What is your plant's safety index?

An event that had the potential to cause harm, and did not.

Turkey's Occupational Health and Safety Law (Law No. 6331) defines a near miss as an event at the workplace that had the potential to harm a worker, the workplace or the work equipment, but did not.

A serious accident is usually preceded by many near misses of the same kind. A recurring near miss points to a recurring root cause. That is why the law asks employers to investigate and record near-miss events.

Reporting on the floor is low. The event is dismissed because "nothing happened". The most frequent event no longer looks unusual to anyone.

The law splits it three ways: the employer, the worker and the safety professionals.

The main duty-holder is the employer. Carry out and update the risk assessment, train workers, provide the required personal protective equipment free of charge, record work accidents and near misses, prepare an emergency plan, provide health surveillance. The duty does not go away by contracting the work out. Responsibility lies in precaution, not in outcome. The absence of accidents does not show the job was done.

The safety specialist and the occupational physician identify hazards, propose precautions and report them in writing. Fixing them is the employer's job. Workers must follow their training and instructions, use the equipment correctly, never disable safety devices and report danger immediately. Faced with a serious and imminent danger, they have the right to refrain from working.

This structure also settles the "he did not wear his helmet, it is his fault" argument. The violation is the worker's responsibility. A culture where entering without a helmet is normal, with no warning and no supervision, is the employer's. Well-run plants aim to keep measurements rather than penalty lists. Which zone, which shift, how often - those are the questions to answer.

Accident counts look backwards. Leading indicators predict the accident before it happens.

Lagging indicators count what has already happened. Work accidents, lost days, occupational disease. They are objective and the basis of legal reporting, but because serious accidents are rare the numbers are small and noisy. A period with zero accidents can be the sign of a safe plant or of a lucky one.

Leading indicators count the behaviours and conditions that come before an accident. Near misses, helmet and vest compliance, unauthorised entries into restricted zones, close passes between forklifts and pedestrians. Their strength is frequency. Dozens of data points a day instead of one accident a year show within weeks whether a change is working.

Leading indicators should be used as a map of where precautions need to go, not as an instrument of punishment.

It means something when counted continuously and consistently.

The law asks for hazards to be prevented at the source first, then through collective protection, and through personal protective equipment last. PPE is the last step, but it is also the most visible rule on the floor. The employer must provide the equipment and monitor its use.

Counting done by people is inconsistent. Who does the inspection, which shift and what time of day change the result. One walk-round a week does not show the real compliance of a hundred-person shift. The more continuous and automatic the measurement, the more reliable the comparison.

The useful reading is by zone and shift. The answer to where and at what hour compliance falls says where the warnings and the training should go.

A single score computed from the plant's own past and current events.

Management does not have time to read ten charts. The index takes the violations in the chosen period, sets them against the time actually spent on the floor and reduces them to a single score. PPE and zone violations enter with their own weights, and how long a violation lasted shows in the score. It does not hide the detail; it opens a door into it: if the score is falling, which zone and which type pulled it down is one step away.

The score is computed over a whole period: a day, a shift, or the last 24 hours. To mean anything it needs a starting point: measure for a few weeks with no warnings and no interventions, then make a single change and keep measuring the same way. Without something to compare against, improvement cannot be proven.