Regulating Environmental Hazards

Richard Wilson

All people want a simple life. People demand of their government the answer to a clear simple question: "is it safe or not?". What is the safe dose for arsenic? is there a safe level of radiation? The problem for any regulator is that there is usually NOT a simple answer. The regulator must then face the issue of what is meant by the simple word "safe".

That there is a difference between a practical safe limit and a theoretical one becomes clearer when one asks a question about a well understood hazard. What is the safe speed of a car? It is clear that a car can kill you by pinioning you to a wall even if it is traveling at only 1 mile per hour. Normally one would use "safe" in the more practical sense. A pundit can add the adjective "absolute" when appropriate to describe the theoretical concept. We might say that a safe speed for a car is 30 mph in urban areas and 65 mph on interstate highways because our governments have set the speed limits at these figures. But we would have to say that no car is "absolutely" safe. Or we could ask: is an airplane safe? Any airplane can crash if a peculiar combination of circumstances (such as equipment failure or pilot drunkenness) occurs. Before about 1949 the average time saved in crossing the USA by air instead of by train was exceeded by the average loss of life expectancy by air crashes. If pressed, everyone would have to admit that no airplane flight is "absolutely" safe. But we have today a practical high degree of safety and regulation that has never been so onerous that it seriously affects the financial well being of the industry.

However the public (and hence regulators) have been hesitant to discuss "practical" safe levels for radiation or chemical hazards even when the probability of adverse effects on health or life is much smaller than the historically determined probability of a car or airplane accident discussed above. A principal purpose of this paper is to discuss ways to avoid this peculiar reluctance.

I first discuss the facts that force scientists to think of safety in new ways. This leads us into discussing the behavior of pollutants and other substances at low concentrations and doses. Then I will discuss how society has begun to come to grips with this problem and how society is still in the midst of raging arguments today.

A new way of thinking

It has become a practice to quote (or sometimes misquote) the physician Paracelsus who said several hundred years ago that "the dose makes the poison". For example arsenic can kill promptly if given in large doses. At low doses it does not - and may in some case be even be necessary for life. Thus one develops very easily the concept of a "safe dose" where prompt death from a poison will NOT occur.

But over the last 100 years society has become aware of a number of other hazards from continuous exposure to a substance at a concentration level too low to cause an acute (prompt) disease or death. During the same period life expectancy has improved markedly. Some diseases that were a major risk to life have been (almost) eliminated and society is able to pay attention to smaller and smaller risks, and risks that only materialize late in life. Of these cancer dominates public concern. Firstly cancers attributed to radiation and later cancers attributed to chemicals and other substances. But lung problems (caused for example by air pollution) have recently become evident.

These chronic effects have a very important feature that distinguishes them from the acute effects. If a group of people are fed a lethal acute dose of arsenic all will die. At one tenth the dose few or none will die. But with chronic effects there is a far shallower dose response relationship. Typically in a group of people no more than 10% will get cancer from an environmental exposure. We are then led inexorably to consider the probability of getting the cancer in a certain time frame and with a certain exposure. In addition the chronic effect (such as cancer) is often delayed tens of years from the cause and causality is hard to establish. With this should come a new way of thinking about the problems. It is no longer obvious that there is a threshold below which no effect will occur. Also the cancers caused by the pollutant are indistinguishable from cancers already seen in the general background. These two facts make a very profound difference which should be (but usually is not) well understood by regulators and politicians. One must be circumspect about using terms like "no risk", "zero risk" or "threshold" or calling for "absolute safety".

The dose response relationship

Once we are forced to discuss the probability of an event rather than merely whether or not it has occurred, one is led to enquire how that probability might change with dose. There are two distinct issues which I here distinguish as the "cellular" dose response relationship and the "population" dose response relationship.

A biologist might ask: "what is the probability that a substance will cause a cellular change that starts a chain of events that leads to a cancer?" and "how does this probability with dose?" Geoffrey Crowther suggested that the probability of such an event initiated by radiation might be linear with dose. This idea has since been extended to chemicals and other substances. But there must be millions or billions of cells "initiated" for every cancer and many biologists insist that normally these are repaired and only above a threshold dose does the repair mechanism fail. But this may apply only for the "cellular" dose response relationship for an isolated cell or group of healthy cells. A population includes healthy and sick, young and old. The "population" dose response relationship may well be different. 30% of the population gets cancer from some cause. Since these cancers are indistinguishable from those caused by the pollutant being considered it seems likely that these are caused by some agent, which acts in the same way as the pollutant acts. Then, almost by definition the threshold has been exceeded. In this way Crump, Hoel, Langley and Peto described some 20 years ago how almost any cellular dose-response relationship for carcinogens can become linear at low doses when background cancers are taken into account. The argument depends critically on the assumption that the pollutant and the background proceed by the same biological mechanism. Crawford and Wilson pointed out the same argument applies to non-cancer end-points also. Examples include a reduction in lung function (which is already declining with age) and consequent increase in death rate due to (particulate) air pollution; reduction in IQ and hence (in extreme cases) mental deficiency due to radiation in utero; reduction of sperm count and hence increase in male infertility due to DBCP exposure.

Few if any pollutants cause new, not previously seen, adverse effects on health. Almost all merely increase the probability of developing a problem which may already be common in society. Thus a linear dose response relationship may not be uncommon in society (which is composed of whole bodies and not merely unconnected cells) but is probably the usual situation. A little thought shows that this is true for much more ordinary hazards also. The probability of a president being assassinated is roughly proportional to the number of years in office. As the number of cars on a rural highway doubles, so does the accident probability increase. For them, a linear dose response is a very real procedure for consideration of the low exposures. This therefore suggests that we think about these more ordinary situations to obtain concepts and words to discuss the problems.

Many scientists have suggested the use of the term "effective threshold" or "practical threshold" to describe a practical rather than an absolute exposure to radiation and chemicals. These terms might be defined in either of two distinct ways.

(1) the level below which it is not possible to detect (find an adverse effect in an epidemiological study or

(2) a level below which the risk is "insignificant" or "negligible".

The detection limit for an adverse health effect is NOT an absolute. With more resources and more sophisticated analyses the detection limit can usually be reduced somewhat. Until recently epidemiologists were reluctant to attribute causality unless the cancer rate doubled (technically defined as risk ratio greater than 2) which in practice meant that lifetime risks of less than 15% were unmeasurable. However when a risk is determined at a higher dose, scientists are willing to accept a lower standard of proof and a risk of a few percent can be considered proven. This suggested a "practical threshold" at the level that produces a 1% risk. For radiation effects two professional societies, the Health Physics Society and the American Nuclear Society, have suggested a threshold for lifetime exposure - which I would here call a practical threshold - at between 10 Rems and 20 Rems which on a linear dose response this corresponds to a lifetime risk of about 1/2 to 1%.. Indeed present radiation exposure "standards" are such that this threshold is unlikely to be exceeded in practice.

Society, led by many scientists, felt that one should be more cautious than this and a "practical threshold" at 1% is inadequate for protection of the public. Without a further restriction mankind might automatically raise pollution levels everywhere until this threshold is reached. This then led to emphasis on the second approach - reducing the risk until it is "negligible". But what is "negligible"? and "How do you calculate risks this low?". These are coupled questions.

"Zero risk" advocates, have demanded that the risk be calculated using a linear dose response, and other "conservative" or "pessimistic" assumptions and that risks be deemed acceptable only if they are less than an extremely low level of one in a million per lifetime. This has dominated regulatory thinking (particularly in the USA) for the last 25 years. But a little thought shows that it is not possible to regulate all societal risks in this way. On my list of risks (tables I and II below) there are many that society decides not to regulate that are far bigger than one in a million. I have argued therefore that the regulation of only some risks so tightly is arbitrary and capricious - but I have not persuaded anyone to join me in a lawsuit on this ground. I argue that there should be a "de minimis" risk not appreciably below a risk level that agencies such as EPA decline to regulate (maybe 1 in 1000 lifetime risk).

I and many others over the last 25 years have suggested determining "negligible" by comparing the risks to other activities that are better understood, and are calculated in a similar manner to the risk under discussion. Because understanding is an individual matter, the comparison has to be made to a number of different risks to assist the understanding of a number of different individuals. For a number of reasons society accepts greater risks in an occupational setting than in a public setting: regulation of occupational exposure and public exposure is therefore different and it is useful to discuss them separately.

Occupational vs Societal (Public) Risks

In Table I, I show a set, similar to those I have presented elsewhere, of occupational risks. I now separate them into a class where they are historically calculated and the linearity at low doses (as discussed cars and airplanes above) is well determined, and a class where we are not sure of a threshold and the risk MAY be zero.

Table I

Occupational Risks

Deaths per year of Risky Activity

Multiply by years of work to obtain the lifetime risk

Historically Calculated (uncertainty about a factor of 2)

US President one in fifty-two

Metal Mining one in three thousand

Policeman on duty one in three thousand

Transportation one in five thousand

Quarrying one in five thousand

Airline pilot (accident risk) one in ten thousand

Government Office Worker one in eleven thousand

Frequent flying professor one in twenty thousand

Involving uncertain dose - response

(uncertainty of slope about a factor of 3 plus the uncertainty of extrapolation)

Coal miner with black lung disease one in two hundred

Asbestos worker at TLV 1/4 of time one in four thousand

Airline pilot (cosmic ray exposure) one in five thousand

Hospital x ray technician one in ten thousand

Benzene worker at TLV 1/4 of time one in thirty thousand

Table II

Some Commonplace Societal (Public) Risks

Action LIFETIME Risk

Historically Calculated (mean values: uncertainty about a factor of 2)

All cancers one in four

Cigarette smoking one in three

Motor vehicle accident (motorist or passenger) one in eighty

Motor vehicle accident (pedestrian) one in four hundred

Home Accidents one in one hundred & twenty

Electrocution one in three thousand

Being hit by meteorite one in twenty five thousand

being hit by falling aircraft one in two hundred thousand

Involving uncertain dose - response relationship

(uncertainty a factor of 3 or more plus the uncertainty of extrapolation)

Air Pollution, Eastern US one in fifty

Drinking water:

(i) with EPA limit of chloroform one in fifty thousand

(ii) with EPA limit of arsenic one in one hundred

School with us average asbestos less than one in a million

EPA regulatory practice one in a million

The calculation procedure for these is obvious in most cases. Being hit by a falling aircraft is just the number of people killed in this way each year divided by the US population, and thus is an average over all Americans. The risk of being hit by a meteor or comet is less obvious. It is based upon measurement of meteorite densities, and the estimate that if a very large meteorite comes in (every million years or so) that half the human race might be killed in the consequent fire, flood, dislocation, famine and disease.

The high figure for air pollution is based upon studies at the Harvard School of Public Health. That estimate could already have been made in 1953 based upon the London air pollution "incident" of December 1952 and a linear dose response. Later work shows that this assumed linear dose response is better justified than any linear dose response for effects of asbestos or radiation which are two over regulated substances. The arsenic risk is easily obtained by noting that high levels of bladder, kidney and lung cancer have been reliably reported in Chile and Taiwan at arsenic concentrations only ten times the EPA standard and that there is no evidence for a threshold close to that level.

Some people reject such comparisons because (for example) automobile driving is voluntary. But the average risk for pedestrians killed (one in four hundred) is surely involuntary.

Although society is trying to reduce the historically calculated risks, the improvement is inevitably slow, and a reduction of more than a factor of 3 in the next century is unlikely. I note that the uncertainty in the radiation and chemical hazard risks is mostly downward, driven by whether or not there is a threshold. It is very unlikely for the risk to be 10 times the magnitude calculated here. It therefore seems reasonable for society to accept a risk of 1/10 of that given by typical historical figures - between one in ten thousand and one in hundred thousand per year or by multiplying by the lifetime of 75, between 3/4 percent and one in 1300 in a lifetime.

Historically chemicals have been regulated in the workplace to avoid acute (prompt) effects. Threshold Limit Values have been set based thereon - often with a Time Weighted Average (TWA) of 8 hours. If one wants to regulate on the basis of chronic effects and assume no threshold, a new concept should be used. Thus acute exposure to benzene has been regulated with a TLV of 10 ppm in the workplace. If a linear dose response is assumed, this leads to an annual risk of one in 30 thousand for a person at the TLV one quarter of the working days - which can be considered high. But it is very unlikely that he would be exposed at this level every day. Thus it is inconsistent and illogical to use the threshold limit value as a long term average dose. One might either retain the Threshold Limit Value solely for acute effects, and find another concept for the chronic effects or one could take the desired lifetime dose limit and divide by the occupancy to obtain the Threshold Limit Value.

In Radiation Protection the concept of "Collective Dose" has been introduced and has been widely use to characterize the exposure of a population. It does not replace an exposure standard supplements it. The collective dose is the summed product of the dose and the number of people sustaining that dose. (usually a lifetime dose).

Collective dose = S [d ´ N(d)]

If a linear dose response is assumed the number of cancers produced by a pollutant is proportional to the collective dose no matter how that dose is distributed among people. If the actual dose response is sublinear and characterized by a threshold the collective dose calculation always gives an upper limit to the number of cancers produced, because the slope of the dose response relationship is anchored by data at the high dose. If a threshold exists the actual cancers would be fewer.

It is important to recognize that using the concept of collective dose for regulation by itself, without any insistence on adherence to standards would be closely related in public policy to emphasizing "the Greatest Good of the Greatest Number" - a well known procedure which when carried out slavishly has assisted in an intolerable oppression of minorities. Therefore all the extensive literature for and against the "Greatest Good for the Greatest Number" can and should be used to address what to do about Collective Dose.

At the time that the concept of collective dose was first introduced, advisory bodies such as the International Commission on Radiological Protection suggested that doses be kept "As Low As Practicable" below the regulatory standard. About 1970, scientists thought this term too vague and introduced another, which I consider to be equally vague: "As Low As Reasonably Achievable (ALARA)". This vagueness causes considerable trouble and may even cause the demise of the nuclear power industry. The only definitions of ALARA that I have seen have been to put this into monetary terms.

Payment to reduce a hazard

Created by Richard Wilson and Michael Polkanov Probably the first government agency to grapple with this problem (albeit incompletely) was the first US Nuclear Regulatory Commission. Following a 2 year long public hearing (inherited from the US AEC) the commission proposed a practical definition of ALARA. Exposures should be reduced if that can be done at a cost of $1000 per Man Rem. Accounting for inflation, this is now updated to $2000 per person-Rem ($200,000 per person Sv). Although this is not very explicit this amount was for exposure to the general public. If one assumes a linear dose response model, this is roughly equivalent to $4,000,000 per cancer averted (or what some economists call a statistical life). For occupational doses (medical and dental personnel) the independent National Council for Radiation Protection suggested doses should be reduced if that can be done at a cost of $10 - $1,000 per man Rem or less ($1000 - $100,000 per person Sv) a number considerably smaller. This accords with a usual practice of accepting a larger risk in an occupational setting than in a public one.

It is noteworthy that the US Environmental Protection Agency in summer 1998 proposed draft guidelines for economic cost/risk benefit analysis in which a number of $4,000,000 per statistical life (for the general public) was suggested which is consonant with the above numbers. It should also be pointed out that the justification for the above numbers implies that if a cost is greater than the above sum the action should NOT be performed.

But the cost per statistical life for many radiation protection regulations in particular, this number is vastly exceeded. Thus (according to a study by the Harvard Center for Risk Analysis) the NRC regulations for radioactive waste from a reactor mandate a cost of $800,000,000 per statistical life - 200 times the amount justified for public risks. A recent article in Health Physics suggested that whereas the federal program to cap uranium mine tailings cost about $500,000 per statistical life and was justified, extension to more remote mines was unjustifed and had cost a billion dollars. In deciding upon compensation for those exposed to fallout from atomic weapons tests, Congress first asked for a set of tables to calculate the "Probability of Causation" for a sufferer from cancer who had been exposed to radiation. But when these tables (and associated dose estimates) showed that few persons would be entitled to compensation, Congress changed the rules and awarded compensation to people with "radiogenic" cancers in certain defined areas (such as the whole state of Utah) regardless of the dose. This has cost $500 million so far with no effect on health!

Some analysts have proposed a "de minimis" level at which the risk should be ignored according to the legal maxim; "de minimis non curat lex" (the law does not deal with trifles). This need NOT be the same as a threshold dose. For example even if we know there is a true threshold dose, one may wish to set a "de minimis" level a little lower to allow for a margin in implementation. On the other hand if a linear dose response relationship is assumed, and a "practical" threshold is taken as the level where the effect cannot be proven (a risk of about 1 - 15%) one may need to assign a "de minimis" level that corresponds to a dose that it would be too expensive to reduce.

I argue that this should be a derivative (secondary) criterion. NCRPM have recommended a de minimis level for an individual dose of 1 milliRem (100 mSv). If this is to be a public dose, the considerations above suggest that this corresponds to an expenditure of $2 to avoid the dose. I do not know of a radiation exposure that can be reduced for this cost, even if the dose is undergone by the whole US population of 225 million people, and the total cost becomes $450 million. . Therefore the acceptance of a "de minimis" dose of 1 milliRem is excessive and could be better replaced by 10 millRems or more cost criterion. It would also be effective in avoiding most of the anomalies. The tritium releases from the French plutonium reprocessing plant at Le Hague would be ignored; as would the C14 releases from incineration of medical radioactive wastes. A particularly bad example of misunderstanding of radiation hazards arose 5 years ago when irresponsible school teachers in New York City dragooned the captive children in their care to demonstrate against an incinerator at Rockfeller University the exposures from which would have been "de minimis" under any proposed level. The incinerator project was abandoned because of this outrageous pressure.

Inherent in all discussions of low doses is that the science is uncertain and always will be. We are dealing with doses where the effect, if it exists is small and not measurable directly. The indirect evidence is more complex and very subject to individual judgement. If society declines to spend excessively on these risks where the victims can only be known statistically, then the amount spent on reducing doses below the "practical thresholds" where effects can no longer be directly measured will be small enough that there will be little disagreement. On the other hand if society insists on demanding "zero risk" there will be problems. It is economically impossible to insist on zero risk, or even one in a million for all pollutants and if it is applied for some arbitrarily selected pollutants there will be pressure from scientists with a responsible concern for public policy, to "declare" (and fight for) the existence of a threshold whether or not such a threshold can be proven even by indirect means, the unseemly and ultimately self defeating spectacle of scientists engaged in pseudoscientific arguments in Congressional committees will continue. One tragic feature of such arguments is that many scientists take positions in front of these committees that they would not dare to take in front of their colleagues at home.

In the above paragraphs the overregulation of nuclear power by demands for excessive demands for reduction of radiation exposure has been outlined. It is useful to consider an opposite case. A possible UNDERregulation of air pollutants, particularly the prime suspect -fine particles (diameter less than 2.5 micrometers). Specialists have argued that air pollution reduces the life (kills) about 50,000 people in the USA yearly. The data seem to be linear with exposure down to the lowest levels and there is a model which satisfies the minimal criteria for biological plausibility (discussed below). Applying the $4,000,000 per life one finds that society might reasonably spend $200 billion a year to eliminate the exposure or $100 billion a year to cut the exposure in half. Of course this is only one half of the cost benefit analysis. The cost to reduce the exposure must be estimated. But clearly this large sum is NOT being spent. The EPA coped with this by setting a standard for fine particles - those less than 2.5 microns - and demanding that communities meet it. But how can a community meet such a standard when most of the particulates blow in from outside, including other states? The states came together 200 years ago so that there was a procedure for discussing such interstate issues - by, for example, demanding reduction of particulate emissions. The US Court of Appeals in the DC circuit recently decided to set aside the PM 2.5 standard promulgated by the US EPA. The court was concerned that the EPA had and has no clear scientific procedure for considering the matter. This court decision makes it urgent that society develop a more formal procedure for coping with these chronic hazards that are not well addressed by demanding adherence to standards.

I argue that society should retain the existing set of standards for exposure to pollutants - such as radiation and chemical substances which are adequate to prevent directly measurable effects in populations. These standards (TLV etc ) should NOT be modified downwards to account for chronic effects such as cancer which might have a linear dose response at low doses. For those chronic effects where a biological mechanism is plausible - in a very general sense - for a linear low dose response, three concepts should be retained. A collective exposure or collective dose to be controlled (probably by emission regulation) at $4,000,000 per statistical life for the public and perhaps $100,000 for occupational exposures, a derivative "de minimis" exposure where no one even writes a regulatory letter and a derivative "de minimis" risk below which no one tries to regulate. If this is done consistently I believe that many of the present acrimonious arguments would disappear.

Provided that the regulation of collective dose is properly done (which is not now the case for radiation) I am in favor of extending the concept to collective exposure from other pollutants. However the proviso is important and the failure to regulate properly using collective dose may be a reason why my opinion is not universally popular. I consider that a calculated risk must be taken seriously for a pollutant exposure even when there is no proof provided that there is a "plausible" model, even though possibly not completely understood. Indeed, if plausibility is taken to require that every step in the mechanism leading to the toxic effect be clearly understood, we will fail to describe (and hence predict and prevent) many risks to health. However, we must ensure that a hypothesis is rooted in available data and knowledge, and be willing to modify our statement of plausibility if and when new data appear, or we could spend time and money in fruitless chases. For this reason I would completely reject the idea that low frequency alternating magnetic fields of 10 milliGauss (1 microTesla) can cause cancer. It is NOT plausible.

Evans J, Ozkaynak J, R. Wilson R. The use of models in public health risk analysis. J Energy and Environ 1(1): 1-20(1982).

Hill AB. The environment and diseases: association and causation. Proc Roy Soc Med, Sec Occup Med 58:272 (1965).

Crowther J. Some considerations relating to the action of x-rays on tissue cells. Proc Roy Soc Lond B Biol Sci 96:207-211 (YEAR).

Crump KS, Hoel DG, Langley CH, Peto R. Fundamental carcinogenic processes and their implications for low dose risk assessment. Cancer Res 36:2973-2979(1976).

Guess H, Crump K, Peto R. Uncertainty estimates for low-dose-rate extrapolation of animal carcinogenicity data. Cancer Res 37:3475-3483(1977).

M Crawford and R. Wilson (1996) "Low Dose Linearity: the Rule or the Exception?" Human and Ecological Risk Assessment 2:305-330.

Policy Statement of the Health Physics Society, 1313 Dolley Madison Boulevard, Suite 402, McLean, VA 22101. (1996).

Position statement of the American Nuclear Society, "Health Effects of Low-Level Radiation", April 1999. Tel: 708-352-6611; website: www.ans.org.

Wilson, R. "Measuring and comparing risks to establish a de Minimis risk level, Regulatory Toxicology and Pharmacology 8: 267-282 (1988)

Wilson R and Spengler, J.D (Editors) (1996) "Particles in Our Air: Concentrations and Health Effects" Harvard University Press Cambridge, MA

A.H. Smith, et al (1998) "Marked increase in bladder and lung cancer mortality in a region of northern Chile due to arsenic in drinking water". Am. J. Epidemiol. 147:660-669.

"Arsenic in drinking water", report of a study by a committee by the US National Academy of Sciences/National Research Council (May 1999).

see for example; Wilson R and Price B.P. (1999) "Regulation of Asbestos Exposure", Canadian Mineralogist in press.

"The ALARA Principle", NRC rulemaking RM-30-2 of 1976 and subsequent federal regulations document.

"Implementation of the Principle of As Low As Reasonably Achievable (ALARA) for Medical and Dental Personnel, National Council on Radiation Protection and Measurements, Report No 107 (NCRPM, Bethesda MD, 1990).

Miller, M.L., et al.,"Number of cancer deaths prevented by UMTRA Project", Health Physics, May 1999.

Wilson R. "Putting balance into safety regulation", in Preparing the Ground for a Renewal of Nuclear Power , B.N. Kursunuglu, S.C. Mintz and A. Perlmutter, eds., Plenum Press, NY (in press).

Tengs T et al, "Five hundred life saving interventions and their cost effectiveness", Risk Analysis 15:369 (1995).

"Report of the National Institutes of Health Ad Hoc Working Group to Develop Radioepidemiological Tables", NIH Publication No. 85-2748.