We have a criterion for putting an incident in the equivalent of your red category (referred to senior management) that can be described as "If this event happened again, would it be suprising if it was fatal?" (Multinational chemical company.) I agree, "It's never happened to us before" makes no sense as a criterion. You cannot judge the likelihood of rare events just on whether they have happened to you in the past. The sample's too small. You would need to look much more broadly at whether similar incidents have resulted in fatality, not just in your company / site but overall.

Estimating the likelihood of an incident can bring in a host of factors; one of which is previous history. However, this must be taken in context. The weight that you give to it depends entirely upon the circumstances. Being struck by lightning on a golf course does not mean that everyone one will, every time they go out, does it? But a high instance of electricians receiving shocks because no one bothers to isolate and test, would indicate that it was pretty likely to happen again - soon. The thing is with history is that high evidence of that event is pretty compelling but no evidence at all actually isn't. Not on its own. If there is lack of history, it only really helps if there is a valid reason as to why there is none. IE you have effective controls. Now, I am not entirely clear in your question if you are talking about the potential for an event to occur, or for the out come of the incident (injury) to be the same again IF the event is repeated. Let's take a car crash."We crashed 100 cars at speed and the results for the occupants were...." This really only works if the crashes are the same. And naturally with driving as an activity being risk assessed you can't actually predict the precise nature of the crash people might have. So, some has to be settled in the assessment itself so that context is clear. Valet parking of cars would be different to business driving mostly on fast road networks. How we approach this is by identifying the general type of incident that could occur - let's say loss of control at speed due to inattention, distraction or other driver's action, resulting in high speed crash. OK? Then we consider not the most common outcome, nor the best we'd hope for, but the reasonable worst. What is not beyond the bounds of credibility? In this case I have to say that fatal injuries of permanent disability would not be unreasonable. This does NOT mean that every crash will result in this, people will walk away at times, but we now understand the potential at its reasonable worst. Whenever there is an incident, we examine what we predicted against what actually happened and analyse why there was a difference, if any and rethink the potential we recorded in the first place. Now to be honest, for high speed driving, unless they make card that turn into massive beach balls on impact, fatal injuries will always be the reasonably foreseeable worst for us. But if you start to look at other tasks then this becomes quite interesting. Let's take a task that could at its reasonable worst result in a fracture to an arm or leg. Typically we experince less that this with only sprains and the like. Periodically we do get a fracture. "Ah hah" we say, "We thought that might happen" But just very occasionally we get something worse. And on examination there are factors that mean the elevated injury outcome were not predictable and in fact quite unique to the outcome on that day. We do record that, but we don't change the injury potential for that risk assessment simply because of that one slightly bizarre event. So in summary. Potential for an incident to occur again can be based on many things not just history unless you can show it occurred because of common control failures still in existence and that nothing has been done to address that Injury outcome can be predicted in advance given what we reasonably know about the universe and can be different to the actual outcome in many cases, but it's worth looking at why they were different. Risk assessment after all is largely about "What do we do?" "How could that go wrong?" and "What are we doing to prevent that from happening?" What isn't helpful though is to confuse likelihood as the likelihood of the INJURY happening again. It should be the likelihood of the incident, which could lead to an injury. In that, I would say the consultant is incorrect. It leads to erroneous conclusions. I have never been killed in a car crash, therefore 'risk' is low. That can't be right. I drive 40k miles per year in all conditions, under pressure, at the ends of the day - the chances of me having as crash are pretty high. And under the sort of conditions that I drive, the type of crash I might have could reasonably result in major injuries or death. That should be enough for my boss to sit up and take notice. He is not going to say, "Ahh but you haven't been killed so I don't buy that" is he? Jericho

unless they make card that turn into massive beach balls Obviously that should have read 'cars'. Can't imagine that card beach balls would really catch on. Where is the edit facility? Jericho

Thanks Jricho and chris, i just wanted to add that, in our risk matrix, if the incident fell in Yellow(normally if that incident happened in industry before) will only get middle level management attention and he may do something about it but taking an incident to Top Management table is based on red category and should be happened before or some bad thing like fatality or disability just occurred ..(our consultant view point)

There are two different likelihoods being talked about here. One is the likelihood of an event happening. The other is the likelihood that this event (given that it does happen) could have a major consequence. It's the second likelihood that is relevant here - this isn't a risk assessment of an activity, it's a classification of an event by its potential after it has already happened. It's quite right not to go on the worst consequence that could imaginably happen (as the papercut example shows). But nor is it enough to go on what happened last time, because that may just have been a lucky escape. Lucky escapes are good things to learn from. If you are left thinking after the event "It's lucky no one was killed", then that makes the potential consequence a major one.

I agree with the opinion that we should not go to worse cases for every single incident, however when we get involved in incident investigation and we find adequate evidences that this could lead to fatality , it should be classified as red, no matter it was occurred in project site or not.. One of the view is that, some of control measures were worked, thats why we did not have fatality( lets not say lucky) but when we have lack of details in investigation report and safety climate is not positive.