160 F. Supp. 2d 1291 | N.D. Ala. | 2001
Elizabeth L. BRASHER, et al., Plaintiffs,
v.
SANDOZ PHARMACEUTICALS CORP., Defendant.
Ruby Quinn, et al., Plaintiffs,
v.
Sandoz Pharmaceuticals Corp., Defendant.
United States District Court, N.D. Alabama, Southern Division.
*1292 Lister Hill Proctor, Wanda Jo Wallace Batson, Proctor & Vaughn, Sylacauga, AL, Denise M. Dunleavy, Ellen Relkin, Catherine T. Heacox, Jerry M. Kristal, Weitz & Luxenberg PC, New York City, for Plaintiffs.
Joe G. Hollingsworth, Katherine Latimer, Bruce J. Berger, William J. Cople, Scott S. Thomas, Neil S. Bromberg, Spriggs & Hollingsworth, Washington, DC, Edward S. Sledge, III, Archibald T. Reeves, IV, McDowell Knight Roedder & Sledge LLC, Mobile, AL, for Defendant.
ORDER DENYING DEFENDANT'S MOTIONS FOR SUMMARY JUDGMENT ON MEDICAL CAUSATION
PUTNAM, United States Chief Magistrate Judge.
Before the court is the defendant's[1] motion for summary judgment on medical causation, filed July 15, 1999, on which the court conducted a Daubert hearing on June 5, 2001.[2] The parties have filed extensive briefs and voluminous exhibits dealing with whether the proposed opinions of plaintiffs' expert witnesses[3]that the strokes suffered by the plaintiffs were caused by ingestion of a drug manufactured by Sandozare scientifically reliable under Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S. Ct. 2786, 125 L. Ed. 2d 469 (1993), and cases following it. After hearing and reading the testimony of experts, and after having struggled through much of the scientific *1293 literature offered by the parties, the court concludes that the expert opinions are scientifically reliable and admissible. Consequently, defendant's motion for summary judgment on the ground that plaintiffs cannot establish the necessary causal link between the ischemic strokes and Sandoz's drug, Parlodel,[4] is due to be denied.
Factual and Procedural Background
A. Ruby Quinn
Plaintiff Ruby Quinn[5] delivered a child by Caesarean section on August 25, 1993. She initially chose to breast-feed her baby. On August 31, 1993, Ms. Quinn returned to the hospital, complaining of stomach pains. While in the hospital, she decided to stop breast-feeding. She was given a prescription for Parlodel on or about September 1, 1993. The Parlodel was prescribed to be taken twice a day for 10 days, although it is not clear whether Ms. Quinn took the Parlodel as prescribed, or by using only one tablet per day for 20 days.[6] There is no evidence that Ms. Quinn had ever used Parlodel before.
On or about September 14, 1993, Ms. Quinn developed headaches, which she attributed to dental problems. Her dentist prescribed Tetracycline, Amoxil, and a pain medication. On September 21, 1993, the plaintiff developed paralysis on her left side and experienced slurred speech. Emergency medical workers recorded her blood pressure as 180/90. She was taken to the hospital, where her blood pressure was recorded as 180/84. An MRI was done, revealing a recent cerebral infarction in the right lenticular nuclear and periventricular region. Other tests showed an occlusion of the right middle cerebral artery. A cerebral angiogram performed on September 27, 1993, showed an abrupt occlusion of the right middle cerebral artery. In layman's terms, Ms. Quinn had suffered an ischemic stroke due to the blockage of blood flow to the brain through the right middle cerebral artery.
At the time of her stroke, Ms. Quinn, who is African American, was 40 years old and was overweight or obese and at least mildly hyperlipidemic.[7] She had not been found to suffer from hypertension before the stroke, and had no family history of stroke. Plaintiffs' experts, Drs. Coyle and Kulig, have opined that Ms. Quinn's ingestion of Parlodel caused her stroke.
B. Elizabeth Brasher
Plaintiff Elizabeth Brasher delivered a child vaginally on February 15, 1994. On the following day, she underwent a tubal ligation. She chose not to breast-feed her baby, and was given a prescription for Parlodel on or about February 15, 1994. The Parlodel was prescribed to be taken *1294 twice a day for 10 or 14 days. She had previously taken Parlodel in 1991 after a pregnancy, and experienced no known adverse reaction.
On or about February 21, 1994, Ms. Brasher experienced left-sided paralysis and was taken to the emergency room. It was determined that Ms. Brasher had suffered a cerebral infarction in the right temporal region. An arteriogram in March of 1994 showed the appearance of fibromuscular displasia ("FMD") in the left common carotid artery.[8] At the time of her stroke, Ms. Brasher, a Caucasian, was 38 years old and a smoker of one to two packs of cigarettes per day. She also drank about half a pot of coffee per day and her cholesterol levels, measured after pregnancy, were elevated. She was not hypertensive, but had a family history of stroke. Plaintiffs' experts, Drs. Coyle and Kulig, have opined that Ms. Brasher's ingestion of Parlodel caused her stroke. It is this basic causation opinion that is at the center of the Daubert challenge underlying the motions for summary judgment in these two cases.
C. Parties' Positions
The defendant argues in support of the motion that the plaintiffs' experts' opinions on causation are scientifically unreliable under Daubert and, therefore, must be stricken. In the absence of testimony establishing a causal link between the plaintiffs' strokes and the use of Parlodel, plaintiffs' causes of action would fail and defendant would be entitled to summary judgment on all claims. Sandoz contends that, absent a scientifically appropriate epidemiological study showing an increased risk of stroke associated with Parlodel use, plaintiffs' experts' opinions are nothing more than unscientific speculation.
Plaintiffs counter that, although there is no statistically valid epidemiological study dealing with the association between Parlodel and cerebral infarction, there is an abundance of other scientifically reliable evidence from which a well-reasoned opinion that Parlodel can cause vasoconstriction or vasospasm severe enough to cause a stroke can be drawn. Plaintiffs further offer expert opinion evidence that Parlodel, and not other potential causes, triggered the strokes suffered by Ms. Brasher and Ms. Quinn. This evidence includes animal studies, case reports, Adverse Drug Reaction reports (ADRs) to the Food and Drug Administration, and the generally accepted notion in the medical community, reflected in medical textbooks and literature, that Parlodel is a risk factor for stroke because of its vasoconstrictive effects. Plaintiffs also point to a study commissioned by Sandoz, which ultimately proved to be too underpowered to be statistically valid, called the ERI study, which revealed evidence of an 8.44 fold increase in the likelihood of stroke in postpartum women consuming Parlodel compared to postpartum women not using it.
The Law of Daubert
In Daubert the United States Supreme Court rejected the argument that the standard for determining the admissibility of scientific opinion testimony was the "generally accepted in the relevant scientific community" test originated in Frye v. United States, 54 App. D.C. 46, 293 F. 1013 (1923). Rather, the Court held that Federal Rule of Evidence 702, promulgated in 1976, supplanted the Frye standard with a more "flexible" approach. The *1295 Court described the old Frye test as "austere," "rigid," and "uncompromising" and signaled with Daubert (reiterated in Kumho Tire Co. Ltd. v. Carmichael, 526 U.S. 137, 119 S. Ct. 1167, 143 L. Ed. 2d 238 (1999)) a more practical and flexible approach to assessing whether a proposed expert opinion has sufficient evidentiary reliability[9] that the fact-finder should be allowed to consider it. While it is true that the Court listed four "factors" for measuring reliability,[10] it made clear, both in Daubert and Kumho Tire, that these were neither exclusive nor exhaustive and that it remains for the trial court to determine what procedures and tools are necessary for it to analyze the "trustworthiness" of the expert's opinion.
The "gatekeeping" role of the trial court requires flexibility and a practical recognition of what can be known and how it is known. If scientific methodologies can validate certain facts, scientifically reasonable inferences drawn from those facts are admissible.[11]Daubert did not erect insurmountable obstacles to the admissibility of expert opinion evidence; rather, it simply holds that before expert opinion evidence should be allowed, the opinion should be based on "good grounds," that is, "supported by appropriate validationi.e., `good grounds,' based on what is known." Daubert, 509 U.S. at 590, 113 S. Ct. 2786. The point of the gatekeeping role is to separate opinion evidence based on "good grounds" from simple subjective speculation masquerading as scientific knowledge.
The assessment process, that is, the process of examining whether "good grounds" exist, focuses on the methodologies the witness used to reach the opinion he or she will express, not the scientific correctness of the opinion. It is not part of the trial judge's gatekeeping role to determine whether the proffered opinion is scientifically correct or certain in the way one might think of the law of gravity. The gatekeeping role is addressed to mere evidentiary admissibility; it is the role of the fact-finder (usually a jury) to determine whether the opinion is correct or worthy of credence. For the trial court to overreach in the gatekeeping function and determine whether the opinion evidence is correct or worthy of credence is to usurp the jury's right to decide the facts of the case. All the trial judge is asked to decide is whether the proffered evidence is based on "good grounds" tied to the scientific method.[12]
*1296 Application of Daubert in this Case
The court is satisfied that the proffered opinionsthat the strokes suffered by Ms. Quinn and Ms. Brasher were caused by cerebral arterial spasms arising from the vasoconstrictive effects of the Parlodel both women were takingare based on "good grounds" tied to the scientific method and that they possess sufficient evidentiary reliability (that is, trustworthiness) that a jury should be allowed to consider the opinions in the determination of the facts of this case.[13] In the words of Daubert, that opinion is "an idea inferred from such facts" as are scientifically known and established through appropriate scientific methodologies.
At the outset, Drs. Patricia Coyle, Denis Petro, and Kenneth Kulig all express the opinion that Parlodel, as an ergot alkaloid, is capable generally of causing both vasodilation and vasoconstriction, depending upon the initial vascular resistance of the subject exposed to the drug. All three point to several animal studies conducted by defendant which showed vasoconstrictive properties in Parlodel. They also point out that Parlodel is an ergot alkaloid, and that ergots have long been recognized as causing vasoconstriction, sometimes severe enough to cause gangrene and the death of tissue due to blood loss. This phenomenon is known as ergotism. Additionally, there are a limited number of case reportsreports of medical events observed in patients by physicianssuggesting that women taking Parlodel and with no other known risk factors had suffered vasospasmotic events such as hypertension, stroke, and myocardial infarction. Finally, there are several well-recognized medical textbooks and treatises that identify bromocriptine as a risk factor for stroke and myocardial infarction.
Although it is true that none of these bits of evidence establish conclusively that Parlodel can cause vasoconstriction and vasospasm, taken together they present a compelling picture, one which can support a reasonable scientific inference. Daubert does not require proof to a certainty, or even proof convincing to the trial judge. The trial judge is not required to find that the proffered opinion is scientifically correct, but only that it is trustworthy because it is tied to good scientific grounds. What Daubert does require is that the expert's opinion be based on sound methodologies of the type used by experts in the field in which the opinion is offered. There can be little question that scientists routinely use animal studies, case reports, and pharmacological comparisons of similar classes of drugs to infer conclusions, which are expressed in peerreviewed journals and textbooks. Unquestionably, epidemiological studies provide the best proof of the general association of a particular substance with particular effects, but it is not the only scientific basis on which those effects can be predicted. In science, as in life, where there is smoke, fire can be inferred, subject to debate and further testing. The court is persuaded that the opinions offered here are reliably grounded on known scientific fact derived from recognized scientific methodologies.
*1297 Further, the fact that no epidemiological studies exist is due to the extreme difficulty, if not impossibility, of framing such a study. Because stroke is a relatively rare event in women of child-bearing years (although not unknown), it would require an extremely large pool of participants to reach statistically acceptable conclusions. Likewise, controlled studies cannot be performed because such would unethically expose test participants to risk of harm or death. As Dr. Coyle testified, when it appears that a drug causes serious, life-threatening reactions, to conduct a controlled study would "be dangerous and not in the best interests of people." Given the practical unavailability of other forms of scientific evidence, reliance on those that are available is all the more reasonable.
As to the opinion that Parlodel specifically caused the strokes in Ms. Brasher and Ms. Quinn, Drs. Coyle and Kulig have utilized a recognized and widely-used technique relied upon by medical clinicians worldwide to identify and isolate the causes of disease so that they may be treated. The differential diagnosis calls for the physician to list the known possible causes of a disease or condition, usually from most likely to least likely. Then, utilizing diagnostic tests, the physician attempts to eliminate causes from the list until he is left with the most likely cause of the disease or condition. These diagnostic tests may include physical examination, family and personal medical history, testing of blood and other bodily fluids; X-rays, CT scans, MRIs, MRAs, and any of a host of generally accepted techniques for eliminating or "falsifying" a hypothesis that the disease arose from a particular listed cause.
In the cases of Ms. Quinn and Ms. Brasher, these testing techniques included physical examination, family and medical histories, existing medical records, MRIs, MRAs, and CT scans. All of the tests performed on the plaintiffs are well-recognized and scientifically accepted techniques for confirming or eliminating particular causes of strokes. Ultimately, following the protocol of a differential diagnosis, the physicians were able to eliminate every possible cause for the strokes except for vasoconstriction. The court has no difficulty finding that those conclusionsthat the strokes were caused by vasoconstriction or vasospasmto be well-supported and on good grounds.[14]
The next step in the causation opinion is that the spasm was caused by plaintiffs' ingestion of Parlodel. Plaintiffs' experts offer the opinion that, because Parlodel has vasoconstrictive characteristics, it is capable of causing a cerebral artery spasm and, in the absence of any other reasonable explanation, it was the most likely cause of the strokes. Defendant attacks this reasoning on two fronts: first, that the evidence that Parlodel can cause vasoconstriction is unreliable and, second, that the conclusions that there were no other causes for the strokes suffered by Ms. Brasher and Ms. Quinn are unreliable.
Without attempting to recite the extensive and voluminous scientific evidence presented on the first argument, the court is satisfied, as expressed above, that plaintiffs' experts based their opinion that Parlodel can cause vasoconstriction sufficiently severe to cause a stroke on sound scientific evidence and methodologies. Plaintiffs' experts cite as a foundation for their opinions animal studies[15] that have *1298 shown ergot alkaloids similar to Parlodel to have a vasoconstrictive effect; the same studies were relied upon and acknowledged in internal Sandoz documents. Additionally, plaintiffs' experts cite case reports, including dechallenge/rechallenge reports, and ADRs reported to the FDA indicating that Parlodel has vasoconstrictive side effects such as stroke, seizure, and myocardial infarction. As plaintiffs' expert Dr. Petro has pointed out, case reports "are essential in assessing drug safety" in the "real life practice of drug safety monitoring and regulation." While limited in number, these case reports and ADRs were further bolstered by literature reviews that identified Parlodel as a risk factor for stroke in the postpartum period. See, e.g., M. Comabella, Bromocriptine and Postpartum Cerebral Angiography: A Causal Relationship?, 46 Neurology, 1996, (Plaintiffs' Ex. 20); E. Janssens, Postpartum Cerebral Angiopathy Possibly due to Bromocriptine Therapy, 26 Stroke, 1995 (Plaintiffs' Ex. 22). Several medical textbooks state that ergot drugs, such as bromocriptine, can cause stroke. See, e.g., H. Barnett, Stroke Pathophysiology, Diagnosis, and Management, 3d ed., 1998 (Plaintiffs' Ex. 23); J. Bogousslavsky & M. Fisher, Textbook of Neurology (Plaintiffs' Ex. 30); M. Ellenhorn, Ellenhorn's Medical Toxicology: Diagnosis and Treatment of Human Poisoning, 2d ed., 1997.; M. Ford, Clinical Toxicology, 1st ed., 2001.
Although defendant is correct that there is no reliable epidemiological study showing an increased risk of stroke associated with bromocriptine,[16] there is adequate evidence of a scientific nature from which a reliable conclusion can be drawn about the association. While an epidemiological study may be the best or ideal evidence, Daubert requires only that reliable evidence be presented, and that evidence here consists of the animal studies, the medical literature reviews, the ADRs reported to the FDA, and the "general acceptance" of the association between stroke and Parlodel, reflected in several neurology and toxicology textbooks and treatises. These all are recognized and accepted scientific methodologies, used for assessing the possible side-effects and hazards associated with particular drugs and the causes of disease. The fact that strokes suffered by one or both of the two plaintiffs were caused by their ingestion of Parlodel can be reliably inferred from the facts known about the vasoconstrictive effect of bromocriptine and the absence of other likely causes.
*1299 The causation opinions expressed by plaintiffs' experts are not made unreliable or inadmissible simply because it may be debatable whether there were other possible causes for the plaintiffs' strokes. Defendants note that Ms. Quinn was deemed obese by at least one doctor, and that she was African-American, factors which, defendant's experts opine, put her at an increased risk for stroke. Similarly, defendants argue that Ms. Brasher was at an increased risk for stroke because she was a heavy smoker and had a family history of stroke. Defendants cite additional factors, such as hyperlipidemia, which they claim put both plaintiffs at an increased risk of stroke. Plaintiffs' experts dispute whether the conditions cited constitute risk factors and/or whether the conditions caused plaintiffs' strokes. That debate creates a question about the weight to be accorded the plaintiffs' experts' opinions, but it does not effect the admissibility. In reaching the causation conclusion, Drs. Kulig and Coyle utilized the recognized and valid technique of the differential diagnosis, which was explained above.[17] Their opinion that Parlodel was the most likely explanation for the strokes suffered by the plaintiffs is reliably grounded on that methodology. It will be for the jury to determine which of the alternative explanations for the strokes is more likely true than not true.
Conclusion
The court finds, therefore, that the expert testimony of Drs. Petro, Kulig, and Coyle, to the effect that Parlodel can cause vasoconstriction severe enough to cause a stroke and that the strokes suffered by Ms. Brasher and Ms. Quinn were caused or contributed to by their use of Parlodel to be scientifically reliable under Daubert and Federal Rule of Evidence 702. Consequently, plaintiffs have shown there to be a triable issue of medical causation. The defendant's motions for summary judgment on medical causation (Brasher document #64, Quinn document #58) are hereby DENIED.
The Clerk is DIRECTED to forward a copy of this Order to all counsel of record
NOTES
[1] Although styled as Sandoz Pharmaceuticals Corporation, defendant changed its name to Novartis Pharmaceuticals Corporation. Defendant will be referred to herein as "Sandoz."
[2] The motion challenges plaintiffs' ability to prove causation in all three cases assigned to the undersigned magistrate judge, Globetti, Quinn, and Brasher. Plaintiff Melissa Globetti suffered a myocardial infarction, or heart attack, after her ingestion of Parlodel, while plaintiffs Brasher and Quinn both suffered ischemic strokes. A hearing in December 1999 focused on the admissibility of expert testimony only in Globetti, and was the subject of an opinion and order denying the motion for summary judgment dated September 6, 2000.
[3] The experts proffered by the plaintiffs are Dr. Patricia Coyle, Dr. Kenneth Kulig, and Dr. Denis Petro. Drs. Coyle and Kulig will testify that Parlodel can cause vasoconstriction and vasospasm in some people severe enough to obstruct blood flow and, as a consequence of that phenomenon, actually caused the ischemic strokes in Ms. Brasher and Ms. Quinn. Dr. Petro will testify that Parlodel can cause vasospasm, but will not opine as to the specific cause of the strokes suffered by these plaintiffs.
[4] Parlodel is the trade name for the chemical compound bromocriptine mesylate, an ergot alkaloid with an added bromine atom.
[5] Plaintiff now goes by the name Ruby Quinn Hatcher, but will be referred to herein as Ms. Quinn.
[6] Ms. Quinn has testified that she was not given any instructions as to the use of the Parlodel and that she took only one per day. However, she told one doctor that she was "not sure" how she took the Parlodel. Ms. Quinn has testified that she "believes" she took one pill per day in the mornings. In any event, plaintiff has offered evidence from Dr. Coyle that Parlodel can be detected even several days after ingestion, and that the drug "seems to have a very prolonged bioactive half-life and its effects linger[] for a significant length of time." Transcript of 6/5/01 hearing, at p. 45. From this testimony, a jury could infer that the Parlodel caused Ms. Quinn's stroke even if she had stopped taking the drug three or four days before the stroke.
[7] Different doctors have characterized her weight as from simply overweight to obese. Similarly, doctors assess her cholesterol in varying ways, but confirm she is at least mildly hyperlipidemic.
[8] The parties dispute whether Ms. Brasher had FMD, or whether the radiographic picture appeared similar to FMD because of the use of ergot alkaloids. Defendant takes the position that Ms. Brasher had FMD, but defendant's expert Dr. Bucholz admits that FMD and vasospasm can be confused on angiography.
[9] See Daubert, 509 U.S. 579, 590 n. 9, 113 S. Ct. 2786, 125 L. Ed. 2d 469 (1993), where that Court stated "[O]ur reference here is to evidentiary reliabilitythat is, trustworthiness."
[10] These factors were testability, peer review and publication, assessing the known or potential rate of error of the proposition, and whether it has found general acceptance in the scientific community.
[11] "The subject of an expert's testimony must be `scientific ... knowledge.' The adjective `scientific' implies a grounding in the methods and procedures of science. Similarly, the word `knowledge' connotes more than subjective belief or unsupported speculation. The term `applies to any body of known facts or to any body of ideas inferred from such facts or accepted as truths on good grounds.'" Id. at 589-90, 113 S. Ct. 2786 [italics added].
[12] This court notes that inherently, the judge's role in a Daubert determination if fraught with conflict. In most cases, if the court bars the testimony of one party's expert witness or witnesses, that party is unable to present an essential element of his or her claim, or to proffer a defense. Accordingly, judges are aware that applying Daubert heavy-handedly has the effect of lightening one's caseload, as a party stripped of its expert often must dismiss the claims or settle the lawsuit. Judges, including the undersigned, must be vigilant not to let the judge's impulse to dispose of cases infiltrate the Daubert assessment. They must avoid even the subconscious temptation to look favorably on a Daubert challenge simply because it will result in the dismissal of a case without the burden of a trial.
[13] Daubert requires that expert opinion evidence be both reliable and relevant. The relevance prong of the test is concerned with whether the proffered testimony "assists the trier of fact" to determine some issue in dispute. In this case, the parties agree that causation is central to the plaintiffs' ability to recover from the defendant. Each plaintiff must prove by a preponderance of the evidence that she suffered an injury because of the Parlodel manufactured by defendant. Thus, the court concludes that the expert causation opinion is relevant. The dispute is over its reliability.
[14] The court notes that defendant offers expert opinions that indicate the cause of Ms. Brasher's stroke was dissection or clot formation arising from FMD. Plaintiffs' experts assert that Ms. Brasher did not have FMD. The same defendant's expert rules out smoking, obesity, hyperlipidemia, and coffee consumption as causes of Ms. Brasher's stroke.
[15] The experts cite animal studies involving the hind legs of dogs, the ears of dogs, the tails of rat and mice, tongues of sheep, and a "spinal cat," all indicating that bromocriptine or like substances causes vasoconstriction sufficient to lead to necrosis. The hind leg study established bromocriptine as both a vasoconstrictor and vasodilator, depending upon the initial vascular resistance. Low initial resistance causes bromocriptine to cause vasoconstriction, while greater vascular resistance leads to vasodilation. This is important because in the postpartum period, vascular resistance is low due to shifting blood volumes and hemodynamics of the female body readjusting itself to a nonpregnant state. Thus, the conclusion that postpartum women may be more susceptible to bromocriptine causing vasoconstriction is consistent with the animal study.
[16] Both plaintiffs' experts and defendant's experts agree that there have been no epidemiological studies sufficient in size or scope to assess a possible link between stroke and Parlodel, Even so, while critical of the studies, plaintiffs' expert Dr. Coyle asserts that epidemiological studies conducted by Sandozthe "ERI study" and "HCIA study"suggest that Parlodel can cause stroke, or at least constitute a "red flag" that Parlodel may be associated with stroke in postpartum women. She points out that the ERI study suggested an 8.44 fold increase in the likelihood of stroke in postpartum women on Parlodel. Defendant's expert, Dr. Bucholtz, takes the opposite view, opining that the Sandoz studies show that the postpartum period itself, and not Parlodel, increases the risk of stroke.
[17] The defendant argues that the Eighth Circuit Court of Appeals' recent decision affirming the grant of summary judgment in favor of Sandoz in Glastetter v. Novartis Pharmaceuticals Corp., 252 F.3d 986 (8th Cir.2001), compels the conclusion that this court also must find that the testimony of Drs. Kulig and Petro lacks scientific reliability. The court rejects that notion, noting that the Glastetter court relied heavily on Ellenhorn's treatise, which is outdated; the more edition relates bromocriptine to seizure and stroke. Furthermore, the Eighth Circuit opinion stands only for the proposition that the district court's finding does not represent an abuse of discretion. It does not necessarily follow that an inconsistent holding by this court would constitute an abuse of discretion. The court further notes that it agrees with the Glastetter opinion from the district court in finding that epidemiology is not necessary to prove causation.