Comments on NIOSH Update of Carcinogen Classifications and Risk Levels

NIOSH is proposing an update of their “Carcinogenic Classification and Target Risk Level Policy for Chemical Hazards in the Workplace”.  My friend and colleague Chris Laszcz-Davis asked that I comment on these changes so that she could forward comments to the American Industrial Hygiene Association.  I thought I would share these comments below on this blog.

The new classification policy proposes using the assessment schemes used by the NTP, EPA and IARC which makes plenty of sense because to do so will enhance harmonization and will keep NIOSH from reinventing the wheel. 

The use of risk based exposure limits (RBOEL) for carcinogens is a step directly into the 21st century for NIOSH.   The chosen benchmark of one in 1000 risk of cancer at the 95th lower confidence limit for a 45 year working lifetime seems imminently appropriate to me.   Mention is made in the document that this risk is at least an order of magnitude higher than the cancer risk permitted in the US for the general public.   What is not mentioned in the document is that, according to the Bureau of Labor Statistic, the risk for accidental death occurring during employment in a working lifetime in the United States is slightly higher than 1 in 1000 over the entire US worker population and very much higher for some classifications of workers (e.g., construction workers, commercial fisherman).  Note to viewers of TV show:  Deadliest Catch.  It really is perhaps the most dangerous occupation in the US.   It is certainly the most dangerous of the occupations that I know of that are tracked.

What is even more interesting is that these accidental deaths of workers represent actuarial data; that is, this the portion of working folks who actually died as evidenced by historical records.   The risk of cancer from exposure to a carcinogen on the other hand is putative and the result of low dose extrapolation of animal data.   The extrapolation also assumes that there is a linear dose-response all the way down to exposures that are many orders of magnitude below those tested on animals.   It also estimates the occurrence of cancer and not the rate of death from cancer.  

Low-dose extrapolation modeling will be the topic of at least one future blog but suffice it to say here that when you extrapolate with one model (linearity) far below the level of data you are essentially presenting an argument without data.

Given all these factors, the criterion outlined by NIOSH for RBOELs for carcinogens seems perfectly reasonably to me.

I also agree with the decision to make the RELs risk-based; that is, NIOSH will no longer consider the technical achievability (i.e., ability to control exposure) in establishing these limits.

I disagree with one area of the proposal; namely, the treatment of RELs set when the reliable analytical quantification limit is higher than an REL set using the above criteria (i.e., the 1 in 1000 quantitative risk level).   Here NIOSH is proposing using a higher REL with an AF notation for Analytical Feasibility.  This policy implicitly ignores the ability of modern exposure science to estimate exposures in essentially any scenario by physical-chemical modeling.    My suggestion would be to have two RELs in this instance.   The first would be the standard REL using the above criteria and the second an REL-AF to reflect the analytic realities. 

For many, perhaps most, carcinogens this will bring down the OEL significantly and this will undoubtedly present problems for some who need to comply with these new lower levels.   A colleague asked me to comment on this issue which I will do in the next blog.

1994 Contract with America: Lessons for Risk Assessment

This happened 20 years ago but I think this episode still provides food for thought today.   Back then I subscribed to a Business Publishers, Inc., Newsletter that was entitled "INDOOR POLLUTION NEWS".   I Googled it recently and seems to be gone; however, in one of their issues from around 1994 they  presented an editorial opinion on Risk Assessment (RA) that I think is still worth discussing in 2014 .  The piece is presented below:

"RISK ASSESSMENT - One Section of the Republicans Contract with

 America calls for regulatory agencies to use risk assessment, that

 is the scientific measurement of risks to be regulated-before rules

 are promulgated.  The focus on risk assessment by business and then

 made into a campaign issue by the Republicans seems clearly a

 subterfuge to gut environmental laws. ...  The science behind much of

 risk assessment is costly and in some cases problematic, rather than

 err on the side of caution for public and environmental health, the

 thrust now will be to negate rules and regulations that cannot be

 backed by ironclad science."

Wow!  This really struck me at the time because it was the strongest and most direct statement I have seen relative to the fears of the governmental, regulatory risk assessment/management community.   It got me to thinking about what I believe regarding this subject.   What I came up with long ago I still believe but with perhaps some of the more youthful naivete removed.  

My opinion is that RA is just a tool but it should be a tool that is used honestly.  If one uses the worst case then one should also present the "best" case and, if possible, the "average" case in a RA.  UNCERTAINTY born of a lack of basic knowledge makes worst cases worse while it broadens the error bands around the evaluation of risk in general.  Ideally and perhaps, in retrospect, naively in 1994, I saw the proposed legislation as asking for the honest reporting of risk knowledge and uncertainty.   Indeed, open reporting does not have to result in the “gutting” of environmental laws but it could allow for better decisions and better allocation of resources.  One-sided RA (only worst case) tends to hide behind the appearance of certainty.  Serious decisions need to have all the "cards on the table face-up" and all the uncertainty revealed in the most open and honest manner we are capable of. 

This blog has been discussing OELs of late.  There is plenty of uncertainty around the setting of OELs but the gist of the message so far has been that we need to do the best we can in rationally describing the process and for me that mean quantitative modeling predicting the range of risk at the OEL.   If the various models are very uncertain they are at least open relative to their particular quantitative assumptions.  It is this openness that can lead to identification of areas for improvement (spelled "research" and "money"); however, if we continue to cling to a somewhat hidden or ill-defined basis for setting OELs, then I am afraid that improvement and progress will come very slowly if at all.

Finally, it should be understood that the technical details and error bands around any RA only represent some of the “cards", we still have plenty of subjective political and cultural values that need to be acknowledge, openly faced, reported and factored into most decisions including OELs.  My vote is  that we take the onus off of technical RA to do everything and recognize it as a valuable tool that needs some nurturing (again that is spelled “research money”) if it is to really make good on its potential to render more light than heat regarding any question of risk.

Uncertainty/Safety Factors in OELs

From my perspective, most OELs are set these days by taking a No Observed Adverse Effect Level in a repeat dose toxicology study (NOAEL or NOEL) and dividing it by an uncertainty or safety factor (SF).  Last week’s blog was about NOELs.   This week it is about uncertainty/safety factors. 

As we discussed last week, NOELs are in reality typically frank effect levels for the adverse health effects from exposure.   The portion of any tested animal population having a frank ill effect from a NOEL exposure has been shown to range from 2-21% depending on the experimental design and shape of the dose-response curve.   Assuming people are as sensitive as rodents, an exposed population of 100,000 workers and a 2-21% response results in 2,000-21,000 workers whose health has been adversely effected by exposure to the NOEL.  Clearly, the NOEL would not make a good OEL and some adjustment needs to be made.  Thus, an uncertainty or safety factor is applied to bring the OEL down to be a fraction of the NOEL.

Various schemes have been forwarded to size these factors.  A paper is available online that does a reasonable job of providing the details and the references for the standard OEL/Safety Factor approach written in 2000 by Dr. Robert H. Ku, Ph.D.,CIH.:  http://www.safebridge.com/pdf_docs/overview_of_setting_oel.pdf

Here is an excerpt from that paper dealing specifically with safety factors:

The number and magnitude of these safety factors depend on the quality of the data. In general, some of these safety factors may include: (1) a factor from 1 to 10 for animal-to human (interspecies) extrapolation (if the NOEL is based on animal data), (2) a factor from 1 to 10 for human-to-human intraspecies variability in response, (3) a factor from 1 to 10 to consider study duration (a long-term study being more helpful than a short-term study), (4) a factor to consider the persistence of the drug in the body (or elimination half-life), and (5) a factor to accommodate for absorption efficiency by different routes of exposure.

Thus, it appears that the overall safety factor can range from close to 1 to a factor larger than 1000. In my experience, they typically come out to be significantly less than 100. 

As a range, I have seen safety factors as low as 3 for fully reversible, local tissue irritation response effects and greater than 1000 when considering highly uncertain data for a dreaded health effect (i.e., cancer).

From viewing it for many years, my sense is that the entire process is somewhat subjective and, in my estimation, to a certain degree driven by the practical ability to control the exposure at whatever limit is decided upon.

   

Indeed, one of the pioneers in our field, Dr. Stanley Roach along with his coauthor Dr. Stephen Rappaport published a paper on this subject in 1990.  The title of this paper and its reference:  But they are not thresholds: A critical analysis of the documentation of threshold limit values, American Journal of Industrial Medicine, Volume 17, Issue 6, pages 727–753, 1990.

Their primary observations and conclusion:

•       “Upon analysis it was found that, where the exposure was at or below the TLV, only a minority of studies showed no adverse effects (11 instances) and the remainder indicated that up to 100% of those exposed had been affected (8 instances of 100%)”

•       “…a surprisingly strong correlation was found between the TLVs and the (workplace) exposures reported in the corresponding studies cited in the Documentation [of the TLVs](emphasis added).”

Perhaps a new study should be done to see how much this situation has changed in the last 24 years.

That fact remains, however, that in most cases the size of the safety factors are driven by guidelines and are not subject to stringent rules.   We are often advised that this is necessary to allow for “professional judgment”.    I think we can all see how such a subjective system can potentially lead to somewhat unconscious bias in which the resulting OELs might become strongly correlated with workplace exposures as the study by Drs. Roach and Rappaport found.

In recent years there has been a scheme to strictly apply safety factors to toxicology data put forward by the REACh regulations in Europe.   This is done in the determination of the recently devised OEL class known as DNELs.  In the interest of time and space I am not going into the details of the DNELs and the safety factors here but suffice it to say that the process in general has cause a lot of consternation primarily because of its prescriptive nature.  

So where are we with all of this?   My sense is that the current system is clearly in need of some improvement.   The National Academy of Sciences seems to agree and has forwarded a rational modeling scheme for determining or estimating the amount of risk that exists at any exposure including the exposure limit for both carcinogens and non-carcinogens.   This is Chapter 5 of the so called “Silver Book” which you can download for free (You can download just Chapter 5 or the entire book) at:   http://www.nap.edu/download.php?record_id=12209

Even though this book has been out since 2009, the OEL setting community has not embraced the recommendations.   Some folks in our community including this blogger are asserting that it should now become an increasingly prominent topic of discussion within our world.

NOELs & LOELs which form the Basis of OELs

In a previous blog I made the statement that, from my perspective, the basic approach of the ACGIH TLV Committee breaks down to taking a NOEL (No Observed Effect Level) or LOEL (Low Observed Effect Level) from a toxicological study in animals and dividing it by a safety factor.    The resulting Occupational Exposure Limit (OEL) is designed to protect “nearly all” workers from the adverse health effect that was seen in the animal model.

Since the NOEL or LOEL are important in this process, it would seem worthwhile to investigate them further.  The NOEL as its name implies is the dose where there is no observable adverse effect on the tested animals.  In a practical sense it is the dose whose results are indistinguishable from the control group.  The comparison between the responses of animal test groups is typically done with statistical tests to determine if there was as statistically significant effect caused by the dose.   In my work with Toxicologists I saw that their pretest estimation of the highest dose that would produce a NOEL was much sought after and the matter of a lot of deliberation on their part.   Indeed, they know that the NOEL (or LOEL) will drive a measure of the allowable exposure to that substance which is what we have in any established OEL.

Some folks make the distinction between the No Observed Effect Level (NOEL) and the No Observed Adverse Effect Level (NOAEL).  The NOEL being where no dose related effect can be discerned.   The NOAEL indicating that no adverse health effect occurs at that exposure.   For the purpose of this discussion let’s consider them the same.

Note that the LOEL (or LOAEL) is a dose in a test group that just missed being a NOEL (or NOAEL).  That is, it was just a tad too high so that a slight but statistically significant adverse health effect was detected.

So when we have a repeat dose toxicology study with a NOEL does that mean that we have truly found a threshold dose below which NOTHING bad will happen to any rat given this or a lower dose?   The clear answer to that question is NO.    The fact of the matter is that a true threshold for any large population (that is, ALL rats or ALL people), if it exists, is essentially impossible to determine given the above methodology of testing groups of animals with 5 to 50 animals in each group.    It is a matter of statistical reality that is best described by the binomial theorem.  

I will not go into the details of this theorem here except to state that it predicts the occurrence of chance events given some level of true effect or a true relationship (e.g., half of all coin flips will be heads).   Casino operators know all about this theorem and it is the reason that they will always have the “house” edge and will always come out on top for any game they offer in the long run.  However, I digress; the cold truth is that it has been convincingly shown that zero response (i.e., the NOEL) in toxicology studies could represent a response level as high as 20%. Indeed, based on a computer simulation study, Leisenring and Ryan (1992) show that the average NOEL for quantal data as described above actually could be a 3 to 21% adverse response level depending on the experimental design and shape of the dose-response curve.

So on the strength of this reality let’s assume that the typical NOEL represents, on average, a 10% effect level in rats.  Well, most of us are clearly not rats and we may be more or less susceptible to the toxic effects of the substance.   Thus, besides the uncertainty born of testing a limited number of animals, we have another reason that safety (or uncertainty) factors are applied to (divided into) NOELs and LOELs with larger factors being applied to LOELs for obvious reasons.   I will discuss the sizing of safety/uncertainty factors in a future blog but I think you can see that it could be considered a tricky business, at the end of which we have an OEL designed to protect “nearly all” but without any reasonable quantitative explaination as to what that means.

Staying with the NOEL for now, one might ask is there a more elegant way to get to the NOEL operationally defined as the estimated 10% effect level?   Fortunately, the answer is YES.  The EPA has developed a very nice piece of PC freeware entitled:  Benchmark Dose Modeling Software (BMDS). The software uses all the dose-response data in the toxicological study, not just the NOEL, and then calculates the estimated BMD or Benchmark Dose along with its uncertainty bounds.  If you set the BMD to 10% you have a more elegant way of estimating the NOEL.  Online training modules on the theory and use of this freeware are available at:   http://www.epa.gov/ncea/bmds/training/captivate/bmd-intro/bmd-intro.html

  

You may or may not want to become a BMDS maven but I urge you to at least look at it.  The take-away from all this is that given typical toxicological data we essentially cannot prove the existence of a true threshold, although one may indeed truly exist.  We can describe the level of predicted risk at any dose at or below any exposure including the exposure at the OEL along with its uncertainty using modeling.   That will also be the subject of a future blog.

Leisenring, A.J., and L. Ryan: Statistical properties of the NOAEL. Regul. Toxicol. Pharmacol. 15:161-171 (1992).

15 min STEL in IH MOD

The blog before last I discussed the subject of point-in-time vs time-weighted average airborne concentration that might occur in an exposure scenario.   I went over how the current version of IH MOD makes life easier in estimating the point-in-time concentration at any time for any model.   Of course, we are typically not that interested in point-in-time concentrations except in situations where we have a Ceiling TLV or perhaps we are concerned about the possibility of obtaining lower explosive limit concentrations.

Because we are often working with 8 hour time-weighted average (8 hr TWA) TLV exposure limits or 15 minute short term time-weighted average (15 min STEL) TLV exposure limits, our primary focus is on the time-weighted average for these time periods.    Just to recap, the TWA is the integrated area under the C,t curve divided by the time interval.  IH MOD comes to the rescue here as well for 8 hour TWA concentrations because it calculates the TWA from the start of the scenario out to any time including 8 hours.  Thus, the 8 hr (480 min) TWA concentration is readily presented by IH MOD.

As I mentioned two week ago, the situation becomes a bit more complicated with you are looking for the maximum 15 min time-weighted average STEL during the exposure scenario.    Indeed, I went over some manual calculations that were necessary to get at that number.    I mentioned the possibility that this might “one day” be included in IH MOD.   Here is the quote from that blog:

“Perhaps someday we can get IH MOD enhanced to provide this calculation more directly.  My guess is that it would not be a trivial task but Daniel Drolet continues to surprise me.  He is a truly gifted programmer and I have no doubt that he can do it.  Indeed, what would be difficult for me could be relatively easy for him.”

Well Daniel took me up on this and did it!   He is truly a programming wizard and we are most fortunate to have him on our side.   In essence, the spreadsheet enables us to take any point in time along the scenario and calculate an approximate 15 minute TWA concentration forward  from that point.    See the screenshot below:

Of course, this allows us to pick the 15 minute interval that provides the maximum TWA concentration.

In this example it shows the maximum 15 minute STEL exposure occurring in the period 6.6-21.6 minutes into the scenario.   One  might (including me) think that it should occur in the period from 5 to 20 minutes.  The truth of it is that the 1.6 minutes after 20 minutes contributes more to the area under the curve than the 1.6 minutes after 5 minutes.  This slight shift was not intuitively obvious to me.   Daniel pointed this out to me when I questioned it.  It is another example of how these models provide insight.   

Daniel is a perfectionist and is working on making it more "elegant".  Future versions are inevitable; however, from my perspective is it good to go right now.    If you want to play around with this relatively untested BETA spreadsheet, please let me know (mjayjock@gmail.com)  and I will send you a copy.    I am sure that Daniel will appreciate any comments you have on it.  

Because of Daniel's tireless efforts we have what has been and continues to be the single best inhalation exposure assessment tool that I know of.

Making Your Own Exposure Limits (OELs)

Exposure limits (OELs) seem to be on everyone’s mind and for good reason; as exposure assessors we find it very hard to live without them!  A prime example comes from Hila Wright, CIH, MPH who sent me the following after I requested ideas for future blogs.

“Would you talk about your approach to how you handle compounds of concern that do not have published exposure limits, but do have a body of available tox and epi data? Would you do your own risk assessment and try to come up with your own exposure limit to compare any exposure assessment results against? What would be your approach (especially since some of the multiplicative factors used in risk assessment seem somewhat haphazard)? How do you distinguish the good research papers from the bad (this is an area that I have trouble with). Or do you stick to a more qualitative approach, such as control banding?”

Hila, I am going to try and answer your questions directly and succinctly; thus, some of the details will not be included but will form the subject of future blogs.   As such, my answers are going to be somewhat shallow but hopefully still helpful. 

I have never been shy about setting working exposure limits when none are published and there is some available toxicology data, especially repeat dose (chronic or sub-chronic) toxicology data on the chemical of interest.    If you would like to do this I would suggest first getting a copy of the Documentation of TLVs from the ACGIH.  It is not cheap but it has all the details of the TLV committee’s deliberations on the hundreds of chemicals for which they have set OELs.   If you read enough of these you start to see how they are doing it.   There are, of course, a lot of details and folks may chide me for making it too simple but I frankly do not see it as being very complicated for most substances.  From my perspective, their approach breaks down to taking a NOEL (No Observed Effect Level) or LOEL (Low Observed Effect Level) from a toxicological study and dividing it by a safety factor.  

I can understand how you would see the “multiplicative factors” in this process “[as]… somewhat haphazard”.  In fact, I have not found any written rules as to how large the safety factors need to be for OELs.  Indeed, this appears to come under the almost magical category of “expert judgment”; however, if you read enough examples in the TLV documentation for similar chemicals to the one you are interested in, you will get the idea of how large a factor to divide the NOEL or LOEL by.  You may even disagree with their assignment of any particular safety factor as I had done on occasion.

The next step I would suggest, especially when you are early in the process, is to seek out the advice and counsel of a toxicologist and explain to him or her exactly what you are trying to do and your thinking about what the OEL should be given the data. 

Companies do this all the time for their unique chemicals.   They typically have a Committee to set OELs.   The committee usually has Occupational Doctors, Epidemiologists, Toxicolgists and Industrial Hygienists. You would be a committee of one, or two if you have a toxicologist.  If you want to engage the company that made the chemical – just share your documentation with them and ask for their review and advice on your work.

Whatever you do, I would advise that you put down all your deliberations in writing; that is, document your process and decision for that particular working OEL.

Deciding good toxicological data from bad is an important issue.  Modern studies are often good because they are typically done under modern requirements for GLP (good laboratory practices).  Studies done by and for large companies or the government are often pretty good.  Again, seek the advice of a toxicologist partner as to the value of any particular study.

It used to be that studies from Eastern Europe were suspect but I am not sure that is still the case.  Clearly old studies from Eastern Europe or the old USSR might be considered suspect.  It all breaks down into having an understanding of what was done and how it was done.  If that information in not available in the report and the laboratory not well known then using the data could be problematic.  Again, consulting with a toxicologist could be very valuable.

I consider control banding something to be done when you do not have enough data to set a working OEL as described above.  That is, its use is strictly from hunger for data and requires a very conservative approach of trading conservatism for this lack of data.   That is, the bands should be appropriately and consciously set to be considerably biased and more overestimating of risk than if you had the data.

I hope readers will find the above quickie explanation helpful.  In any event, it has generated the following topics to be explored in a future blog; namely,

•          Meaning of NOEL and LOEL
•          Sizing a Safety Factor
•          Safety Factor Approach versus Estimating Risk Levels at the OEL

Point-in-time versus time-weighted average exposure

This week we go back to some of the nuts and bolts technical stuff of modeling.  Specifically, the difference between point-in-time and time-weight average concentrations and exposures.

The following curve shows a typical pattern of airborne concentration versus time for a process that starts at time zero and ends at time = 45 minutes.   Notice that after 45 minutes the source is cut off and the airborne concentration decays to essentially zero in about 80 minutes.

The point-in-time airborne concentration anywhere along the curve is easy to see.  For example, as indicated on the above curve at 7.2 minutes it is 109 mg/m³.   Note that in IH MOD it shows concentrations in both a graphic (the image above is an IH MOD cut and paste of its graphic output from this exposure) and in a table of concentration, time data points.

As it turns out, we are not typically interested in the specific point-in-time concentrations except in situations where short term peaks are important.   Indeed, they are important for exposure limits that have a “C” designation for ceiling concentration.   If the “C” exposure limit for the chemical in the graph was 150 mg/m³ the peak of about 240 show above indicates an overexposure.

There are very few chemicals with “C” exposure limits.  Most occupational exposure limits are either 8 hour (TWA) or 15 minute (STEL) time-weighted averages.   Here we need to understand the worst case time-weighted average exposure in the appropriate averaging time of interest.

Just for general information, the time-weighted average concentration for any time interval is the area under the point-in-time concentration curve, divided by the time interval used. 

For 8 hour TWAs IH MOD makes it relatively easy.    If you check the box on IH MOD that says “TWA on chart” it will calculate the time-weighted average for any time from zero out to the “maximum time for simulation” you put into the model.  Thus if you run the model out to 480 minutes (8 hours) you get the 8 hour TWA exposure which is 23 mg/m³.    If you run it out for 1440 minutes you get a 24 hour time-weighted average exposure from this scenario.

If you are working to a 15 minute short term exposure limit (STEL) it gets a bit more difficult to calculate in this example.  What we want is the time-weighted average exposure during the worst 15 minute period during the scenario.   This is the number we will compare to the 15 minute STEL exposure limit.

Perhaps the easiest way to do it is to find the peak point-in-time exposure and take the point-in-time values at 15 minutes before the peak to calculate the time-weighted average.   We would assume linearity in this relatively short period such the average exposure between each of these intervals represents the time-weighted average during this 15 minute interval.     

If we look at the predicted concentration output in IH MOD we see the peak point-in-time concentration occurs at 45 minutes (when the source dies) with a concentration of 240 mg/m³.   The point-in-time value at 15 minutes before the 45 minute mark (i.e., the 30 minute mark) shows an estimated point-in-time exposure of 225 mg/m3.   The average concentration occurring between the end value of 240 to the beginning point-in-time value of 225 is 232 mg/m³.  This is the time-weighted average concentration over the 15 minutes before the peak.   This is also the 15 minute time-weighted average exposure to be used for comparison to the 15 minute STEL. Notice that it is almost 10 times higher than the 8 hour time-weighted average exposure of 23 m³.

Perhaps someday we can get IH MOD enhanced to provide this calculation more directly.  My guess is that it would not be a trivial task but Daniel Drolet continues to surprise me.  He is a truly gifted programmer and I have no doubt that he can do it.  Indeed, what would be difficult for me could be relatively easy for him.

There is another important lesson in all this; namely, you should never monitor any batch operation for a STEL by starting the monitor at time = 0.   This example clearly shows that it takes time for the concentration to build to a peak.   If the source stops abruptly, start monitoring about 20 minutes before its end and end monitoring at the end of the source or shortly afterward.   You can also get a sense of where to monitor a decreasing evaporative source in a similar manner by first running IH MOD and seeing where the peak is predicted.   Please note that constant rate sources can be monitored for STELs any time after they achieve steady state concentration and IH MOD can help you with that estimation as well.