Epitoxicology

Dissecting Hazard Identification

When the Redbook risk assessment paradigm has been applied, the hazard identification step has typically been a subjective determination that initiates the more formal process of conducting an exposure assessment and characterizing the dose-response relationship for a specific cause-effect pairing should begin.  When hazard identification itself is thought of as a formal process, the iterative nature of risk assessment becomes apparent.  This is especially true when human epidemiological studies are pivotal in establishing that there is a causal relationship:  Hazard identification needs a weight of evidence evaluation, and a weight of the evidence evaluation a dose-response evaluation.  In addition, working with epidemiological data often requires estimation of the dose that the subjects in the cohorts received, which means the hazard ID may also require exposure assessment.  In other words, in order to conduct a proper hazard identification, one must do a risk assessment.

 

However, it probably won’t be exactly the same risk assessment as one needed for the decision at hand.  For one thing, the exposures will certainly be different for a cohort that was selected precisely because they are known to have unusually high exposures.  There also may be differences in other causal influences on a particular disease or health measure in a cohort that was studied and the population for which a risk estimate is needed.  Nonetheless, one would generally expect that a dose-response analysis used to establish that there is causal relationship should resemble the one used to estimate the risk.

Building Weight of the Evidence

So it would seem that the consideration of evidential weight and the dose-response assessment cannot be treated as entirely separate processes.   For example, Suter and Cormier (2011) characterized environmental epidemiology as risk assessment in reverse, where both epidemiology and risk assessment share a common need for weighing evidence and building the case for a causal relationship.  But, the weighing of evidence itself isn’t really reversed.  Yes, one should look for alternative plausible explanations when conducting a causal assessment, but the need to characterize uncertainty should make that part of the risk assessment as well.  For example, the fact that an association may be plausibly explained by both causal and noncausal relationships may be an important source of model uncertainty.  Where epidemiology does, or should, work in reverse is during study design.  Since the expectation of how the data are going to be used drives what data are to be collected, knowing how data becomes evidence will determine what a study looks for.  An epidemiology study designed to support regulatory decision making should look like a risk assessment.

Human Toxicology

But, most environmental epidemiology studies are not designed to support risk assessment.  The primary reason for this is pretty obvious and deep rooted: In spite of the fact that it is exactly what Hill (1965) advised them not to do, most environmental epidemiologist and their statisticians are trained to think that the weighing of evidence is accomplished with statistical significance test, and just about any test will do.  Find a p that is less than 0.05, publish the paper in a journal, and then call the press office to report that an association has been found.  Demonstrating a causal relationship is not on the radar screen.  The regulatory agencies need to deal with that problem (or so I’ve been told):  But they can’t do that very well at all because the studies weren’t designed for that purpose.

So, unless it serves some other purpose I don’t know about, as it is presently conceived, environmental epidemiology is broken.  Here is how it can be fixed:

• Ban p values.  There is no legitimate scientific argument that cannot be made without a statistical significance test.  P values are bad for public health, bad for the environment, and bad for the economy.  They are a stupid toy – take them away.  A psychology journal recently banned p values; epidemiology journals need to do the same thing.  Then just maybe epidemiologists will be motivated to build a case. 
• Share data.  Besides the p value, another legacy if statistical significance testing is the idea that decisions will revolve around a single experiment.  With environmental epidemiology, it almost surely won’t.  Putting data will allow data to be pooled and analyzed together.  It can be reanalyzed again every time a new study comes on line.  Not sharing data hides weak conclusions.  Sharing data will allow stronger cases to be developed.
• Theory Matters.  Unless it is a very very strong association, just demonstrating an association of some sort isn’t enough.  A dose-response relationship has to be at least theoretically plausible.  Likely is even better.   For environmental studies concerned with the effects of chemicals, toxicology and epidemiology cannot work effectively as two entirely separate disciplines. 
• Multivariate analyses.  When it is necessary to distinguish the causal influence of multiple variables on a particular health outcome, far greater caution is needed.  One bad apple can spoil the whole bunch; if even one variable is modeled incorrectly (i.e. with a mathematical form that isn’t likely or plausible) then all the parameter estimates from a multivariate regression will be off.  In many circumstance, to may be preferable to establish the causal relationship for each variable independently, perhaps using different cohorts or studies for each.  For example, the studies used to characterize the effect of smoking on lung cancer should be different than the ones used to characterize the additional effect of arsenic on lung cancer.

In fact, maybe epidemiology shouldn’t be a stand-alone discipline at all.  Instead, perhaps it should be viewed as a form of human study design and data interpretation that can be used to augment any biological or medical discipline.

References

Hill, Sir Arthur Bradford (1965).  The Environment and Disease: Association or Causation?  Proc Royal Soc Med 58:295-300.

Suter, GW and Cormier SM (2011).  Why and how to combine evidence in environmental assessments: Weighing evidence and building cases.  Science of the Total Environment 409:1406–1417.

Official Post Soundtrack

Gerrard L and Bourke P (1998).  The Human Game.  In: Duality, Track 6.

Post Note

Thesis Post #44.  Part of the solutions thread, but also an epilogue for the epidemiogical thread

2015 EPA Interim Risk Assessment Guideline

Less is More

The 1986 Cancer Risk Assessment Guidelines were only about 20 pages long, giving only a broad outline about how to proceed.  In over 200 pages, the 2005 Cancer Risk Assessment Guidelines describe a far more codified process that doesn’t result in a risk estimate at all.  As an interim replacement guideline, this is much better:

Thou shalt estimate the risk.

Sure, some more general guidance could be proffered, but 20 pages should be plenty.  Besides leaving Points of Departures in the dust, another concept that can be retired is the idea that quantitative risk assessment is just for cancer.  If a potential health effect is worth caring about, it is worth quantifying.  It does not matter whether the endpoint is cancer or not.

Usually, the hardest part of producing a public health risk assessment is the dose-response analysis.  With a little more thought and some better data, it can always be done better.  For that reason alone, prescriptive guidelines are a bad idea.  But, here is a useful guideline:

Keep it simple, improve as necessary.

Risk assessment isn’t an academic exercise.  It won’t produce any great lasting scientific truths.  At best, it will distill and make use of current knowledge for the purpose of informing current policy decisions.  Answer the question first, then work on a better answer after that.  The EPA Benchmark Dose modeling program is good for starters; instead of the benchmark dose estimates, make use of the model parameter estimates.  

How good the risk assessment or the dose-response analysis needs to be can vary widely depending on the decision at stake.  Trying to do it perfectly the first time is a dumb idea.  It will never be perfect anyway, and it probably doesn’t need to be.   It doesn't have to done all at once, either.  You can estimate the risk for one health end point now, and get around to another one later.  Risk assessment is an iterative process where some revision may be necessary every time it is submitted to peer reviewers or the public for comment.  The only thing that stops it is a final decision.

Uncertainty Analysis

Just about every one of the many treatises on public health risk assessment written over the last 30 years has paid homage to the importance of characterizing uncertainty.  The first reason for that is there is usually lots of it.  The second reason is that without a credible characterization of the range of plausible interpretation it is virtually impossible to produce a risk estimate that isn’t politically biased in some way.  But that advice has not been fully heeded, so a little extra guidance just might spur things on a bit:

1. The Uncertainty Analysis IS the Risk Assessment.  It is not something to be tacked on later.   In fact, a good way to proceed to start by producing a range of how big the risks might be, and then work on filling in the probability distribution in between after that.  Even if it is not possible to include every conceivable source of uncertainty, the important ones should be, because otherwise they won’t count.   Sensitivity analyses that offload the uncertainty into a separate analysis that doesn’t figure into the decision don’t really count either.
2. It’s Not Just Statistics.   The general perception of probability is that it is just one of the many flavors of statistics, and therefore the way to characterize uncertainty is to hand the data off to a statistician so the uncertainty can be quantified.   In fact, some of the uncertainty can be represented that way, but it quite often happens that the major uncertainties are something else entirely.
3. Theoretical Probability.  The other great source of uncertainty arises when there are two or more theories that may be used to explain the data or describe reality.  The shape of the dose-response function is the most common occurrence of this type of uncertainty in public health, but there can be other instances of it as well.  Since it is a product of scientific reasoning, the main responsibility for characterizing this type of uncertainty has to lie with a scientist rather than a statistician.  Theoretical probabilities can be represented with probability trees which involves giving each competing theory a probability that is based the relative “weight of the evidence” for each theory.  While tree probabilities are not statistical or even mathematical, if probabilities are assigned so that they sum to one, theoretical probabilities can be mixed and matched with statistical probabilities.

Problem and Solution Formulation

The latest treatise on risk assessment from the National Academy of Sciences (2009) emphasizes the importance of identifying the regulatory issues that a risk assessment needs to address.  If the risk assessment is really intended to provide useful information, this should not be difficult.  Every formal risk assessment is preceded by a subjective one, so regulatory issues and a rough idea of the likely answer should be known before it begins.

It may be even more important to identify how the problem might be solved.  If there is a significant risk, how can it be eliminated or reduced?  The risk assessment may then be designed to evaluate how effective different regulatory intervention efforts may be. 

Monte-Carlo Simulation

Even though the 2005 Guidelines contain some very useful discussion of weight of the evidence evaluations, the guidelines conclude with instructions for writing a risk characterization “narrative”.  If the intent were to guide the production of a risk assessment, instead of trying to prevent one, that wouldn’t be necessary.  What you really need for a risk characterization is a computer programmer.  The exposure and dose-response analyses will produce two or more model bits that need to be put together to obtain an estimate of which will have some associated uncertainty.  You can calculate a worst case, best case, and a most likely estimate without a computer, but a Monte-Carlo simulation or something like will give a more complete picture of what is currently known.

Since it is a public health risk assessment, a one dimensional simulation may not be enough.  You can produce an estimate of the frequency of disease in a population where the only distribution is uncertainty.  But, a two-dimensional simulation where one dimension describes variability in the population, while the other characterizes the uncertainty is far more informative.  Transforming disease frequency measures into life expectancy or some other measure that can be interpreted as an effect of an individual can help make this happen.   

In any case, after the calculations are complete, the estimates may be tabulated, graphed, and explained.  A narrative should not be necessary.  However, it is not unlikely that there will more questions that need to be answered.

References

National Research Council (2009).  Science and Decisions: Advancing Risk Assessment. National Academy of Sciences Press, Washington, DC.

USEPA (1986).  Guidelines for Carcinogen Risk Assessment.  EPA/630/R-00/004

USEPA (2005).  Guidelines for Carcinogen Risk Assessment.  EPA/630/P-03/001F.

Official Post Soundtrack

Focus (1972).  Answers? Questions! Questions? Answers!  In: Focus 3, Track 6.

Post Notes

Thesis Post #43.  Disclaimer: This is not an official EPA publication.  Do cite, quote, or plagiarize.

Reformation

Deharmonization

It’s kooky and inane, but RfDeontology has spread like a contagious disease over the last three decades.  Not only has it thoroughly infected the EPA where it was founded, it has also spread throughout the Public Health Service.  In addition, it has thoroughly corrupted the academic discipline of toxicology, and it also adversely affects the practice of many quantitative disciplines, particularly statistics, epidemiology and risk analysis.   Perhaps worst of all, it has infected academic curricula in toxicology and public health and, as a result, given many students a distorted picture of what both science and government are really about.

RfDeontology only flourishes because the government pays for it.  Believing that a policy is a fact is a career for many people.  They might still have a career without it, but it wouldn’t be the same.  A really good first step towards the eradication of RfDeontology would be to rewrite the bible that still resides on an EPA server.   This document was written to make it sound like the procedure used by the agency to evaluate chemical safety is entirely scientific.  It needs to be replaced by a document that makes it clear that it isn’t.  First and foremost, for God’s sake, stop calling the Reference Dose a “risk assessment”.   Risk assessment is something else entirely.

Repackaging

Safety Assessment could be called something else.  I have referred to it as safety assessment throughout this collection of essays both because that is what it was called when it was invented and because I think it connotes the element of policymaking that is associated with it.  Something like “Chemical Safety Evaluation Procedure” would work too.  But, I will still call it Safety Assessment for now.

The most important thing is to rename the end result.  Maybe the “Acceptable Daily Intake” isn’t quite right for EPA purposes, but the “Reference Dose” has to go.   The Agency for Toxic Substances and Disease Registry calls their number a “Minimal Risk Level” (MRL), but that also makes it sound like some determination of risk has been made when it hasn’t.  Perhaps the best alternative, which comes from the bible itself (EPA, 1993), is the phrase “Regulatory Dose” that appears in section 1.4:

That is, after carefully considering the various risk and nonrisk factors, regulatory options, and statutory mandates in a given case (i), the risk manager selects the appropriate statutory alternative for arriving at an "ample" or "adequate" margin of exposure [MOE(i)]. As shown in Equation 2 below, this procedure establishes the regulatory dose, RgD(i) (e.g., a tolerance under FIFRA or a maximum contaminant level under SDWA), applicable to the case in question:

RgD(i) = NOAEL / MOE(i). (Equation 2)

How cool is that?  Each program office can work out their own procedure for determining an ample or adequate margin of exposure and boom, each program gets to have its own procedure for making policy.  In fact, the Office of Pesticide Programs already does this – they call their number the “Population Adjusted Dose”.  So, why do we need the “Reference Dose” again?  Well, we don’t.  Die, RfD, Die.  Long live the RgD (and Equation 2).

Another main theme of the bible that needs to be ditched is the idea that Safety Assessment is applicable to everything but cancer.   The only thing that really makes cancer special is the Delaney clause.  Cancer isn’t the only nasty disease, it isn’t the only “stochastic” disease, and it isn’t the only disease to worry about after years of exposure.   A safety factor of 100 or so applied to a dose that has no observable effect is still a lot, even for cancer.  However, an extra factor for the severity of the disease might be a good policy for at least some applications.

Of lesser importance, giving the multitude of factors applied in the conduct of a Safety Assessment more specific names might be good idea too:

• Adjustment Factor.  A factor applied to account for known differences, like calculating human equivalent doses from studies with laboratory animals.
• Variability Factor.  A factor designed to address concerns for individuals who may be more susceptible.
• Precautionary Factor.  A factor designed to provide a greater margin of safety for a specific reason, e.g. an extra factor applied for inadequate data or severity of the disease.

The main reason for doing this is that different types of factors might be appropriate for different statutes.  For example, a variability factor might be appropriate for a "may render injurious" evaluation, but not for an "ordinarily Injurious" evaluation.  A precautionary factor may be justifiable for a premarket approval evaluation, but not for naturally occurring or legacy chemicals.

Refocusing

The other thing that really needs to happen is that it needs to be made clear that Safety Assessment is not the one and only possible response to issues involving chemicals in food or the environment.   Safety Assessment is aptly thought of as a form of civil law that is specifically designed to codify procedures for premarket approval.  It is far less useful for decisions and statutes that do not involve premarket approval like the Clean Air Act, the Safe Drinking Water Act, Superfund cleanup, and unintentional contaminants in food.  The most common and best use for these applications is for screening purposes, where the goal is to identify chemical exposures that are NOT a problem.  In fact, this is the recommended use for the MRL which is primarily intended for be used by the Superfund program (ASTDR, 2014):

MRLs are intended to serve as a screening tool to help public health professionals decide where to look more closely. They may also be viewed as a mechanism to identify those hazardous waste sites that are not expected to cause adverse health effects.

One of the presumptions built into Safety Assessment is that the eventual goal is to achieve a specific level of exposure.  Even in premarket approval applications, this is only possible when use is very tightly controlled.   For other applications it may be next to impossible.  While it may be possible to reduce exposures across the board, guaranteeing that everyone will have an exposure below a specific level that is generally considered to be safe may not be.  Is that really a problem?  You won’t get an answer to that question using Safety Assessment, because providing information is not the intended purpose.  If you want to look more closely, you are going to need a risk assessment.   

References

ATSDR (2014).  Minimal Risk Levels (MRLs). 

USEPA (1993).  Reference Dose (RfD): Description and Use in Health Risk Assessments. 

Official Post Soundtrack

Rundgren, T (2004).  Mammon.  In: Liars, Track 6.

Post Notes

Thesis Post #42.  Yeah, the zombies finally ran me out.  Besides being part of the Public Heath Risk Analysis Thread, this is a proposed solution to the problems outlined in the "Protection" essay.  Soundtrack link is for entire album; Mammon starts at 24:30.

Joined At the Hip

Toxicological Testing

Whether the potentially marketable chemical is a drug, a pesticide, a food additive, or an industrial chemical, toxicology studies done for the purpose of getting government approval are conducted to meet guidelines that are directed by regulatory policy.   Studies for drugs and food additive are designed to get FDA approval, while studies for pesticides and industrial chemicals are done to get EPA approval.  As a result, it is pretty much known exactly how the results are going to be used before the study is ever done.  Some scientific reasoning presumably goes into the design of the guidelines, but once that is done, special permission is required to do something that is not by the book (e.g. USFDA, 2007, USEPA, 2015).

But not all toxicology is like that.  In particular, studies on “unintentional” chemicals that are not governed by premarket approval regulations are usually far less structured.  Even if the same testing guidelines are used, it is often not possible to predict the impact that a study will have on any action the government may take.   For one thing, it will depend on what is possible.  Therefore, the science of and policy for unintentional chemicals do not necessarily move in concert.

Science-Policy

The White House has an Office of Science and Technology Policy (OSTP).  The primary mission of OSTP is as follows:

The mission of the Office of Science and Technology Policy is threefold; first, to provide the President and his senior staff with accurate, relevant, and timely scientific and technical advice on all matters of consequence; second, to ensure that the policies of the Executive Branch are informed by sound science; and third, to ensure that the scientific and technical work of the Executive Branch is properly coordinated so as to provide the greatest benefit to society.

In short, the primary function of OSTP is to provide scientific information to federal decision makers.  

The US Environmental Protection Agency has an Office of Science Policy (OSP) that at first glance seems to have a similar mission.  But instead of “informing policy”, OSP seeks to incorporating ORD science and technology into regulatory and non-regulatory actions taken by the agency.   It oversees the Office of Research and Development (ORD) that is responsible for most of the scientific research conducted by the agency.  But, ORD also is responsible for what the agency refers to a Human Health Risk Assessment that is directly involved in formulating agency regulatory policy.  Here, it is hard to distinguish the science from the policy; instead the subject becomes hyphenated “science-policy” where the technical jargon and the regulatory jargon are inextricably intertwined.   The primary reason for this is simple: Ever since its inception, the EPA has primarily relied on a decision making regimen (The Safety Assessment Paradigm) that was designed for premarket approval that dictates how scientific information will be used.  When EPA changed the terminology associated with the Safety Assessment Paradigm in 1986 (e.g. the ADI became the RfD), one of the reasons was because different programs were doing safety assessments in different ways (Barnes and Dourson, 1986, see section 1.2.2.2.4.):

In addition to occasionally selecting different critical toxic effects, Agency scientists have reflected their best scientific judgments in the final ADI by adopting factors different from the standard factors. For example, if the toxic endpoint for a chemical in experimental animals is the same as that which has been established for a related chemical in humans at similar doses, one could argue for an SF of less than the traditional 100.  On the other hand, if the total toxicologic data base is incomplete, one could argue that an additional SF should be included, both as a matter of prudent public policy and as an incentive to others to generate the appropriate data.

Since the use of the safety assessment paradigm is justified as a matter of statute and agency policy, it is actually not unreasonable for different programs to do safety assessments somewhat differently.  For example an extra safety factor for inadequate data may make sense for a chemical subject to premarket approval (e.g. a pesticide or a chemical released in to the environment)), but it may not for a naturally occurring element like arsenic or oxygen.   Furthermore, for some regulatory decisions, the Safety Assessment Paradigm may not be a good fit at all.

Dehyphenation

In theory, the solution is simple.  If Safety Assessment Paradigm isn’t working, do a risk assessment instead.  Well, actually no, since the EPA has called the Safety Assessment Paradigm “risk assessment” for the last 30 years, let’s say we need a “risk analysis” instead.  The point is the same: instead of a regulatory decision that takes the form of an acceptable level of exposure, the assessment needs to deliver information about what the risks are.   The broken cog in the risk analysis wheel is pretty well recognized within the agency; while exposure assessment has improved tremendously over the last 30 years, dose-response modeling has gone almost nowhere (EPA, 2012):

Although dose-response analysis is an integral part of human health risk assessment, it has been decades since there have been any major fundamental changes in how dose-response is characterized. The combination of increased demands on risk assessment and the recent explosion of scientific knowledge presents unique opportunities to modernize the practice of dose-response analysis. This has been echoed in several NRC recommendations to advance dose-response analyses, particularly in the areas of increasing the throughput of chemical assessments, characterizing uncertainty and variability, quantifying incremental risk and addressing susceptibility.  During the October 2010 Human Health Risk Assessment Colloquium, risk managers indicated that advancing dose-response analysis would be useful for their decision making needs.

The most obvious reason why dose-response analysis has gone stagnant is that the current noncancer and cancer EPA guidelines both discourage it.  So, part of the solution is to revise the guidelines so dose-response analysis is welcomed.  But there is a fundamental underlying cause that needs to be overcome – the toxicology testing that goes into scripted regulatory approval processes are often designed by toxicologists for toxicologists.  A quantitative risk analysis changes all that.  The value of a study will depend on its information value rather than it’s conformity to protocol.  In addition, since a technocratic decision process is being replaced by a democratic one, many regulatory players in academia and the federal government will have less control over the decision process.  That’s the inevitable result when the public gets more information.  Therefore, when they realize that their ox is being gored, complaints about the unreliability of modeling from the powers that be are just as sure to follow.

The other science-policy obstacle is that most chemicals worthy of quantitative risk assessment have gained their notoriety from human epidemiological studies.  Yet, most human studies are not designed with dose-response characterization in mind.  This is also attributable to decades of quantitative malaise.   In the Safety Assessment Paradigm, statistical significance is often equated with regulatory significance.  For a dose response analysis more weighty evidence is needed.   Because informing regulatory decision is the main purpose for the conduct of many epidemiological studies, establishing a regulatory market for studies that can establish a biological gradient is prerequisite to actually getting them.

References

Barnes DG and Dourson ML (1988).  Reference Dose (RfD): Description and Use in Health Risk Assessments.  Regul Pharmacol Toxicol 8:471-486.  Also at http://www.epa.gov/IRIS/rfd.htm

USEPA (2012).   Human Health Risk Assessment.  STRATEGIC RESEARCH ACTION PLAN 2012-2016. EPA 601/R-12/007

USEPA (2015).   Harmonized Test Guidelines.

USFDA (2007).  Toxicological Principles for the Safety Assessment of Food Ingredients: Redbook 2000.

Official Post Soundtrack

Doors (1967).  Break on Through (to the Other Side).  In: The Doors, Track 1.

Post Notes

Thesis Post #41.  Introductory thesis for a Public Health Risk Analysis thread.

Obscure Risks

No News is Good News

Compared to chemicals that are intentionally added to food, the FDA has very little authority to regulate unintentional contaminants in food.  I therefore always thought that providing public information about the risk of contaminants in food to be a very important part of my job.  On paper at least, so did the agency since “risk characterization” was always part of my job description.  But even though I managed to publish analyses in academic journals that very few people read, anytime an analysis even started to become a public agency position, political interference arose.

Although there are many secondary reasons, the primary reason is that everyone wants to believe that the food they eat is perfectly safe.  But, I didn’t have to work at the FDA for very long at all to figure out that it isn’t, it never has been, and it never will be.  Sure, the agency does a pretty good job of protecting most people from big risks, in no way does the Center for Food Safety and Applied Nutrition protect everyone from every little real or imagined risk.  But that great truth would never pass through the lips of the press office.  About the only time the agency is willing to acknowledge a risk is when there is a regulatory policy that needs to be justified.  For example, a fish risk benefit analysis probably would have never made it to the agency web site if there weren’t accompanied by a fish consumption advisory.

As you might expect, suppression of bad food news is often strongly supported by the food. industry.  But, I think they are short-sighted.  Word gets out anyway; there is this thing called the internet these days.  When risks go uncharacterized, people often end up thinking the risks are bigger than they really are, which leads to the next reason.

Trivializing the Risk

It’s not as big as the food industry, but consumer advocacy keeps a lot of bodies busy these days.  There are big chunks of federal and local government involved, non-governmental organizations, and the media of course.   But, the backbone of consumer advocacy is in the academic sector.    Researchers will raise the presence of a chemical as an issue, conferences are sponsored, uncertainties are identified, research needs identified, and research is funded.  Then the issue goes away until the papers get published and grant money runs out, and then new conferences and uncertainties pop up.  This can go on for decades without anyone ever trying to figure out big the risk might really be or what might be done about it.  Since a chemical can become a career, the result of that is that many researchers have a financial interest in NOT reaching a decision.

Consumer advocates are also likely to be unappreciative of efforts to quantify risks when there are already public health programs to address them.  For example, fish consumption advisories have been issued by just about every state public health department without anyone trying to estimate what the risk is.   Risk characterizations that might either make ongoing efforts seem unjustified or give consumers information that would allow them to make their own decisions are therefore unwelcome.

Too Much Uncertainty

It isn’t really a reason, but a rationale often given for not quantifying a small risk is that there is too much uncertainty.  The effects are generally unmeasurable, so of course, there is uncertainty.  That doesn’t mean you can’t come up with a reasonable determination of how big the effects could be.  It also doesn’t mean that there can’t be some scientific disagreement about what the plausible range of effects could be or what the best estimate is.  However, it does mean that whatever the uncertainty is, it can be discussed and quantified, and notable uncertainties can be included in the characterization of the risk. 

Even though the too much uncertainty argument is often couched as a scientific argument, it’s really just a form of political censorship intended to protect the status quo – which can be virtually any career or financial interest that might be disrupted by new information.   A good constructive scientific criticism may be cause for a revision that will result in a better risk assessment; it won’t kill it all together.  Risks worth caring about are worth quantifying.  Certainty is a luxury rather than a necessity. 

Message Control

Good managers try to control what they can control.  Also, being in control is the sport that makes Washington tick.  As a result, most agencies don’t want to release information without carefully gauging how the public will receive it.  The FDA is no different.  If new data or a new issue arises they will study the problem until they feel they are ready to take control of the issue.  But, the problem with unintentional contaminants in food is that the FDA has very little control; certainly far less than what the public generally expects.  This is at least partly a self-inflicted problem because the agency and/or the department never wants to admit how little control it has.  So, when a public issue involving an unintentional contaminant arises, usually one of two things happen:

1. Delay. The problem is analyzed in perpetuity or until everyone forgets about it, whichever comes first.  Waiting for new research that probably won’t affect the risk estimate, if there is one, will usually buy a few years.  
2. Declare Victory.  The agency determines that nothing needs to be done, but still spins a message that sounds like something is being done.

I think the agency should give up on the control thing and just focus on providing trustworthy information, which is the sort of thing a Civil Service should be in a position to provide.  But, they won’t.  Oh well, I guess I’ll just do it without them.  Maybe if I set a good example they will see the advantage of not trying to maintain appearances.

Official Post Soundtrack

Barbieri R (2004).  Path Not Taken.  In: Things Buried, Track 9.

Post Note

Thesis Post #40. A second introductory essay for the Personal Risk Assessment thread. I eventually plan of following with some actual risk assessments, with the first one being a reworked version of the fish risk benefit analysis.

Personal Dietary Risk Assessment

Public vs Individual Health

A public decision is about public welfare.  Decisions made by the Federal government, where I used to work, are invariably about the welfare of the U.S. population in general.  It’s not about you.  Yes, you have a voice, but so do a lot of other people who may want something different than what you want.   If you have a loud voice or you have enough money to buy a politician or three, your opinion will be considered more.  That is as true in the Public Health Service as it is in the rest of the Federal government.  It’s called politics.

But, there is another reason why public health risk assessments tend to focus on populations; it’s easier.  Information about potential health risks usually come, at least in part, from epidemiology studies that compare what happens in different populations with different variables that may be causally linked to a particular disease.  Discerning a causal relationship for any individual in a study is usually impossible.  Therefore, risk assessments based on epidemiological data almost always provide risk estimates that either involve the frequency of a disease in a population, or the average magnitude of an effect in a population.  But, you are not an average person, and neither is anyone else.  Everyone knows that, but an estimate for a fictional average person is far easier to produce that an estimate for each and every individual.  It’s not about you. 

Consumption Advice for Chemical Contaminants in Food

For a chemist, food is made up of chemicals.  Some are natural, some are not.  Virtually any chemical constituent, including water, can be poisonous if consumed in large enough quantities.  Yet consumption advice for toxicological reasons is rare.  The main reason for this is that any chemical that is added to food (I.e. is not natural) must be approved first.  If there is a risk worthy of consumption advice, it won’t be approved.   Naturally occurring chemicals are another story.  They are not as tightly regulated because they can’t be.  Often the choice is whether or not the food that the chemical is in can be sold or not.  For example, a handful of raw bitter almonds that contain high levels of cyanogenic glycosides can be lethal, and as a result cannot be sold in the United States.  But, there are also unintentional chemical contaminants that are present in food that wouldn’t be considered safe if they were added.  Yet, they also aren’t poisonous enough to warrant banning the food.  So, consumption advice may be the only viable option.

But, if someone like the U.S. Environmental Protection Agency, the U.S. Food and Drug Administration, or Consumer Reports advises you to modify your diet, you would like to know why, right?   At least I would if I didn’t already.  I don’t necessarily want to be “safe” or “protected”, especially if I’m really the one who has to do the protecting.  If there is a risk of eating too much of something, I want to know what the risk is.  Is it big like smoking or driving on the beltway, or small like getting hit by a meteor?  If it’s the latter, I’m probably going to eat it anyway.  I take my chances with meteors.  Hell, back when I had to get to work, I even drove on the beltway several times a week.  I’m cutting back on that now.

Providing individual risk information to consumers is rare, but it can be done.  The impact of smoking on life expectancy is a familiar example.  Jha et al (2013) who estimated that a lifetime of smoking reduced life expectancy by about 10 years, which is useful information when considering the possibility of starting the habit of smoking.  Furthermore, the risk will be reduced if you stop.   Shaw et al (2000) provided an alternative way of presenting information about reduced life expectancy by estimating a reduced life expectancy of 11 minutes per cigarette.   

An Individual Dietary Risk Assessment Paradigm

So, if you really want to know what your risk is, you are going to have to figure it yourself.  Since I’m retired and don’t have anything better to do, I’ll help.   Here’s the process:

1. Hazard Identification.  As far as I’m concerned, the grocery store aisles are a sea of risks.  Take your pick.  Actually, as far as me helping goes, I only plan on putting up some risk assessment software (probably Excel worksheets, but maybe I can do something web-based) for methylmercury in fish and arsenic in rice because they are the subject of current consumption advice.  Maybe I’ll add more if there is some demand for it. 
2. Exposure Assessment.  This is where substantial personalization is possible.  You know what you eat and what your body weight is, I don’t.  I can help with the amount of the chemical in the food.  That will allow us to estimate the amount of the chemical you consume, otherwise known as your “dose”.
3. Dose-Response Modeling.  A dose-response model is a mathematical equation that describes the causal relationship between exposure and a bad outcome that would be nice to avoid.  Usually, these are models based on population studies that may not be quite right for you.    Therefore, since we don’t know if you are more or less susceptible than an “average” person, most of the population variability that anyone knows about, will end up being an uncertainty for you.  Also, since the risks are small and unmeasurable, educated guesses are the best thing going.  Unless you want to volunteer for a toxicology experiment, that’s the best anyone can do.    So, expect lots of uncertainty.  There, you have been warned.    
4. Risk Estimation.  Plug the dose from the exposure assessment into the dose-response model and then you will have it -- a highly uncertain risk with a specific magnitude. Welcome to my world.
5. Risk Management.  In general, the issue is this:  The health effects that might occur from even the worst chemical contaminants in food, like methylmercury in fish and arsenic in rice are “marginal”, which means somewhat more than negligible, but far short of alarming.   They may be worth considering, but don’t lose any sleep over it.  In fact, the stress might be the bigger risk -- but that’s not my area.  In any case, you decide.  Because this time, it is about you.

References

Jha P, Ramasundarahettige, C, Landsman, V, Rostron B, Thun M, Anderson RN., McAfee T and Peto R (2013).  21st-Century Hazards of Smoking and Benefits of Cessation in the United States.  N Engl J Med 368:341-350.

Shaw M, Mitchell R, and Dorling D (2000).  Time for a smoke? One cigarette reduces your life by 11 minutes.  BMJ 320: 53.

Official Post Soundtrack

Oingo Boingo (1982).   Nothing to Fear (But Fear Itself).  In: Nothing to Fear, Track 7.

Post Notes

Thesis Post #39.  This starts the personal risk assessment thread, and is also another main starting point for the entire collection of essays.

Three Is a Crowd

Risk Analysis

The Society for Risk Analysis (SRA) was established in 1980, three years before the NRC (1983) Redbook.   While the founding constitution and bylaws were modified from those of the Society of Toxicology, the vision statement reads as follows:

The Society for Risk Analysis is a multidisciplinary, interdisciplinary, scholarly, international society that provides an open forum for all those who are interested in risk analysis. Risk analysis is broadly defined to include risk assessment, risk characterization, risk communication, risk management, and policy relating to risk, in the context of risks of concern to individuals, to public- and private-sector organizations, and to society at a local, regional, national, or global level.

There are two aspects of SRA that stand out.  First, the primary goal is synthesis.  Instead of creating a new independent field of study, it draws from other fields.  In fact, risk analysis draws from just about every field of study known to man, including biology, chemistry, economics, epidemiology, law, mathematics, philosophy, physics, psychology, public policy, and statistics.  As might be expected, the actual academic disciplines involved vary with the actual issue, which leads to the other distinction: Risk Analysis is an applied field that deals with issues that are of current interest.

The Third Wheel and the Big Wheel

In the Redbook Risk Assessment Paradigm, all the various disciplines boil down to two main areas of discourse: Risk Assessment and Risk Management.  The scientific and mathematical disciplinesthat are concerned with what we know all figure into risk assessment in some way.   On the other hand, any field that is concerned with what ought to be done pertain to risk management.  That certainly includes the study of public policy.   But interdisciplinary arguments can arise over what’s what.  For example, I think of economics or any other exercise that assigns values to outcomes as being largely associated with risk management, but many economists will argue that they are pursuing the objective truth of the value of currency.   Some disciplines, like risk analysis, are naturally concerned with the interaction between risk assessment and risk management.  Philosophy is a prime example; while metaphysics and epistemology are concerned with what we know, ethics is part of the “ought” dialog. 

But the generally accepted view of Risk Analysis for food safety these days is that there is a third main component comprised of Risk Communication (e.g. USDA and USFDA, and FAO):

Since the whole point of SRA is to facilitate communication between different disciplines, identifying Communication as a separate endeavor is quite odd.   Interdisciplinary or interagency communication is not easy.  There is indeed an abundance of technical and bureaucratic jargon to wade through.  The employment of different concepts of probability (Kaplan, 1997) can quickly turn a discussion about risk into the Tower of Babel.   A very talented multi linguist who is familiar with all of the subjects could surely could facilitate public policy debates involving highly technical subjects:

But, that’s not it.  More often than not, what Risk Communication is really about is communicating with the public (WHO, 2005).  That could mean a dialog where the public is educated about the technical issues and included in the public policy discussion.  But, that’s usually not quite it either.  As an academic field risk communication is a branch of psychology that studies consumer response to various messages about risk (Slovic, 2006).  As an applied field, risk communication is a form of political propaganda.

Politics

Make no mistake, risk analysis at a regulatory agency is political.  In fact, the main function of risk management is political resolution of the value judgements involved in making a decision.  As a separate process, risk communication is a process for promoting the decision after it has been made.  But, at least in a democracy, good risk management can’t work that way.  Being aware of how the public will react needs to be an integral part of the decision process, rather than something that is tacked on later.  So, at best, Risk Communication is an integral part of Risk Management – which means is doesn’t need to be a third category all by itself.  So, why is it?  Here are two potential bad reasons:

• The Public is Being Sold a Bad Decision.  Nobody needs trust more than someone who doesn’t deserve it.  This may involve hiding a risk, selling a risk that really isn’t there (e.g. a weapon of mass destruction) or hiding ineffective risk management.  Or, it may involve hiding the fact that no one considered the interest of the public at all.  The latter possibility is especially likely when the agency decision process is formulaic.
• The Decision is Still Being Made.  Here, the risk communication exercise really is the risk management process.  Except for the pretense of risk management that really hasn’t happened yet, this reason isn’t so bad.  Like reason #1, this is symptomatic of an irrational or inefficient bureaucracy, but at least there is some semblance of responsibility for a decision.  Late is indeed better than never.

Risk Paralysis

Left unchecked, an exclusive emphasis on politically effective messaging can result in the evil twins of risk analysis:

• Risk Manglement.  Answer no questions; tell no lies.
• Risk Caressment/Harassment.  Persuade the analysis to support the message, or at least provide a technical distraction.  If a carrot doesn’t work, use a stick or hire another consultant.

The cure for that is dialogue.  One of the problems with the Redbook (NRC, 1983) formulation of the risk assessment paradigm is that it depicted the process as a monologue, where the risk assessor identified the problem, carries out the assessment, and informs the risk manager about the risk.  But, it doesn’t really work that way.  Risk analysis is iterative process that begins subjectively, and risk managers are always a part of it.  Formal analysis tends to bring more people, with an interest in the issue (“stakeholders”) into the decision process, including laypeople who may also vote.  In fact, that is a main reason for doing it. So, yes, communication is very important every step of the way, but it needs to be a conversation.  Monologues can kill a conversation, and they can also kill a risk analysis.

References

Kaplan, S (1997).  The Words of Risk Analysis.  Risk Anal 17:407-411.

Slovic, P (2006).  Risk Perception and Affect.  Current Directions in Psychological Science 15: 322-325.

World Health Organization (2005).  Effective Media Communication during Public Health Emergencies.  A WHO Handbook.

Official Post Soundtrack

Pink Floyd (1975).  Dogs, Pigs (Three Different Ones), and Sheep.  In: Animals, Tracks 2-4.

Post Notes.

Thesis Post #38, on the Risk Assessment Paradigm thread.