Chapter 8: Better assessment
8.1 You don't need maths
Reinforcing the message of Chapter 7b these words from HSG 65, the HSE’s key document on ‘Managing for health and safety’ make it clear:
For small businesses, with few or simple risks, a suitable and sufficient risk assessment can be a very straightforward process based on informed judgement and using appropriate guidance.
Unless your workplace is covered by specific legislations, such as COMAH (Control of Major Accident Hazards Regulations) or Offshore Installations (Safety Case) Regulations, nothing in the general legislation or guidance requires you to use maths.
The example risk assessment templates on the HSE website do not include any columns for scoring, asking instead about what you are doing already, what else you need to do, who is going to do it and by when. The suggestions in this chapter align with that approach.
Box 8.1: How to read this book from here
For small businesses with straightforward operations you can read this chapter, skip Chapters 9 and 10, and move directly to Chapter 11. Spend less time assessing (step 2), and more time controlling risk (step 3). Within this chapter Sections 8.2 to 8.4 cover examples where you might need numbers, but those numbers won’t be drawn from a matrix, and the rest of the chapter looks at how to prioritise without using numbers.
If you work for an organisation that insists on a matrix, use Chapter 9 and Chapter 10 to make suggestions as to how you can use a matrix to better effect. If your organisation won’t let you adjust the matrix you have, use the tests in Chapter 10 to make the business case for changing the matrix. Or, use the tests to demonstrate to yourself that in your case, you have a matrix that works (in which case, please contact me with the details).
Decibels can be measured
Vibration can be measured (illustration from HAVi)
8.2 Measure only what can be measured
Sometimes, it is useful to add numbers to a hazard. But only when there is something measurable.
If you might have a workplace noise problem, have the noise measured at different locations and times. Where noise levels exceed 80 dB(A), find out how long people spend in each area to calculate their daily or weekly exposure. Even if everyone’s exposure is below the lower exposure action value (LEAV) you need to reduce the risk so far as is reasonably practicable, for example by replacing noisy machinery with quieter versions.
You can use a similar process for vibration – measure the vibration from your tools and compare the numbers with the statutory exposure values. If full-time use of a tool can exceed those levels, you need to find out how much time people are spending using the tools. And then control the risk. Buy better tools, change work methods to reduce time spent on tools, improve maintenance and so on.
The HSE have downloadable spreadsheets which make it easier to calculate exposure to vibration and noise from collected data.
If people could be exposed to ionising radiation, there might be a requirement for them to wear personal dose monitors or for equipment to be assessed periodically for leakage (see L121, the approved code of practice and guidance for work with ionising radiation). Electromagnetic Fields (EMF) might need to be measured if your workplace includes significant sources such as MRI scanners, radio transmitters and some welding operations (see HSG 281, a guide to the Control of Electromagnetic Fields at Work Regulations). Radon might need to be measured if you have a workplace in a high-risk part of the country, particularly if you have a basement or other underground worksite.
Some hazardous substances have numeric values against which you can compare measurements. The HSE publication EH40 lists over 500 workplace exposure limits (WELs) measured in ppm (parts per million) or mg/m3 (milligrams per cubic meter). These are averaged over 8 hours (long-term WEL) and over 15 minutes (short-term). As with noise and vibration, you still have a duty to reduce risk so far as is reasonably practicable.
The impact of exposure might need to be assessed through health surveillance. This can include some measurable effects such as the levels of lead in the blood (see L132, the approved code of practice and guidance for the Control of Lead at Work Regulations), hearing thresholds at different frequencies for noise exposure, and lung function measurements for respiratory hazards.
However, do not spend your resources getting increasingly detailed measurements when you could put those resources into reducing the risk at source. In 2018 I spent weeks refereeing a debate between the providers of vibration measurement tools as to the most effective way of measuring the risk of hand-arm vibration. On the tool? On the hand? On the wrist? Chris Steel, a Specialist Inspector for noise and vibration at the HSE cut through the debate with a simple statement that “assessment should be a relatively short process which allows dutyholders to establish the level of risk and move their focus on to control procedures” An academic interviewee for the same article, Professor Michael Griffin of Southampton University, also set measurement into its correct context:
“Measurement, evaluation and assessment do not control risks: they are only useful if they lead to a better decision than would otherwise have been possible.”
Ever more sophisticated tools have become available since then to measure other hazards. For example, AI-supported computer vision to score posture during manual handling. However, the same principle applies. If measurement helps to identify a hazard, and to assess the benefits of an intervention, that’s great. But don’t spend all your resources on measurement, with nothing left to reduce risk.
8.3 Proxy measures
Sometimes there are things you can measure which are not a precise measure of the risk itself, but are nevertheless objective, and give you a good indication of how well risk is being managed.
Legionella bacteria grow most quickly in moist conditions between 20 °C and 45 °C (L8, HSE approved code and practice and guidance for the control of legionella bacteria in water systems). Therefore, keeping a record of the temperature of water (at sentinel taps, and in storage tanks) at appropriate intervals gives you an indication of when the likelihood of legionella bacteria growth has increased. For example, if the temperature in the hot water system drops below 50 °C, you might need remedial measures such as pasteurising the water (see HSG 274 part 2).
Speed, harsh braking and acceleration recorded by in-vehicle telematics systems are proxy measures that give an indication of driving behaviour and vehicle accident risk. AI-supported telematics provide the possibility of more detailed monitoring – of whether drivers have their hands on the wheel, if they yawn or close their eyes. While these might provide tools for reducing road risk, be careful you don’t fall into the same over-measurement paralysis warned of for vibration measurement tools.
Water temperature can be measured
8.4 Expert scores - Asbestos and COSHH examples
If you are responsible for any workplaces with asbestos-containing materials (ACM), you should have an asbestos survey where expertise and experience have been used to apply two scores for each incidence of asbestos:
Material assessment score – based on the asbestos and product type, the extent of damage and any surface treatment of the material. The score can be between 2 and 12.
Priority assessment score – based on the location of the asbestos, and how activities in that area could impact the asbestos. This should have been assessed in consultation with users of the location, not by an external consultant alone. The score can be between 0 and 12.
HSG 264, the HSE survey guide on asbestos, explains these scores in more detail.
These numbers should not be added or multiplied together. A material assessment score above 10 indicates a higher potential to release fibres if disturbed, but if the likelihood of being disturbed is extremely low (that is, the priority assessment is very low) then it might be better to spend resources on removing or remediating asbestos with a moderate material score that has a high priority score (for example, because the asbestos is in a school classroom). Alternatively, you might be able to reduce the priority score through other controls. If there is asbestos in the back of a cupboard with high scores, instead of removing or remediating the asbestos to reduce the material score, you could reduce the priority score by sealing the cupboard shut and storing things elsewhere. You need to consider the scores separately to determine the best action to take given the work context.
The HSE guide COSHH essentials: Controlling exposure to chemicals – a simple control banding approach explains how to identify the control approach needed from factors such as the Hazard statements (H) from the Safety Data Sheet (SDS), the concentration used and the physical properties (dustiness or volatility). The system is necessarily more complex than guessing on a scale of 1 to 5 how hazardous something is or how likely it is to harm someone. But as a result, it provides more specific recommendations on how to limit harm from the substance, pointing towards specific HSE control guidance sheets. The structured approach makes the control recommendations more consistent than relying on individual judgement alone.
Box 8.2: Don’t lose the detail
Don’t try and convert any of the numbers mentioned here back into scores on your risk matrix. Keep records in dB(A) or m/s2 or ppm or °C or whatever other measure you used. It can be useful to include measures of time as well – so how long during the day was the factory operating at that volume, on how many days was there a cold tap recording temperatures above 20°C, and how many people had measured exposures uncomfortably close to a pre-determined limit? But unless you have a really good reason, don’t try to convert ratio numbers, or expert judgements, into ordinal risk matrix numbers.
8.5 How to assess risk without numbers
From step 1 you should have:
The scope: a clear idea of what is being risk assessed, the people, locations, equipment, substances and so on involved.
The hazards: a list of well-defined hazards relating to the people, locations and so on in the scope.
Rather than trying to weigh up at this stage how effective current, proposed, hypothetical or imagined controls are, let’s focus on what could go wrong. On ‘who might be harmed and how.’ Can you produce a statement (not a number) that describes the most likely serious scenario (or scenarios)?
That phrase needs unpacking. We’re trying to assess the severity of the consequences here, and yet I’ve brought in a probability phrase “most likely”. Here’s an example to explain.d
Ask yourself:
Imagine someone falls from the window of a two-storey building. Write down at least five different outcomes of the fall. Now arrange them in order of consequence. Now, think which one is the most likely.
This person could be immensely lucky and escape with no injury, or just bruising. In a freak scenario they could fall on a party of infant school children and kill themselves and ten small infants. The most likely serious scenario is that they suffer a major injury, with broken bones, and possible long-term effects such as paralysis. If we had data for 1000 falls from 1st floor windows we could present the data as a bar chart. We place the consequences of the falls along the x axis, from nothing happened, to an accident causing multiple deaths. It might look like Figure 8.1.
From this, you can see that the most likely serious scenario is major injury to a single person (the person who falls). The single death, although not the most frequent accident, is also worth noting. So I could produce a statement to describe “Who might be harmed and how”:
Assessment of harm:
If a person falls from the first-floor window they could suffer major injuries such as broken bones, leading to long-term injuries, paralysis and in some cases, death.
Too many risk assessments leave out this stage.
In the case of the window, the controls are likely to be common regardless of the outcome (other than keeping infant school children away from windows people are liable to jump from). But in other cases, you might need to take account of multiple outcomes as the controls will be different.
Figure 8.1 Hypothetical bar chart for the consequence of falling from a window
Returning to the driving for work example I used in Chapter 7 we can write four statements outlining who could be harmed and how, with a description of the severity::
- If a driver is involved in a motorway accident there could be multiple fatalities, including the death of our employee.
- If a driver is shunted by another vehicle, or shunts another vehicle, they could suffer whiplash.
- If our drivers spend long hours driving without a rest they can suffer from musculoskeletal pain which might lead to time off work.
- If a car hit a pedestrian, the pedestrian could be seriously injured or killed. Our employee could suffer from whiplash, and psychological harm.
Figure 8.2: Four views of outcomes of driving for work
Ask yourself:
Consider the hazardous event ‘slipping on a wet surface’. Where might this happen in your workplace? Would the severity vary depending on the location, or on who slips? Write down at least three statements of harm.
Is it possible to decide which statement describes the most likely event? Is it possible to think about which is the worst consequence?
Does it inspire more ideas about possible controls than putting ‘slips and trips’ in the hazard column and numbers from 1 to 25 in the scoring columns? See Appendix 2 for some suggested harm statements.
8.6 Uncertainty
In Chapter 7 I raised the difficulty of representing uncertainty in a risk matrix. Figure 7.5 showed one way of representing different estimates for the risk score for hazards. The approach of writing harm statements helps us to deal with this uncertainty. You can write multiple statements as we did for the examples in Section 8.5, or include alternative options within a single statement. This is also helpful where the impact of something could be negative or positive (something you can’t show on a risk matrix).
For example, imagine you are about to introduce a new shift pattern which you hope will improve wellbeing and productivity for staff. Correctly, you have decided that you need to assess the risk of this new way of working. Your harm statement should admit to this uncertainty, for example stating:
The results of this pattern of working could be improved wellbeing and reduced errors, but there is also a danger that for some people it will reduce wellbeing and increase short-term sickness absences.
8.7 Conclusion
If there is an objective way of measuring a hazard (like noise in decibels) then if it is useful to measure it, do so. If there is an expert scheme for assessing components of a hazard, like the priority and material assessments for asbestos, then use those (but don’t average, sum or combine them). If there isn’t an agreed way of measuring or ranking a hazard, agree clear statements of the most likely serious scenarios, and then move on to focus on the controls.
If you’re happy with this as an approach, I’m going to suggest something you won’t find in many books. Ignore the next two chapters. They are for people who really like matrices (or work for organisations that insist in them). Once you’ve got your scope, your hazards, and your assessment of harm statements clear, read Chapter 11 to focus on controls.
However, if you still think a matrix is a good idea (or you need to understand the arguments to convince others) then read Chapter 9 and Chapter 10.
Otherwise feel free to skip those chapters, and go straight to Chapter 11 to learn more about controlling risk,
You can use the Contact form to send me feedback. If you’d like to receive an email when I add or update a chapter, please subscribe to my ‘book club’
Alternatively, go back to the book contents page
Appendix 1: Case studies by year
Appendix 2: Answers to questons posed in each chapter
Appendix 3: Lost HSE references