In the US the term is also referred to as a "close call" and is defined by the Occupational Safety and Health Administration (OSHA) as an incident "in which a worker might have been hurt if the circumstances had been slightly different" (bit.ly/2kwpZ3X).
These definitions would lead logically to a fire that destroyed a building but did not harm anyone (perhaps because it occurred at night) being classified as a near-miss. The Buncefield explosion at a Hertfordshire oil terminal in 2005 is estimated to have caused more than £1bn of damage but, with no injuries or deaths, would be a near-miss.
Other definitions overcome this gap by considering material damage to property or the environment as well as harm to people. The European Agency for Safety and Health at Work (EU-OSHA) definition is "an unplanned event which did not result in injury, illness, or damage -- but had the potential to do so" (bit.ly/2kloe7E).
Determining that an event "had the potential" to result in harm or damage might depend on your imagination. Each time you pour boiling water into a mug there is the potential to scald yourself. On the other hand, an unimaginative technician might assume that, despite having a drill slip several times because of a dull drill bit, their skill leaves no chance of a harmful accident.
Broad definitions of near-misses, along with organisational requirements to report "all near-misses" lead to reporting systems that don't work. Incident investigators become overwhelmed by a welter of reports of people almost slipping, tripping or bumping their heads. Sorting the situations in which you can act to prevent future accidents from those where you cannot becomes an impossible task.
In turn, those reporting the hazards receive limited or no feedback. Their instinct may then be to stop reporting incidents, and important information that could provide evidence for change goes unheard.
Determining that an event "had the potential" to result in harm or damage might depend on your imagination
In the 2015 publication, Incident [accident] Investigations: a guide for employers, US OSHA (bit.ly/2jAR4DX) provides a narrower definition than the one described earlier. It limits near-misses or close calls to incidents "that could have caused serious injury or illness but did not, often by sheer luck".
However, tightening the definition to exclude incidents where damage is caused -- which should be investigated as accidents -- and where the result could only have been trivial, misses any consideration of the value of investigating the near-miss.
The Heinrich triangle suggested a link between a large number of near-misses at the base of the triangle and a (hopefully) small number of fatalities or serious accidents at the apex. Investigate your near-misses, so the theory went, and you can remove the causes of the major accidents. If the near-misses are diminishing, the risk of a serious accident is reduced. This works for similar types of accidents. If people report that they nearly slipped on a wet floor, an investigation can discover that condensation from equipment was the underlying cause. The equipment can be fixed and a more serious fall avoided.
However, controlling slips and trips from near-misses will not control unrelated major hazards. This was stated clearly in the Baker report on the BP Texas City Oil Refinery explosion in 2005 (bit.ly/2ku6UzQ) in the finding that "BP mistakenly interpreted improving personal injury rates as an indication of acceptable process safety performance at its US refineries".
This finding does not suggest near-misses should be ignored or that the triangle model has no validity. BP's mistake was in lumping together all near-misses and assuming they related to all accidents. The Baker report pointed out: "Many of the process safety deficiencies are not new but were identifiable to BP based upon lessons from previous process safety incidents, including process incidents that occurred at BP's facility in Grangemouth, Scotland, in 2000."
Investigating the right sort of near-misses could have prevented the accident, and we should be guided by our own risk assessments in choosing which near-misses we want reported and investigated -- and which we can be brave enough to ignore.