Blueprint for General Causation Analysis

in Toxic Tort Litigation

 

William O. Dillingham

Patrick J. Hagan

Rodrigo E. Salas

 

I.

Introduction

This article discusses some of the legal and scientific principles that are appropriate to the step-by-step judicial determination of what type of general causation expert testimony should be admitted in toxic tort cases.  Specifically, it provides a blueprint for analyzing general causation in the context of toxic tort litigation and examines the extent to which Daubert I and its progeny, when taken to their logical extensions, guide the admissibility determination.[1]

Since 1923, when the United States Court of Appeal for the District of Columbia Circuit decided Frye v. United States,[2] courts have been concerned with the introduction of “junk science” in the courtroom. Rigorous examination of expert testimony,

 

(1) assur[es] that those persons most qualified to assess the validity of the scientific technique would have the determinate voice, (2) provid[es] a “minimal reserve of experts” to critically examine each technique in a particular case, [and] (3) promot[es] uniformity of decision based on finding consensus in the scientific community . . . .[3]

 

Additional compelling reasons exist for limiting the encroachment of junk science into the courtroom: a “posture of judicial caution in this area [is] supported by the considerable weight jurors tend to give to scientific evidence presented by experts with impressive credentials,” a “misleading aura of certainty . . . often envelops a new scientific process, obscuring its currently experimental nature,” and “scientific proof may in some instances assume a posture of mystic infallibility in the eyes of a jury.”[4]

Two key legal concepts of causation have emerged as courts have attempted to keep junk science out of the courtroom: general causation and specific causation.  Courts define general causation as “the capacity of a product to cause injury,” and specific causation as “proof that the product in question caused the injury of which the plaintiff complains.”[5]  General causation may be thought of as a scientifically established cause-and-effect relationship.  To satisfy this burden, sufficient hypotheses and testing must be demonstrated to establish that a disease or condition can arise from exposure to a certain substance.  Specific causation, on the other hand, involves a variety of factors including level, duration and proximity of exposure, also known as “dose,” which tend to show that the person’s alleged exposure, in fact, caused his or her condition.[6]

Accordingly, a toxic tort plaintiff’s burden is twofold: first, to show that the substance to which he was allegedly exposed is capable of causing his injury; and second, to show that said exposure was the actual cause of the injury.[7]  Toxic tort plaintiffs must prove the admissibility of their expert testimony in both general causation and specific causation context by a preponderance of proof.[8]  To determine whether a toxic tort plaintiff can satisfy his or her burden of proof, courts must make “a preliminary assessment of whether the reasoning or methodology underlying the testimony is scientifically valid and of whether that reasoning or methodology properly can be applied to the facts in issue.”[9]  This article only addresses judicial intervention and edification relating to general causation.[10]

The United States Supreme Court in Daubert I, when asked to determine what kind of expert testimony or evidence was admissible under the Federal Rules of Evidence (“FRE”), enumerated several factors to assist judges in performing their gate-keeping function.[11]  The United States Court of Appeals in Daubert II expounded on that gate-keeping function, holding that trial courts are to undertake a two-part analysis: (1) whether the experts’ testimony reflects “scientific knowledge,” i.e., whether the experts’ findings are derived by the scientific method; and (2) whether the experts’ testimony is “relevant to the task at hand,” i.e., that it logically advances a material aspect of the proposing party’s case.[12]  More recently, the Supreme Court clarified that the gate-keeping responsibilities imposed by FRE 702 extend to the testimony of all experts.[13]

The Supreme Court has also acknowledged that, in disease causation cases, “conclusions and methodology are not entirely distinct from one another.”[14] The Court warned that,

 

[N]othing in either Daubert or the Federal Rules of Evidence requires a district court to admit opinion evidence that is connected to existing data only by the ipse dixit of the expert.  A court may conclude that there is simply too great an analytical gap between the data and the opinion proffered.[15]  

 

A “methodology-based” inquiry, as described herein, properly considers every aspect of the reasoning/scientific process, from the selection of the data, to the interpretation of that data and, finally, to the conclusions that may be derived from the data.[16]

            As discussed hereinafter, when assessing expert medical causation testimony, a methodology-based inquiry requires, at a minimum, consideration of the following factors:

 

·        Replicated human epidemiological studies to prove causation in toxic tort cases;

·        Epidemiological studies that address the agent at issue;

·        Epidemiological studies that address the disease at issue;

·        Epidemiological studies that must be statistically significant; while,

·        Opinions that rely solely on in vivo or in vitro studies ought to be inadmissible.

 

The foregoing list of disease causation evidentiary requirements demonstrates that valid epidemiological evidence is the lynchpin of general causation and illustrates the depth to which analytical courts have probed in evaluating general causation testimony in toxic tort cases.  The following step-by-step general causation analysis insures that there will be no bridging of the scientific method by “leaps of faith” that might undermine the reliability of the entire scientific reasoning process or the conclusions derived therefrom.

II.

Epidemiology: The Lynchpin Of General Causation

            Epidemiology is “the study of distribution of disease in populations and the risk factors associated with particular diseases.”[17]  Epidemiology studies[18] are the most useful and conclusive methodology used to validate causal relationships in toxic tort cases.[19]  “Epidemiologists use an analytical tool known as the null hypothesis, which postulates that there is no association between a specific exposure and a particular outcome.”[20]  “The goal of an epidemiology study is to determine whether one can reject the null hypothesis and conclude that, in fact, there is an association between the exposure and the outcome.”[21]  Accordingly, numerous courts have held that epidemiological evidence is the prime scientific undertaking required to prove general causation.

For example, in Grant v. Bristol-Myers Squibb,[22] the court excluded the testimony of the plaintiffs’ experts due to their failure to explain why their opinions should overcome the twenty epidemiological studies that either found no association or no statistically significant association between breast implants and systemic disease.[23]  In fact, it has been stated that “epidemiological studies provide ‘the primary generally accepted methodology for demonstrating a causal relation between a chemical compound and a set of symptoms or disease.’”[24]

A. Reference to Single Epidemiological Study Not Sufficient

Courts have repeatedly held that the epidemiological evidence necessary to prove general causation must be replicated,[25] which requires that the association in an epidemiological study be observed in other studies among different subjects and variables.[26]  Replication of results is a basic requirement of the scientific method as scientists generally cannot be assured that the results of any individual study are not due to bias, confounding[27] or chance.[28]  The inability to replicate a study calls into question the data therein, the analytical model utilized, the conclusions reached and, at times, all aspects of the study.  Thus, absent evidence from valid and replicated epidemiological studies, a single study showing a cause-and-effect relationship does not establish that relationship for purposes of general causation.

 

III.

Epidemiological Studies Must Address Agent At Issue In Litigation

In order to establish general causation, the epidemiological studies offered as a basis for expert opinion ought to address the agents at issue in the litigation.[29]

In Siharath v. Sandoz Pharmaceuticals Corp.,[30] two plaintiffs, each of whom suffered hemorrhagic strokes after taking a bromocriptine drug, brought a products liability action against Sandoz Pharmaceuticals.  Both plaintiffs were postpartum women who suffered a stroke after taking Parlodel, which was prescribed to suppress lactation.  The issue before the court was whether the drug was capable of causing and whether it, in fact, caused the strokes or whether the temporal association of taking the drug and a subsequent stroke was merely coincidental.[31]  The Siharath plaintiffs argued that bromocriptine, the active ingredient in the drug Parlodel, caused hypertension, seizures and ischemic strokes.  The plaintiffs further contended that Parlodel, through bromocriptine, could also cause hemorrhagic strokes.

On a motion to exclude the plaintiffs’ expert testimony of causation, the determinative issue was whether the epidemiological evidence relied upon by the plaintiffs’ experts was sufficiently reliable and relevant to establish causation.  The Siharath court explained, “the burden is on Plaintiffs to show that well-conducted epidemiological studies do show a statistically significant relationship between Parlodel[ ] and seizures and stroke.  It is not defendant’s burden to show the lack of such relationship.”[32]  The Siharath plaintiffs brought forward four epidemiological studies which investigated a possible association between Parlodel and stroke.  After examining the plaintiffs’ epidemiological evidence, the Siharath court concluded that each of the studies, for separate reasons, lacked the requisite reliability to support the plaintiffs’ arguments that Parlodel could cause seizures and hemorrhagic strokes.[33]

Since “no evidence [had] been offered of an increase in postpartum strokes after [Parlodel] was approved for suppression of lactation [and] no evidence [had] been offered of a decrease of postpartum strokes after the approval for suppression of lactation was withdrawn,” the court questioned whether “the causation opinions of plaintiffs’ experts [were] merely speculative and not based on scientific knowledge.[34]  Thus, it was improper to simply extrapolate from one chemical to another, or to lump chemicals together: studies where exposure to the chemical is unverified, unquantified, unidentified, or otherwise speculative cannot provide reliable foundational support for a causation opinion concerning that chemical.[35]

As Siharath demonstrated, in order for epidemiological evidence – or any other type of reliable scientific knowledge – to be relevant, it must, at a minimum, address the specific agent at issue.   Where experts offer general causation opinions, the scientific literature they rely upon to formulate those opinions should include epidemiological studies that concern the same agent at issue in the litigation; otherwise, the proffered evidence ought to be inadmissible as it is unreliable.

IV.

Epidemiological Studies Must Address Disease At Issue In Litigation

In order to establish general causation, the epidemiological studies offered as a basis for expert opinion should also address the specific diseases at issue in the litigation.[36]  As the Siharath court wrote, “the burden is on Plaintiffs to show that well-conducted epidemiological studies do show a statistically significant relationship between Parlodel and seizures and strokes,” and “. . .  Plaintiffs’ [sic] have not pointed to a single case report involving a postpartum woman who suffered a hemorrhagic stroke.”[37]

The Texas Supreme Court in Havner also reiterated the requirement that, according to accepted scientific methodology, epidemiology studies must address the disease of concern.  At issue in Havner was the cause of a birth defect that resulted in Kelly Havner being born without fingers on her right hand.  During the course of her pregnancy, Kelly’s mother had taken Bendectin to relieve nausea and other symptoms associated with morning sickness.  Alleging that Bendectin had caused their daughter’s limb reduction defect, the Havners sued Merrell Dow, the maker of Bendectin, on theories of negligence, defective design and defective marketing.  On a motion challenging the legal sufficiency of the plaintiffs’ causation evidence, studies on Bendectin were submitted, which demonstrated statistically significant results but that dealt with birth defects other than limb reduction defects.  In determining the utility of these studies on the causation relationship before it, the Havner court wrote, “these studies cannot of course support a finding that Bendectin causes limb reduction defects.[38]

It is improper to simply extrapolate from one disease to another, or to lump diseases together.  Studies that analyze a different disease or a different form of a disease do not, and cannot, provide any support for making a causal link between the disease at issue and the agent in dispute.[39]  As one court has stated, “[o]ne simply cannot assume that just because a substance causes a particular kind of cancer, it will cause another type.”[40]  Therefore, studies that do not address the disease at issue lack any indicia of scientific reliability and methodology.[41]  For these reasons, when experts offer general causation opinions, the scientific literature they rely upon to formulate those opinions should include epidemiological studies that concern the same disease at issue in the litigation.  Otherwise, the proffered evidence is unreliable and ought to be inadmissible.

V.

Epidemiological Studies Must Be Statistically Significant

            In order to disprove the null hypothesis, i.e., in order to establish an association between a specific exposure and a particular outcome, epidemiologists must establish whether those individuals exposed to the particular agent have a greater risk of developing the particular outcome than those not so exposed.  In epidemiological terms, the difference in risk of contracting the disease is known as the “relative risk.[42]  As the Siharath court explained:

 

A relative risk of 1.0 means that the agent has no effect on the incidence of disease.  When the relative risk reaches 2.0, the agent is responsible for an equal number of cases of disease as all other background causes.  A relative risk of 2.0 implies a 50 percent likelihood that an exposed individual’s disease was caused by the agent in question. . . .  Thus, in the world of epidemiology, the threshold for concluding that an agent was more likely than not the cause of a disease is a relative risk greater than 2.0.[43]

 

Therefore, to establish a causal relationship, the plaintiffs’ experts must be able to opine a causation factor greater than 50-50 and must establish that the purported chemical-disease association is statistically significant.   Faced with the plaintiffs’ experts who could not and would not opine a causation factor greater than 50-50, in Jones v. Ortho Pharmaceutical Corp,[44] the California Court of Appeal declared:

 

If the experts cannot predict probability in these situations, it is difficult to see how courts can expect a jury of laymen to be able to do so. [¶]  This requirement does in some instances place extraordinary burdens of proof on claimants.  But once the theory of causation leaves the realm of lay knowledge for esoteric scientific theories, the scientific theory must be more than a possibility to the scientists who created it.[45]

 

The court further reasoned that since “[t]he testimony of plaintiff’s experts [was] such that it was equally probable that the development of the carcinoma in situ was due to a cause for which defendant could not be liable,” it could only conclude “that plaintiff did not establish a prima facie case and that the motion for nonsuit was properly granted.”[46]

Absent evidence that the relative risk is equal to or greater than 2.0, there is a fifty percent or more chance that the “disease was not associated with the exposure.”[47]  Therefore, when the relative risk is less than 2.0, the null hypothesis cannot be disproved.

In addition to measuring relative risk, an epidemiological study must also measure the probability that a particular risk is due to chance, a concept expressed in terms of a confidence interval.[48]  Under generally accepted epidemiologic methods, a confidence interval indicates how much that observed results might be expected to vary due to chance.  Specifically, a confidence interval shows the range of relative risk values within which the result could be expected to fall ninety-five percent of the time due to random variation or chance.[49]  When the confidence interval is ninety-five percent and contains a number greater than one, chance is considered an unlikely explanation for the observed result and the result is considered statistically significant.  When the confidence interval is ninety-five percent and contains the number one or a number less than one, the observed result is more likely than not due to chance and, therefore, is not statistically significant. [50]   Only when the confidence interval is ninety-five percent and contains a number greater than one, may the observed association be deemed a true association.[51]  Only when chance is eliminated as a potential cause of the observed association may the null hypothesis be disproved.

VI.

Opinions Relying Solely On In Vivo Or In Vitro Studies Should Be Inadmissible

            Evidence of in vivo[52] and in vitro [53]studies alone is insufficient to establish causation, since there is no generally accepted method for extrapolating from that type of evidence to reach conclusions about similar effects in humans.[54]

            Courts overwhelmingly hold that drawing causation conclusions in humans from animal studies, in the absence of confirming human epidemiology, is generally insufficient to satisfy admissibility standards.[55]  The main reason for these concerns is that “[e]xtrapolations from animal studies to human beings generally are not considered reliable in the absence of a credible scientific explanation of why such extrapolation is warranted.[56]  The use of animal studies to prove causation in humans has, at a minimum, two distinct disadvantages:

 

First, extrapolating from animals to humans is difficult because “differences in absorption, metabolism, and other factors may result in interspecies variation in responses.” …   Second, “the high doses customarily used in animal studies requires consideration of the dose-response relationship and whether a threshold no-effect dose exists.”[57]

 

In addition, “laboratory animal studies . . . are generally viewed with more suspicion than epidemiological studies, because they require making the assumption that the chemicals behave similarly in different species.”[58]

            Thus, to ensure that an expert’s conclusions based on animal studies are reliable, there must exist ‘a scientifically valid link between the sources or studies consulted and the conclusion reached.’”[59]  Expert testimony concerning general causation in humans, which is solely derived from animal studies, is unreliable and ought to be inadmissible in a court of law.

            Similarly, reliance on in vitro studies to opine general causation in humans is insufficient to establish admissibility.  Courts have consistently held that laboratory studies on Petrie dishes or test tubes are not generally accepted as predicative of the human experience.[60]  The same concerns relating to in vivo studies are present (and are amplified) in in vitro studies: species specificity (differences in absorption, metabolism), route of exposure, dose, etc.  Accordingly, expert testimony concerning general causation in humans, which is solely derived from in vitro studies, is unreliable and ought to be inadmissible in a court of law.

VII.

Conclusion

            In determining the admissibility of expert causation testimony, trial courts should not only adhere to the judicial process but also scrupulously analyze the proffered testimony to ensure that the scientific process has been followed.  In performing their gatekeeping function, judges must strictly scrutinize each step of the analytical process utilized by the expert in question to determine the reliability of the proffered evidence or opinion testimony. This article has outlined a few of the more prevalent scientific concepts that courts have used to assist in determining the admissibility of expert testimony offered on general causation.

            Of note, there are additional, more traditional concepts, which also guide the general causation determination.  These were expressly recognized by the United States Supreme Court in Daubert I, and include: (1) whether the experts’ theories or techniques have been tested; (2) whether the theory or technique has been subjected to peer review and publication; (3) in the case of a particular scientific technique, what is the known or potential rate of error; and (4) whether the theory or technique has been generally accepted.   Those concepts also have been subject of ample commentary and judicial decisions and, therefore, have not been addressed here.  Those admissibility requirements in conjunction with the instant “blueprint” should guide the resolution of the first causation threshold, general causation.

            Only after reliable, relevant, and admissible evidence concerning general causation has been put forth by a toxic tort plaintiff should the issue of specific causation be reached.


Endnotes

 



           Submitted by the authors on behalf of the FDCC Toxic Tort and Environmental Law Section.

[1]           Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579 (1993) (“Daubert I”).  For an earlier treatment of Daubert I and the admissibility of expert testimony in California, please see Patrick J. Hagan & Pamela Winston Bertani, State Variations on the Daubert Theme: California, 51 Fed’n Ins. & Corp. Couns. Q., 27 (2000).

[2]           293 F. 1013 (D.C. Cir. 1923).

[3]           People v. Leahy, 882 P.2d 321, 325 (Cal. 1994) (citing People v. Kelly, 549 P.2d 1240, 1244 (Cal. 1976) (adopting Frye)).

[4]           Kelly, 549 P.2d at 1244-46.

[5]           Siharath v. Sandoz Pharms. Corp., 131 F. Supp. 2d 1347, 1352 (N.D. Ga. 2001) (citing Wheat v. Sofamor, S.N.C, 46 F. Supp. 2d 1351, 1357 (N.D. Ga. 1999) (product liability action excluding testimony that failed to establish (1) that Parlodel is capable of causing stroke and (2) that Palodel did in fact cause plaintiffs’ strokes)).

[6]           Siharath, 131 F. Supp. 2d at 1352. See e.g. D. T. Ralston, Toxic Tort Causation – Not Just Chemical Exposure Plus Symptoms, Maely’s Daubert Rep., Vol. 4, No. 5, at 15-25 (2000).

[7]           See e.g., Raynor v. Merrell Pharms., Inc., 104 F.3d 1371, 1376 (D.C. Cir. 1997).

[8]           Daubert I, 509 U.S. at 593 n. 10; see e.g., Carnegie Mellon Univ. v. Hoffmann-LaRoche, Inc., 55 F. Supp. 2d 1024, 1030 (N.D. Cal. 1999).

[9]           Daubert I, 509 U.S. at 592-93.

[10]          As the name implies, specific causation inquiries are fact and case-specific.  This paper will focus solely on foundational principles dealing with general causation and, therefore, will not explore admissibility requirements concerning specific causation.

[11]          Daubert I, 509 U.S. at 593-95 (factors include (1) whether the experts’ theories or techniques have been tested; (2) whether the theory or technique has been subjected to peer review and publication; (3) in the case of a particular scientific technique, what is the known or potential rate of error; and (4) whether the theory or technique at issue has widespread acceptance).

[12]          Daubert v. Merrell Dow Pharms., Inc., 43 F.3d 1311, 1315 (9th Cir. 1995) (applying California law) (“Daubert II”).

[13]          Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137 (1999).

[14]          Gen. Elec. Co. v. Joiner, 522 U.S. 136, 146 (1997) (“Joiner”).

[15]          Id.

[16]          The scientific method calls for the formulation of a scientific hypothesis, designing a study that can answer it, collecting objectively verifiable evidence that will address the hypothesis, and drawing only those conclusions supported by the evidence. See Landrigan v. Celotex Corp., 605 A.2d 1079, 1088 (N.J. 1992) (with respect to proffered testimony regarding the causation of plaintiff-decedent’s colon cancer, holding that the trial court “should examine each step in [the expert’s] reasoning”); Hall v. Baxter Healthcare Corp., 947 F. Supp. 1387, 1401 (D. Or. 1996) (“This court need not and should not ignore any step in that process, but must ensure that in each step, from initial premise to ultimate conclusion, the expert faithfully followed valid scientific methodology.”).

[17]          Wade-Greaux v. Whitehall Labs., Inc., 874 F. Supp. 1441, 1451 (D. V.I. 1994).  See e.g., Brock v. Merrell Dow Pharms., Inc., 874, F.2d 307, 311 (5th Cir. 1989) (“Epidemiology attempts to define a relationship between a disease and a factor suspected of causing it . . . . ”).  Epidemiology is “the branch of medical science that deals with the incidence, distribution, and control of a disease in a population.” Webster’s Collegiate Dictionary 389 (10th ed.1998).

[18]          In re Agent Orange Litigation, 611 F. Supp. 1223, 1231 (E.D.N.Y. 1985) (citing Dore, A Commentary on the Use of Epidemiological Evidence in Demonstrating Cause-in-Fact, 7 Harv. Envtl. L. Rev. 429, 431 (1983) (“Epidemiological studies rely on ‘statistical methods to detect abnormally high incidences of disease in a study population and to associate these incidences with unusual exposures to suspect environmental factors.’”); see also Siharath v. Sandoz Pharms. Corp., 131 F. Supp. 2d 1347, 1356 (N.D. Ga. 2001) (“Epidemiologists employ cohort studies, case control studies, and ecological studies to determine whether individuals exposed to an agent have a greater risk of developing the disease in question.”).

[19]          See Allen v. Pa. Eng. Co., 102 F.3d 194, 197 (5th Cir. 1996) (quoting Brock, 874 F.3d at 311 (“Undoubtedly, the most useful and conclusive type of evidence in a [products liability] case is epidemiological studies”); Lopez v. Wyeth-Ayerst Labs., Inc., 1996 WL 784566 *1, *3 (N.D. Cal. 1996) (“epidemiological evidence is one of the most valuable pieces of scientific evidence in causation”); Casey v. Ohio Med. Prods., 877 F. Supp. 1380, 1385 (N.D. Cal. 1995) (considering lack of epidemiological study an important factor in determining reliability).

[20]          Wade-Greaux, 874 F. Supp at 1451-52.

[21]          Id. at 1452.

[22]          97 F. Supp. 2d 986 (D. Ariz. 2000).

[23]          See also Hall v. Baxter Healthcare Corp., 947 F. Supp. 1387, 1403 (D. Or. 1996) (excluding plaintiff’s expert’s testimony on causation between breast implant and systemic disease); Brock v. Merrell Dow Pharms., Inc., 874 F.2d 307, 315 (5th Cir. 1989) (affirming District Court’s exclusion of testimony of plaintiffs’ experts finding that they did not present sufficient proof that ingestion of Bendectin during pregnancy caused Poland’s Syndrome); In re Breast Implant Litig., 11 F. Supp. 2d 1217, 1224 (D. Colo. 1998) (excluding testimony of plaintiffs’ experts for failure to use valid, statistically significant epidemiological studies showing an association between breast implants and auto-immune diseases); Allen, 102 F.3d at 194 (affirming District Court’s exclusion of plaintiffs’ experts for their failure to present valid, statistically significant epidemiological studies showing an association between ethylene oxide and brain cancer); Wade-Greaux v. Whitehall Labs., Inc., 874 F. Supp.1441, 1484 (D. V.I. 1994) (excluding testimony of plaintiffs’ experts and granting summary judgment for defendant where plaintiffs’ experts failed to conduct or rely on valid, consistent epidemiological studies showing an association between over-the-counter asthma medications and skeletal birth defects); Haggerty v. Upjohn Co., 950 F. Supp. 1160, 1168 (S.D. Fla. 1996) (granting defendant’s motion for summary judgment while holding that plaintiff did not show a genuine issue of material fact with respect to medical causation of a prescription sleeping medication and its claimed psychiatric side effects); and Lynch v. Merrell-Nat’l Labs., Inc., 830 F.2d 1190, 1196-97 (1st Cir. 1987) (affirming summary judgment for Bendectin manufacturer where plaintiff offered only criticism of defendant’s epidemiological evidence with no epidemiological evidence of her own).

[24]          Siharath v. Sandoz Pharms. Corp., 131 F. Supp. 2d 1347, 1356 (N.D. Ga. 2001) (applying Daubert II) (citing Conde v. Velsicol Chem. Corp., 804 F. Supp. 972 (S.D. Ohio 1992), aff’d, 24 F.3d 809 (6th Cir. 1994)).

[25]          See Blum v. Merrell Dow Pharms., Inc., 705 A.2d 1314, 1323 (Pa. Sup. Ct. 1997) (“Replicated epidemiological studies consistently finding a strong association are necessary to establish causation . . ..”); Merrell Dow Pharms., Inc. v. Havner, 953 S.W.2d 706, 722 (Tex. 1997) (“Havner”) (observing that an isolated study finding a statistically significant association would not constitute legally sufficient evidence of causation).

[26]          Havner, 953 S.W.2d at 727 (“[I]t is important that any conclusions about causation be reached only after an association is observed in studies among different groups and that the association continues to hold when the effects of other variables are taken into account.”).

[27]          Wade-Greaux, 874 F. Supp. at 1452 (“A ‘confounding factor’ is an exposure to one agent that is temporally associated with exposures to other agents, such that they appear to be having the same effect.”)  A well-conducted epidemiology study will use statistical techniques to eliminate the possible effects of confounding factors. Id.

[28]          See id. (“[A]bsent a large single study with an overwhelmingly high rate ratio, epidemiologists generally require several individual positive studies linking a specific exposure with a particular outcome before they draw any conclusions as to an association.”).

[29]          See Mitchell v. GenCorp, Inc., 165 F.3d 778, 782 (10th Cir. 1999) (“Without scientific data supporting their conclusions that chemicals similar to benzene cause the same problems as benzene, the analytical gap in the experts’ testimony is simply too wide for the opinions to establish causation”); Lofgren v. Motorola, 1998 WL 299925, at *17 (Ariz. Super. June 1, 1998) (under Frye, “extrapolation from studies of different chemicals does not constitute accepted methodology in the absence of a scientific and accepted basis for doing so”); Cavallo v. Star Enterprise, 892 F. Supp. 756, 766 (E.D. Va. 1995) (Fed. R. Evid. 702 requires “that the expert demonstrate a scientifically valid basis for projecting the findings of a study identifying a different chemical-illness relationship to the proffered causal theory.”), aff’d in part, rev’d in part on other grounds, 100 F.3d 1150 (4th Cir. 1996), cert denied, 522 U.S. 1044 (1998); Savage v. Union Pac. R.R. Co., 67 F. Supp. 2d 1021, 1038 (E.D. Ark. 1999) (stating that the issue of whether exposure to coal products in general can cause basal cell carcinoma is altogether different from the limited issue of whether exposure to “creosote” can cause that disease).

[30]          131 F. Supp. 2d 1347 (N.D. Ga. 2001).

[31]