May 6, 2002

Mr. Joseph A. Mendicino, Jr.
Director of Administrative Services
The American Law Institute

By E-Mail and Fax: (215-243-1636)

The following suggested revisions to Comment c to § 28 are prompted in part by a letter to Professor Green dated December 3, 2001, signed by a group of 10 scientists and public health experts. The letter was made available at the December 2001 ALI Council meeting. Although that letter was addressed to an earlier version of Comment c, and in response there have been changes in Tentative Draft No. 2, many of the comments in the letter are still relevant to the current draft. The electronic text of the scientists’ letter is attached hereto as Attachment A. The responsibility for these amendments, however, is entirely mine.

The last full paragraph of Comment c(1) on p. 101 presently reads as follows:

Much of the law addressing proof of causation in toxic-substances cases has been developed in assessing the admissibility of expert-witness opinions on causation, a subject not addressed in this Restatement. However, the specific aspects of proof of agent-disease causation, the sources of evidence appropriate to prove those elements, and their adequacy outside the admissibility context are matters of substantive tort law, and they are addressed in this Restatement. Regardless of the legal categorization within which the discussion arises, the law regarding proof of agent-disease causation is the most significant development on the subject of causation in tort cases since the Restatement Second of Torts.

To clarify the relationship between proof of causation and admissibility of expert-witness opinions on causation, I move that the last paragraph of Comment c(1) (reproduced above) at p. 101 be replaced with the following paragraphs:

The great bulk of the law addressing toxic-substance causation has been developed in decisions assessing the validity and reliability, and thus the admissibility, of expert-witness opinions on causation. However, admissibility of evidence is not a subject of this Restatement. Admissibility is a question preceding the determination of sufficiency of proof, which is the subject of § 28. Hence, cases on admissibility of scientific evidence and the validity of expert-witness opinions on causation are cited in this Comment solely for illustrative purposes. Their holdings should not be taken as the majority rule on the sufficiency of proof of causation in cases involving toxic substances and disease.

Causation proof in cases involving toxic substances and disease is subject to the same rules of law as is causation proof in all other cases. The evidence in toxic substance and disease cases, however, may involve questions that are beyond the scope of this Restatement, including (1) what physicians and scientists treat as adequate proof of causation, and (2) the admissibility of expert causation testimony, whether under the special rules employed by the federal courts or under traditional rules. The division of reasoning on the latter question is exemplified by, e.g., Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), and Frye v. United States, 293 F. 1013 (D.C. Cir. 1923), and analogous state cases.

Courts evaluate the sufficiency of causation evidence in toxic substance cases under the same rules of law that apply to torts in general. While there may be some issues that are normally assumed in tort cases, such as general causation, and that require specific proof in toxic substance cases, courts utilize the same legal standards and, in determining whether reasonable minds can differ as to whether a toxic substance more likely than not caused the plaintiff’s injuries, are informed in the same manner by the quantity and quality of the evidence, the credibility and qualifications of the witnesses, and other judgmental factors. Furthermore, reviewing and weighing evidence is a subjective process. Thus, courts can come to different conclusions on apparently similar evidence as in, for example, the decisions cited in the Reporters’ Note at pp. 152-53. Among those cases, some courts insisted on a doubling of the incidence of disease in group studies as a minimum for proof of specific causation, while others recognized that causation can be established with less than a doubled incidence. While these cases are illustrative of how different courts have resolved the evaluation of particular evidence, they do not constitute rules of law.

Because of numerous citations in this Comment to decisions dealing with admissibility of evidence, the distinction between the sufficiency of proof of causation and admissibility of evidence should be made clear and be adhered to throughout § 28. This is particularly important in that many of the cases cited come from courts that are uniquely required under Daubert to act as vigilant "gatekeepers" for expert opinions, and do not represent a general American rule. Many state court systems approach these questions in different ways, and the majority follow Frye, which is largely restricted to challenges of either "new" science or machines and devices, such as lie detectors. Similarly, Frye challenges are generally not recognized for medical expertise, while Daubert challenges can be. And, even those state courts which follow Daubert do not embrace all aspects of the holding. Furthermore, although the authorities cited in support of this Comment in the Reporters’ Note are predominantly federal, the great bulk of tort litigation takes place in state court. Given the differences among the state rules and the fact that the great majority of tort cases are litigated in state courts, it should be made clear that, as to "admissibility" rules of law, there are different regimes involved and that one single set of rules does not exist.

Add the following language at the very end of Comment c(1), page 101:

It must be remembered that American state courts holding jury trials seldom intervene in questions of cause in fact unless there is an issue of the sufficiency of the evidence proffered by the party with the burden of proof. Such an issue is typically raised through either a motion for judgment as a matter of law or a motion for summary judgment. The forbearance of the state courts is typically required by their state constitutions, statutes, or court rules based on the common-law right to trial by jury that is the law of virtually every American state.

There are very few references in Tentative Draft No. 2 to juries, to trial by jury, or to the common-law right to trial by jury that is the subject of state constitutional provisions in 48 states. See Paul B. Weiss, Reforming Tort Reform: Is There Substance to the Seventh Amendment?, 38 Catholic U. L. Rev. 737, 739 n.11 (1989).

In most litigation, American state courts do not enter into factual disputes unless their intervention is legitimately demanded by one of the parties. Most of the instances when a court is called upon to make some judgment about factual presentations arise when the court is addressing sufficiency of proof in the context of a motion for summary judgment or a judgment as a matter of law. Otherwise, the courts are bound by constitutional provisions, by state statutes, or by rules of court to leave resolution of factual disputes to the jury. These restrictions on judicial fact-finding in jury trials apply whether the issue is causation in a toxic substance case or causation in any other tort case.

The most prominent exception to this longstanding practice now arises in federal courts, which are authorized by the U.S. Supreme Court’s Daubert decision, as well as by the current Federal Rule of Evidence 702, to intervene in matters of expert evidence. Under Daubert and its successor cases federal courts are called on to review expert opinion testimony, the methodology employed by the expert in reaching conclusions, and sometimes even the adequacy of the general scientific literature to support admission of such an opinion, and to act as a "gatekeeper" to keep the jury from hearing testimony deemed unreliable by the court.

Court interventions under the Daubert doctrine are made overwhelmingly in the federal courts. Most American state courts adhere to the usual American practice of leaving factual determinations to juries, including factual determination of causation.

Add the following two paragraphs at the very end of Comment c(1), p. 101:

Because of the limited applicability of federal court decisions in the area of tort law, federal decisions are cited in this Comment only to illustrate points of discussion. Many of the federal cases cited interpret application of the federal Daubert rule on expert testimony. Those cases should not be taken as representative of a majority American rule on sufficiency of proof of causation in tort cases.

In the past decade, owing to attempts to integrate the federal Daubert rule into state law, some state courts have considered changes to their regimes on proof of causation. Their decisions have reached a variety of conclusions. Because many of those cases are decisions on admissibility of evidence, not on the law of proof of causation with which this Restatement is concerned, no attempt has been made to categorize the different results reached in the state courts. For that reason, references herein to the varying approaches taken by the state courts are made for illustrative purposes, and they should not be taken as statements of a rule on the sufficiency of proof of causation among the state courts.

Tort law is state law, and the state courts handle the overwhelming body of tort litigation in the United States. Further, in considering matters of tort law, federal courts are required by Erie R.R. v. Tompkins, 304 U.S. 64 (1938), to apply the substantive tort law of the state in which they sit unless choice of law rules require them to apply some other state’s tort law. Yet Comment c provides limited discussion of state court decisions. This leaves the Restatement open to the interpretation that the federal cases cited represent the majority rule.

Citation of predominantly federal cases, many of which are Daubert inquiries concerning admissibility, leaves the impression that the Daubert decisions are the appropriate rule of law. While Daubert and its sequelae have constituted a significant development in the last decade, it is a development in federal evidence practice, not in tort law, which comes from the state courts.

In the state courts the Daubert doctrine has been received with varying degrees of enthusiasm. Some states have begun to follow Daubert or have cited Daubert as similar to their state rules or precedents, and some have ignored it. Some continue to use Frye as their model, and some reject Daubert.

A good summary of this diversity in the state courts appears in an extensive paper prepared by Professor Leo Whinery for an ALI-ABA CLE course held in April 2001. (Professor Whinery serves as the Reporter for the Uniform Rules of Evidence drafting committee and teaches evidence at the University of Oklahoma College of Law. Whinery organized a large number of state court decisions into six categories according to their responses to Daubert. The largest group consisted of 19 states that still adhere to Frye. Leo H. Whinery, Expert Testimony Trends in State Practice and the Uniform Rules of Evidence (1999) (April 26, 2001), available on Westlaw at SF78 ALI-ABA 149, 176-182. These 19 take a more conservative approach to admissibility questions than that approved in Daubert, which is essentially the approach advocated in Tentative Draft No. 2. They include a number of very populous states: California, Florida, Illinois, Michigan, New Jersey, New York, and Pennsylvania. See, e.g., People v. Hill, 89 Cal. App. 4th 48 (2001); David v. Amtrak, 801 So. 2d 223, 228 (Fla. 2d Dist. 2001); Rodriquez v. Feinstein, 793 So. 2d 1057 (Fla. 3d Dist. 2001); U.S. Sugar Corp. v. Henson, 787 So. 2d 3 (Fla. 1st DCA 2000); and People v Wesley, 83 N.Y.2d 417, 422 (1994). Note especially a very recent decision, Donaldson v. Central Illinois Public Service Co., 2002 WL 254042 (Ill. 2002). In Donaldson, the Illinois Supreme Court rejected Tentative Draft No. 2’s stand on the legal sufficiency of epidemiological analyses. See also Jennings v. Baxter Healthcare Corp., 14 P.3d 596 (Ore. 2000).

While Reporters’ Notes are not a subject for floor action, I also note concern about some of the generalizations contained in those Notes about how state courts address toxic substance causation issues. I raise this because, in the absence of specific reference in the Comment, courts might infer state court rules from the reference in the Reporters’ Note. Examples of the general statements to which I refer are:

• Many state courts, whether they formally adopt Daubert as the standard governing expert witness testimony, continue to adhere to the standard provided in Frye v. United States, 293 F. 1013 (D.C. Cir. 1923), or decide to employ some modification of these standards, are increasingly aggressive in their review of expert witness testimony involving toxic substances and diseases.

(§ 28, Reporters’ Note to Comment c(1), p. 140.) This language appears to be promoting "aggressiveness" in matters of expert testimony. The use of "many" suggests that this may be the majority approach. Whether courts adopt the Daubert doctrine, continue to follow Frye, or take another route, presumably they do so because of governing case law or rules of court, not out of a generalized desire to be more "aggressive" toward expert testimony. The statement invites the misinterpretation that this approach may have been taken by a majority of states—a notion that is unsupportable, given Professor Whinery’s research for ALI-ABA.

• Courts frequently assess the state of the scientific record and only when it meets a sufficiency threshold is an expert witness permitted to testify about the existence of agent-disease causation. Legal standards for assessing the admissibility of expert testimony are, nevertheless, not addressed in this Restatement.

(§ 28, Reporters’ Note to Comment c(1), p. 140.) Again, the use of "frequently" suggests that this may be the majority approach, when it is not. In fact, states which ostensibly follow Daubert often reject the notion of testing proffered expert witness testimony against some threshold. See, e.g., Schafersman v. Agland Coop, 631 N.W.2d 862, 877-78 (Neb. 2001); Adroin v. Mills, 780 So. 2d 1265, 1272-73 (La. App. 3 Cir. 2001); New Haverford P’trship v. Stroot, 772 A.2d 792, 800 (Del. 2001); People v. Shreck, 22 P.3d 68 (Colo. 2001); and Ex parte Taylor, 2002 Ala. LEXIS 24.

In view of Professor Whinery’s research, and given the minimal state law analysis in Comment c, the Comment should indicate clearly that assertions about state law are made for illustrative purposes only, and should not be taken as indicative of a majority viewpoint.

Comment c contains a number of statements about matters of science concerning, for example, the meaning or significance of certain scientific evidence, the inferences that can (or cannot) be drawn from certain kinds of scientific evidence, the necessity for certain kinds or degrees of scientific evidence, and how physicians use certain diagnostic techniques. While some of these examples relate to how evidence has been evaluated when deciding burden of proof issues, and others relate to admissibility decisions, they are not rules of law. Nor are these matters within the expertise of most members of the Institute or even within the expertise of the courts that produced the cited decisions.

Some of such references in the reported decisions are, at best, only dicta. As the Reporters have acknowledged, others are either misguided or plainly wrong. Incorporating these statements in the Restatement in ways that suggest or which can be read as rules of law is, I submit, bad policy for the Institute. Not only does it run the substantial risk of misguiding courts and practitioners, but it would run counter to scientific inquiry which allows for continual challenge, reexamination, and change as necessary.

I therefore move that, for the reasons stated below, the following matters be changed in Comment c(3) and (4):

Even when satisfactory evidence of general causation exists, such evidence generally supports proof of causation only for a specific disease. The vast majority of toxic agents cause a single disease or a series of biologically-related diseases.(Of course, many different toxic agents may be combined in a single product, such as cigarettes.) When biological mechanism evidence is available, it may permit an inference that a toxic agent caused a related disease. Otherwise, proof that an agent causes one disease is generally not probative of its capacity to cause other unrelated diseases.

This statement goes too far. There are a number of toxic agents that can cause a wide variety of diseases. One example is lead, which, at toxic levels, can damage the nervous system, the blood supply, and the kidneys, complicate pregnancy, and injure the fetus. In adults, it can contribute to high blood pressure, damage reproductive organs, and even lead to death. Lead is just one example. Multiple toxic effects are caused by numerous other agents.

Other examples of widely used substances that cause multiple toxic effects include: tobacco smoke (which can cause or accelerate damage to the lungs and the cardiovascular system, peptic ulcer disease, reproductive disturbances, esophageal reflux, hypertension, fetal illnesses and death, and delayed wound healing. Tobacco and its various components also increase the risk of several types of cancer—especially cancer of the lung, mouth, larynx, esophagus, bladder, kidney, pancreas, and cervix; benzene (which can cause anemia, leukemia, suppression of the immune system, among other things), polychlorinated biphenyls (PCBs) (which can cause acne-like skin conditions in adults, and neurobehavioral and immunological changes in children); vinyl chloride (which can cause cancers in the liver, brain, lung, and hematolymphopoietic system); and arsenic (which can cause or aggravate skin cancer, liver cancer, intestinal and urinary bladder cancers, and meningioma, as well as affecting the vascular system and promoting the development of diabetes).

(B) (1) Strike Illustration 1, pp. 109-110 and delete the following language in Comment c(4), at p. 106 of Tentative Draft No. 2:

When proof of specific causation is an issue, courts confront the question of what evidence is required beyond the existence of general causation. Courts generally permit juries to infer specific causation from a group study when the study finds that exposure to the agent causes at least a doubling in the incidence of disease in a group exposed to the agent compared to a group that was not exposed. The doubled incidence means that of those exposed to the agent, the agent is inferred to have caused half of the disease found in that group.

(2) Replace the language quoted in (B)(1) above with the following language:

When proof of specific causation is an issue, courts may confront the question of what evidence is required beyond the existence of general causation. Their approach varies considerably from court to court, and from case to case. In the past, some courts have permitted inferences of specific causation when the party with the burden of proof offered an epidemiological study concluding that exposure to the agent caused a doubling in the incidence of disease in a group exposed to the agent compared to a group that was not exposed. However, other courts have not required proof of a doubled incidence, and instead looked to the entirety of the evidence presented in determining the sufficiency of causation proof. Most epidemiologists and biostatisticians themselves would not reject a finding of specific causation merely because the incidence of disease was not doubled.

Illustration 1 sets forth an absolute rule-of-law threshold that would require a "doubling" in the incidence of disease (i.e. relative risk of at least 2.0) to sustain the burden of proof on specific causation for exposure to a single toxin. The referenced text promotes the same idea by saying what courts permit when there is evidence of a doubling in the incidence, without any corresponding reference to how courts do or should treat such evidence when there is less than a doubling of incidence. Discussion of that question can be found in a recent article in which two environmental scientists reviewed all cases decided between 1982 and early 1999 that discussed "doubling," and found that courts required proof of doubled incidence in only 12 of the 31 cases they identified. Carruth & Goldstein, Relative Risk Greater Than Two in Proof of Causation in Toxic Tort Litigation, 41 Jurimetrics 195, 199 (2001). (Professor Goldstein was the lead author of the toxicology chapter in the Federal Judicial Center’s Manual on Scientific Evidence.)

The Reporters’ Note (at pp. 152-153) cites a number of cases allowing proof of causation without proof of a doubled incidence. However, the absence of any such references in the Comment itself could mislead courts and practitioners into concluding that "doubling" was a required rule of law—especially when the text was read in conjunction with Illustration 1. Moreover, the entire treatment could be read as elevating a certain quantum of scientific evidence to an absolute rule of law.

The mechanistic approach to epidemiological studies favored by some courts has no basis in science. The vast majority of epidemiologists and biostatisticians reject such an approach, and do not ignore relative risks of less than 2. See, e.g., Sander Greenland, The Relation of the Probability of Causation to the Relative Risk and the Doubling Dose: a Methodologic Error That Has Become a Social Problem, 89 Am. J. Pub. Health 1166 (1999); Sander Greenland and J.M. Robins, Epidemiology, Justice, and the Probability of Causation, 40 Jurimetrics 321 (2000); and J.M. Robins and Sander Greenland, The Message of Counterfactual Causality, 95 J. Am. Stat. Assn. 431 (2000). (Dr. Greenland is Professor of Epidemiology at UCLA School of Public Health and Research Professor of Preventive Medicine at the University of Southern California School of Medicine. He is considered a leading authority on quantitative methods and statistical theory in epidemiology. He is co-author of the leading textbook on epidemiology, Modern Epidemiology (2d ed. 1998)).

Clinical physicians use differential diagnoses not to determine the cause of a disease but to make a diagnosis of an illness or disease by ruling out other possible illnesses or diseases that might account for a patient’s symptoms. In most cases, the cause of a disease is of no clinical significance, and physicians do not seek it.

This statement is not about the law and it does not explain medical practice correctly. The human body has a limited number of ways of manifesting damage or pathology and thus physicians may not always need to know the underlying cause of the disease or damage to treat correctly. Usually, and particularly when harmful exposures may be continuing, the cause of the damage is critical for correct and effective treatment and in those cases, discovery of the cause is an extension of the differential diagnostic process as carried out by the treating physician. The language that should be deleted could undercut the valid practice of using court testimony by treating physicians to give opinions about the cause of their own patient’s illness, which not infrequently is based on techniques used to establish a differential diagnosis.

Respectfully submitted,

Larry S. Stewart

December 3, 2001

Professor Michael D. Green

Wake Forest University School of Law

P.O. Box 7206

Winston-Salem, NC 27109-7206

Dear Professor Green:

We, the undersigned scientists, understand that you are the chief draftsperson for the section of the Restatement of the Law of Torts dealing with liability for physical harm. Although we are not lawyers, and would not presume to critique legal scholarship, we are writing to express our views about some of the many scientific statements contained in this document.

For your convenience, our comments are keyed to particular pages and line numbers.

The difficulty that courts confront is that the line between reasonable inference and prohibited speculation is one of the more indistinct lines that exists in law. Different courts draw those lines at different points at different times; comparison of cases is very difficult because modest differences in the evidence can substantially affect the power of an inference. Thus, it is very difficult to state specific rules that locate the line between permissible inference and prohibited speculation.

(Emphasis is ours.)

We think you underestimate the problem. To say that something "is very difficult" is also to say that an attempt should be made to achieve the result. However, delineating specific rules is not simply "very difficult;" it is completely impossible. We, as scientists, believe judges should not be led to believe that clarity and perfection is susceptible of always or even often being attained. In this respect, law may be much like science. Science, more often than not, involves human judgments and assumptions, not mere mathematical certainties, as well as the appreciation of the possibility of multiple truths, not the realization of a single one.

With respect to epidemiology in particular, the eminent and oft-quoted public health professional, Sir Austin Bradford Hill, has cautioned that

we [cannot] usefully lay down some hard-and-fast rules of evidence that must be obeyed before we can accept cause and effect, [because all] scientific work is incomplete — whether it be observational or experimental. All scientific work is liable to be upset or modified by advancing knowledge. That does not confer on us a freedom to ignore the knowledge we already have, or to postpone action that it appears to demand at a given time.

Austin Bradford Hill, The Environment and Disease: Association or Causation?, 58 Proceedings of the Royal Society of Medicine 295, 299-300 (1965). Similarly, as the editor of the Journal of The American Public Health Association has pointed out, "conclusiveness in inferring causality—in epidemiology as with all studies of free-living human beings—is a desire more often than an accomplishment." Mervyn Susser, Rules of Inference in Epidemiology, 6 Regulatory Toxicology And Pharmacology 116, 127 (1986). As a consequence, those who seek in science the immutable truth they find lacking in the law are apt to be disappointed:

One notable similarity [between law and epidemiology] is the dependence of both fields upon subjective judgments.

* * *

In the end, a quality which lawyers should understand better than any—judiciousness—matters more than any. Scientists use both deductive and inductive inference to sustain the momentum of a continuing process of research. . . . The courts of law, and the courts of application, use inference to reach decisions about what action to take. Those decisions often cannot rest on certitudes, most especially when population risks are converted into individual risks.

Ibid. at 116, 128. See Mervyn Susser, Casual Thinking in The Health Sciences: Concepts And Strategies in Epidemiology (1973).

In the end, the line between permissible inference for the jury and impermissible speculation is one that must be determined based on the specific facts of the case and the jurisdiction’s general procedural approach to allocation of decision-making authority between judge and jury.

Consistent with what we write above, we think the "line" in question is so uncertain that the decision should simply be left in the province of the jury.

By contrast [to the causes of traumatic injuries], the causes of disease are much less well understood. The causes of many diseases are unknown and, even for others for which there are known causes, the known causes only explain a fraction of the incidence of disease, with the remainder due to unknown casual chains. Even when causes are identified, most often this occurs in group studies that are able only to determine an increase in disease among a group exposed to the agent. Because the biological mechanisms by which disease develops are more complicated, more difficult to perceive, and therefore less well understood than traumatic injuries, there remains substantial ignorance about the other casual factors that are necessary, along with the agent, for disease to occur in any individual.

(Bracketed words have been added to keep the paragraph in context.)

We think that this is ambiguous. Perhaps the ambiguity stems from the term "disease," which you use to describe both a common disease process, such as lung cancer, and a particular cause of that lung cancer in a single individual. The etiology of lung cancer is multifactorial, ranging non-exclusively from tobacco smoke inhalation to exposure to certain industrial substances, such as asbestos, arsenic, uranium, and diesel fuel, to certain non-industrial environmental factors, such as radon. (Sources: National Center for Biotechnology Information (NCBI), http://www.ncbi.nlm.nih.gov/disease/Lung.html, and the Alliance for Lung Cancer, http://www.alcase.org/education/detection/atrisk.html.) Given the many causes of lung cancer, no scientist can be absolutely certain that a given case of lung cancer in a particular person is caused by substance X. Nevertheless, if one is confronted with an individual with lung cancer who has a thirty year history of smoking three packs a day, scientists will conclude, in the absence of other specific knowledge, that this is consistent with the fairly high percentage of lung cancer in adults in industrialized countries attributable to tobacco smoke. For example, the NCBI reports that "About 90% of male lung cancer deaths and 80% of female lung cancer deaths are attributable to cigarette smoking." Ibid. The American Lung Association confirms this consensus view of scientists in stating that: "Lung cancer occurs most often in people over fifty who have long histories of cigarette smoking." (Source: American Lung Association, "Facts About Lung Cancer,"  http://www.lungusa.org/diseases/lungcanc.html#what.) Under these circumstances, in reference to the paragraph above, would it still be reasonable to assume that the etiology of this individual’s lung cancer is less well understand than say a sleep disorder related to a closed head traumatic injury?

The first step in proof that a toxic agent caused a disease is a demonstration that the plaintiff was exposed to the agent. There are three different means of exposure: inhalation; absorption; and ingestion.

We think this is incomplete, as individuals can be exposed to biological agents in various other ways, such as perinatal exposure of a baby while passing through the birth canal to various viruses that the mother may carry, exchange of body fluids during sexual intercourse, injection of biohazards during blood transfusions, and introduction of foreign substances in medical, cosmetic, or remedial procedures. Individuals can also be exposed to harmful radiation in various forms and through various means, ranging from ultraviolet rays (from the sun), infrared radiation (from heat lamps, microwaves, and radio waves), and ionizing radiation (from numerous industrial processes, medical tests and procedures, and environmental sources).

More frequently, however, the evidence of general causation that is available consists of scientific studies that attempt to identify toxic agents based on comparisons of the incidence of disease in groups of individuals or animals with different levels of exposure to the agent being studied.

(Emphasis is ours.)

At the very least this is incomplete, as you emphasize cohort studies and ignore the fact that epidemiologists often may infer causation on the basis of other types of studies, including case-control studies. From our point of view this should at the very least be added.

Also, this section on its own leads the uninformed reader to misunderstand the efficacy and purpose of animal testing. Given the realities of how animal testing needs to be conducted, it necessarily is designed to discern risk rather than incidence.

This is both an unfortunate and inaccurate comment. First, we are very troubled by the primacy given to epidemiological studies. Under certain specific circumstances epidemiological studies may be thought to trump other types of studies, but they are widely regarded as being very "insensitive" (i.e., often they don’t even see fairly strong effects that are actually there). If an epidemiological study demonstrates a positive effect, that study might trump other data and other kinds of studies (in vivo, in vitro, etc.). If epidemiological studies do not show positive effects (or just a "weak" effect), that could just be because the studies were quite insensitive. These kinds of "non-positive" results should not be put on an equal footing with ones that show an effect (the intrinsic asymmetry of the studies must be acknowledged; simply put, positive studies should count more than non-positive ones). For these reasons, while positive epidemiological evidence can definitely help establish causal connections, the absence of positive epidemiological studies, or the existence of multiple epidemiological studies that do not demonstrate a significant effect, can never be used to exonerate a suspected toxin. The most that scientists can say about epidemiological evidence that does not show positive effects is that such evidence fails to establish causation; scientists can never definitively forecast that other studies, both epidemiological and non-epidemiological, will not be able to do so—nor that the particular studies that were conducted were fully capable of measuring the entirety of the putative risk. If we may use a legal analogy, the absence of evidence of guilt does not prove innocence.

Epidemiologic studies are expensive and can take considerable time to design, gather and analyze data, and publish the results. Some plaintiffs may become ill and be forced to litigate long before epidemiologic research is complete. Sometimes epidemiologic evidence is impossible to obtain.

(Emphasis is ours.)

We think it might be more accurate to say: "can be completed." This is not only true as a matter of cost, but also because of the considerable latency for certain diseases—for example mesothelioma related to asbestos can easily take 20-30 years to develop.

Even when satisfactory evidence of general causation exists, such evidence generally supports proof of causation only for a specific disease. The vast majority of toxic agents cause a single disease or a series of biologically related diseases.

(Emphasis is ours.)

This is not accurate and we do not believe that there is any scientific basis for this view. Dioxin, PCBs, arsenic, benzene, kepone, DDT, vinyl chloride, asbestos, TCE/PCE, and many other toxic agents can cause a wide variety of diseases. Indeed, consider the example of lead. Too much lead in the body can seriously injure the brain, nervous system, red blood cells, and the kidneys. It disrupts the functioning of almost every brain neurotransmitter. It can cause permanent effects on the fetus and cause delays in early sensory-motor development. High lead exposures can cause a baby to have low birth weight or be born prematurely, or can even result in miscarriage or stillbirth. In adults, lead poisoning can contribute to high blood pressure and damage to the reproductive organs. Severe lead poisoning can cause subtle loss of recently acquired skills, listlessness, bizarre behavior, lack of coordination, vomiting, altered consciousness, and, as with children, seizures, coma and death. Poisoning without severe brain effects can cause lethargy, appetite loss, sporadic vomiting, abdominal pain, and constipation. This is just one example, however we could easily discuss the multitude of toxic effects of many other chemicals in addition to those listed above.

When proof of specific causation is an issue, courts confront the question of what evidence is required beyond the existence of general causation. Courts generally permit juries to infer specific causation from a group study when the study finds that exposure to the agent causes a doubling in the incidence of disease in a group exposed to the agent compared to a group that was not exposed. The doubled incidence means that of those exposed to the agent, the agent is a cause of half of the disease found in that group.

We recognize that for purposes of general causation a very few courts have limited testimony to that based on studies that showed at least a doubling. However, we are unaware that courts "generally" take this position and we are sure that a mechanistic approach to doubling has no basis in science. Although certain courts may reject studies lacking a doubling, and although judges may exclude the testimony of scientists who rely or even consider such studies, you should emphasize in the text (and not just the notes) that the vast majority of epidemiologists and biostatisticians consider such views to be completely wrongheaded. Epidemiologists do not ignore studies that show an RR of less than 2.0 and do not reject scientists who consider such studies to be relevant and useful. (For a rejection of such views, see, for instance, a variety of articles by Sander Greenland, the co-author of the leading textbook on epidemiology and a professor at UCLA, in which such a view is criticized, including Sander Greenland, The Relation of the Probability of Causation to the Relative Risk and the Doubling Dose: a Methodologic Error That Has Become a Social Problem, American Journal of Public Health, 89:1166-1169 (1999); Sander Greenland and J.M. Robins, Epidemiology, Justice, and the Probability of Causation, Jurimetrics, 40:321-340 (2000); and J.M. Robins and Sander Greenland, The Message of Counterfactual Causality, Journal of the American Statistical Association, 95:431-435 (2000).

We are also troubled by the fact that your comments may be read to say that doubling constitutes a prerequisite if you want to use those studies "in addition" for "specific" causation. There are a variety of difficulties about this idea and some additional (hidden) assumptions. These are discussed in the "Reporters’ Note" portion of your section, but the fact that those assumptions are critical to the intended interpretation as given in the main text portion of the section is not made very clear. The main text does give some reasons why specific plaintiffs should not be held to the doubling requirement and this is helpful. We urge you to clarify some of the assumptions that you reference in the notes so as not to inadvertently lead judges to bar litigants from using studies where the RR is less than 2.0 for general causation or for many instances of specific causation where not enough is known about the mechanism to even sustain the reasoning behind the RR = 2.0 argument. Moreover, the similarity, or lack thereof, of the plaintiff to the "average" in the study group is critical, (it is discussed in the text) but it is not made clear enough that this also should call into question any bright-line nature of the RR = 2.0 criterion, which we think is its main attraction for judges. Nor is there a proper referent to the justification for elimination of confounding factors as they relate to specific causation. While an RR might only be 1.5, for example, this is not to say in any fashion that when specific causation is analyzed in the light of all other factors that a reasonable conclusion could not be reached that the plaintiff was more likely than not to be the one in three who was affected.

Clinical physicians use differential diagnoses not to determine the cause of a disease but to make a diagnosis of an illness or disease by ruling out other possible illnesses or diseases that might account for a patient’s symptoms. In most cases, the cause of a disease is of no clinical significance, and physicians do not seek it.

Again, we believe this to be wrong. First, there is no real distinction between the techniques used in differential diagnosis to diagnose and treat a disease and the techniques used to determine the cause of a disease. The draft’s discussion of "differential diagnosis" introduces an artificial distinction that is not recognized by clinicians. Moreover, the idea that physicians use the differential diagnosis concept to "make a diagnosis of an illness" but not to "determine the cause of the disease" is clearly without foundation, as is the notion that physicians do not seek to find the cause of disease. Many physicians, such as cardiologists, allergists, immunologists, dermatologists, otolaryngologists, and experts in occupational, environmental, and infectious diseases must necessarily establish the cause of disease in order to effectuate treatment modalities. Indeed, for such physicians, establishing the cause of an individual’s symptomatology facilitates and may be absolutely critical in determining a proper course of treatment. Finally, as currently drafted, this section might have the unfortunate effect of undercutting the long-recognized validity of treating physicians appearing in court and giving their opinions about the cause of their own patient’s illness.

We appreciate the difficulty of your task and the obvious effort that went into producing this draft. We hope our comments are helpful.

Sincerely yours,

David Kriebel, Sc.D.
Department of Work Environment Epidemiology and Occupational Health
University of Massachusetts at Lowell

Mary O’Brien, Ph.D., Staff Scientist
Science and Environmental Health Network
Ames, IA

David Ozonoff, MD, MPH
Professor and Chair, Department of Environmental Health
Boston University School of Public Health

Frederica P. Perera, Dr.P.H.
Professor and Director
Columbia Center for Children’s Environmental Health
Joseph L. Mailman School of Public Health
Columbia University

Anthony Robbins, M.D.
Chair, Department of Family Medicine & Community Health
Tufts University School of Medicine
Boston, MA

George C. Rodgers, Jr. M.D., Ph.D.
Professor of Pediatrics and Pharmacology/Toxicology
University of Louisville School of Medicine
Louisville, KY

Ted Schettler MD, MPH
SEHN Science Director
Department of Internal Medicine
Boston Medical Center
Boston, MA

Joel A. Tickner, Sc.D.
Department of Work Environment
University of Massachusetts at Lowell

David Wallinga, M.D., MPA
Senior Scientist
Institute for Agriculture and Trade Policy
Minneapolis, MN

Langdon Winner
Professor of Political Science
Rensselaer Polytechnic Institute
Troy, NY