By Nick Bell
Over the years I have met several training delegates and auditees who told me a similar story: “Our safety department won’t let us reduce the severity rating on risk assessments”.
These people worked in a host of different industries: Construction, rail, manufacturing and the public sector. They were using risk assessment methodologies that required them to score severity and likelihood.
It happened again recently and I decided it is time to call out the industry on this issue.
I urge readers to think about their own stance: Is it sometimes possible to reduce severity of harm as well as or instead of reducing the likelihood of harm occurring?
Anyone with a reasonable grasp of the principles of prevention, a cornerstone of how we manage risk, will hopefully say “yes” and have a number of examples. I offer just a few of my own.
I once undertook a health and safety survey at a galvanizer which had historically used a strong bleach solution to clean metal work prior to dipping it into molten zinc. In the past, staff had come into frequent contact with the solution and suffered damage to their skin. The galvanizer had redesigned the bath, changed the dipping process and altered the PPE regime to reduce the chance of people getting the solution onto their skin (i.e. reduced likelihood of harm occurring).
The managers trialled different concentrations of bleach solution and found that a much weaker (and much cheaper) concentration was equally effective at cleaning the metal work. In the event that the liquid came into contact with people’s skin it was far less harmful (i.e. severity was reduced). This was supported through improved welfare arrangements enabling operatives to promptly wash off any solution.
We can also tackle energy sources on a number of fronts. For example, I was inspecting a manufacturing site and noticed a very noisy motor (feeding air handling equipment) above a production area. It could not be moved outside (it would cause a nuisance) and it was uneconomic to replace it with an intrinsically quieter motor. A noise assessment was instructed and presented a range of options. The motor could be suitably maintained and have acoustic treatment to reduce the amplitude of noise it produces. .
Duration, frequency or amplitude of noise exposure could be reduced by effective organisation of the workplace (e.g. positioning the motor a suitable distance away from longer duration jobs, such as stacking, wrapping and packing or vice versa). The employer might then consider the use of PPE. Routine monitoring of the arrangements and audiometry testing would identify if controls are slipping. Collectively, these measures will serve to reduce exposure to the hazard, the likelihood of harm resulting from that exposure and the potential extent of that harm (i.e. severity): We would not reasonably foresee rapid and significant hearing loss. 
The whole premise of fall protection systems, such as nets, is that while they do not reduce the likelihood of someone falling, they reduce the consequences of a fall. Someone could obviously still be injured but far less severely than if they fell onto a concrete floor.
Behind this instruction may be a well-meaning sentiment that we should reduce the likelihood of a hazardous event occurring rather than protecting against the consequences. I concede there are situations where the foreseeable outcome will always be dire (e.g. a building catching fire).
However, a dogmatic stance that severity cannot be reduced is more than just a technical inaccuracy: It undermines effective and proportionate risk management. These are some of my concerns.
If you are at loggerheads over this issue, it may be useful to talk with workers and managers to get their views and collect examples. A few photographs, perhaps turned into mini-case studies, could become useful discussion starters and training aides. This will help us adopt the role of credible, solution-focussed enablers and coaches. Potentially, there’s a discussion to be had about how useful it is to numerically score risks but that’s one for another day.
You must be logged in to post a comment.
Nick – your bleach example is spurious, as you have not reduced the severity of the original risk that was identified. By changing the substance from the original, more harmful bleach; you’ve simply replaced one for another. the severity of harm from the original, stronger bleach was not reduced. When working at significant height, where falling and hitting the ground could lead to death; stopping someone from falling firstly and hitting the ground secondly is what is required. Stopping them from doing so, does not reduce the severity if the risk controls failed.
If a significant hazard is identified and the risk outcome is assessed, it is managing that risk and preventing the harm from the hazard identified which is crucial. Substituting battery or lower voltage electrical power, does not reduce the severity of contact with 240v electricity. It is simply best practice, so long as the equipment can operate as effectively as the higher voltage type.
I suspect the complaints were from people who were forced to use quantitative risk assessments, which I hate. The time spent arguing about the calculations is a waste of time, when competent and experienced personnel can almost always work out the hazard and potential severity, then what controls are needed, without having to resort to numbers.
Risk assessment should be a simple, cost efficient and practical method to manage harm within the workplace. Unfortunately, too many OH&S professionals try to take things to the Nth degree and turn everyone off the process. Keep it simple and get everyone involved and encourage personal risk assessment as part of the daily routines.
The confusion stems from how the risk is stated. If the risk is stated as fall from height, then the net would reduce severity. However falling does not cause injury, the injury is caused because of a fall from height. Hence the risk should be correctly stated as “hitting the ground after a fall from height”. Hence a net would reduce the likelihood of hitting the ground. However one does want to prevent even the fall from height , hence the harness. So the implementation of sequential controls over time where the probability of the first control failing creates the risk for the second control.
Severity controls is mitigation, like first aid after the incident. Or vaccination that may reduce likelihood of a disease occurring after even after an exposure but sometimes is given after an incident like an anti-tetanus shot.
So it depends on which risk you are controlling. If we are analysing across a span of time, the later controls appear to reduce severity of the original risk but they only reduce reduce the likelihood of the later risks, depend on which risk you have stated in your risk assessment.
Therefore carcinogen exposure, radioactivity or noise exposure where the more severe consequences accumulate over time, often stating that controls reduce severity makes more sense than rewriting new consequences that arise from the original hazardous activity over time and new likelihood reduction measures to be put in place over time.
I agree with Bob regarding the original risk still existing.
In many of these examples, the original risk (with the higher hazard exposure level) still exists, and a new risk (with a lower hazard exposure level) has been introduced. Both should be analyzed and controlled appropriately.
I agree with Sumita regarding other risks also existing.
In many of these examples, the original risk exists, and a separate later risk also exists. Both should be analyzed and controlled appropriately.
Underlying both of these comments is the identification of multiple hazardous situations where each of these should have independent analysis, assessment, and evaluation (including reasonable risk reduction, identification of residual risk, and identification of benefits that outweigh the risk(s)). Identifying hazardous situations correctly is critical to managing risks.
The practices below potentially compromise the completeness and accuracy of risk identification, assessment, control, and evaluation:
1) Using a risk control to convert one hazardous situation into another (in addition to or instead of the normal risk reduction impact) when the hazard exposure levels and risk controls may be different.
2) Combining different risks into a single analysis, when the hazard exposure levels and risk controls may be different.
Bob also seems to be touching on the fact that reasonable risk reduction should not be wholly (or even significantly) subservient to a process that generates numbers on a chart that are potentially invalid (inaccurate or consciously/unconsciously manipulated). I agree with this point as well.
Engineers, Industry, and Regulators should generally know what reasonable (SotA) risk reduction measures are via compliance to applicable product safety standards, comparison to design of predicate/equivalent/similar/competitive products, or original research for new technology. This is true regardless of what the residual risk is assessed as (and regardless of how this residual risk compares to the benefits of the device). We are expected to assess risk (ie determine what the risk is in relation to other risks). Trying to make the assessment of risk (what a risk is) be the same as the evaluation of risk (how a risk is accepted, factoring in reasonable risk reduction and Benefit vs Risk) compromises the completeness and accuracy of risk control and risk evaluation.
Blindly using these numbers (without consideration for what is reasonable risk control) can incentivize both waste and negligence.
This problem has been exacerbated in recent years by a disturbing trend in the development of product safety standards. This trend has been replacing specific requirements that industry has established over time (and that industry should continue to develop or modify over time) with allowing manufacturers to simply “risk assess” their way out of compliance with specific requirements. This actually defeats the purpose of product safety standards in identifying SotA risk controls and this incentivizes manufacturers to leverage numbers on a chart instead of engineering expertise on what is actually reasonable risk control.
Hi John,
I totally agree with you on the risk assessment issues, just one thing that most of safety professionals do wrong, especially when you create a Risk Assessment on a product or plant/equipment you manufacture and sell to a customer is to use the terminology “Residual” Risk” In legal terms when you submit that Risk Assessment to a customer means there is still risk remaining on the product you supplied. Rather use the word “Controlled” risk in the document, which means that your risk assessment team has put controls in to as low as reasonably practical. This we had in a fatality hearing in court where the judge said Residual risk means you did not controlled all the risk! The other point that the company was nailed on was the email that supported the Risk Assessment to the customer read “if you need anymore information on this Risk Assessment, please do not hesitate to contact us” on which the judge replied that this also means that you did not give all the information to the customer. There was lots more that needed to be changed on our risk assessment formats after that case, but to much to explain here.
using safety net u simply prevented the fall to the ground. the original risk was falling from height and hitting ground.
If you’re calculating residual risk from the likelihood of controls failing, then all risk assessments should result in extensive severity and frequent likelihood outcomes. For example, I would never allow an operator to use a harness and lanyard if I understood there was a potential for the anchor point, or the PPE to fail. That risk would be captured and mitigated against. Humans also wilfully breach control measures; shall we then say that likelihood can increase due to the unforseeable nature of humas and human factors?
Hi Bob. Thanks for the feedback. I think quantitative risk assessments could be Marmite for our profession! I certainly agree that we need to keep things practical and use the knowledge and experience of the people involved. The point I was making with the bleach is that I believe the company chose to (in the words of the principles of prevention) “replace the dangerous by the non-dangerous or the less dangerous”. COSHH also asks us to consider whether there is a safer form of the product that we can use. My concern is that, despite these principles, people are being told that reducing dangerousness is not possible or is not a valid strategy.
I think it is really useful to discuss the possibility of controls failing: What if someone accidentally ordered the previous, highly concentrated bleach solution? What if someone had not installed the safety nets properly and the inspection regime failed or what if someone had maliciously damaged the nets? A good risk assessment will – excuse the pun – tell us how our important and reliable our safety nets are and how many we need. As you say, a bad risk assessment will take this to the Nth degree.
I tend to follow the simple mantra, that severity can change, but likelihood is more probable overall when looking at most of the control options over a whole general risk assessment. As long as the 5 steps, suitable and sufficient and principles of prevention (including specific versions in other hazard regulations like COSHH) are adhered to and that the controls work, job done! The rest is academic indulgence… by which time your employees have started to pull out all their remaining teeth!
I often find people conflate the meaning of likelihood and severity together.. Unless you have altered the intensity and original origin of the hazard, severity will stay the same. This accounts for issues such as adjustable guards failing, and the risk of the moving blade shattering or being still potentially exposed. Therefore such a guard reduces likelihood of contact, not severity.
I accept that with control measures, you can introduce new ones e.g. based on the MHO interpretation of avoidance if the pure elimination can not be introduced; that the use of mechanical aids might ‘avoid’ or at least considerably reduce manual handling elements (likelihood and severity); but now you have introduced mechanical aids that might kill or cause serious injury. But the collective approach on balance, might be more desired, based on the higher likelihood of harm occurring from manual handling injuries than mechanical failure etc.So you would undertake a machine or crane assessment separately to cover this problem.
Equally, if you use nets, and scaffolds, you are addressing both likelihood and severity. But there is room for error in assembly too. But that is why a basket/hiearchy of measures are used. The administration controls are there to ensure the more collective and effective technical solutions are used properly. Otherwise you are going to create many rabbit holes and endless evaluation of risks.
With the chemical example above, of course severity has been reduced! As you have combat the risk at source, and replaced the dangerous with the non/less dangerous. Likelihood has not changed from contact though. If you wanted to dilute the product instead, rather than a non toxic based option, it would only be Reduce (as apposed to an admin control), if this was done at the supplier end. i.e. it arrives in a different form, not diluted by someone at work who might make an error in dilution! If we can not agree on this, then the existing regulatory principles of prevention and hierarchies in MHO and W@ H; COSHH , become meaningless.
I agree with your comments, particularly the marmite I do feel a lot of people engaged in health and safety bring our profession into question when not engaging a proportion of common sense
Nick, very good article and a well presented argument and could not have out it better myself. An obvious example of a reduction in severity ratings can be seen with car design. The introduction of seat belts in the 1980’s saw an almost overnight reduction in the type of injuries sustained. As a front line medic I witnessed a drastic reduction in major chest and lower limb trauma. Similarly, the air bags now fitted have reduced severity of injury further. Vehicles collisions that previously had the potential for a fatal outcome we now find car occupants walking away from them. However, we must of course look at this as a hierarchy and treat controls that reduce the likelihood of occurrence in preference to the controls that reduce severity. In my context, technology that eliminates any risk of vehicle collision will supersede the need for air bags and seat belts.
I also think we need to be careful with the term quantitative risk assessments because in their true sense, quantitative risk assessments are an essential tool in safety engineering and based on calculated failure modes. If we are referring to risk assessments with a numerical score system (as an excuse for quantitative element), these may be useful as reassurance to the lay risk assessor if using a simple 5 X 5 matrix, but other than that, complicated and in particular tripartite scoring systems (hazard x likelihood x severity) are often nothing more than ‘guesstimate’ smoke and mirrors processes to cover what is a simple process.
HSE themselves have stated in previous guidance, that it’s important not to “get too hung up on the numbers, it’s the process of doing the risk assessment that’s the important thing, not the outputs”. Everyone’s viewpoints differ slightly, so unless one person is doing all the risk assessments (and the person best placed to assess risk is the person doing the job) there will be variance anyway. I’m no fan of the 5×5 matrix (or any odd sum matrix where you have a middle option) because I prefer there to be an active choice to place the assessed risk “above the middle or below it” just to further assist with prioritisation. I also agree with the premise of the article that of course you can reduce severity. It’s why pressure vessels are tested using water or other hydraulic media rather than gas.
Couldn’t agree more!
I have had this same , sometimes heated, discussion with ‘Safety Proffesionals’ and some just don’t seem to get it.
Bob, you state, “you’ve simply replaced one for another. the severity of harm from the original, stronger bleach was not reduced.”
It was reduced, because it’s not there anymore! The risk asessment is to identify how to make the TASK safer, not the (in this case) substance. Nick has shown that. They looked at a task that HAD to be carried out , improved PPE , methods, first aid provisions etc. etc.
But in the end, there is still the CHANCE of coming into contact with the substance. So by substitiution with a less harmful substance the possible severity is reduced.
Surely that is obvious?
Lets take that a stage further… for argument sake lets say the original fluid used was hydrofluoric acid? That stuff will kill you. Eat through your skin, through your PPE , the floor etc. So it has high risk rating yes? The severity is likely to be fatal.
We replace it with citric acid. However we DON’T CHANGE another thing?
Same process, same handling methods, same PPE, same training…
So the POSSIBLE SEVERITY is reduced by substitution of a less harnful substance.
But the LIKELIHOOD of coming in to contact is still the same.
The task has been assessed and the task has been made safer.
The LIKELIHOOD however can still be the same.
Seems obvious to me…
100%agree. An even simpler example of reduced severity is the use of a shock absorbing lanyard. The message to managers doing risk assessments should be apply your hierarchy of controls. If you find that you are unable to reduce the likelihood before you get to PPE (which does happen), then choose PPE that will reduce the severity. A shock absorbing lanyard is intended to reduce the consequence of the fall, even if you have been unable to reduce the probability of the fall occurring. Nebosh concede this to be true in their course material.
Nick
A good piece and a point always debated, I think the important part here is if we don’t change anything then the outcome remains the same, however if we go back and look at what we can do to change the process or what mitigation’s can be implemented to improve/reduce risk is good. At that point you can then look at the likelihood / severity that was set at the first pass and see what can be changed. What a lot of people tend to do is look at the task without looking at mitigation in place.The bleach example is a good one …. if the risk is metal falling into bleach and splashing personnel … have we stated that the personnel are kept outside a “splash zone”, they have full ppe on there are barriers in place. etc etc… What I have seen is that we look at the risk as a headline without looking at the full task with current mitigation’s in place and setting the likelihood/severity at that point. Then you can ask what can be further implemented to try and reduce the likelihood etc …. I guess in short its the continuous improvement process.
Nick, it is very interesting that in each scenario that you use, you actually reduce the potency of the chemical and the noise of the machine. Now, if you reduce sulphuric acid to vinegar, then yes the severity will of course reduce. If you reduce the noise (by whatever means) to ALARP then again, of course you will reduce the severity.
Here’s one to consider, a engineering firm has a guillotine machine, the staff are protected by state of the art mechanical guarding and robust managerial processes. Does the severity reduce?
Steve I would say of course not.
But I dont think that is what Nick is trying to say. He is making the point that severeity CAN reduce in certain circumstances and with appropriate controls. The old belief (and I’ve heard this uttered by CMIOSH members) is that severity CAN NEVER reduce.
That’s blantantly wrong and such an intransigent position only restricts and stymies the development of control measures.
Safety management is not something set in stone. It changes all the time, is fluid and adapts to different circumstances.
Mechanical guarding will reduce the likelihood of harm through contact with the guillotine. The severity of harm will remain the same if the guard is bypassed or fails.
For example: An individual walks across a construction site and is struck on the head by a falling object (no head protection) resulting in serious injury. The person behind them is struck by a similar object (head protection worn) and suffers a less serious injury. We have reduced the severity of harm through the use of personal protective equipment (the last resort) We should have reduced the likelihood of objects falling in the first place through not storing at height, placing barriers to stop objects falling……These measures will reduce the likelihood of harm as it is then less likely that objects will fall. Head protection (PPE) would be used in these circumstances to cover the residual likelihood that a barrier or management control i.e. a rule could be ignored. We have reduced both likelihood and severity of harm in this example through the use of simple qualitative assessment
Severity remains the same. The control measures reduce the likelihood of persons being exposed to the hazard, but should these fail, they will still be exposed to the same hazard; in this case.
Are we saying that a hard hat will not reduce the severity of harm through exposure to the same hazard? The severity of harm will be lower if the item of PPE is worn correctly; in this case.
It reduces the likelihood of harm, but there is still the possibility that the hard hat fails to sufficiently cushion the impact. Just because it makes it more likely that a less severe outcome is the result of an incident, does not make it impossible that the most severe outcome occurs. It just makes the most severe outcome less likely. The severity would only be changed if the falling material is changed (e.g. if a block of concrete is replaced with a block of foam).
I fully agree with Nick that one can not only reduce the likelihood of an event but also its severity.
This is a fundamental principle of safety management – it’s even taught on introductory courses such as the IOSH Managing Safety – so how on earth can a “safety department” argue that the severity cannot be reduced?
It seriously calls into question the basic competency of the people running such “safety departments”. If they give the wrong advice on this key topic then what else are they getting wrong? How much damage are they causing to the profession of health & safety management?
If the controls fail, the severity is the same. By implementing controls you are looking at reducing the likelihood of the event occurring
I have to say that the arguments put forward by ‘Bob’ seem to be arguments for the sake of arguments. Even the most basic of risk assessment training courses teach that reduction in overall risk will be achieved in one of three ways:
1. Reduction in likelihood
2. Reduction in severity
3. Reduction in both
If a dangerous substance is replaced is replaced by a less dangerous one then the severity has been reduced. while I agree that the severity of the original bleach has not been reduced it is not being used having been replaced by a substance with LESS severity. Makes sense to me but then I’ve only been in the safety industry for 28 years so what did I know
Health and Safety risk management is a practical subject reflecting the real world of work. We impose theories and their associated definitions in an effort to communicate our ideas about how to create and maintain a workplace that is safe and free from danger. We define risk as severity of outcome multiplied by frequency of exposure. The notion that you can’t alter the “severity” in a given risk scenario is a truism. In the examples given I would say that the risk scenario is changed in the first (chemical) and the frequency of exposure is changed in the remaining two by implementing controls, i.e. noise damping and fall arrest equipment. The severity of outcome, deafening or collision with a concrete floor remains however the likelihood is lessened by the application of suitable controls. In practice the reality is that failure of the controls will potentially lead to the sever outcome identified in the risk assessment and not any lesser outcome. The onus of care is to ensure that the controls are implemented for the duration of the exposure. Great query though.
Great points Nick. There is a huge gap in the understanding of risk assessments and they are too often treated as purely documents to be filed away.
I have had the experience of what you describe where the safety department of a large infrastructure contract insisted that because I had reduced the risk of electrocution due to contact with 25kV cables, I had to also show a reduced severity in the risk matrix otherwise they would not sign off the risk assessment. Their logic was that before controls, touching the wire would cause death but after the controls to minimise the risk of reaching the wire, touching it would only cause serious or minor injury. I could not follow that at all. As I walked away from the job because of the impasse I don’t know what happened there. I suspect that they found someone more biddable.
Nick,
A tad confused, perhaps? The whole point of asserting — absolutely correctly — that severity cannot be reduced is to oblige the assessor where appropriate to re-address the hazard. Only by eliminating, diminishing or substituting the hazard will severity be reduced. The replacement hazard may or may not have less severe consequences associated with it. But the fact remains that had the original hazard been allowed to persist, reduction in severity would not have been attainable. To be effective the assessment process must be iterative in this way and not lose focus on the hazard. That’s why we assert that for any given hazard severity cannot be reduced. Remember the good old hierarchy …
Nick, I wholeheartedly agree with you. I have had this stance for a number of years and have managed to change many view points over the years among my fellow professionals and new work colleagues at various locations and industries.
This subject for many H&S professionals seems to have been a case of “the emperor’s new clothes”, I’m really glad that you have aired this subject and look forward to seeing the feedback from those who standby the stance that you can’t reduce the severity.
Excellent article!
Regards
Paul
Hi Nick
I agree with your observations. Working in engineering construction for many years, its surprising to find that even now I come across many HSE professionals who simply cannot accept that reduction in severity is achievable.
When debating the issue I tend to use the ‘fall from height’ scenario.
Let’s assume a worker is on a second storey roof. Of course it is always better to prevent the worker falling but let us say that there was no edge protection in place or fall prevention/arrest harness being worn by a worker. The worker trips and goes over the leading edge falling from height. Upon impacting the concrete below it is very likely that the worker would sustain life changing injuries or even die as a result. A very high severity rating. In this case you can install control measures that don’t prevent the fall but will reduce the impact should that person fall. Safety catch nets and drop bags for example. As I say, these controls do not prevent the worker from falling but reduce the severity of the fall by catching the person before impact or greatly cushioning the impact.
It’s a simple scenario that challenges the mindset of the non-believers and usually gives them food for thought during future assessments.
Great article and stimulating it seems… thanks
I agree partly with Nick’s views, but I am quite surprised to hear that there are some advisors who state that severity cant be reduced. I have always taken the stance that addressing probability is much easier than addressing severity – especially if talking about task-based risk assessments (as opposed to during the earlier design phase – when severity is a bit easier to address. However, when it gets to the task risk assessment, it is really a case of trying to manage aspects that couldn’t be designed out – and often, it is the probability that can be addressed at that stage. However, I have never myself, stated that reducing severity is impossible its just that it is a lot more difficult to do at that stage. Regarding the possibility of ending up with all the residual risks as high or medium, that is also sometimes due to unsuitable risk matrices being used (e.g. 3 x 3 matrices – which don’t allow enough flexibility to handle inherently hazardous activities, such as work at height, live electricals etc). I find a 5 x 5 matrix provides enough flexibility to differentiate residual risks so that the ALARP ones can be separated from the Low residual risks – which is useful in focussing peoples minds on the aspects that may really hurt if the expected controls break down in practice. Regarding the fall arrest example as reducing severity, it can also depend on the wording used to describe the hazard. If the hazard is falling, with the potential effect of fatality, then I would say it can be argued that a fall restraint will reduce the probability of death, although it wont however reduce the likelihood of falling. Having said this however, the key point in doing the risk assessment is in identifying the hazards and reducing the risk. Whether it is the severity or the probability could sometimes end up being a moot point to a degree – the net benefit is that the hazard is identified & reasonably practicable controls are identified & most importantly, put into place & actually used. As sometimes happens of course, the risk assessment is seen as an exercise to do, before just getting on with the work & sometimes not all of the identified control measures get implemented as expected. A piece of paper by itself, hasn’t often reduced injuries – regardless of what was written on it. (e.g. Chamberlin’s “Peace in our time” statement).
Thanks everyone for the comments/feedback (and for confirming that some health and safety practitioners are adamant that severity cannot be reduced).
I can see that in some scenarios, we can’t make the hazard less potent. However, if I take an example like the Foodles Production (UK) Ltd prosecution, I can see how even machinery can be made less dangerous. If I was designing the Millennium Falcon doors I could make sure they came down at a leisurely pace, I’d construct them predominantly from soft wood or aluminium rather than steel, I’d have a rubber pad on the bottom and could put a sensor in the pad so that the door stopped if it encountered an obstruction such as the floor or Han Solo’s leg. The likely outcome of contact would be bruising rather than a break which would take our hero out of the Intergalactic fight for many weeks.
Then we can play the ‘what if?’ game…what if all those controls failed? Imagine we were erecting scaffold in a public area and some standards were going to be exposed, and could be walked into by members of the public. If we don’t think severity can be reduced we should tell contractors only to wrap them in high visibility tape and not to bother with foam padding because that doesn’t really reduce the severity of harm. That seems unwise to me.
If we go down the line of always arguing that controls don’t work we will never, ever be able to reduce risk: “You say that you have reduced the risk of a fall by replacing ladders with mobile tower scaffolds…but how do you know people aren’t using ladders anyway or erect/use it correctly or there is damage/defect that has been missed etc. – I can’t possible see how you have reduced the likelihood of a fall”! We would then talk about our training, inspection and monitoring regimes etc. but what if they fail too?
I think I am understanding what the cause of the differences of opinion might be:-
People who believe severity cannot be reduced might be assessing the hazard.
The law says that we must assess the risks. This requires us to imagine how we can come into contact with the hazard and what the results could be (sometimes this is called ‘the hazardous event’). If I assess risks, rather than the hazard, then there will be many instances when I can foresee the outcome of contact with a hazard will be less severe.
This is Risk Assessment 101!
If we are still convinced severity cannot be reduced because we cannot trust the controls (or people), I wonder if there is something deeper going on about how we see our role…Fundamentally, as a profession are we there to always assume the worst and find fault?
I’m really looking forward to the forthcoming Podcast on the Battle of the Ideas – I think some of these ideas might emerge.
I believe the point you are missing is that if you fundamentally change part of the process/machinery/substance as a result of the original risk assessment, then the task will need to be reassessed and your risk assessment updated taking into account the changes that you have introduced. At this point severity in your pre-control measures risk rating will be reduced because you are assessing a different process/machinery/substance.
The principle should still be to always reduce the likelihood to as low as possible regardless of the severity rating.
Totally agree Nick. It would appear that people are concentrating purely on the hazard and whether or not it can be reduced as opposed to the risk i.e. quantifying both the likelihood and severity of harm and applying control measures using the principles of prevention in an attempt to reduce both, as far as is reasonably practicable. As mentioned, in many cases it is much easier to reduce the likelihood of harm than the severity (potential outcomes)
Yes, we can reduce the severity of harm of a corrosive chemical (the hazard) to the skin through replacing the dangerous with the less dangerous but we can also reduce the severity of harm through the use of PPE, such as protective gloves. In most cases it would be reasonably practicable to utilise a number of control measures to reduce both the likelihood and severity of harm The assessment process should be holistic in approach with the assessor being able to appreciate the potential consequences and the applicable controls.
In conclusion: We must not over complicate what is, and should be a simple process. You will only find examples of simple qualitative risk assessment on the HSE website for this very reason.
Dear Steve,
Sorry to involve in your comment .
Could you please explain the way to reduce the severity of risk in the risk assessment process with use of PPE directly.
If you applying direct PPE there is some clause have to full fill.
e.g The selection,use,maintenance and storage{if anything fails(slips and lapses) the severity won’t change}
Moreover PPE is the last resort in the risk assessment process.
Hi Rajesh, This depends very much on the task itself. For example, wearing cut-proof gloves can reduce the severity. I rarely reduce severity in a risk assessment as in the construction industry the task is almost always decided in advance with no opportunity of reducing severity. I have been working on a task today where a void in the 1st floor was covered by timber which was held in place. To carry out the task the timber had to be removed. It was replaced by a smaller piece which was knocked down the void onto the ground floor, near missing someone. My point is that in that situation, no matter how many controls we had in place, if a control failed then the timber could have hit the person with the same severity as if the control measures were not there. I understand that the timber should have been bigger to prevent it falling but once the hazard was realised ie the timber fell, then we were relying on the control measures to reduce the severity. Failure of any of these could have lead to an accident which would have been just as bad with no controls.
A well considered article. Reducing the severity of the outcome is sometimes possible and professional practitioners will always try to achieve this where possible.
I had to smile at some of the comments regarding the bleach, It is so easy to get bogged down in these sort of discussions. In the case study covered the use of a bleach to clean the metal was a component of a wider process and it was this risks associated with this process that were being assessed. Substitution of a substance with a less harmful alternative eliminates the original substance from the process and reduces the potential level of harm to operatives which should be recorded on the risk assessment. It no longer matters in the context of the process in question what the level of harm was with the original bleach as it has now been substituted.
I also agree that risk rating matrixes can be unhelpful. Many organisations want them as they tie in with their corporate risk register systems. The down side is protracted discussions over the numbers turn everybody off. I still remember one of my early tutors Chris Burt stating “the level risk is either acceptable or not acceptable”
Simplistic? perhaps, but it makes you think.
Nick et al, a thought provoking article and comments even if it is an old chestnut. One of the problems with our industry is that the terms we use are interchangeable which often creates confusion and sometimes nothing more than semantics. I would be very surprised if an experienced h&s practitioner could not rationalise the concept of risk applied to risk assessments i.e. severity, likelihood and frequency.
Another confusing aspect of risk assessments is all the different types of templates, some with a matrix and some without. If ever there was an industry good cause it would surely include a standard RA template across industries, with the ubiquitous 5×5 matrix confined to room 101 – I hope!
Wearing a helmet on a motorbike doesn’t reduce the likelihood of falling off the bike and hitting your head. It reduces the severity of the outcome of hitting your head. If you want to reduce the likelihood of falling off your bike and hitting your head then drive a car.
Hope you don’t have adequate idea about the risk assessment process.
First point is that PPE is the last resort in the risk assessment process.
Helmet only protects the specific part (Head) but not whole part of the body and what about to other part of body when accident happens while driving.
Consider the selection(degree of area covered when selecting the Helmet),use,maintenance and storage of PPE.
Expected your valued comment.
I’d argue that using a water based low VOC paint would be far better for health than using high VOC paints. Reducing severity of harm – sorted!!!
You have changed the task which would then require a new risk assesment, and the severity of that would change. The severity of using high VOC paints cannot change
Everyone on this thread is correct because each is speaking to different way of thinking about risk. Taking a point risk for which the outcome is specified (it’s a constant) means risk control can only reduce likelihood. Taking a population of outcomes means risk control can change the distribution of the outcome population. Overall severity (area under the curve) can be reduced. Those using matrices are sometimes using point outcomes and sometimes population outcomes as this thread shows. Interesting and illuminating thread at so many levels. I don’t use matrices. When I need to quantify I use Monte Carlo methods.
Dear Gents;
I am really happy to join this fruitful discussion. Many people insist on severity couldn’t be reduced. It is not correct. Techniques of the hierarchy of control help greatly reducing the severity along with reducing the probability as well.
Techniques/Technology of Manufacture helps greatly providing products that share reducing the severity of harm.
Adequate Safety Management (Administrative control measures) share reducing the severity of harm, too.
Adequate HSE Management system components (Manual, plans, programs, procedures) along with the competency of the adhered key persons will result in adequate safety performance that means reducing both probability of hazard occurs & severity of harm through Pro-active Risk Assessment.
Good Luck..
Eng./ Sayed Fathy
ESH Director and Authorized International Tutor
Really enjoyed reading this article and everyone’s stance on this key subject. The debates I get into revolve around this in a different way. People send me RAMS for approval and mark down severity with only likelihood affecting controls or sometimes they mark down both with only likelihood affecting controls. I’m all for reducing severity when it can be achieved (And I believe it can and agree with examples in the comments) but there are a lot of people out there writing RAMS with 5 x 5 assessments that just don’t get it.
I can see the logic in each argument however, from my point of view the severity is exclusively linked to the initial problem statement. The severity of harm from being burnt buy concentrated acid will forever be a 10. to mitigate by using a weak acid is not adjusting the severity of the original concern, it reduces the occurrence/likelihood of the concentrated acid causing harm; in this case to zero. On the next round of risk assessments, that original hazard can be removed.
The argument is similar for the person falling; the severity remains the same but occurrence goes down. To be clear, occurrence here is specific to the original hazard that is “falling to the floor” not simply falling. safety nets reduces to zero (if the nets never fail) the occurrence of falling to the floor.
A hazard always remains a hazard hence the severity. The likelihood of your interaction with it reduces the hazard. A shark for example does not change from being a shark to cotton wool. It remains a shark. So the control measures could be to stay on the esplanade, on the beach or in a cage if you want to swim with them. Basically the closer you get to the hazard the more potential there is to be affected by it. The only way to reduce the hazard completely is to not have the hazard, but then there would be no need to risk assess against it. Adding controls to reduce exposure doesn’t make the hazard less harmful if exposure is realised it just reduces the likelihood.
Although I can argue the same that Nick’s example is not so well put by saying the risk identified is no longer the original risk. I can provide a better example.
lets say a man is performing work on a tall structure, considering that this man just matriculated and has no work experience and control measures are non existing at this stage. The chances of this man to fall is high, the consequences are fatal, resulting in a high risk rating. now, after implementing control measures, such lifelines, safety harnesses, training, supervision, safety catch nets, SWP & PTO, etc. In doing so you have reduced the severity of the residual risk, since he can no longer fall to his death, the harness will hold him, or the net will catch him.
I also agree with the argument many will bring forth saying that: “What if” he did not secure himself to the lifeline and he falls and the net was not securely installed, thus ending in a fatality. my answer to this is simple. your control measures were inadequately implemented, your initial score rating of high still counts. a residual risk rating only is valid if all your control measures are well implemented.
Mr. Max ,
Simple logic,
What ever you are taking the preventive measures while assessing the risk it may only reducing the likelihood but REMEMBER there is the foreseeable risk fall from height still existing so that severity will remains same in these case.Or you have to give value ZERO severity.
Thank you
It is not the fall that is the harm. It is the impact. The risk associated with the fall is injury. The potential harm is fatal, maximum severity. You have reduced the potential for this harm with the net. You have not reduced the cause (fall) but have reduced the occurrence of harm. The severity of the impact harm remains maximum, the likelihood of occurrence has been reduced.
Fall from a different height, while all other conditions remain the same, will change the severity of the risk.
amended version as i was unable to change original one!
As a risk manager, professional and academic i love this debate!
The examples you have used above are in relation to the Hazard. The hazard remains the same and cannot change and as such the severity calculation within a standard RA is based on a person coming into contact or being directly affected by the hazard.
In regards to the risk assessment process, you are measuring the likelihood of an individual coming into contact with that specific hazard. If you change the hazard you change the assessment, because if you don’t, you are comparing apples with oranges.
Some assumptions are made that the hierarchy of control has determined that the hazard still needs to be engaged with in some capacity. Therefore changing the bleach is a good move to a less harmful chemical but now you should be calculating the severity of the new bleach i.e. Original bleach is a 4 on a 5×5 matrix and new bleach is a 2. Yes you have changed the severity because you have changed the product. So the RA you receive from the person completing the form should state Hazard-Bleach, SxL=2×4 (control measures) SxL=2×1. Severity remains the same, control measures have reduced likelihood. If you are calculating 4×4 and then reducing to 2×1 and claiming you have reduced severity to 2 because you’ve changed the bleach. You haven’t used the hierarchy of control and you are not aware of effective risk management practices before you request a risk assessment review. Make sense?
It’s like saying I’m going to go swim with great white sharks and i might get killed but it’s ok because I’ve decided to swim with goldfish. Still fish, still swimming completely different scenario and hazard engagement.
So, when a site HSE professional states that you cannot alter the severity on a risk assessment form they would be correct. The caveat to this is when you are planning the work and considering the hierarchy of control which may have identified an alternative process.
So, to summaries, it depends on the circumstances of hazard conversations/planning/engagement. If you are discussing the design/technical safety aspects of the job then you would expect to have discussions surrounding alternative methodologies such as eliminating, engineering out, etc. etc. If you are beyond this phase then any hazards identified are probably unmovable at that stage and as such you are basing your final control measures around the identified hazard and unable to change its severity.
Hope this helps!
Hi, Please clarify? You wrote -“If you are calculating S×L 4×4 and then reducing to 2×1 and claiming you have reduced severity to 2 because you’ve changed the bleach. You haven’t used the hierarchy of control” … – How is the act of changing to a less dangerous substance not the act of applying the “substitution” element of hierarchy of control? And also, let’s consider a different scenario :- When you install a fall-net around a highrise building, you are not reducing the likelihood of an unplanned fall event. You are not reducing the “likelihood of there being a consequence of a fall”. But you are reducing the severity (consequence) of a fall, regardless of how likely that fall might be. It is not logical to state that a fall-net has reduced the likelihood of the fall from occurring given that the unplanned event can still occur. Also, you have applied the hierarchy of controls; it is just that you have reached the engineering options level of the hierarchy and those options will reduce the severity of the unplanned event. Note that in this scenario the hazard remains the same.
So my reference to not using the hierarchy of control is in relation to the risk assessment form that you would use. If you are using the hierarchy of control prior to the risk assessment form then yes you are substituting the bleach, but the form you are now completing and submitting for final appraisal is on new bleach not old bleach. So the severity is on new bleach or in my example a 2. Using your example of a net is correct in that you have reduced the severity of fall from let’s say 100m to 10m due to the net. So if 100m is a 5, 10m might be a 3/4. So you might decide to base your risk assessment form on the basis of 10m. So the risk assessment needs to consider different things as a hazard. Nobody dies from a fall, it’s the impact that generally kills people or injuries sustained from the fall.
Returning to your example you are right that the net won’t stop a fall, but it would prevent someone falling ‘further’. Therefore, if the net fails the person falls the full 100m.
So, if you are putting equipment and processes in place as your mitigations then should these barriers fail you end up at the original severity.
You’re 10m has just become 100m. This is different if you are working from a fixed platform such as a scaffold etc. So the falling aspect or risk of falling is where you are working from an unprotected edge.
The landing or impact from a fall is generally where the calculation comes from. So what would be the impact of falling into a net? Twisted ankle, hip, broke. Arm?? Very different if you fall the same distance to concrete.
This topic is a great thought proving one and it’s not about snobbery or oneupmanship. It’s about understanding where in a process you are deploying your motivations and in what environment. As a PC you might review contractor RAMS prior to work commencing. You would expect that at this stage the Hazards have been through the Hierarchy of control and reduced as far as possible. Therefore the control measures in the form you receive should be what they as the contractor intend to use. Based on the effectiveness of the control measures you can determine if they were to fail what would be the likely most severe outcome.
It’s the deployment of hazard and risk management that changes depending on the role you play and the environment you work in that changes the interpretation of the severity debate!
Collective fall protection such as netting obviously reduces the likelihood of an injury whilst falling but the bottom line is if that control measure failed (loose netting, worn or frayed netting, supports failure etc.) then the severity would be the same, therefore CANNOT be reduced.
Still changing the design in the wider scheme. A case of what constitutes the design. I would consider all aspects so changing the bleach or drop height is actually changing the design because severity related to the previous design concept.
Tiger is a man eater and the severity is fatal. Once, it is caged, the probability of harm comes down. When we talk of Risk reduction measures, we should never distinguish between an aggressive Tiger and an aged old Tiger to reduce Severity. The same is true for high voltage and low voltage electricity, where Severities are different right in the beginning. The severity does not change in a given situation but only probabilities could be reduced to reduce risk.
If the severity is not reduced and only the probability …Please explain how a fall net (let’s say in good condition) when placed around the edge of a multi-storey building only reduces the probabilitiy of the unwanted event, such as a fall. When we talk about risk reduction we cannot talk about the liklihood of the severity being reduced because then it becomes 100% subjective. We talk about reduction in the probability of the unwanted event occuring OR the reduction in the severity (consequences) of the unwanted event. We can’t state that a fall net is reducing the probability of a fall, because it isn’t. The only thing being reduced is the consequences of that fall if and when it happens ….surely???
When we use a fall net, we are introducing a layer of protection against an undesirable event(fatality) and this LOP reduces the Overall Risk by reducing the probability. Risk is a product of Severity and probability. Severity with respect to fatality from fall does not change but the consequential risk gets reduced.
Caging a Tiger reduces the probability of the unplanned event to pretty much zero. That is not the case with a seatbelt or fall net …because the likelihood of the unwanted event ….fall or crash …is still the ssame.
Seat belt and fall net are layers of protection introduced to reduce the risk of fatality from impact. Here again, the Severity remains same but the consequential risk gets reduced as the probability of the undesired event(fatality)gets reduced drastically(probability of failure of the Seat belt snapping or failure of the air bag getting inflated and coming in between the steering and the driver) on a sudden crash scenerio .
Great article Nick and brilliant debate, it’s something we have all likely discussed at some point. My view is you can alter both the severity and likelihood it’s simply a matter of what choice of controls you decide. By keeping the same hazard (chemical) and installing controls or altering the work method you can reduce the likelihood, however, if we change the hazard (chemical) we can also reduce the severity. I don’t think there is any science to this just different mind sets.
A risk assessment can reduce severity of outcome by deploying the heirarchy of control. Once you have established that you have reduced your hazard as low as reasonably practicable (ALARP) and you identify your risk assessment form and method of calculation (e.g. 5×5 matrix) and subsequent mitigations. The severity remains the same on the residual calculation. This is because you have, or should have, already reduced the severity to this lowest point. If this happens to be working at 100m high at the edge of a building for example as there is no othr safe way. Then the controls/mitigations reduce the likelihood and not the severity. Due to Bobs earlier comments about mitigations failing, I totally agree the severity will be the most severe. So this debate is really about context and the phase you are at, when performing a risk assessment, and mainly contentious when completeing the particular form you use, which has been clearly misunderstood by many HSE professionals and continues to spark debate.
I conducted a case study on a school road in London, which had experienced a few vehicle and child collisions over a few years. In order to further reduce risk, the local authority reduced the speed limit to 20mph. This resulted in reduced risk perception in the children, and more incidents occured as a result, but the overall data of incident severity was much lower than previously recorded. An increase in likelihood, and a reduction in severity.