Incident & Near-Miss Log

What this tool does

CM8-37 is a working incident register. You log what happened, who it happened to and why; it works out your injury frequency rates, shows which root causes keep repeating, tracks the corrective actions to completion, and prints a report you can take to a board meeting, a client audit or an insurer.

Everything runs inside this single file. There is no account, no upload and no network request of any kind, so names, injuries and site details never leave the computer you are using. That matters here more than in most tools: an incident log contains health information about identifiable people.

What to log

Log everything that could have hurt somebody, not only what did. A register containing nothing but injuries is a register that is being under-reported, and it will tell you nothing until somebody is hurt.

• Near misses — nobody was harmed, but the same event on a different day would have harmed somebody.
• Injuries — from a plaster to a hospital admission.
• Occupational illness — conditions caused or made worse by the work: hearing loss, dermatitis, upper limb disorders, respiratory conditions.
• Dangerous occurrences — a collapse, a failure, an uncontrolled release, a vehicle incident, where nothing was hit only by chance.
• Property and environmental events — they usually share their causes with the ones that hurt people.

Write the description as a sequence of events, in plain words, without adjectives. "Operative stepped back and slipped on oil that had leaked from a forklift" is useful. "Operative was careless" is not, and it will read badly if the log is ever produced in evidence.

Incident types, and why they matter

The type drives the rates, so it is worth getting right. The tool uses the classifications that most national schemes share, in increasing order of consequence:

• Type — Meaning — Counts as
• Near miss — No harm, no damage — Neither
• First aid only — Treated on site, no professional treatment — Neither
• Medical treatment — Professional treatment, returned to normal work — Recordable
• Restricted work — Returned, but not to the normal job — Recordable
• Lost-time injury — At least one full shift not worked — Recordable and lost-time
• Occupational illness — Condition caused by the work — Recordable and lost-time

The tool refuses two combinations, because they are the classification errors that quietly ruin a rate: days lost recorded against something that is not a lost-time injury, and a lost-time injury with no days against it.

Frequency rates

A raw count of injuries tells you nothing about whether you are getting safer, because it moves with how much work you did. Rates fix that by dividing by hours worked.

Two conventions dominate, and they are not interchangeable:

• Per 1,000,000 hours — the usual basis for a lost-time injury frequency rate, common across Europe, the Middle East, Australia and much of Asia.
• Per 200,000 hours — roughly 100 full-time workers for a year, the basis behind the recordable-incident rates used in North America.

The same performance produces a number five times larger on one basis than the other, so always state the basis when you quote a rate. The tool prints it on the report and shows the full calculation in the Rate calculation table so anyone can check the arithmetic.

Small organisations should treat rates with care. With 50,000 hours worked in a year, a single lost-time injury produces a rate of 20 per million hours and none produces zero. Neither number means very much on its own — the trend over several years does.

Getting the hours right

The hours figure is the one people get wrong, and it changes every rate you publish. Include everyone the log covers: employees, agency workers, and contractors if their incidents would be recorded here. Use actual hours worked, including overtime, not contracted hours; exclude holiday, sickness and unpaid breaks.

If you only have headcount, a common approximation is average headcount × 2,000 hours for a full year, adjusted for part-time staff. Note in the report that it is an estimate.

Above all, keep the hours consistent with the incidents. If the log covers one site for one year, the hours must be that site for that year — and the date filter must match. Filtering the register to one quarter while leaving the annual hours in the settings will divide a quarter's incidents by a year's hours and produce a rate four times too low.

Root cause, not blame

"Human error" is where an investigation stops, not where it finishes. Someone made a mistake — but why was the mistake possible, and why was it not caught? Keep asking why until you reach something you can change: a guard, a procedure, a layout, a workload, a supply of the right equipment.

The root cause list is deliberately about conditions rather than people. If the honest answer is that somebody was rushing, the cause is human factors, and the corrective action addresses why the job was rushed — not that the person should try harder.

The root cause chart ranks causes by frequency and marks the ones that produced lost-time injuries. That ranking is the most useful thing in the tool: it tells you where the next injury is most likely to come from.

Corrective actions

Every incident can carry a corrective action with an owner, a due date and a status. The Reports tab lists the outstanding ones by due date and marks those that have slipped.

Distinguish the immediate action — what you did that day to make it safe — from the corrective action that stops it happening again. Cleaning up the spill is immediate. Fixing the leak is corrective. Logs full of immediate actions and empty of corrective ones explain why the same incident keeps appearing.

The near-miss ratio

The long-standing observation behind incident triangles is that serious injuries sit on top of a much larger base of minor injuries, and those on a larger base again of near misses. The exact ratios are disputed and vary by industry — do not treat any particular pyramid as a law. The useful part is the direction: if your log shows few near misses relative to injuries, you are not seeing the events that come before harm.

A near-miss count that goes up after you encourage reporting is a good sign. It usually means people trust you enough to tell you. The number to watch alongside it is whether injuries fall.

What incidents cost

Set your own cost for a day lost on the Settings tab and the tool values the lost days. Include cover or overtime, lost output, retraining, investigation time and administration — the uninsured costs, which are usually a large multiple of any insurance claim.

It is a deliberately simple figure and it understates the total: it does not attempt to price investigation time on incidents with no days lost, damage, delay, or the effect on the person injured. Use it to make the case for spending money on prevention, not as a full cost of loss.

Reporting and records

The "reported to a regulator" checkbox records that you made an external report. It is a record only: this tool does not report anything to anybody.

Which events must be reported, how quickly, by whom and on what form differ by country and by industry, and so do the periods for which records must be kept — often years, and longer where health surveillance is involved. Establish what applies to you and keep exports accordingly.

Because the log holds health information about identifiable people, treat exports as confidential. Consider recording initials rather than full names if the file will be shared internally, and restrict who can open the machine it is stored on.

Printing and sharing

Print Report produces a report from whatever the current filter shows: header, headline figures including both rates, the four charts, the rate calculation, the outstanding actions, the full register and your closing notes. Print to PDF to circulate it.

The scope line under the title states the filter in force. Clear the filters before issuing anything described as the full period, and check that the hours worked in the settings match that period.

Saving your work

Incidents, settings and the report header are written to this browser's local storage as you type, and the toolbar shows the time of the last save. That storage belongs to one browser on one computer: another browser, a private window, a second machine or a clean-up tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. Export CSV gives you the register for spreadsheet work and includes every filtered record, not only those drawn on screen. Reset asks twice, then erases everything this tool has stored. There is no undo.

Accuracy & disclaimer

This tool records what you enter and calculates from it. It cannot tell whether an incident was classified correctly, whether the root cause is the real one, or whether the hours worked are accurate — and every rate depends on all three.

Reporting duties, classification schemes, rate conventions and record-retention periods differ by country and by industry. This is an internal record-keeping and calculation aid, not legal advice, not a regulatory submission, and not a substitute for a competent investigation.