations, and Ideagen follows them to add to the discussion. It gives EHS managers a way to test their own reporting data for gaps, and a clear answer on which instruction to follow.

A near miss is an unplanned event that did not cause injury or damage but could have. Near miss underreporting is the gap between the near misses that happen and the ones that reach the log. A low or flat near miss count does not prove a safe site. Treat it as a data-quality signal until the checks below say otherwise.

Why do workers stop reporting near misses and how can an EHS manager tell?

Workers stop reporting when it costs them something and returns nothing. The cost might be a supervisor's frown, an argument on the line or a long form. The missing return is a report that disappears without a reply. The log shows the result before anyone says it out loud. Three checks on the data itself show where reports are going missing.

Reports per supervisor: where one crew goes quiet

Pull the last 12 months of near miss reports and group them by supervisor or crew. Divide by headcount or hours worked so a large crew is not compared unfairly with a small one. A crew with no reports in a year, working beside crews that report steadily, is a data gap rather than proof of a safe crew. Look at the spread as well. A supervisor whose crew logs half the site median is the first conversation to have.

Near miss to recordable ratio: what it means when injuries rise and reports do not

A recordable is a work-related injury or illness that meets OSHA's recordkeeping criteria and goes on the OSHA Form 300 log. Events that almost cause harm are more common than events that do, so near misses should outnumber recordables. Track near misses per recordable by site and by quarter. A ratio close to one, or one that falls after a recordable, suggests events are being filtered before they reach the log. No universal target exists, so compare each site against its own history and against sister sites doing similar work.

Time to close: what instant or endless closure tells you

Measure the days from report to closed corrective action. A median that keeps growing teaches workers that reports go nowhere. Reports closed within hours as no action required, or closed in bulk before an audit, show reports being cleared rather than investigated. Check the line itself too. A monthly count that stays almost identical while headcount, hours and seasonal work change points to a cap or a habit rather than the site.

How reporting gets suppressed without anyone intending it

Suppression rarely needs a formal order. A safety board that counts days since the last incident makes a report look like a reset. A bonus tied to low incident numbers makes a report look like a cost to the crew. A long form makes reporting a chore. Then there is the plain instruction: sort it out on the floor and do not log it.

An instruction not to log an event discourages reporting, whatever the reason behind it. OSHA's guidance on reporting procedures that must not deter or discourage employees sets out the regulator's own standard for a reasonable reporting procedure and its rule against retaliation. That guidance covers work-related injuries and illnesses, so it does not set a rule for near misses. The logic still carries over. Workers do not sort their silence by category, and a worker told to stay quiet about a near miss has learned that staying quiet is safer.

What to do when the standup and a supervisor give conflicting instructions

The site's stated procedure stands unless the EHS manager changes it in writing. A supervisor's verbal instruction does not override it. These steps keep the data intact while the conflict gets resolved:

• Log the event first. A report can be corrected or merged later, while an unlogged event leaves no trace.
• Note in the report that a conflicting instruction was given, so the pattern can be counted without naming blame.
• Give workers a route that skips the line supervisor, such as the EHS inbox, a safety representative or a mobile form.
• Confirm the rule in writing. One sentence from the site manager that every near miss gets logged and nobody is disciplined for logging one removes the ambiguity.
• Meet the supervisor privately and ask what is being protected. The answer is usually a target, a shift record or a belief that minor events are noise.
• Treat any discipline linked to a report as a possible retaliation issue and escalate it to HR and legal counsel.

How supervision and production targets shape what gets reported

Supervisors run output and handle safety reporting at the same time. When a target sits on one and a reporting rule sits on the other, the target usually wins. A supervisor measured on rework and downtime has a reason to treat a near miss as noise.

For the wider habit and supervision analysis, see Ideagen's post on why people follow safety rules only when watched. The point here is narrower: reporting data reflects how supervisors are measured. Line up reports against production pressure such as peak shifts, month end and audit weeks. A dip in reports during the busiest weeks shows targets crowding out reporting. The fix is to review supervisors on reports raised by their crews and time to close, with no reward for a low count.

Reading a quiet near miss log: four patterns and four fixes

A quiet near miss log is a question your data can answer

A low or flat count tells an EHS manager very little until it is tested. Reports per supervisor show where a crew has gone quiet. The near miss to recordable ratio shows whether the log is keeping pace with injuries. Time to close shows whether reports lead anywhere. When a supervisor and a standup disagree, the event gets logged and the conflict gets resolved separately.

Readers who want to compare notes on testing their own reporting data can continue the discussion in Ideagen's Reddit community.