Most health and safety professionals have come across Bird’s Triangle, which theorises a relationship between the number of near misses and injuries. Alan Field considers whether it is still relevant in controlling incidents.
Accident triangles — often called Bird’s (or Heinrich’s) Triangle — are based on a theory of industrial accident prevention, and particularly a human factors approach to safety.
The above is a typical pyramid or triangular representation of the theory. Essentially, it shows a statistical relationship between the number of major, minor and near misses, with the implication that if the number of minor accidents is reduced then the number of serious accidents will correspondingly reduce. In this iteration, for every six hundred accidents with no injury or damage there is likely to be one involving serious or disabling injury.
It should be remembered that accident or Bird's Triangle are often generic terms for updated or business specific versions of triangles that have been created, ie they don't necessarily use the original probabilities suggested by Heinrich or Bird, ie there might not be a serious injury accident for every 10 minor injury accidents. To complicate matters further, some triangles will also show a separate figure for fatalities and some will show a distinction between unsafe acts and near misses.
The number of different versions is explained by their history. The version of the Triangle was first postulated in 1931 by Herbert W Heinrich, in his book Industrial Accident Prevention, A Scientific Approach. Then, in 1966, this was updated by another American, Frank B Bird, who then refined this model over a number of years along with other theories relating to accident causation.
While we will refer to Bird's Triangle in this article, the analysis will apply to other versions and iterations that can be found in the generic accident triangles.
Is Bird's Triangle still relevant today?
The simple answer is yes.
However, the longer answer involves some qualification.
A significant advantage to Bird's Triangle is that it provides a strong visual representation of the relationship between near misses, minor accidents and the potential for more serious incidents (bearing in mind that incident is a term that is more likely to be used today).
It can be used for any level of staff and as a training aid, clearly communicates that even relatively minor actions can have serious consequences.
Accident triangles can influence staff and wider organisational thinking on risk exposure, particularly where there might be an unspoken toleration of unsafe acts in some work environments. It provides a talking point for further risk assessment and understanding the complex factors that leads to safety incidents arising. Most importantly, Bird's Triangle shows a link between a certain volume of relatively trivial incidents and a more serious one. In other words, accident triangles can lead to positive change. They can also, extrapolating statistics from serious incidents, indicate that near miss reporting needs to be improved.
Limitations of Bird's Triangle
Yes, Bird's Triangle does have limitations.
Firsty, the statistics on which Heinrich based his theory are hard to verify now. Bird's statistics are based on insurance statistics available to his employer in the 1960s. That means this ratio of near misses to serious accidents is, today, indicative at best.
Even if the statistics were valid, the triangle assumes that an individual organisation's datasets are complete, which is only true in some cases. In other words, serious incidents will almost certainly be known, but perhaps not all minor injuries will be recorded and certainly complete transparency of all near misses cannot be assumed. Might two serious injury accidents over a defined period of measurement with only a 100 or so near misses reported indicate an under-reporting of near misses?
Secondly, datasets can also be skewed by the fact that not all near misses or unsafe acts are of the same gravity. So, in an individual case, if an accident had occurred — rather than the reported near miss — would the outcome have been a minor accident, a RIDDOR-reportable incident or perhaps even a fatality? It assumes we lump all near misses together, regardless of possible outcome.
Of course, sometimes, it is difficult to tell how serious a near miss could have been, especially if the near miss triggered a chain of events that did lead to a serious incident. This domino effect of failures is the so-called “Swiss cheese” event which, again, Bird's Triangle doesn't overtly take into account.
Thirdly, Bird's Triangle is about actual event outcomes. It doesn't particularly consider risk potential, ie that the possibility of certain serious incidents occurring may not necessarily be indicated by a higher volume of near misses. Equally, reducing the number of minor incidents doesn't necessarily, in practice, reduce fatalities.
Fourthly, Bird's Triangle assumes human factors in safety are the predominate, probabilistic indicators of accidents and also assumes this is chiefly based on the behaviours of workers. However, this does not take into account factors such as management systems and management decision making. For example, a serious incident might arise due to, say, a design flaw in a machine or a failure of management to have a maintenance contract for the machine — neither of which are related to operator non-compliance at all. This is a significant limitation, although some modern iterations of the triangle attempt to include management activity. So, when looking at a Triangle for the first time, check whether all staff activity is included within it.
In fairness to Bird, his later work did consider theories such as multi-causal theories of incidents. This refers to the fact that there can be more than one cause of an accident, ie another, unrelated event occurs at or near the same time as the first, which turns a minor incident into a major category. This is sometimes referred to as a “Swiss cheese” incident, reflecting the way that the failures in safety systems and controls can line up like the holes in a Swiss cheese.
So, Bird's Triangle assumes accidents arise in a sequence of events that can be influenced by an intervention — rather like removing a domino or two from a shuffle — the single cause domino theory of accidents. Bird's Triangle has little impact on multi-casual events although some might argue if the first event is stopped then that might disrupt the whole chain of the two or more events that leads to a disaster.
Multi-causal theories are influenced by different risks, so risk assessments need to look at a possible combination of unrelated or simultaneous events. The key point is the accident triangles don't really consider multi-causal events at all.
Bird's Triangle (and its many variants) is still a useful tool today, especially for communicating the importance of safety.
Bird's Triangle has limitations, particularly in terms of how accurate — in real terms — the statistical correction between near misses and serious injury accidents is. It is a guide rather than an absolute.
Bird's Triangle does not consider failure of management systems as part of its data, ie all accidents are the cause of operator non-conformance (be this lack of training or adherence to procedures).
Bird's Triangle assumes accidents arise in a sequence of events that can be influenced by an intervention — the single cause domino theory of accidents. They have little, if any, direct impact on multi-casual theories of incidents, ie where an unrelated event occurs at the same time which turns a minor incident into a major one.