If the control measure that you introduce changes the severity, then I suggest 'yes'.

My understanding of the standard matrix for risk assessment is that it is based on 1) the degree of severity of harm AND also 2) the likelihood of harm (ranging from inevitable through probable and possible to remote) The matrix I use for training courses and guidance documents has 'degree of harm' as its vertical axis and 'likelihood' as its horizontal axis. I've found that splitting severity of harm into 3 main categories is the easiest for people to understand, i.e. The severest category is 'Death/Permanent Injury' (e.g. chainsaw removing a leg), the middle category is 'Injury requiring medical/hospital treatment' (which may require surgical operations and/or take a long time to heal, but the person eventually returns to the pre-injury state. The lowest category is 'Minor injury', one which requires no more than local first aid. Thus, situations/activities which land in the top left hand corner of the matrix box are most likely to occur and pose the greatest severity of harm. Ones which land near the bottom of the box and towards the right hand corner do not involve significant risk and therefore merit little or no effort in the way of measures to prevent or control them. It can be argued that each activity or situation with significant risk has TWO places in the matrix box. The first is its position without any precautions being taken, while the second is its position after identifying and applying all appropriate precautions. It could also be argued that reviews of assessments for controlled risk situations/activities should include trying to identify better feasible precautions/control measures than the existing ones. For example, using mains voltage inspection lamps in a commercial garage poses a high risk of harm i.e. death through electrocution which is fairly likely because the lamp cables are notably vulnerable to damage by vehicle tyres. The likelihood of such harm can be reduced by using RCD protection for the power supply to the lamps. In matrix terms this measure drags the actual risk well away from the top left hand corner of the box. However, as RCDs are fallible, there is still a slight residual risk even though the likelihood of harm has been greatly reduced. Therefore, the best way of managing the electrocution/shock risk is to eliminate it by using lamps which run at 12 volts rather than 240 volts. Hope the above makes some sense, especially with regard to having two matrix box positions - potential significant risk and actual controlled risk - for each activity/situation being assessed.

I agree with achrn that control measures can often reduce the severity of harm as well as its likelihood. In the case of activities with a risk of falling from height, the use of suitable harnesses plus other appropriate control measures, notably prompt rescue arrangements, are very likely to prevent falling persons from actually hitting the ground and therefore also prevent the related deaths or permanent injuries. In terms of the harm severity scale for my matrix, the measures realistically take the risk out of the death/permanent injury category probably to the lowest (minor injury) category. It's also worrying to read of achrn's experience that some people don't seem to understand risk management and therefore rejected his method statements and risk assessments on spurious grounds. Hopefully, the people involved were not OS&H trained or qualified.

All depending on: Correct use of harness (training) Suitability of anchor point Condition of harness distance of fall vs length of lanyard Just because a harness is provided does not mean that impact with the ground is eliminated!!!!!!!!!!! If you hit the ground following any failure from the above then the severity of the consequence is the same only the likelihood is affected.

There's a surprise...Chris discussing chemical exposure. :) There was a thread only recently where this subject was discussed to death. That said, I am going to go against the grain and state as a rule the severity remains the same, unless you have a RA with and without controls. In other words, the severity will normally remain the same if the control measures fail and to my knowledge most RA matrices use the severity without controls for a good reason. That is, not all control measures have the same impact. For example, a brick dropping from height landing on a person's hard hat will save them from a much worse injury, but a 1 tonne concrete slab falling from the same height will undoubtedly negate the hard hat and kill the person. Similarly, a hi-vis vest will protect the worker through better visibility, but it will not protect them if the hazard should materialise and they get run over by a 20 tonne excavator. Therefore the hi-vis vest reduces the probability (likelihood) of an injury, which is not the same as the aforementioned brick scenario. So, when you see a list of controls against the hazard it is clear not all of them have same impact, hence I believe the correct 'severity' is the one without controls and remains constant. RAs are often a lot more complex than some people fully appreciate when you get down to the nitty gritty. They are of course usually 'dumbed down' wittingly, most probably unwittingly, by the author. Now putting my hard hat on for responses - LOL.

???

Too be fair, it is not in most cases but only in the remaining cases where elimination or substitution takes place will consequence be changed. By the way the probability is not the risk it is the likelihood.

Not accordingly to the law of achrn

That provides an example of what is wrong with H&S today. I was using the word 'risk' in it's normal English sense, as defined by the Concise Oxford Dictionary: "Risk:- a chance or possibility of danger, loss, injury, or other adverse consequence..." No mention of the severity of the consequences. In my world one in the same, unless I am speaking swaheli, what do you think is the severity of a consequence Chris

Achrn stated "A fall arrest harness is, I think, an illustration of this - it doesn't affect the likelihood of falling off a high place at all, but it means that if you do fall off you don't die (probably - assuming rescue arrangements are in place fast enough). Think about what you said a bit deeper. Many a time I have seen guys with harnesses on but not connected to fall arrest systems or blocks, or life lines getting tangled as they move about, or they move outside the safe angle to actuate properly. Let alone guys hooked on with the fall arrest below them - give me strength!! The fall arrest system depends on the human understanding and complying. The human factor is one of the weakest controls we could possibly use. So my answer? The consequence will remain the same (high) if the control fails or is not used. David

No, not formally qualified. We work in construction, principally as designers. We regularly have teams going onto sites to do specialised inspection or survey, in order that the design of the solution properly addresses the situation to be addressed. We work in niche areas, and from experience relying on the main contractor to gather the information leads to gaps. It also doesn't identify the unknown unknowns - that is to say we might get the answers to all the questions we asked, but there might be something that's obviously relevant to us when we're on site, but we didn't know to ask about, and it wasn't obvious to the contractor or client that it would be relevant to us. We provide method statements to the main contractor, but they are normally approved (or otherwise) by the contractor's staff on site, who have an absolute veto (ie, if the main contractor says the method statement is inadequate and we can't do the work on site, obviously we can't do the work on site). If the resident chap on site has preconceived notions about what a risk assessment should look like or should cover or should not do, then we basically have the choice of go along with what they want or not do the work. So we make sure it has sufficient for what we need, and then has whatever else the contractor demands as well (be that formatting, layout, numbering, high/medium/low, whatever). The down side to this is that our people need to accommodate any format of RA. We have a suite of standard forms (three main types - a numeric, a numeric with explicit before&after scoring, and a qualitative) and normally one of those is close enough to the contractors preconceived notions, but occasionally we need to produce something laid out completely differently, or work to arbitrary rules like 'no control shall alter the severity score'. Both the person producing the MS and RA, and the people working to it need to be prepared to work with any format or layout of document. But I don't think this is particularly unusual - I would expect that anyone who does work for a larger organisation finds they end up having to modify their own processes and paperwork to satisfy the demands of the client. That's life.

I dislike RA's that give risk ratings before controls are put in place and after controls and after controls are put in place. What's the point? However, if you are reviewing a RA because additional controls have been put in place following the outcome of the RA then the answer depends. Often it is only the likelihood that changes but sometimes it is the severity that changes too. A fall from a ladder is a fall from a ladder despite control measures such as safe work practices, all you can do is reduce the likelihood. Hearing damage is still hearing damage even if you reduce the noise levels and therefore the likelihood. However, using a fall prevention measure does reduce the severity as does waering a seatbelt.

To answer the question, you need to be clear about how you use a risk matrix. Do you use it to highlight: * the most likely outcome, * the worst severity case, or * the scenario of greatest risk? I don't think many people are very clear on this issue. To take the fall arrest harness example. It does not prevent falls and the harness cannot be 100% reliable (even if used properly). Hence, it does not totally eliminate the highest severity outcome. But it does significantly reduce the likelihood of that outcome to a point that may be shown on your matrix to be very low. If you then look at a lesser severity outcome you may find it has a higher likelihood and comes out with a higher risk ranking on your matrix. This does not mean the potential severity has been reduced but that according to your matrix the risk to be concerned about is the one associated with a lesser severity. I think the point I am making is quite well illustrated in section 3.2.1 of this document from HSE http://www.hse.gov.uk/research/rrpdf/rr025.pdf Taking Chris Packham's scenario. If the less hazardous substance is equally effective at the job then the only reason to not use it if the greater cost is grossly disproportionate in comparison to the reduced risk. If the less hazardous substance is less effective I would argue you are actually changing the activity and so comparing risks is more complex than just looking for squares on a matrix.

Of course the discussion has taken the inevitable twists and turns, all good stuff, but surely the key issue of the original question is whether you can manage/reduce a risk by reducing the severity of the hazard or conversely whether reducing the severity of the hazard you can reduce the risk? i.e. can we change things to reduce hazards? While I agree and accept that many controls are concerned with reducing the likelihood (many of which rely on human performance) for me, reducing the severity, or the risk ‘at source’ should be at the very heart of our considerations. Yep, it’s Friday!!