That got your attention, I saw on a recent thread on accidents a reference to the Heinrich model of accident causation which was the grounding for most behavioural based safety initiatives. Most of us will also know or recognise his accident pyramid which goes along the lines of: 1 Major incident 30 major injuries 300 minor injuries / first aid 3,000 near misses 30,000 at risk behaviours and various themes, different numbers and categories which have since cropped up. Bread and butter stuff for the aspiring safety advisor, no? But did you know its all a fallacy? Well that may be subject to debate and be industry specific, but it reminded me of a recent article I read only last week published by the national safety council: "Safety professionals generally agree that Herbert William Heinrich had a significant impact on the practice of safety, but whether his influence was positive or negative remains a subject of debate. He was a pioneer in bringing attention to workplace safety; however, to describe what he did as ‘research’ is questionable. Heinrich is best known for his 1931 book, “Industrial Accident Prevention: A Scientific Approach,” in which he said 88 percent of accidents are caused by “unsafe acts of persons” and put forth what often is referred to as Heinrich’s accident triangle or pyramid: In a group of 330 accidents, 300 will result in no injuries, 29 will result in minor injuries and one will result in a major injury. ...I do think what people don’t realize is that it was flawed data to begin with... ...It has misled people running safety programs into thinking that if they work on minor incidents, major incidents will go away..." Read more here: http://www.nsc.org/safet...ngthefoundation1011.aspx My personal opinion is that in some organisations behavioural based safety can result in capturing and correcting unsafe acts which could lead to major injuries. However it is easy to focus too heavily on these behavioural risks, lose sight of the underpinning root causes (not often the individuals fault) and forget to control the major hazards out of the individuals control. Too much for a friday so I thought I would get this in early, hope you find the article interesting. Des

I have recently commented on another thread about the H/B pyramid. It's strange, but whilst I was writing the post I wondered if the metaphor had done any good or whether it had done more harm than good. Incidentally, it is amazing how many non-safety people ie project managers, agents, engineers, etc know of the H/B pyramid. I guess it must be included in the IOSH managing safely course or similar. It could be argued the pyramid has been responsible for this incessant desire to capture near miss data. Whilst we all appreciate there is a correlation between fatalities, major accidents...near misses, one would only have to look at the HSE accident data to be aware of that! So, does the H/B hypothesis stand up to scrutiny? It has its uses but I don't think it's particularly relevant in the world of accident causation.

The Swiss Cheese Model has now been seriously questioned. It purely looks for a management failure. If prevention is the aim of an investigation then you need to look closer to the antecedent behaviour. I would suggest reading the recently released book, Redressing the balance: A common sense approach to causation. This was written following a ten year study by the Institute of Industrial Accident Investigators. If you still believe the SCM is for you then fine, you may have a different view.

I saw the same thread - didn't comments but have thought more about it, so here we go. I have always found the numbers used in the triangle a little hard to except as I can't imagine any organisation is aware of all it's near-misses and therefore corrolating the near misses to accidents would be impossible to do. However, ignoring the numbers, the principle I have always taken from the triangle (which is supported to an extent by the swiss cheese diagram) is that you are lucky more times than you are not - if you can identify when you have been lucky you can make the defences better to prevent the time when you are unlucky from occurring, which is why near miss reporting is vital (IMHO). The same principle applies to the swiss cheese model - when the holes don't line up exactly, that is where your near misses lie and there are many more combinations of these than there are possible combinations for accidents - so I suppose in that sense it is more apt to say that near-misses are more probable rather than a luck issue.

Fallacy may be a bit harsh but certainly nothing more than a theory. As with many things, this theory has morphed into something that it never was intended to be. The simple message from the triangle has always been that if you ignore the apparently inconsequential things then you miss the chance to minimise losses, not prevent them. This message has been lost in the development of the theory into ‘if you focus on the inconsequential then you will better prevent high consequence-low probability events. Therefore you no longer need to manage that end of your spectrum as closely because your ‘data’ says no problems.’ So from that point of view alone I agree with the opinion of the referenced article that it has become a constraint rather than a tool and should not have the prominence in thinking that it apparently still has. As a slight aside and hopefully not to derail the topic, I was very interested to read the following as one of the responses in the referenced article. “While not downplaying the importance of research, Main cautioned against arguing about the probability of a hazardous event occurring. “When you do risk assessments quite a bit, you figure out that there is often much less controversy over how we’re going to reduce the risk than there is over the ratings,” he said. “If you can get over that hump and focus on risk reduction, you can avoid the discussions and all the hand-wringing.” Now that is an opinion that would make an interesting theory of practice don’t you think? P48

Pete, I have always though that probability is irrelevant or at least not a significant factor due to the variables in probability. Sometimes the frequency of a task gets confused with probability, it follows the more you do something the more likely an adverse event will occur. However, that is not the same as probability, which in itself is a random event and could strike at any time.

I started a thread once before about Heinrich. I can't find it but I received a very interesting and enlighting reply that the triangle is different for different hazards. For example it would be a wide shaped triangle for slips, trips and falls, and a rather narrow pyramid for high voltage electricity. I wonder if this could be taken further such that each hazard has it's own shape - not strictly a triangle. So, falls from a very high level would probably be an upside down triangle where as falls on a level would probably be a normal way up triangle. Manual handling might be an upright elipse shape with few minor injuries and few fatalities but lots of over-three day injuries in the middle. Just a thought!

Oddly enough I have seen people concentrate on the top and bottom of the pyramid and leave the middle to itself. The most worrying triangle / pyramid I have ever seen had lots of near misses at the bottom, a couple of major injuries at the top and almost nothing in the middle. Looked like they were doing a good job of collecting near miss and minor injury data and then doing nothing about it!!