Response to Commenters
Frank B. Cross, Professor
University of Texas
Dept. of Management Science & Information Systems
Campus Mail Code: B6500
Austin, TX 78712
University of Texas
Dept. of Management Science & Information Systems
Campus Mail Code: B6500
Austin, TX 78712
I must begin by thanking the commenters for their time and Dr. Calabrese for soliciting their efforts. My commenters are among the most prominent and well regarded legal experts in this field, and I greatly appreciate their efforts. I am slightly perplexed by many of the comments, because they don't seem to respond to the article that I meant to write. My article was meant to be a thought experiment on how the law could theoretically implement hormesis and its implications, presuming that the theory should prove valid. Some of the comments focus on the presumption, i.e., the validity of hormesis and others are positive evaluations of its prospects for adoption, which tells me that I didn't effectively write the article that I meant to write. Nevertheless, I think they are quite illuminative.
Let me begin this response with one presumption that we have little if any direct evidence of the actual health effects of ambient low level exposures to pollutants. It appears that most, if not all of the commenters agree with this presumption. Professors Heinzerling and Lechleider expressly point out that every available approach requires extrapolation into areas where scientific knowledge is limited. Given this absence of evidence, we must rely on default presumptions, which ideally should be grounded in what limited evidence is available. With that backdrop, I will address some of the important themes of the comments, focusing on those critical of using hormesis, and conclude with a discussion of the one feature of many comments that I find troubling the double standard applied to consideration of hormesis.
I. Scientific Validity of Hormesis
I did not spend much time on the scientific evidence for hormesis in this article, because that was not my purpose and because I am not an expert, while many of the readers of this journal are themselves expert in the relevant science. I am personally inclined to find the evidence for hormesis persuasive enough for consideration. Several of the commenters seriously questioned the validity of the evidence for hormesis. While there is something odd about lawyers debating this out before an audience of scientists, ours is a profession not renowned for its humility, and I'll proceed to discuss the issue briefly.
Professors Latin, Poulter, Walker, Heinzerling, Lechleider, and Applegate all question the validity of hormesis or at least consider the evidence supporting hormesis to be insufficient for public policy. I am not trained to evaluate this matter conclusively, but it appears to me that the evidence for hormesis is considerable and at least equals that for linear risk extrapolation. I have written an article on the subject for an audience of lawyers in which I summarized some of the scientific evidence for hormesis.1 The BELLE Newsletter has been the source of much of my evidence, and I understand that there is material evidence of a hormetic effect for non-natural synthetic substances as well as natural ones.
Some of the commenters are more agnostic about the validity of hormesis but suggest that it has not been sufficiently proved for regulatory consideration. My short answer to this is that none of the models for extrapolating risk at low levels have been proved, so this argument is only relevant in a comparative context (e.g., the evidence for hormesis vs. the evidence for linearity). I'll address this comparison further in my concluding section.
II. Realistic Prospects for Use of Hormesis
Some commenters have suggested that even if hormesis were valid, government regulators would be very unlikely to employ the method. This descriptive issue was beyond the purview of my original article but is worth considering. Professor Poulter suggests that EPA is very unlikely to embrace hormesis, even if it should prove true as a scientific matter. Similar opinions are expressed by Ms. Kunzer, Ms. Peters and Professor Walker. Professor Wagner aptly notes that science does not rule regulatory decisionmaking, such that demonstrating the scientific validity of hormesis would not necessarily translate into regulatory change.2
It would surely take a jolt to the system to provoke consideration of hormesis, and I noted that EPA has been highly resistant to the concept. Such a jolt may well be on the horizon, however. As Professor Weiner and Ms. Kunzer observe, the United States Supreme Court is considering a case, American Trucking Associations, in which the D.C. Circuit held that the application of the Clean Air Act was constitutionally illegitimate because the law provided no limiting principle for the discretion of the agency. While most parties have proposed cost considerations as a balancing principle, hormesis offers an alternative limiting principle. Advocates of environmental regulation might logically prefer consideration of hormesis to consideration of compliance costs, so the EPA might well have an incentive to embrace the theory. Professor Marchant's discussion of the ozone standard illustrates how courts may directly compel consideration of hormesis. Ms. Peters notes that courts might compel at least the consideration of evidence of hormetic effects.
The remarks on the consideration of unusually sensitive populations cuts to the heart of my intended article, and requires more explication than I gave it. Various commenters accurately observed that humans are not uniform in their responses to hazardous substances, so that a given level of exposure might produce beneficial hormetic effects for some, while that same level produced negative adverse health effects for others. The sensitive populations question is one that hormesis may complicate.
Agencies have already dealt with the issue of sensitive populations in hazardous substance regulation. While statutes and courts have offered little guidance, the EPA has consistently considered health effects suffered by unusually sensitive populations, to at least a degree, in its regulations. The hormesis curves in my diagrams are averages, and it is probable that different individuals have different curves, such that a given exposure level may have beneficial effects for some and adverse effects for others.
If exposures are in this hormetic zone for some but not others, consideration of hormesis plainly complicates standard-setting. It raises the question of whether we should set standards that protect some, even if they simultaneously harm others. That question raises difficult value issues that I largely ducked in my original article. If hormesis is true, though, the value issues should be confronted. There is little justification for remaining consciously blind to such effects, and, as Ms. Kunzer points out, interpretation of the Clean Air Act currently commands consideration of beneficial effects. Professor Wagner fears that such tradeoffs are fraught with policy judgments that may be difficult, if not impossible, to resolve. The answer to this is that they are inevitably resolved whenever a standard is set, so the only issue is whether we should resolve them blindly or openly. Ms. Peters notes that the Americans with Disabilities Act might inform this resolution, but I am insufficiently expert in this area to predict how.
Professors Latin, Applegate and others wisely focus on regulatory implementation concerns but, I believe, set up a straw man standard for the evaluation of hormesis. They seem to believe that recognizing hormesis requires that we calibrate standards so that individual exposures be precisely set at the optimal level. This is not correct. We currently use linear, no threshold risk assessment methods, but do not demand precision in application. While linear extrapolation recognizes no threshold, it has not been used to require that individual exposures be set at zero. Recognizing hormesis simply means deploying it in risk assessment when the government seeks to determine the amount of risk at a given exposure level. Hormesis complicates the matter somewhat by requiring consideration of other sources of exposure, but such consideration is already undertaken by EPA in a variety of contexts and may be required under the FQPA.
The commenters are surely correct that there are considerable difficulties associated with identifying individual exposure levels, accounting for cumulative exposures, and defining the precise dimensions of a hormetic curve. But similar problems likewise plague existing environmental regulation. The implementation issue is a comparative one. Agencies already perform risk and exposure assessments to determine health benefits, and hormesis would simply require an adjustment in the risk assessment extrapolation curve. Once the curve is altered, hormesis needn't complicate implementation of environmental statutes (though I think that it should provoke an alteration in the regulatory structure). Professor Latin seems to prefer emission standards to ambient standards, but that is a separate issue.
Professor Latin notes that we cannot adjust standards for each individual or geographic square foot. But hormesis does not require this. In the case of ambient standards, we already have some geographic variation in standards, and hormesis needn't alter this. Nor need hormesis alter feasibility-based emission standards, which are already based on averaging. Professor Latin observes that the agency may have difficulty evaluating cumulative exposures and translating discharges into dosages, but the EPA already relies on exposure assessments and they are not intrinsically complicated by hormesis. His criticisms are more relevant to my suggestions for future regulatory change such as the risk cup, but hormesis does not intrinsically require this (should it prove unduly difficulty), and such concepts are already being implemented with some success in the context of the FQPA, emissions trading, bubbles, etc.
Professor Wagner similarly fears that hormesis will add more complexity to an already overwhelmed system. She may well be correct that it would complicate the risk assessment process somewhat, but I think that EPA is up to it. Moreover, both she and Professor Latin seem to believe that I am calling for exactly perfect assessment of risks. Heinzerling and Lechleider likewise seem to believe that I am calling for the calibration of pollution controls precisely. I am not. In fact, current extrapolations provide only a very rough assessment, and the use of hormesis likewise would provide only a rough assessment, and a key issue is which rough method is likely to be closer to reality. The commenters' concerns also seem to be focused on assessment in the hormetic zone, when in fact many of the risk assessments will surely involve exposures on the right side of the exposure curve, outside the hormetic zone. Professor Walker rightly notes that there may be a substantial cost in establishing that a given chemical has a hormetic effect, but there is a comparably substantial cost in establishing that it has no threshold. While I believe that we should employ the most scientifically accurate method, the threshold of proof should not necessarily be higher for hormesis. The implementation issue is a comparative one. The question is whether using a hormetic curve in risk assessment would yield standards that provided greater benefits without unduly adding to the costs of acting. While I think that Professor Baram also exaggerates the difficulties of reliance on hormesis in risk analysis, I find that his comments offer an excellent discussion of many relevant considerations. Professor Weiner's discussion of priority-setting also represents an excellent contribution to the analysis.
I am troubled by one aspect of the comments, which I believe imposes an unfair and irrational double standard. Ultimately, any regulatory decision is not simply a matter of science but also requires consideration of values. Scientific evidence is very important, though, because the key value at issue (protection of human health) requires the use scientific evidence to achieve the desired consequences.
Professor Wagner suspects that hormesis is contrary to the precautionary principle and the laws' tendency to call for regulators to err on the side of public health. Professor Applegate declares that the primary cost of error using the linear assumption is economic rather than risk to human health, while the risk of error with hormesis is risk to human health. But this assertion begs the question. If hormesis is in fact scientifically correct, the risk of error from linear extrapolation would be the one posing the risk to human health.
Those who favor strong environmental regulation seem to oppose consideration of hormesis, a reaction that I believe to be misguided. They seem to believe that prevailing ambient exposures clearly fall outside the hormetic zone, where additional exposures clearly cause adverse health effects. If so, using hormesis may justify stronger regulation and will never justify weaker regulation. If exposures are indeed high, hormesis should support stricter standards. Professor Weiner may be correct that the resistance to hormesis is psychological and grounded in a belief that exposure to manmade substances is necessarily bad. But if hormesis is indeed more health protective, this psychological reaction is a perverse one.
I devoted a considerable portion of my article to try to establish that considering hormesis was not simply an excuse for weaker environmental regulation but that in many circumstances it would support more stringent regulation, yet many of the commenters ignored this analysis. Professor Latin called my reasoning "speculative" without so much as an adjective of explanation. Professors Heinzerling and Lechleider acknowledge my argument but suggest that it is relevant only when the governing standard is cost/benefit analysis. The formalistic answer to this criticism is that hormesis could also justify more stringent regulation under a significant risk standard. The more important answer is that agencies plainly continue to consider relative costs and benefits even when that is not the statutory standard.3 Making regulations appear more beneficial, and therefore cost-beneficial as well, should justify strengthening standards. Finally, there is a possibility that cost considerations may be legally imposed on all margin of safety determinations, in American Trucking Associations.
Professor Heinzerling and Lechleider make a valid point with their thought experiment about dioxin in cereal, though I think they have rigged the scenario a little (as dioxin in particular provokes emotional responses). Under my scheme, dioxin in cereal would be allowed only if other sources of exposure were so low that cereal consumers would fall in the hormetic zone. Should this be the case, I am not horrified by the prospect. Heinzerling and Lechleider cut close to the crux of the issue in asking whether the case for controlled, informed, voluntary ingestion of substances with hormetic effects is stronger than the case for uncontrolled, uninformed, involuntary exposure. But they have plainly rigged the language to drive their preferred answer (the cereal eating need not be uncontrolled, uninformed, or involuntary). My equivalent thought experiment might ask whether consumers should be denied the benefits of chlorination or fluoridation or medical x-rays because there is evidence that these substances are carcinogenic at high exposure levels.4 And I would add the question of whether those consumers should be denied the benefits of the more stringent regulation that could result from applying the hormetic extrapolation to situations on the right end of the exposure curve, where it could warrant tougher standards for dioxin and other hazardous substances.
My initial effort analyzing the legal implications of hormesis was concededly preliminary and tentative, and so it remains. Yet I find it worthwhile, even if I am wrong, because it has provoked such thorough analysis by knowledgeable commenters. Professor Calabrese has produced a breakthrough in getting legal experts to think about what hormesis might mean as a legal matter.
1 Frank B. Cross, Incorporating Hormesis in Risk Regulation, 30 Environmental Reporter 10778 (2000).
2 I mean for my analysis to apply only to the regulatory determinations of public law and am unprepared to express an opinion on whether the evidence for hormesis would meet the demanding Daubert standards for expert testimony in such cases as toxic torts.
3 See, e.g., George L. Van Houtven & Maureen L. Cropper, When Is a Life Too Costly To Save?, Resources for the Future Discussion Paper CRM 93-02 (July 1993) (finding that cost considerations appear to be pervasive in EPA rulemakings, even when prohibited by statute); Curtis Travis, et al., Cancer Risk Management: A Review of 132 Federal Regulatory Decisions, 21 ENV. SCI. & TECH. 417 (May 1987).
4 Professors Heinzerling and Lechleider argue that chlorination and fluoridation are not truly hormetic, because the low
level benefits are not the reciprocal of the high level risks, which I meant to acknowledge when I called them "like hormesis."
In any event, the policy judgment seems to be much same, whether the low level effects are protection from cancer or
some other public health benefit.