Raising The Standard By Lowering The Benchmark

June 8th, 2022 • News

Raising The Standard By Lowering The Benchmark

Exposure monitoring – What do you as the employer need to do?

As the duty holder, you must determine if monitoring your employees’ exposures to hazardous substances is needed. Let’s say you have gone through HSE document HSG 173 ‘Monitoring Strategies for Toxic Substances’ and you have gone through COSHH Regulation 10 ‘Monitoring Exposure at the Workplace’, and determined via your risk assessment that exposure monitoring is necessary to confirm that the controls in place are effective.. You find yourself a competent  occupational hygienist to carry out the monitoring. And so we, the occupational hygienist, arrive to your site, a clipboard in hand and with our kit at the ready. We are there for (typically) one day. Although, as we will discuss later in this article, more time spent on site would be preferable. Our aim; to carry out exposure monitoring to assess if workplace exposure limit (WEL) compliance for that hazardous substance(s) has been met. And also, most importantly, we carry out workplace observations to assess if the risk controls in place are effective.

COSHH Regulation 10 Approved Code of Practice (ACOP) (Figure 1)

We need to know what the risk is, when does the problem occur, who may be affected, how and to what extent. We watch carefully, enquire, and use qualitative tools to judge the control measures in place. We assess if the principles of good practice for the control of exposure to substances hazardous to health have been applied (see COSHH Regulations Schedule 2A). It is a lot to do in one day- even in the simplest of workplaces. But the need for exposure monitoring stands, should the criteria outlined in figure 1 above show it to be necessary. See specifically where this ACOP states ‘measurement is required to be sure that a WEL… is not exceeded’. But how are we ‘to be sure’? Given we are only there for one day and will likely have only taken a small number of samples, this is a heavy judgment to make based on limited data. And indeed it may be very difficult to be 100% sure, but let’s look at how we can get close enough.

N.b. ‘’WEL’’ is the UK’s term for an Occupational Exposure Limit (OEL) but this may be termed differently in other countries.

How can one day of monitoring represent a whole year?

From this one-day site visit, our monitoring results will need to be representative of the exposure levels that all of your workers will typically encounter. And if we are only there for one day per year, this represents about 0.03% of operational time. That’s it! So how do we ensure our monitoring results will accommodate for this? How do we ensure no worker has unacceptable exposures? Understanding your measurement results is an important step in coming to a reliable conclusion.  But exposure monitoring results are inherently variable. Exposure levels may differ between people doing the same job and from day to day depending on the workload or conditions at the time. Let’s take a look at how these exposure results might actually look.

The below graph is an example which shows 100 measurements taken from a packing plant. We see here the number of samples with results at various concentrations. It may look odd at first. How can there be such a spread of results from the same process? Is there something amiss at this work process? Did the sampling method or the lab’s analysis go wrong? This spread of data is fairly typical of an exposure scenario. We can see from the graph that the data’s distribution forms a sort of curve. It is asymmetrical, with the majority of measurements towards the lower values and a long tail with fewer higher values. In statistical terms, this exposure distribution is highly skewed and is called a ‘log-normal distribution’. This type of distribution is close to what is often found in typical occupational hygiene measurements. (There are means to test for lognormality, but it is usually assumed that occupational exposure data follows a lognormal curve).

Image Credit:  Monitoring for Health Hazards at Work, Cherrie et al (Figure 2)

N.b. If we take the logarithms of the exposure levels for a distribution such as the above and plot them against the number of measurements, the distribution would take the form of ‘normal distribution’. The mean of such a log-normal distribution is called the ‘Geometric Mean’ as opposed to the ‘Arithmetic Mean’ for normal distribution. This is an important distinction as we shall see.

What are the implications of a log-normal distribution of data?

Let’s take a step back first. Think of the graph above as if it were a jar of sweets. You have a larger portion of green and yellow sweets (the bulk of the results being skewed to the lower values) and a smaller portion of red sweets (fewer higher values tailing off to the right). If we reach into the jar, what are the chances of pulling out a red? You could potentially pull out a green or yellow or red. But there is a lower probability of picking out a red with your first reach. The more times you reach into that sweets jar, the more likely you are to get a red sweet. If you only reach into the jar a few times, you’ll likely end up with a green or perhaps a yellow. This is why it is important to understand your full spread of results. The more times you reach into the jar, the better you can form a coherent picture of that exposure profile. You may be asking ‘How many sweets do we need to pull out to be reasonably sure that we have a good idea of the portion of different colour sweets? Without having to count the whole jar?’

Every sample result seen on the graph above is a true and valid representation of the worker’s exposure (with exposures which exceed the WEL breaching of the law). The outliers (infrequent high results) we see on this graph (the ‘red sweets’) may not necessarily be due to control failure, but these may happen because there is a statistical chance that the many variables affecting exposure monitoring can sometimes combine in a way which produces an apparently outlying result. This can be a problem partly because the WELs are defined as sharp-cut off values which may not normally be exceeded. The law requires exposures to be controlled below a WEL, but in reality, occasional exceedances may inevitably occur. Trained HSE enforcers should take into account if good practice was applied and look beyond the simple number to the reality of the controls in place.

All that being said, let’s continue to look at the statistics behind WEL compliance. If we only reach into the sweets jar a few times then it is difficult to come to a conclusion about the spread of coloured sweets from this limited information. So how do we understand our exposure profile based on a limited number of samples?

If you reached into this jar only a few times, what are the chances of getting a red? (Figure 3)

(Note: This distribution would not look the same for all exposure data. It depends on the WEL and how well-controlled the process is)

What is the ‘95% criterion’?

If we only reach into the sweets jar a few times and get a green sweet, that tells us very little. A few sample results below the WEL cannot assure us WEL compliance is always met. Not unless the results are very low indeed. Your first few reaches into the sweets jar may give you greens, but the next reach may give you a red. This high result is still a representative measure of exposure and is still in breach of the law. This measurement could be the one made by an enforcing authority! You may be thinking, but with so many variables how can I possibly be achieving WEL compliance 100% of the time? It is recognised that even in the most well-controlled environment there may never be 100% certainty that exposures are below the WEL. Here is where the ‘95% criterion’ comes in. If the employer has a sampling programme in place which demonstrates that there is a less than 5% chance of any exposure exceeding the WEL, that is generally recognised to be good enough in practice and that WEL compliance is achieved. Provided of course that the principles of good practice are also met (and ALARP applied where required).

How far below the WEL must I be and how many samples must I take to ensure I have met this ‘95% criterion’?

This is a simple question with a complicated answer. To which is detailed in full in BS EN 689 and the document ‘Testing Compliance with Occupational Exposure Limits for Airborne Substances’ published jointly by the British Occupational Hygiene Society (BOHS) and Dutch Occupational Hygiene Society (NVvA). These documents enable the duty holder to demonstrate that the WELs are statistically complied with.

Following this strategy, we start by breaking down the workforce into SEGs. SEGs are similarly exposed groups. These are groups of workers having the same general exposure profile for the hazardous substance(s) being monitored because of the similarity and frequency of the tasks they perform, the materials and processes with which they work and the similarity of the way they perform those tasks. Once the SEGs are determined, we start with a ‘screening test’. We take three representative samples at random from workers within each SEG. If all three of these exposures are less than 10% of the WEL, it can be assumed that the WEL is complied with. If any of these samples are above the WEL, then the WEL is not being complied with and immediate corrective actions are needed to meet the legal requirement of WEL compliance. If any of these samples are greater than 10% of the WEL, but less than the WEL, then WEL compliance cannot be assured to the 95% certainty outlined above. More samples would be required through the next step which is the ‘group compliance test’. (Precisely how many more samples and why three samples are taken initially is explained in the guidance document).

Do I have a legal obligation to comply with BS EN 689?

Like all European standards, EN 689 has no regulatory status unless some national or European law requires it to be used. But it is expected to be influential as an international expert consensus of the state of the art. If challenged about the validity of monitoring data, demonstrating compliance to this standard would become important. Otherwise in a court of law, for example, the results could be challenged which could result in successful litigation or even prosecution.

The exposure monitoring strategy laid out in the BOHS/NVvA document ‘Testing Compliance with Occupational Exposure Limits for Airborne Substances’ (Figure 4)

What if the number of samples I can take within each SEG is not exactly 3?

Taking enough measurements to know accurately where the 95th percentile is can be difficult to achieve in reality. This problem is eased with the BOHS/NVvA guidance which outlines the WEL testing compliance strategy, starting with the ‘screening criterion’. This ‘screening criterion’ of 0.1 x WEL (per SEG of 3) is conservative, but it is a simple way of ‘passing’ or ‘failing’ an exposure based on a small number of samples. The more measurements that are made, the greater the confidence that can be placed on interpreting the results. A minimum of three samples should be taken and, unless are all below 0.1 x WEL, more will be required under the guidance

Taking more than three samples within the SEG can be done and will allow for a higher benchmark to show WEL compliance. For the screening test, four SEG samples can show WEL compliance at 15% of the WEL. And five SEG samples can show WEL compliance at 20% of the WEL. We have found though that typically three samples within a SEG is achievable. Most sites have a wide range of tasks and materials that the workers are involved with. And so the SEGs become quite small. With the exception of certain workplaces such as large-scale manufacturing or production lines where the work undertaken is similar across larger groups of workers. In this case a higher number of samples could be taken per SEG.

EEUK Group adopts the ‘10% of WEL’ benchmark as we believe this will cover the majority of sites we visit and typical scenarios of smaller sized SEGs. We usually would only be able to sample a few workers within each SEG. However, each site will be assessed on a case-by-case basis and the sampling scope adjusted if needed. If say you could only sample for one or two people within an SEG, then the benchmark should be adjusted accordingly (lowered further). (Or we could consider sampling again on another day). However, EEUK Group recognise that a ‘10% of WEL’ benchmark is tight and that tightening this even further for smaller sized sample numbers may cause difficulties for the employer to be able to achieve this figure. EEUK Group emphasise that the focus be given to the principles of good practice for the control of exposure to substances hazardous to health, as this is paramount.

I have the sampling results. Now what? How do I interpret these values?

Let’s test this out! Here is that log-normal distribution again (below). The data shows the results of several personal exposure samples and the dotted line is the log-normal distribution which fits the curve. Say we took our three samples within an SEG (the ‘screening test’). The results were 26 mg/m3, 10 mg/m3, and 39 mg/m3 (as 8-hour time weighted averages). The WEL is 150 mg/m3 8hr-TWA.  We see that two of these results are above 10% of the WEL. Thus WEL compliance cannot be assured and further sampling may be needed (unless control ineffectiveness is apparent in which case focus should be given to improving controls).  If the sample results had been, let’s say, 14 mg/m3, 10 mg/m3, and 12 mg/m3 (8hr-TWA), then all three results are below 10% of the WEL and compliance is confirmed for the survey. The next step would be to carry out routine monitoring when warranted (see HSG 173 and COSHH Reg10).

Image Credit:  Rappaport and Kupper (2008)  (Figure 5)

What statistical tools can I use? And what is the ‘UTL95,70’ ?

It may sound complex. But there are easy ways to interpret this data. Statistical analysis tools are available. One such is BWStat which is designed to work with the BOHS/NVvA strategy. Figure 6 below shows a screenshot of the ‘conclusions’ outcome of our first example above (results at 26, 10 and 39 mg/m3 8hr-TWA) using BWStat. This concluded that these three samples within the SEG were not all below 10% of the WEL and so we do not have confidence in our WEL compliance and further sampling may be needed.

Let’s go back to the ‘95th percentile’ for a moment. We know that if we take only a small number of samples then we might get all low or all high or a wide range of results. And if we took two sets of samples on different days, this would almost certainly get different results for the 95th percentile. So we need to make allowance for the fact that the result for the 95th percentile may different depending on the day. There needs to be a way to calculate an upper limit on the 95th percentile so that we have a good level of confidence that the results are conservative. Luckily, statisticians have devised a way for us to specify the percentile and the degree of confidence all in a single calculation.

You’ll see from the BWStat results below, this also shows us the ‘UTL 95,70’. The UTL is the ‘upper tolerance limit’. This is an important parameter which essentially allows us to specify a 95th percentile (to account for the variability of exposure) with a degree of confidence (to account for the randomness of the sampling process). With70% confidence being widely accepted in the UK. If the UTL 95,70 is above the WEL then the conclusion is that compliance with the WEL cannot be confirmed. We do not need to be too concerned with the UTL 95,70 when it comes to smaller sample sizes though as it then becomes less accurate and precise. Our focus should stay with the 10% benchmark (for an SEG of 3) when it comes to decision making.

Figure 6: Screening test (results at 15, 10, 22 µg/m3 8hr TWA) shows non-compliance to WEL. Using BWStat statistical analysis tool.

What other statistical tools can I use?

There are several OH statistical tool available. Such as IHDA, IHStat and Expostats. IHStat, for example, shows a ‘log-probability plot’. If the data is indeed approximated by a log-normal distribution, then it should fall on a straight line on this plot. To demonstrate, figure 7 below shows the outcome of plotting results at 2.2, 1.8 and 3.4 mg/m3. Let’s say that the WEL is 5 mg/m3 8hr-TWA (so the sample results are all below the WEL). The y-axis showing probability and the x-axis showing concentrations of the exposure results. Let’s use this plot to estimate the 95th percentile point on the distribution. If we follow from ‘95%’ on the y-axis (this is the 95th percentile criterion we need to meet) to the line of the plotted results, follow down to the concentration, we see that the concentration point exceeds the WEL in this scenario.

Figure 7: Screening test (3x samples within SEG) shows non-compliance to WEL. Using IHStat statistical analysis tool.

EEUK Group can, if requested, share the output of these tools into the monitoring report we produce for our client on the back of our exposure monitoring visit to site. This can help visualise how the sampling results were interpreted and how the conclusion was reached.

Let’s recap what we’ve learned so far…

Exposure results may be highly variable. The distribution of occupational exposure results are typically lognormal. You can’t evaluate exposure confidently with one or two samples. You can use a ‘screening criteria’ and use three samples to screen for compliance, but these samples must all be less than 10% of the WEL. The natural variability of exposure is accounted for by comparing the ‘95th percentile’ against the WEL. There is uncertainty in our estimate of the 95th percentile due to small sample numbers, this lack of confidence is covered in the ‘UTL95,70’ which gives 70% confidence. Implementing good practice is paramount. Pouring time and money into a robust sampling regime is non-sensical if the principles of good practice are not abided by to begin with.

But why does this change need to happen? What benefit does it give me? What was so wrong with using a 50% of WEL benchmark?  

The 50% has been used as a benchmark by many occupational hygienists for some time. This has some (limited) statistical rationale to it and has been applied as an action level, but there is a significant chance that cases of overexposure would be missed using this benchmark. And hence movement away from this benchmark is much warranted.

This rule of thumb may have originated from the Control of Lead at Work (CLAW) regulations which states exposures to lead exceeding half of the occupational exposure limit to be ‘significant’. When this definition of ‘significant exposure’ is triggered, the CLAW regulations apply and certain actions in particular e.g. protective clothing, airborne monitoring, medical surveillance. But this ‘50%’ figure was never defined within the CLAW regulations as a means to statistically prove WEL compliance. This figure is stated as a means to identify where ‘significant’ exposures are occurring and when further actions are required to control said exposure to Lead. It is in effect a quick rule of thumb.

This 50% benchmark is seen in the HSE document HSG173 which states ‘If you have used relatively unsophisticated techniques, such as chemical indicator tubes during an initial appraisal, do not place too much confidence in the level of compliance when an employee’s exposure is within plus or minus 50% of the WEL.’ Although here we see that it is pointing out the lack of confidence in a result at 50% of the WEL.

The Control of Lead at Work regulations were published nearly 20 years ago. Statisticians have made huge developments in this time (advancing their understanding on how exposure behaves and improving sampling strategies to deal with this), with the BS EN 689 standard ‘Strategy for testing compliance with OELVs’ being revised and published a couple of years ago. We therefore have a clearer stance when it comes to analysing our sampling data for WEL compliance purposes. It should be said though, occupational hygienists have been working with statistical principles for quite some time and this is not something new that we are suddenly waking up to. The principles are merely growing over time and are becoming more widely accepted. Also, the British Occupational Hygiene Society has adopted this approach through their ‘Testing Compliance with Occupational Exposure Limits for Airborne Substances’ document (which gives guidance on measurement strategies for determining WEL compliance).

This ‘10% of WEL’ benchmark is not written into legislation, it is not written officially into HSE guidance. However, EEUK Group have adopted this stance as we see this as the future norm and is generally in line with industry best practice. And as the best way to show our clients what their exposure monitoring data truly means and how WEL compliance can be met. Rest assured, our change in stance is to help you comply with the law!

Overreliance on the results… where should the focus really be?

As we’ve discussed, occupational exposure monitoring can be highly variable and it only represents a small portion of actual operational time. If the monitoring was conducted with poor planning, poor observations, poor contextual information, poor interpretation of the data, then this report will be unhelpful and the client may not know what to do. What our exposure monitoring must entail is not just the results of sampling, but what was going on that lead to those results. Were control measures in place (LEV, RPE, workplace segregation, job rotation)? Are these controls appropriate? Was there adequate training given around these controls? Were the workers given information on the hazards within their workplace and the risks they face? Do the workers understand the importance of the controls? Is health surveillance needed and, if yes, is it in place? Do the results from previous monitoring records show a concern? Is the risk assessment suitable and sufficient?

Let’s say the answer to any of the questions above was ‘no’. Ah… but the monitoring results were ok! Well, this is not good enough. What the COSHH regulations require you to do is to adequately control exposure. This is achieved in three parts:

  • The principles of good practice for the control of exposure to substances hazardous to health (COSHH Schedule 2A) are applied;
  • Any WEL for that substance is not exceeded;
  • For a substance which is a carcinogen, mutagen or sensitiser (see exact definition in COSHH Reg 7(7)), exposure must be reduced to as low a level as is reasonably practicable (ALARP).

The Hierarchy of Control (Figure 8)

Emphasis must be given for the need for exposure control, supported by critical thinking and situational analysis. Rather than sampling to show compliance with a WEL. Even with robust statistical analysis of the monitoring results where you may be able to show WEL compliance, but this may not necessarily mean that current controls are adequate or that further controls are not required. Let’s say you are a bakery, handling flour dust. Flour dust is a sensitiser, and so exposure must be controlled to ALARP, as this is where the legal duty sits. Regardless of WEL compliance.

Saying all that, monitoring is may often be needed and certainly has a solid place in the repertoire of an occupational hygienist. But, observation and critical thinking are key. The monitoring results are there to support observations gathered, not visa versa. A few simple observations, using qualitative or semi-quantitative means of analysis, may be all that is required to show the controls are ineffective.

What if my sampling results exceed WEL? Have I broken the law?

Workplace exposure limits are intended to prevent excessive exposure to specified hazardous substances by containing exposure to below a set limit. It is the maximum concentration of an airborne substance averaged over a reference period to which employees may be exposed by inhalation. It is a legal obligation for the duty holder to ensure that a WEL is not exceeded. This is written under COSHH regulation 7 ‘Prevention or control of exposure to substances hazardous to health.’  If a WEL is exceeded, the duty holder is in breach of the law and corrective actions must be taken.

What if my sampling results slightly exceed 10% of the WEL? Have I also broken the law? Do I need to invest more money in controls?

Being just above 10% of the WEL (if you have taken 3 x samples within each SEG) does not necessarily mean that the WEL has been breached. But we cannot be sure that the WEL has been complied with and further sampling may be needed to prove this. So what does a result >10% of the WEL (from an SEG of 3) really mean? In practical terms, one or more results over 10% of the OEL but below 25% of the OEL is still very likely in compliance. We just don’t have as high a degree of confidence as we would like, we only have a fairly high degree of confidence. So let’s say you have a few samples with results between 10% and 25% of the WEL. You are likely still within compliance of the WEL. But without all 3 samples at <10% of the WEL, further sampling may need to be considered to increase your confidence in the results.

If let’s say your results were over 25% of the WEL, or worse yet over 50% of the WEL, then additional samples would be more warranted. And improvement controls may need to be considered sooner rather than later. This approach does not strictly follow the BOHS/NVvA guidance, but we state this here to help explain that not all exposures over 10% of the OEL are equally concerning, both practically and statistically.

Also, this grey area between 10% and 25% should take into consideration the circumstances as this will affect our judgement on WEL compliance and risk.

Let’s not be too distracted by the exposure results though, you may at this stage need to have a closer look at your controls. Use this as an opportunity to trigger further investigation into their effectiveness. If from there it is still uncertain if the controls are performing adequately, then further sampling may be warranted.

As occupational hygienists, we need to be careful that our results do not show green flags when really red or amber flags should be waving. We must ask… were the controls in place effective? Were the workers using these controls appropriately? Were there controls in place at all? Was the relevant risk assessment suitable and sufficient? Did health surveillance results show any indication of ill-health effects caused by the work? Are workers currently experiencing any symptoms? These are all important questions which a simple number cannot entirely answer. So we ask our clients to not focus so much on the numbers, but on applying the principles of good practice regardless if WEL compliance has been met or not.

What about the laboratory’s limit of detection?

As our benchmark drops, we need to be careful to ensure our analytical methods are sensitive enough to keep up with this. We do not want to be in a situation where, for example, the WEL is 0.05 mg/m3 but the result came back to us as <0.007 mg/m3. Thus we would only be able to say that exposure was <14% of the WEL, and not that it was below 10% of the WEL. We need to consider the sampling time, sample volume, flow rate and the laboratory’s analytical limit of detection. The equation below shows how to work out the minimum sampling time needed to achieve an LOD of atleast 10% of the WEL.

  • Let’s go through another example. Say we are on site sampling for hexavalent chromium from welding fume. The WEL is 0.025 mg/m3 8hr TWA. The laboratory’s LOD is 0.01µg. And the flow rate for an IOM head is 2.0 l/min.
    • Minimum sampling time = (10 x 0.01 µg x (0.001 mg / 1 µg)) / (0.025 mg/m3 x 2.0 l/min x (1 m3 / 1000 l)) = 2 mins.
    • No problem at all there! We will still aim to sample for 4 hours or so, according to guidance, in order to obtain a representative sample of that person’s working day.
  • Let’s go through another example. Say we are on site sampling for formaldehyde. The WEL is 2.5 mg/m3 8hr TWA but may soon lower to 0.37 mg/m3 8hr TWA. The laboratory’s LOD is 0.1µg. And the flow rate for low-flow sampling with sorbent tube is 0.2 l/min.
    • Minimum sampling time = (10 x 0.1 µg x (0.001 mg / 1 µg)) / (0.37 mg/m3 x 0.2 l/min x (1 m3 / 1000 l)) = 5 mins.
    • No problem at all there! We will still aim to sample for 4 hours or so, according to guidance, in order to obtain a representative sample of that person’s working day.
  • Let’s go through another example. Say we are on site sampling for total inhalable dust. The COSHH Trigger Value is 10 mg/m3 8hr TWA. The laboratory’s LOD is 0.5mg. And the flow rate for an IOM head is 2.0 l/min.
    • Minimum sampling time = (10 x 0.5 mg) / (10 mg/m3 x 2.0 l/min x (1 m3 / 1000 l)) = 250 mins.
    • We will still aim to sample for just over 4 hours. Which according to guidance should normally (depending on the circumstances) be enough to obtain a representative sample of that person’s working day.

What will the new RAG system look like?

On the note of colour coding, we do intend to keep a RAG system in our reports (Red, Amber, Green). With a view in most cases to apply a cut off for a light green up to 10%, darker green from 10% to 25%, amber for 25% and up, and red for sample results exceeding the WEL. See figure 9 below. The need for and specifications of a RAG system can be flexible, as this depends on what we are wanted to demonstrate with the client e.g. demonstrating WEL compliance, identify areas where exposures are higher, showing areas of potential control measure failings.

The new RAG system, adopted by EEUK Group (Figure 9)

So although a 10% cut off for green is what we deemed as a conservative and progressive approach to ensuring WEL compliance in many situations. We also understand that there may be logistical limitations to achieving a 10% benchmark, which could cause the RAG system to become superfluous. We will deal with these situations on a case-by-case scenario.

For example, let’s go back to ‘’What about the laboratory’s limit of detection?’’ above. Say we are on site sampling for respirable crystalline silica. The WEL is 0.1 mg/m3 8hr TWA. The laboratory’s LOD is 0.01mg. And the flow rate for an IOM head with Puf insert head (to sample the respirable fraction) would be 2.0 l/min. The minimum sampling time = (10 x 0.01 mg) / (0.1 mg/m3 x 2.0 l/min x (1 m3 / 1000 l)) = 500 mins = 8.3 hours. Our options are as follows; sample for the full duration (8 hours of sampling may be logistically challenging to achieve), find a lab with a more sensitive method, use a sampling method with a higher flow rate (such as a cyclone head at 2.2 l/min although this would only take the minimum sampling time down to 7.6 hours). Barring us finding a lab with a more sensitive method, or being able to sample for the full duration, the 10% benchmark would be unobtainable. Aiming for 20% of the WEL though, we could achieve this with a 250 minute sampling period. Which is much more reasonable and roughly how long we would have sampled for anyways (to obtain a sample representative of the full working shift).

The purple colouring is set to indicate where emphasis should not be focused solely on WEL compliance. COSHH Regulation 7 (7) states that control of exposure shall only be treated as adequate if any workplace exposure limit approved for that substance is not exceeded as well as the principles of good practice for the control of exposure to substances hazardous to health (set out in Schedule 2A) are applied, in all cases. And in the case of substances which are assigned specific risk phrases (May cause cancer, May cause inheritable genetic damage, May cause cancer by inhalation, May cause sensitisation by inhalation, or May cause sensitisation by inhalation and skin contact), any other substance which the risk assessment has shown to be a potential cause of occupational asthma, or it is listed under COSHH Schedule 1, then ALARP applies.

We see there are other points to meet. Not just WEL compliance. Purple coloring gives emphasis to those substances which fall under ALARP and so a higher level of control is required which may not be reflected in the exposure sampling result (albeit at a ‘low’ level shown under the RAG colouring). We have selected purple to cover for ALARP substances but also for substances which ‘broadly equate to ALARP’ in their control requirement. For example, welding fume, process-generated silica and manganese are highly hazard but do not fall strictly under the definition of ALARP as described under COSHH 7(7) wording. If we look at COSHH Schedule 2A part C however, this states ‘’Control exposure by measures that are proportionate to the health risk’’. And so for these examples it can be argued that a high level of control is required, a level which broadly equates to ALARP. In those circumstances we also apply the purple colouring to emphasis the high risk (despite what the sample result figures show us) for that substance/process. This will be backed up with our observations and qualitative analysis undertaken whilst we are on site.

As discussed above under ‘What if the number of samples I can take within each SEG is not exactly 3?’. The RAG system we show above is based on a 10% benchmark from 3 samples taken within an SEG i.e. the screening criteria. This RAG system is therefore based on an SEG of 3 samples. And so colour-coding individual results to show WEL compliance has been undertaken with caution, as this criteria is meant to assess compliance across the group. We do emphasis that, according to the BOHS/NVvA guidance, if even 1 of your 3 SEG samples is above the 10% benchmark in your screening test then that shows non-compliance for the group. We will therefore endeavour to arrange the sample results table into groups of SEGs where possible so that this is shown more clearly. For example, if two results are shown as green but the third sample in that same SEG is not green then the whole of the SEG is taken to mean non-compliance to the WEL.

But what about the 10% mentioned in the BOHS/NVvA guidance?

Let’s not forget the decision logic behind the BOHS/NVvA guidance. As summarised:

  1. In your SEG of 3 samples, if one or more of the results is over the WEL this means non-compliancee. the probability of exceeding the OEL is unacceptable.
  2. In your SEG of 3 samples, if all 3 of the results are <0.1WEL this means compliancee. the probability of exceeding the OEL is acceptably low.
    • This benchmark can be adjusted if the SEG samples are more than three. E.g. four samples would need to be <0.15WEL to show compliance.

If either scenario A or B are not applicable, then compliance or non-compliance cannot be decided (via the criteria outlined under the BOHS/NVvA guidance). And additional exposure measurements may be needed to determine compliance and increase your confidence in the results. Via the BOHS/NVvA guidance, a statistical test requires at least 6 valid exposure measurements for one SEG. The test confirms, with 70% confidence, whether less than 5% of exposure measurements in the SEG exceed the OEL.

EEUK Group have considered this in-depth and have decided to follow the RAG system shown in Figure 9 above. We believe this will aid our clients in complying with the law, show a truer reflection of what your exposure monitoring data means (as opposed to applying the previous 50% rule of thumb), whilst balancing the practicable/logistical aspects of needing to meet this new lower benchmark. Remember, one or more results >0.1WEL but <0.25WEL is still very likely in compliance. Being above the 10% benchmark doesn’t necessarily mean that the WEL has been exceeded.  The 10% figure is about demonstrating compliance for low concentrations with a minimal number of samples. THE BOHS/NVVA guidance itself does not mention colour coding or drawing cut-off lines at 10%. It is quite possible to have sample results ranging above 10% of the WEL and still be compliant with the limit. We just don’t have as high a degree of confidence as we would like to in this ‘dark green’ area as we would get in the ‘light green’ area.

Where should my focus really be then?

An occupational hygienist’s job is to achieve good control of the risk. We have talked about how to get a reliable answer re WEL compliance. But… getting this reliable answer means the benchmark needs to tighten and we may need also to take more measurements than we are used to. Therefore, it’s important we put compliance testing in its proper place in regards to achieving adequate control, which is ultimately what the law requires. Testing for WEL compliance must be considered in the wider context of assessing and controlling the risk. There is no point in investing the time and money into sampling unless good practice is also being applied. And so we will leave it here on this final and most crucial note… the Principles of Good Practice for the Control of Exposure to Substances Hazardous to Health (shown below in Figure 10).

To reiterate, the BOHS/NVvA guidance is about statistically demonstrating compliance with a WEL. It is purely about statistics and deliberately does not take into account any contextual information. The BOHS/NVvA strategy is also logistically challenging as it can involve needing to attend site on three separate days and taking at least three samples on each day. This BOHS/NVvA guidance is aimed to demonstrate WEL compliance, and does not extend beyond the remit to the other reasons for undertaking monitoring. And so, we should not get too hung up on the 10% of WEL benchmark. It is quite possible for an occupational hygienist to draw conclusions from visual assessments alone.

Remember, measuring exposure does not in itself decrease it!

Good Practice for the Control of Exposure to Substances Hazardous to Health (Figure 10)

References

  • ‘Monitoring for Health Hazards at Work’, Cherrie et al
  • BOHS ‘Exposure Magazine’, Issue 4, 2018, Trevor Ogden
  • ‘Testing Compliance with Occupational Exposure Limits for Airborne Substances’, BOHS / NVvA, 2011
  • BS EN 689, ‘Strategy for Testing Compliance with Occupational Exposure Limits’, revised 2018
  • COSHH Regulations (L5)
  • BOHS document ‘Progress on the European Standard on Testing Compliance with OELs’
  • Presentation by John Ingle ‘The Statistics of Exposure to Chemical Substances: Compliance with exposure limits’, March 2016
  • Presentation by Kelvin Williams ‘Statistics: The one day survey’, Sept 2019