Monday, April 28, 2014

Opps! How we learn from Surprises – Part 1

I am a modeler.  I just love to model to predict exposures but that does not mean I do not understand and appreciate the value of getting real data.  Actually measuring what is happening in the real world is very often a serious eye opener.  Just when you think you know what the situation is – Mother Nature throws you a real curve ball.

One of the biggest surprises of my career happened when my friend and colleague, Bill Shade, and I were constructing an exposure laboratory.    We were the only exposure assessors in a Toxicology Laboratory and convinced our management to give us a room large enough to construct our lab.  The idea was to build a wood-framed room 8’ x 8’ x 8’ with gypsum board walls, ceiling, a small window and a door.   We also put a duct into the room with adjustable exhaust air-flow so that we could vary the ventilation rate within our little room.   We almost always used fans to assure good mixing of this small volume.  This space was large enough to conduct emissions studies that included having a person in the room performing tasks.  I will go into the details and value of this setup in a future blog; however, I only mention it here to set up what we found when we were doing a “shake down” and calibration of this facility.

I covered the details of using a tracer gas technique to evaluate ventilation rate in a previous blog so I will not go over them here.   We decided to use carbon dioxide as a trace gas.   It is cheap, relatively non-toxic and readily available in a large tank which worked out well since we did not require it to be portable.  At the time there was only about 340 ppmv ambient CO2 in the atmosphere (Now because of our constant emission of this gas it is about 400 ppmv).    We had an IR detector that would measure CO2 in real time and our plan was to put in a few thousand ppmv, measure the fall off with time and calculate the mixing air changes/hour.

Our first test was to see just how “tight” our room was when it was closed up and all ventilation from the duct turned off.    We pumped the CO2 into the room to about 5000 ppmv and measured the fall off in concentration.   To  our surprise the CO2 disappeared quite fast!   Indeed, we calculated about 5 air changes per hour in a closed room! 
We went inside the room, turned off the lights within the room and caulked all the cracks we could see.  When we got done with it,  it was like a tomb!    You could sense that the room was very well-sealed.   We re-ran the test and got essentially the same result!

At that point we were guessing that there was a CO2 “sink” in the room such that the gas was being absorded into the walls surface rather than being exhausted from it.   

In our next test, we turned on the CO2 gas and let the concentration rise to about 2000 ppmv while we left the C02 ON for 8 hour while monitoring the resulting CO2 levels in the room.    If my memory serves me correctly, the room concentration of CO2 essentially stayed the same for 8 hours.   Thus, there was definitely a sink within the room and it was a fairly deep sink in that we did not exhaust it in the 8 hours we ran our tests.

Our next step was to prime and paint the interior of the room and re-run the tracer gas test.   We got about 0.1 mixing air changes/hr which is what one would expect from a highly sealed room.    Clearly the unpainted wallboard was a sink for CO2 gas and painting it sealed it enough that it was no longer functioning as a sink.

A year or so later I was at a meeting with a colleague who knew a lot about gypsum board and I mentioned our experience.   He told me that there often is a fair amount of residual calcium oxide (CaO) in the gypsum board.  As such, the thesis is that calcium oxide (CaO) is fairly reactive with the CO2 to form calcium carbonate (CaCO3), thus we believe that we found the deep, strong and one-way sink of fresh wallboard for C02. 

This certainly explains what we measured.   It also is a good reason to test your models in the real world.

Years later another friend and colleague, Bob Morrison, did his Master Degree project measuring the sink effect using various tracer gases and chamber wall material.  Someday I hope to convince him to publish these very interesting data.

Sunday, April 20, 2014

Risk Assessment in a Small Autobody Shop

This week’s blog does not get into the technical aspects of modeling but more into a lesson in the human aspects of worker exposure/risk assessment that I received as a young man.

When I was a graduate student, many years ago, I attended a neighborhood New Year’s Eve Party.   While doing some social drinking and hanging out with my neighbors, I mentioned that I needed to find a real world project to complete my Master’s Degree program at Drexel University.   They all knew that I was in the field of Industrial Hygiene.   One of the folks at the party said: “ You should  come over to our autobody shop.  We get high a few times a week on the paint fumes. “   He did not have to suggest it twice!    I spent time in that shop every day I could for the next year, including most of my vacation days, measuring airborne vapors, particulate and metals along with noise.  

It was a two-man, jointly owned and operated “bump and paint” shop.  It was extremely difficult work, almost all of it was stoop label, spent bent-over the surface of an automobile cutting, hammering, applying resin or sanding.   Relatively little time was spent doing actual painting.  It was all “piece work” – that is, they were paid by how much work they got done.   As such, they worked hard with few breaks for 10-12 hour days during the week and half a day on Saturday.

The never saw anyone from OSHA but they did have to have insurance and for that they had to have a good hood for painting whole cars.   The hood was large enough to hold a car with enough room for a painter to spray paint it.  It had a band of filters and a large fan that provided 100-150 linear feet per minute flow over the filters comprising the entire wall at its far end.   Spraying an entire car took less than an hour and when the hood was running the breathing zone concentrations of paint vapor or particulate were well below any of the then current OELs; however, they could not always run the hood.

All of the air going through those filters was exhausted outside.   When the outdoor temperatures got below 50F, the heating system in the garage (100,000 BTU/hr from the main heater + 50,000 BTU/hr  from two kerosene space heaters) could not keep up with the loss of warm air and the owners would run the hood either intermittently or turn it off completely.   Suffice it to say that the airborne and breathing zone concentrations went up dramatically.  Indeed, it was not unusual to see the painter emerge from the booth visibly intoxicated.   What was particularly noteworthy to this young investigator was to see him prolong the experience by then drinking a beer!

One day, before I went to my work, I hung an integrating sound monitor on one of the workers who advised he was going to do a “chop job” that day.    This apparently involves cutting two wrecked cars in half and joining the good halves of each to make one good car which I assume wound up on a used car lot somewhere. (Buyer beware!).  This is done with a pneumatic device called a “chipper”.   When I evaluated the dosimeter that evening he had received 160% of an allowable daily noise dose (8 hr/90dbA Standard – 5 dB doubling rate).

The bottom line for these workers and their risk was that they were subjected to intense levels of potential overexposure but of limited duration.    The use of a reasonably well fit half-faced mask with the appropriate canisters and filters for a few hours a week would have protected them from overexposure to vapors or particulates. Similarly, the use of hearing protection during the limited times where noise levels were high would have also protected their hearing.

I was eventually, able to get this work published but my first experience with the American Industrial Hygiene Journal peer reviewers was personally crushing.   They were so critical and dismissive of the work that if I believed their comments I would have left the field of Industrial Hygiene.    I submitted the work to the journal of the British Occupational Hygiene Society and their reviewers were much kinder to a young author.

If you are interested in a copy of this paper, please send me a request at

Toward the end of the study, my wife and I invited both of my neighbors and their wifes over for a roast-beef dinner.  After dinner I presented them with a draft copy of the report and told them what they could do to protect their health from these potential overexposures.   I would love to report that they took all the advice but I think you know that is not a realistic expectation.  


Sunday, April 13, 2014

Hormesis for Setting OELs: Are you kidding me?

I have been interested in the topic of hormesis for a long time.   Even though I am not a toxicologist, some years ago I was responsible for running an inhalation toxicity laboratory in a large chemical company.   When I mentioned hormesis to my toxicologist colleagues in those days the standard response was for them to roll their eyes and dismiss it as nonsense and definitely non-science.  Their minds were closed to the ideas presented within this concept.  I am bringing it up in this blog in the hope that you will be more open to the possibilities it might present.

I am reproducing the graph I put into the blog last week for convenience.   It shows hormesis as one of 4 possible models presented for low-dose extrapolation. 

Indeed, hormesis is an old concept and is actually referred to in the famous 16th century quote from Paracelsus that has been paraphrased over the years as:  “The dose makes the poison”.   His actual quote is:

Poison is in everything, and no thing is without poison. The dosage makes it either a poison or a remedy. [emphasis added]

I challenge you to think of any medicine that is NOT; by definition and design, good for you at a reasonably low therapeutic dose and a poison at high dose.  A common, but very real, example in most of our medicine chests is acetaminophen (common trade name: Tylenol).  This is a ubiquitous OTC drug, which works quite well for aches and pains and fever at the recommended dose, but is deadly to our livers and lives in doses that are only a 3-5 times higher.

All micronutrients also follow the hormesis curve.  For example, consider the effect on your eyes of not having enough vitamin A in your diet or as a separate supplement.

Now consider a very common industrial chemical: ethanol.  Many of us enjoy the acute effects of ingesting ethanol (aka ethyl alcohol, drinking alcohol) as the active ingredient in beer, wine and hard liquor.  Most of us know that drinking too much can be acutely and chronically toxic resulting in alcohol poisoning and liver disease.  What some may not realize is that, in moderation (i.e., a “low-dose” of 2 drinks per day), one can enjoy the mood altering features of this chemical while also benefiting from an increase in cardiovascular health over those who do not drink any alcohol.  Most Doctors admit that this is true but have fallen short of recommending that nondrinkers start drinking for various reasons.  I will not go into the details of their reasons or the scientific evidence here but the benefits have been fairly well established to be true by research.

There have also been paper published in which low doses of very strong carcinogenic substances have results in less cancer (versus untreated controls) in animal experiments at low-dose. 

Professor Edward J. Calabrese, School of Public Health, University of Massachusetts, Amherst has devoted a considerable portion of his professional career to the study of hormesis.   He has written extensively about it and has organized workshops in which he always invites all sides to this discussion.   Most of these discussions are summarized and presented on the web site:

Dr. Calabrese once asked me to present my opinion of hormesis in low-dose extrapolation and risk assessment for a publication and one of Dr. Calabrese’s workshops:

My conclusion from almost 10 years ago remains the same today; specifically:  “… we will only be able to move forward with hormesis as a default hypothesis [for low-dose extrapolation] after the development and use of tools from the realm of molecular biology.” 

The science of the “omics” (genomics, proteomics, etc) has come a long way in the intervening years but from what I can tell it has not explicitly addressed the above issue.   Indeed, unless and until we really understand the effect of low dose at the human tissue level, we will not know which low-dose response curve is the best as either a default model or specifically applied to any substance.  

Until that happy day I believe that we need to rely on a science-informed political process to do the best we can to describe the actual risk at the OEL along with the frank admission of the uncertainty in that description.

Sunday, April 6, 2014

Low Dose Extrapolation for OEL - Part 2 (Many Different Models)

Last week we discussed the general topic of low-dose extrapolation for the purpose of setting Occupational Exposure Limits (OELs).   This week we will get more specific; namely, how do we go from high risk rat data to low risk extrapolated estimates for humans?   

As we established in previous blogs, in  most cases the data ends at an effect level in the animals at around 10% but we want an OEL in humans at a level of effect much lower than this value. Theoretically and conceptually, we want to protect “nearly all” who might be exposed.   Low-dose extrapolation is designed to inform us as to what “nearly all” might mean in operational terms.

A basic problem exists that many models will fit the actual rat data quite well but give VERY different predictions when extrapolated to low risk at low dose.   A slide that I have used in teaching for many years is pasted below: 
Please note that the normal range of available toxicological data occurs all the way to the right of these curves and at high dose.   All FOUR of the very different models on this graph can do a perfectly good job of describing the available data but all are quite different in their predictions of response/risk at low dose.  

Consider the curve labeled “False” Threshold (supralinear).  Like all the curves it starts at zero dose with a background level of the toxicological effect under consideration.   Let us for sack of this example, assume it is cancer which has a relatively strong background level in humans and rats.   As soon as the dose come up from zero the putative risk predicted from this model rises rapidly until it reaches a local peak.  Then it declines to very close to background (i.e., the "false" threshold) and finally, proceeds upward into the normal range of toxicological response.  This may seem somewhat fanciful; however, this “false” threshold curve has been proposed by some to describe what may be happening with endocrine disruptors.

Let us now look at the point labeled “threshold”. Here the putative effect of dose reaches the background line at a relatively high dose.   Below this dose it is assumed that there is NO harmful effect from exposure for anyone or at least for the vast majority of the population.   Indeed, much of the conventional wisdom in toxicological circles today is that this situations exists for a majority of chemicals especially those considered to be non-cancer causing agents.

Now let’s examine, the “No threshold (sublinear)” line and model.   This is where the conventional wisdom currently puts carcinogenic (especially genotoxic carcinogenic) risk assessment.   The assumption being that even a single molecule can mutate DNA and cause cancer and that 2 molecules presents twice the risk.

My personal favorite model and line on this slide is labeled “Hormesis”.   Here there is actually a LOWER risk predicted versus background as the dose comes up from zero.    It eventually, climbs up the curve into the normal range of toxicological data; however, I think you can see what a dramatic effect this curve might have on risk assessment if it were generally true.  I am going to devote an entire blog to the subject of hormesis but I wanted to include it here just to show the comparison with the others.

I have studied this dilemma for a long time and in my opinion we are not in position today to scientifically prove that any of these curves are true representations of reality for most chemicals under consideration.   ALL of these models/curves/predictions are in effect, arguments without data.

I have heard it said that “There are many questions that cannot be answered but that must be decided.”    I believe that it is up to the best science available to inform these political decisions as to which model(s) we use and where we draw the line in setting OELs. Even after a model is chosen, the uncertainty around the predictions of that model needs to be disclosed so that those viewing and using the OELs understand exactly where they came from.

The next blog is designed to open your mind relative to possible and seemingly strange reality of hormesis and offer my ideas on want needs to happen to determine just what is occurring at low doses in human tissue.