Joseph Sanders &
D.H. Kaye

Note: This article appears in Jurimetrics Journal, Volume 37, pages 113-128 (1997). 1997 Joseph Sanders & DH Kaye

On December 26, 1996, a federal district court in Oregon made legal history. In Hall v. Baxter Healthcare Corp., (2) Judge Robert E. Jones excluded all testimony to the effect that silicone-gel breast implants (SGBIs) cause autoimmune system disorders. The court reached this result after calling on four independent experts to advise it on the state of scientific knowledge. "This is just an extraordinary, major blow to plaintiffs' lawyers," crowed a senior vice president and general counsel of Bristol-Myers Squibb Co., one of three defendants in the cases.(3) Lawyers around the country representing women with breast implants were less exuberant. "This judge is playing super-scientist or super-juror," complained a New York plaintiffs' lawyer. "That isn't his role."(4)

This article locates the Hall opinion within the context of the ongoing dispute surrounding the scientific evidence linking implants to autoimmune system disorders, describes the process that led to the ruling in Hall, and it examines the analysis of causation given in the court's opinion. It identifies limitations in the expert reports and gaps between the reasoning in the opinion and those reports. The opinion, we argue, is factually dense--it responds to specific features of the expert testimony that plaintiffs produced. This is an appropriate if not essential aspect of a court's treatment of a factual question, but it leaves room for other courts to reach a different result with different expert witnesses. We anticipate that a national panel of experts already appointed in other cases will reach more generally applicable conclusions based on a more complete analysis of the scientific literature, and that those conclusions will shape the course of the implant litigation.


Silicone-gel breast implants are a source of concern to millions of women.(5) They have joined asbestos and the Dalkon Shield in the ranks of the very large mass tort cases.(6) Like these other mass tort cases, the silicone-gel breast implant litigation has led to the withdrawal of the product from the marketplace(7) and the bankruptcy of a manufacturer.(8)

The federal cases are being handled under the Multidistrict Litigation Act.(9) By October 1994, the Judicial Panel on Multidistrict Litigation had transferred approximately 9,600 cases to the Northern District of Alabama for pretrial proceedings.(10) In September 1994, Chief Judge Samuel Pointer certified a limited class for the purposes of a global settlement. The defendants offered to settle for $4.25 billion.(11) As a result of the bankruptcy of Dow Corning and the emergence of more plaintiffs than had been anticipated, however, within a year the settlement had unraveled.(12) The litigation would not go away.


The fact that women with implants experience medical problems does not come close to proving that the implants cause these problems. After all, women without implants experience many of the same problems. Ascertaining whether SGBI is even associated with disease requires careful scientific study. Even if an association is found, establishing that it is a cause of disease necessitates still more information. Finally, demonstrating that it is the cause of a particular woman's disease demands additional medical and scientific expertise. Implant manufacturers vehemently deny that their products cause most of the conditions that plaintiffs attribute to them.

In 1996, several courts sought independent experts to advise them about the questions of causation. Following the lead of a group of judges in New York,(13) a state and a federal court in Oregon enlisted the help of four such experts. The process began with a medical doctor and biochemist who "assisted the court by screening dozens of potential appointees and ultimately selecting four totally unbiased and uncommitted experts . . . ."(14) The district court then appointed those selected--an epidemiologist, a rheumatologist, an immunologist-toxicologist, and a polymer chemist--"to assist in evaluating the reliability and relevance of the scientific evidence."(15) These specialists served as "technical advisors" rather than as experts appointed under Rule 706 of the Federal Rules of Evidence, which provides for court-appointed experts.(16) Along with counsel and the court, the advisors questioned the parties' witnesses.(17) After viewing videotaped summations from counsel, the advisors submitted written reports. Counsel then had the opportunity to question them.(18) After considering the advisory experts' reports and the other submissions, the court granted defendants' motions "to exclude expert testimony concerning causation of any systemic disease or syndrome."(19)

Later in 1996, Judge Pointer appointed in the multidistrict litigation MDL-926 a Rule 706 panel consisting of four experts in the areas of immunology, epidemiology, medicine, and toxicology. Judge Pointer charged the panel to address the extent to which "existing studies, research, and reported observations provide a reliable and reasonable scientific basis for one to conclude that silicone-gel breast implants cause or exacerbate any of the conditions described . . . ?"(20) The order asked the panel to state the extent to which the opinions it reaches are subject to sufficient dispute that other qualified experts could express contrary opinions that would be viewed by others as "representing legitimate and responsible disagreement within your profession."(21)

Unlike the Oregon advisors, the national panel of experts has yet to act. Nonetheless, it has affected SGBI litigation. The district courts in New York cited the prospect of a report from the national panel as a reason not to "rush to judgment,"(22) and the Oregon district court deferred the effective date of its ruling in Hall "until the findings of the Rule 706 panel are available."(23)


Plaintiffs have claimed a variety of injuries due to implants. Some cases involve ruptured implants, and affected plaintiffs often seek compensation for pain, inflammation, and other local effects, as well as the expense of having the implant replaced or removed.(24) These claims raise serious legal questions,(25) but no especially difficult causation questions.

The more difficult scientific evidence issues arise when a plaintiff claims that an implant has caused a systemic disease. Most cases have involved claims that the implant caused an injury to the individual's immune system.(26) Thus, most scientific admissibility questions in the breast implant litigation relate to the alleged causal connection between implants and autoimmune diseases. Even today, there are only a handful of rigorous epidemiological studies of a possible association, and the base of pertinent medical knowledge was even more circumscribed when the cases of the 1980s were tried and many of the pending SGBI cases filed.

Hopkins v. Dow Corning Corp.(27) is the first published opinion specifically discussing the admissibility of plaintiff expert testimony under the interpretation of the federal rules of evidence adopted by the Supreme Court in Daubert v. Merrell Dow Pharmaceuticals, Inc.(28) The plaintiff was diagnosed with mixed connective-tissue disease three years after she received implants. Several years later, she underwent surgery due to complications from her implants, which had ruptured. She sued on fraud, breach of express and implied warranties, and strict products liability theories. The jury returned a verdict of $840,000 compensatory and $6.5 million punitive damages.(29)

On appeal, Dow Corning argued that the district court erred in admitting testimony that was not based on scientifically accepted principles,(30) and the company sought to overturn the verdict on the ground that without this testimony, there was insufficient evidence of causation.(31) The Ninth Circuit simply wrote that "[t]he record reflects that Hopkins' experts based their opinions on the types of scientific data and utilized the types of scientific techniques relied upon by medical experts in making determinations regarding toxic causation where there is no solid body of epidemiological data to review."(32) Plainly, Hopkins rejects the view, often advanced by defendants, that medical science requires a solid body of human epidemiologic studies to reach a valid conclusion of causation.(33)

Hall v. Baxter Healthcare Corp.(34) challenges the reasoning of Hopkins. Unlike the trial court in Hopkins, Judge Jones appointed four expert advisors and conducted an "intense" Rule 104(a) hearing.(35) The Hall court asked the advisors, who viewed and questioned the parties' experts at the hearing, to address various questions about the scientific literature and the methodology and conclusions of the parties' experts.(36) After receiving the expert advisors' reports(37) and the parties' responses, the judge granted defendants' motions to exclude plaintiffs' expert testimony concerning any systemic autoimmune disorders.(38)

The court assessed the reliability of the plaintiffs' experts' testimony with an eye toward the four factors in Daubert--falsifiability, error rate, peer review and publication, and general acceptance. In addition, it considered a factor added by the Ninth Circuit on remand in Daubert--whether the experts are proposing to testify about matters growing naturally out of research, or whether they have developed their opinions explicitly for the purpose of testifying.(39) Although these criteria are general, courts in science-intensive cases must respond to the particular positions and statements of the experts before them. Much of the Hall court's opinion is in this mold, and that may limit easy extrapolation to other cases with other experts.

Thus, the Hall court made short shrift of proposed testimony from rheumatologists that "silicone exposure is the probable cause of plaintiffs' atypical constellation of symptoms"(40) partly because plaintiffs' epidemiologist, David Goldsmith, had spoken of "trying to frame a hypothesis to test"(41) to establish a link between SGBI and atypical connective-tissue disease [ACTD]. Seizing on this tentative language, defense counsel had argued that plaintiffs' expert had "merely a hypothesis--not proven--not science." The court agreed. "By definition," the court wrote, "ACTD is not one of the classical autoimmune diseases," but is "at best an untested hypothesis."(42)

However, the opinion makes no reference to the portion of the epidemiologist-advisor's report that had cautioned the court against making too much of the terminology.(43) According to the epidemiologist-advisor, defendant's characterization reflected "a serious misunderstanding" because plaintiffs' epidemiologist's reference to a "hypothesis" meant that "he has begun to take the possibility of an association seriously."(44) Even so, there is an obvious lacuna between taking a possibility seriously and testing it thoroughly.(45) Indeed, some have argued that ACTD is not merely untested, but that the vagueness of the description of the putative disease makes it impossible to gather data that would permit a meaningful epidemiological study.(46) In the language of Daubert, the theory cannot be falsified.

Moving from the relatively speculative ACTD to the classic autoimmune diseases, the court concluded that the link to SGBI was dubious because of a perceived contradiction in the testimony from plaintiffs' experts. According to the court, plaintiffs had Dr. Goldsmith testify at the pretrial hearing that epidemiological and other scientific data showed a slightly elevated risk, while plaintiffs also offered transcripts of testimony in other cases from another epidemiologist, Shanna Swan, that no valid epidemiological studies regarding the relationship of silicone breast implants and disease had been completed as of August 1996.(47) The court excluded Dr. Goldsmith's proffered testimony because, it said, he had conceded under questioning from the epidemiologist-advisor that "he was not willing to testify . . . that silicone more likely than not could cause disease in women."(48)

Dr. Goldsmith's concession, however, was made in response to a question about ACTD, and not about more well-defined connective-tissue diseases.(49) The testimony quoted in the opinion simply does not address the possibility of causal relationships between SGBI and all diseases. As to this question, Dr. Goldsmith testified that "there is a statistically significant increased risk of autoimmune disease among women with implants."(50)

The court refused to admit Dr. Swan's critique and "reanalysis" of existing epidemiological studies because it found that testimony was "unreliable and no longer 'fits' plaintiff's theory of the case."(51) Hall cites certain Bendectin cases--including Daubert--in which Dr. Swan's reanalysis of a particular study of that drug had been found inadmissible or insufficient, and draws the sweeping conclusion that "her 'reanalysis' approach" [is] unreliable."(52) But the statistical analysis and reanalysis of data is a legitimate and desirable scientific enterprise. "Reanalysis" merely consists of analyzing previously obtained data differently than the original researchers did. Like the original analysis, a reanalysis can be good or bad. Hall does not explain why "reanalysis" itself is invalid. In fact, it is not obvious that Dr. Swann performed any reanalysis in Hall. She simply criticized defendants' claim that a solid body of epidemiologic research demonstrates that SGBI is not a risk factor for autoimmune diseases, maintaining instead that existing research is inconclusive.(53)

Nevertheless, the court offered more substantial criticisms--that her reading of the literature "has never been subjected to peer review" and that "her theory has never been espoused by any other scientist whose work has been subjected to the peer review process."(54) In short, the court deemed Dr. Swan's critique of the epidemiological literature as inadmissible because of a lack of general acceptance.(55) But even if Dr. Swan's position is aberrational, it easily could be argued that her critique rests on well-known concepts such as statistical power. The problem of distinguishing methodology, which must satisfy Daubert, and conclusions, which are outside its scope and need not be generally accepted, looms large here.

The district court also rejected the testimony of the plaintiffs' immunology and toxicology experts. The decision rests heavily on the conclusion of the court's immunology-toxicology advisor, who surveyed and assessed the scientific literature on "the adjuvant potential of silicone gel implants, the potential for immune stimulation of T cells by silicone gel implants, altered natural killer cell activity, and immune system cancer formation in rodents."(56) The advisor's report deems most of the published studies to be methodologically sound, but questions whether they can lead to the conclusions that plaintiffs' experts reach.(57) The court concludes that:

[A]n evaluation of whether scientific methodology is valid for Daubert purposes should include an examination of how the proffered conclusions relate to the bases upon which the expert relies. The court's neutral technical advisor has advised that [plaintiffs' immunology and toxicology experts have] made too great a leap from the underlying data to his conclusions. In other words, those conclusions are themselves not the result of the faithful application of valid scientific methodology.(58)

To establish biological plausibility, plaintiffs sought to use chemistry experts who would testify that silicone degrades in the body to silica, a more damaging substance.(59) The court's chemistry advisor agreed that one article on the biological reactions of silica "seems scientifically sound," but concluded that although it "seems possible" that silicone could be converted to silica, no papers "clearly demonstrated" this conversion.(60) The court took this to mean that "the opinions plaintiffs' experts proffer regarding the in vivo degradation of silicone to silica are currently unsupported by the scientific literature"(61) and that there is "no scientifically valid evidence to support the conclusion that silicone gel degrades to silica in the human body."(62)

The court also condemned the methodology of a treating physician who was prepared to testify both that silicone gel implants can cause disease in women and that implants caused injury to his patient, plaintiff LeaAnn Hall. According to the court, the physician relied solely on his differential diagnosis to arrive at conclusions as to both general and specific causation. The court concluded that "a single differential diagnosis is a scientifically invalid methodology" for proving general causation(63) because "differential diagnosis assumes that general causation has been proven for the list of possible causes it eliminates."(64) As a corollary to this proposition, it follows that the differential diagnosis could not tend to prove specific causation because something that is not demonstrated to be a general cause cannot be a specific cause. The opinion makes this point elliptically, with the observation that "his conclusion is inconsistent with the epidemiology for classical diseases."(65) In addition, the court complained that the physician had "not testified as to how he eliminated other potential causes . . . ."(66)

Judge Jones tried to distinguish Hall from Hopkins on the ground that the district court in that case admitted the testimony under Frye, depriving the court of appeals of "a record . . . shaped by Daubert's elucidation of the court's gatekeeping function."(67) "Moreover," Judge Jones stated, "the Ninth Circuit [did not] delve into the methodology underlying the scientific data . . . ."(68) These general observations notwithstanding, it is quite difficult to reconcile Hall and Hopkins, and if the SGBI cases reach the Ninth Circuit, we can expect that court to address the causation issue more carefully than it did in Hopkins.


In discussing Rule 702's requirement that scientific evidence must "assist the trier of fact to understand the evidence or to determine a fact in issue," Daubert reasoned that:

This condition goes primarily to relevance . . . . The consideration has been aptly described by Judge Becker as one of 'fit.' 'Fit' is not always obvious, and scientific validity for one purpose is not necessarily scientific validity for other, unrelated purposes.(69)

The "fit" requirement always involves a determination of whether the expert's chain of reasoning contains an inferential gap that is too wide. Fit may be understood to refer to the relationship between the general question asked by a line of research and the question in the case.(70)

The trial court in a Texas case, Luevano v. Baxter Healthcare Corp.,(71) apparently employed a "fit" analysis when it took the remarkable step of granting the plaintiff's motion to exclude defense epidemiological evidence --the Mayo Clinic Study(72) and the Harvard Nurses' Study.(73) Because the number of substantial, peer-reviewed epidemiologic studies of SGBI is small, a brief description of these studies may be in order. The first study is a retrospective, case-control study of all women given breast implants in a Minnesota county from 1964 through 1991.(74) The researchers compared the women receiving SGBIs to twice as many women of about the same age who did not receive implants. Connective-tissue disease was rare in both groups. After an average of eight years of follow-up, five of the 749 implant recipients (0.067%) and ten of the 1,498 non-recipients (0.067%) had contracted these conditions. Since there is no difference in the proportions, the chance of a connective-tissue disorder seems not to depend on the implants at all.(75) It is obvious why the defendant sought to bring this study to the jury's attention.

The second study drew on a prospective, cohort study, known as the Nurses Health Study, which tracks the health status of a large cohort of women health professionals, many of whom are or were nurses. Of the more than 100,000 women left in this group in 1992, more than 80% replied to questionnaires asking about breast-implant surgery, and researchers confirmed that nearly 1,200 had undergone such surgery before 1991. Thus, the incidence of connective-tissue disease in the groups with and without implants could be compared. One part of this analysis is shown in Table 1.

TABLE 1. Incidence of Connective-Tissue Disease
in Women with Breast Implants and
Women Without Implants (Nurses Study)(76)

Case Type No implant Implant Relative risk

(86,318) (1,183) (age-adjusted)
Self-reported disease 5054 (5.9%) 32 (2.7%) 0.7
Definite disease 513 (.59%) 3 (.25%) 0.6
Self-reported symptoms 1277 (1.5%) 17 (1.4%) 1.5
Documented symptoms 898 (1.0%) 6 (.50%) 0.7

The women with implants had a lower incidence of connective-tissue disease. To the extent that the two groups are comparable in other respects, this part of the retrospective study might be thought to suggest that implants are not harmful, but beneficial.(77) Again, it is perfectly clear why defendant would want to introduce this study.

Plaintiffs argued, however, that their diseases were "atypical" and were not specifically included in the studies' protocols.(78) The Texas district court agreed that these studies did not speak to the specific question presented in the case and excluded the evidence. Even without being informed of the negative results of the two studies, however, the jury returned a verdict for the defense.

In Hall v. Baxter Healthcare Corp., Judge Jones rejected some of the proposed testimony of plaintiffs' experts for a lack of fit. For example, he excluded the testimony of a chemist on how silicone behaves chemically inside the human body because that behavior "cannot, in itself, establish that silicone gel breast implants cause signs and symptoms of disease in women in the absence of any epidemiological, rheumatological, or immunological /toxicological evidence linking those breast implants to disease. Thus, such testimony no longer logically advances a material aspect of the proponent party's case."(79) We anticipate more "fit" rulings in the future as defendants assert this ground for excluding some in vivo and in vitro research as irrelevant to the issues presented in the implant litigation.


In breast implant cases, as in most civil matters, the plaintiff must prove each element of the cause of action by a preponderance of the evidence. With respect to the causal question, the plaintiff must show not only that implants are capable of causing harm in some people (general causation) but also that the plaintiff's individual harm was caused by implants (specific causation). Proof of general causation cannot resolve the specific causation issue.(80)

Hall excluded a treating physician's proposed testimony that implants specifically caused a plaintiff's "systemic sclerosis sine scleroderma, manifested by her pulmonary fibrosis."(81) The court held the proffered testimony inadmissible because the physician did not explain how he eliminated other possible causes of the disease, because the conclusion is inconsistent with the epidemiology for classical diseases like scleroderma, and because testimony of specific causation is inadmissible without proof of general causation.(82)

Where courts are less willing to resolve the general causation issue against plaintiffs or the determination of the etiology is explained more convincingly, the role of a so-called differential diagnosis can be more controversial.(83) It would seem that if the scientific evidence shows a relationship between an exposure and a disease--but only a weak relationship--then well-founded testimony from a medical expert as to specific causation is necessary. Indeed, if the substance is only a weak risk factor or if the condition often arises from unknown causes that cannot be ruled out, then even a clinician's differential diagnosis may not go very far toward establishing the actual cause in the case at bar. That, in essence, was the situation with SGBI and generally recognized systemic diseases, as described by the court in Hall. In such cases, courts may not allow a plaintiff to prevail even if an expert is prepared to testify that, based on a differential diagnosis, the cause of plaintiff's injury was defendant's product. In fact, some courts seem to have adopted a numerical rule--they believe that specific causation cannot be found if the relative risk is less than 2.0.(84)

As regards SGBI, the major published, peer-reviewed epidemiological studies have reported relative risks for generally recognized autoimmune diseases below 2.0. In fact, the Minnesota study and the Nurses Health Study were unable to detect relative risks exceeding 1.0.(85) A more recent study--the most extensive to date--examined the incidence of the symptoms and diseases allegedly caused by silicone implants in an even larger cohort of women.(86) The researchers received completed questionnaires from over 400,000 female health professionals, known as the Women's Health Cohort, contacted for a study of heart disease and cancer.(87) The questionnaires provided medical information on many topics, including breast implants and six connective-tissue diseases. In this large sample, women with implants appeared to be at slightly greater risk for connective-tissue disease; overall, the relative risk was 1.24.(88) For a few specific connective-tissue diseases, the relative risks were in the vicinity of 2.0.(89)


Appellate opinions reviewing trial court admissibility determinations under Daubert are rare in the breast implant litigation. As the conflicting rulings in Hopkins and Hall suggest, courts are just beginning to scrutinize the quality of the scientific evidence supporting assertions that implants cause autoimmune ailments. Were the silicone implant litigation to develop as past mass torts have, over the next few years we would expect the courts to wrestle with the difficult scientific evidence admissibility issues that emerge when the causal relationship between an exposure and an injury is not immediately open-and-shut.

However, the implant litigation appears to be headed down a different path. With the appointment of expert panels, particularly the MDL-926 panel, a growing cadre of courts have overcome the historical reluctance to appoint Rule 706 panels.(90) The reports of court-appointed experts will have a substantial impact on future admissibility rulings in the implant litigation. Indeed, the influence will reach beyond questions of admissibility. Dow Corning's final bankruptcy reorganization plan may well be shaped by the panel's conclusion. A report of the MDL-926 panel attributing autoimmune diseases to implants would likely spur settlement offers substantially in excess of the $4.25 billion settlement that collapsed in 1995, while a panel report finding little or no credible evidence of an association would greatly reduce the value of many claims and lead to much smaller settlements.


1. Joseph Sanders is A.A. White Professor, University of Houston Law Center (e-mail: David Kaye is Regents' Professor, Arizona State University, College of Law (e-mail: This article is condensed and adapted from Chapter 30 of Modern Scientific Evidence (David Faigman, D.H. Kaye, Michael Saks & Joseph Sanders eds., forthcoming 1997). The authors are grateful to Judge Robert E. Jones and his staff for making available transcripts and submissions of the parties in Hall v. Baxter Healthcare Corp., 947 F. Supp. 1387 (D. Or. 1996). [BACK]

2. . Civ. No. 92-182-JO (lead), 1996 WL 730693 (D. Or. Dec. 18, 1996) (to be reported at 947 F. Supp. 1387). [BACK]

3. . Richard B. Schmitt, Women in Silicone Implants Case Are Dealt a Major Blow by Judge, Wall St. J., Dec. 19, 1996. [BACK]

4. . Id. [BACK]

5. . It is estimated that between one and two million women have received implants to enlarge or reconstruct their breasts. See, e.g., Charlotte Allen, Jurisprudence of Breasts, 5 Stan. L. & Pol'y Rev. 83, 84 (1994). But see Ralph R. Cook et al., The Prevalence of Women With Breast Implants in the United States--1989, 48 J. Clinical Epidemiology 519 (1995) (reporting that a random survey of 40,000 households in the United States shows that in 1989 there were approximately 815,000 women with implants). A substantial fraction have experienced subsequent medical problems of one kind or another. It is well known that implants can leak, rupture, or cause painful contractures. Compare Eugene P. Goldberg, Evaluating the Health Risks of Breast Implants, 335 New Eng. J. Med. 1154 (1996) (letter), with Marcia Angell, 335 New Eng. J. Med. 1156 (1996) (reply). The association with other conditions is much less clear. The first reported breast implant case, Klein v. Dow Corning Corp., 661 F.2d 998 (2d Cir. 1981), concerned a ruptured implant. [BACK]

6. . There are between 440,000 and 500,000 claimants. In re Dow Corning Corp. (Lindsey), 86 F.3d 482, 485 (6th Cir. 1996); Francis E. Mcgovern, An Analysis of Mass Torts for Judges, 73 Tex. L. Rev. 1821 (1995). [BACK]

7. . In April 1992, the Food and Drug Administration limited the availability of silicone-gel breast implants to women desiring them for reconstruction after cancer surgery. The FDA did not find that the implants were unsafe; rather, it proceeded on the ground that manufacturers had not collected required safety data. David A. Kessler, The Basis of the FDA's Decision on Breast Implants, 326 New Eng. J. Med. 1713 (1992). This action was criticized as overly paternalistic and unnecessarily alarming. Marcia Angell, Breast Implants--Protection or Paternalism?, 326 New Eng. J. Med. 1695 (1992). [BACK]

8. . Dow Corning filed for relief under Chapter 11 of the bankruptcy code on May 15, 1995. BNA Prod. Liab. Daily, May 16, 1995. [BACK]

9. . 28 U.S.C. 1407 (1996). [BACK]

10. . In re Silicone Gel Breast Implants Prod. Liab. Litig. (No. 926), 793 F. Supp. 1098 (J.P.M.L. 1992). Meanwhile, over 15,000 state court claims had been filed. William W. Schwarzer et al., Judicial Federalism: A Proposal to Amend the Multidistrict Litigation Statute to Permit Discovery Coordination of Large-Scale Litigation Pending in State and Federal Courts, 73 Tex. L. Rev. 1529, 1530 (1995). [BACK]

11. . Dow Corning was to contribute somewhat over $2 billion, with the eight remaining parties contributing the remainder. In re Silicone Gel Breast Implant Prod. Liab. Litig. (Lindsey), No. CV 92-P-10000-S, 1994 WL 578353, at *24.(N.D. Ala. Sept. 1, 1994). [BACK]

12. . BNA Prod. Liab. Daily, Sept. 18, 1995. In early 1996, several defendants, not including the now-bankrupt Dow Corning, made a revised settlement offer. Id., Feb. 8, 1996. [BACK]

13. . In April, under the leadership of U.S. District Judge Jack Weinstein, three judges in New York designated several experts from various disciplines as special masters and asked them to locate experts for appointment to a larger expert panel that would consider general principles for establishing cause and effect in the SGBI cases, giving "particular attention . . . to claims respecting immune system dysfunction and connective-tissue and rheumatic disease." The three experts were a law professor, a lawyer-physicist, and a theoretical biologist. Eliot Marshall, New York Courts Seek "Neutral" Experts, 272 Science 189 (1996); Mark Hansen, Panel to Examine Implant Evidence, A.B.A.J., June 1996, at 34. Before fully evaluating the testimony and argument from a Daubert hearing held before the three judges and a magistrate judge, the two federal courts involved denied, at least temporarily, defendants' motions for summary judgment. In re Breast Implant Cases, No. 92 CV 7821 (lead) (S.D.N.Y. & E.D.N.Y. Oct. 23, 1996) (Amended Preliminary Memorandum). The joint memorandum of Judge Weinstein and Judge Harold Baer, Jr., does not specify what role, if any, the New York panel of experts played in this hearing or memorandum. The reasoning of these courts is discussed below. [BACK]

14. . Hall v. Baxter Healthcare Corp., Civ. No. 92-182-JO (lead), 1996 WL 730693, at *2 (D. Or. Dec. 18, 1996). Dr. Richard T. Jones, whom the court appointed as a special master, is Professor Emeritus, Department of Biochemistry and Molecular Biology, Oregon Health Sciences University. [BACK]

15. . Id.. [BACK]

16. . Id. at *76 n.8. Indeed, the court denied motions to appoint experts under Rule 706, believing that this would insulate its advisors from "depositions and testifying at trial." Id. The court reasoned that it had "inherent authority" to appoint such advisors. Id. at *2. [BACK]

17. . The chemist was appointed after the evidentiary hearing. Id. at *2. The testimony was organized by field; that is, plaintiffs presented their experts in a particular field, and defendants' witnesses in that field then testified. Id. The parties also provided statements from experts via telephone and affidavits. [BACK]

18. . Id. [BACK]

19. . Id. at *3. The New York district courts also relied on transcripts and documents from the Oregon Daubert hearing. In re Breast Implant Cases, No. 92 CV 7821 (lead) (S.D.N.Y. & E.D.N.Y. Oct. 23, 1996) (Amended Preliminary Memorandum). [BACK]

20. . In re Silicone Gel Breast Implant Prod. Liab. Litig. (MDL-926), No. CV 92-P-1000-S (N.D. Ala. Oct. 31, 1996) (Order No. 31E). The panel was asked "to consider the relationship, if any, between implants and the following: "classic" connective-tissue diseases, such as systemic lupus erythematosus, Sjogren's syndrome, et cetera; "atypical" presentations of connective-tissue diseases or symptoms; immune system dysfunctions." Id. In an appendix to his order, the judge listed forty diseases, symptoms, conditions, and complaints that some have asserted as possibly associated with silicone implants and asked the panel to comment, where appropriate, on the scientific basis of a claimed linkage between implants and the items on the list. Id. The panel was instructed not to consider purely local complications arising from implants, such as breast disfigurement or capsular contracture. [BACK]

21. Id.. [BACK]

22. In re Breast Implant Cases, No. 92 CV 7821 (lead) (S.D.N.Y. & E.D.N.Y. Oct. 23, 1996) (Amended Preliminary Memorandum). [BACK]

23. . Hall, at *3. [BACK]

24. . Plaintiffs typically allege a breach of express or implied warranty. They also allege that their implants are defective in design or manufacture and that they were inadequately warned about the potential that the implant might rupture or otherwise be in need of repair. [BACK]

25. . See Desmarais v. Dow Corning Corp., 712 F. Supp. 13 (D. Conn. 1989) (learned-intermediary doctrine may present a barrier to any argument that the manufacturer had an obligation to warn the consumer directly); Artiglio v. Superior Court (McGhan Medical Corp.), 22 Cal. App.4th 1388, 27 Cal. Rptr.2d 589 (1994) (holding that "medical devices" such as breast implants are "unavoidably unsafe" under the Restatement of Torts (Second) 402A, cmt. k); Robert S. Adler & Richard A. Mann, Preemption and Medical Devices: The Courts Run Amok, 59 Mo. L. Rev. 895 (1994) (arguing that federal law regulating medical devices might preempt state law); Medtronic, Inc. v. Lohr, 116 S.Ct. 2240 (1996) (suggesting no preemption). [BACK]

26. . A few have claimed the implants caused or accelerated a cancer. See, e.g., Livshits v. Natural Y Surgical Specialties, Inc., 35 Fed. R. Evid. Serv. 433 (S.D.N.Y. 1991). [BACK]

27. . 33 F.3d 1116 (9th Cir. 1994). [BACK]

28. . 509 U.S. 579 (1993). [BACK]

29. . Hopkins, at 1119-20. [BACK]

30. . Id. at 1124. Dow also contended that the district court erred in not holding a hearing on the qualifications of Hopkins' experts to testify as to causation. [BACK]

31. . Id. at 1123-26. [BACK]

32. . Id. at 1124. [BACK]

33. . For discussions of Hopkins, see Marcia Angell, Science on Trial: The Clash of Medical Evidence and the Law (1996); Thomas Koenig & Michael Rustad, His and Her Tort Reform: Gender Injustice in Disguise, 70 Wash. L. Rev. 1, 45 (1995); Heidi Li Feldman, Science and Uncertainty in Mass Exposure Litigation, 74 Tex. L. Rev. 1,20 (1995); Deborah Maliver, Out of the Fryeing Pan and into Daubert: Trial Judges at the Gate Will Not Spell Relief for Plaintiffs, 56 U. Pitt. L. Rev. 245 (1994); Donald Lawson, Silicone and Science, 37 Jurimetrics J. 53 (1996) (concluding that "the Hopkins decision cannot be justified"). [BACK]

34. . 947 F. Supp. 1387 (D. Or. 1996). The fifteen cases involved in this litigation had been remanded to the Oregon court for trial following pretrial discovery in MDL-926. [BACK]

35. . Judge Jones acted in concert with Multnomah County Circuit Judge Nely Johnson, who was handling similar complaints filed in state court. [BACK]

36. . The court propounded five questions along with a set of questions posed by the parties: (1) "Is the expert's opinion supported by scientific reasoning and methodology that is generally accepted in the expert's particular scientific community or otherwise qualified"? (2) "Is the expert's opinion based upon scientifically reliable data?" (3) "If epidemiological studies have not been done or are inconclusive, what other data, such as animal studies, biophysical data, clinical experience in the field, medical records, differential diagnosis, preliminary studies, general scientific knowledge, and medical literature can justify, to a reasonable medical probability, a conclusion concerning the cause of the syndrome or disease at issue?" (4) "Do the methodology and data support the expert's conclusions?" (5) "Does the scientific data relied upon by the expert apply to the syndrome or disease in issue in these cases? For instance, are epidemiological studies directed at other typical or classical diseases relevant to an atypical disease?" Hall, at *2-*3. All but one of the plaintiffs was diagnosed with atypical autoimmune disorder. Id. at *11, *76. [BACK]

37. . The reports are reproduced as an appendix to the opinion. They suggest that the experts knew little about silicone and its effects in the human body prior to their involvement in the litigation. E.g., id. at *48 (Report of Ronald W. McClard) ("Having been supplied with more documents and videotapes than a human could possibly evaluate critically in the short time given me, I have endeavored to get at the heart of the matter . . . ."). The detail and depth of the assessments of the relevant scientific literature in the reports vary. Some are quite detailed, but aside from a reference to the sample size of one study, the epidemiologist's report lacks any specific analysis of the strengths and weakness of pertinent epidemiological studies. It speaks of complications due to "a clash between the culture of science and culture of the legal system" (id. at *24) and draws an analogy between the usual requirement of statistical significance at the 0.05 level and proof beyond a reasonable doubt. For reasons explained in, e.g., David H. Kaye & David A. Freedman, Reference Guide on Statistics, in Reference Manual on Scientific Evidence 331 (Federal Judicial Center ed. 1994), that analogy fails in its details, and it is arguable that, especially as applied to a collection of studies, the requirement of statistical significance at the 0.05 level is not particularly demanding. See, e.g., Paul Meier et al., What Happened in Hazelwood: Statistics, Employment Discrimination, and the 80% Rule, 1984 Am. B. Found. Res. J. 139, 152 ("If a difference does not attain the 5% level of significance, it does not deserve to be given weight as evidence of a disparity. It is a 'feather.'"). [BACK]

38. . "[T]he court will sever plaintiffs' local injury claims from their claims for ACTD [atypical connective-tissue disease] or any systemic illness or injury. [In the] cases that do proceed to trial . . ., I will exclude as irrelevant any testimony or evidence of the following: ACTD; any systemic illness or syndrome or autoimmune disorder of any kind; any emotional distress claims arising out of any alleged fear of developing any systemic disease or injury or fear of cancer." Hall, at *22 (notes omitted). However, because Judge Pointer appointed a national panel of experts for the MDL-926 cases during these hearings, Judge Jones deferred the effective date of his opinion until the findings of that panel are available. Id. at *3. Judge Jones held out little hope of a change that his opinion would change, however. He stated: "I wish to make it abundantly clear that while I will evaluate the Rule 706 panel reports before finalizing my decision, I am unlikely to amend these findings and conclusions absent substantial and compelling developments in the scientific arena." Id. [BACK]

39. . 43 F.3d 1311, 1315 (9th Cir. 1995). Before assessing the grounds for exclusion of the plaintiffs' experts' testimony, the court addressed the Supreme Court's comment in Daubert that the focus of admissibility decisions should be on the expert's methodology, not conclusions. This issue has perplexed a growing number of courts forced to explore the interstices of the Supreme Court's pronouncements about science and law in Daubert. See In re Paoli R.R. Yard PCB Litig., 35 F.3d 717, 746 (3d Cir.1994); Lust v. Merrell Dow Pharmaceuticals, Inc., 89 F.3d 594, 596 (9th Cir. 1996); Claar v. Burlington Northern R.R. Co., 29 F.3d 499, 502 (9th Cir. 1994). Judge Jones concluded that "there appears to be no clear demarcation between scientific methodology and the conclusions it generates." Hall, at *10. The court quoted with approval Judge Becker's observation in the Paoli Railroad Yard PCB Litigation that:

Daubert's requirement that the expert testify to scientific knowledge--conclusions supported by good grounds for each step in the analysis--means that any step that renders the analysis unreliable under the Daubert factors renders the expert's testimony inadmissible. This is true whether the step completely changes a reliable methodology or merely misapplies that methodology.

Paoli, 35 F.3d at 745. [BACK]

40. . Hall, at *11. [BACK]

41. . Id. at *13. [BACK]

42. . Id. [BACK]

43. . Id. at * 26 (Report of Merwyn R. Greenlick). [BACK]

44. . Id. [BACK]

45. . When questioned by defense counsel on his report, Dr. Greenlick explained that "saying that a hypothesis generation is not science simply doesn't make sense." Transcript of 104 Hearing, In re Breast Implant Litig. (Hall), D. Ore. Sept. 13, 1996, at 21-22. Counsel conceded that "it is part of science," and Dr. Greenlick agreed that "you can't address a hypothesis as a conclusion." Id. at 22. [BACK]

46. . E.g., Matthew H. Liang et al., Letter to the Editor, 333 New Eng. J. Med. 1424 (1995). ("[t]he features of a unique connective-tissue syndrome have not been put into a coherent, valid, or reproducible case definition, which severely limits scientific study."). [BACK]

47. . Hall, at *14. It is not clear that Dr. Swan questioned the validity of the major studies of known autoimmune diseases. She contended that these studies were not designed to detect diseases with large latency periods and that they lacked the statistical power to be convincing evidence of the absence of associations with the classical autoimmune diseases. Merlin v. MMM Co., No. CV-N-95-696-HDM (D. Nev. Nov. 20, 1995) (Transcript of Daubert Hearing Pursuant to Motions in Limine). [BACK]

48. . Hall, at *14.[BACK]

49. . The opinion proceeds to analyze Dr. Goldsmith's willingness to testify that "such causation" (involving, one would think, only ACTD) is "more likely than not" as a result of an abstract of an unpublished study that he inquired about by telephone. That abstract refers to "UCTD," and the opinion does not define what that might be, but it offers various criticisms of the study. It presents that study as having little "epidemiological significance" and as being out of line with all previous research involving classical connective-tissue diseases. Id. Thus, the court concludes that Dr. Goldsmith's opinion, taking the latest study into account, is inadmissible because "the methodology supporting this changed testimony [is] unreliable under Daubert I and Daubert II." Id. [BACK]

50. . Transcript of Pretrial Hearings, In re Breast Implant Litig. (Hall), D. Ore. Aug. 6, 1996, at 162. [BACK]

51. . Hall, at *14. [BACK]

52. . Id. [BACK]

53. . Transcript, supra note 49, at 18. [BACK]

54. . Hall, at *14. [BACK]

55. . Judge Jones concluded that "[i]n this litigation, Dr. Swan's well-traveled opinions are no more than educated guesses dressed up in evening clothes." Id.. at *15. It is difficult to see what "guesses" this animadversion pertains to. In testifying that the epidemiologic literature on SGBIs is inconclusive and that epidemiology itself warrants no inferences as to specific causation, Dr. Swan made no guesses, educated or otherwise, about SGBI and disease. [BACK]

56. . Id. at *16. [BACK]

57. . For example, with respect to the argument that the silicone in implants acts as an adjuvant, the report states that, "[f]orming the conclusion that elicitation of autoimmune and/or inflammatory disease occurs in women with SBI based on the evidence that silicone gel acts as an adjuvant when emulsified with antigen is unsupported by the data since peer-reviewed studies failed to show evidence of any autoimmune-mediated disease." Id. [BACK]

58. . Id.. [BACK]

59. . Plaintiffs' experts also discussed other possible chemical changes in the silicone in gel implants. The court's chemistry advisor found aspects of these, mostly unpublished studies "inadequately documented and of clearly debatable value," but "based on appropriate methods." Id. at *19. The court excluded testimony on these matters for reasons of "fit." See infra Part IV. [BACK]

60. . Id. [BACK]

61. . Id. [BACK]

62. . Id. [BACK]

63. . Id. at *22. [BACK]

64. . Id. at *21. [BACK]

65. . Id. at *22 [BACK]

66. . Id. [BACK]

67. . Id. Hall also determined that the admission of the physicians' testimony in Hopkins did not require the admission of the differential diagnosis here. Id. [BACK]

68. . Id. [BACK]

69. . 509 U.S. at 591 (citing United States v. Downing, 753 F.2d 1224, 1242 (3d Cir. 1985)). The fit requirement often overlaps the "scientific validity" requirement). See Cavallo v. Star Enter., 892 F. Supp. 756 (E.D. Va. 1995), aff'd on this ground, rev'd on other grounds, 100 F.3d 1150, 1159 (4th Cir. 1996). [BACK]

70. . "Fit" was interpreted somewhat differently in parts of Hall. The district court stated that "Dr. Swan's testimony has no 'fit'" because, given the court's other rulings on expert testimony, it did not "logically advance[] a material aspect of the proponent party's case." Id. at 15. This analysis of "fit" is nothing more than an analysis of relevance under Rules 401 and 402. Such an inquiry is, of course, fundamental, but it entails no special considerations due to the scientific basis of the testimony. In contrast, the court did not invoke the scientific "fit" rubric--but could have--with respect to plaintiffs' proposed testimony on immunology and toxicology. "Fit" was an issue in the extrapolations from rodents to humans and from studies involving crystalline silica to the biological effects of silicone. Id. at *14. [BACK]

71. . No. 94-04-08755-CV (Texas Dist. Ct., Zavala County, Oct. 23, 1995). [BACK]

72. . Sherine E. Gabriel et al., Risk of Connective Tissue Diseases and Other Disorders After Breast Implantation, 330 New Eng. J. Med. 1697 (1994). [BACK]

73. . Jorge Sanchez-Guerrero et al., Silicone Breast Implants and the Risk of Connective-Tissue Diseases and Symptoms, 332 New Eng. J. Med. 1666, 1666 (1995). [BACK]

74. . The famous Mayo Clinic and other medical facilities, which have accessible, linked medical records of most county residents, are located in this county. Id. at 1697-98. [BACK]

75. . A spokesman for the National Plaintiff's Steering Committee in the class action filed against the manufacturers challenged this interpretation. He explained that "[t]his study looks at the wrong disease. It doesn't look for the disease we all know exists," which is "an atypical connective-tissue disease with some neurological complications that has pieces of all the other autoimmune diseases, but is not like any of them." He complained also that the latency period for "silicone-induced diseases" was so long that about half the control subjects "may not yet have developed symptoms." Study Finds No Risk in Implants, The Oregonian, June 16, 1994, at A1, A19 (quoting Michael L. Williams). [BACK]

76. . This table is adapted from Sanchez-Guerrero et al., supra note 72, at 1668 (Table 2). [BACK]

77. . However, connective-tissue diseases are rare, occurring in less than a percent of the nurses, as determined by a blinded review of their medical records. If the diseases strike within each group at random, another large sample of women easily could have rather different ratios. For example, instead of only three women with implants and confirmed cases of connective-tissue disease, there could be six; that would change the incidence from 0.25% to 0.51%, and the unadjusted relative risk would rise from 0.4 to 0.9. There are more precise procedures to gauge the likely extent of such random fluctuations, but it is obvious that the exact numerical results in this study are fragile.

Moreover, although the nurses study covered a longer time period than the Minnesota study (the mean duration of an implant was about ten years, and the individual periods ranged from one month to over 40 years), it did not address the possibility that a long time must pass before the diseases emerge. The researchers did not purport to exclude the possibility that women who have silicone breast implants are at greater risk for known connective-tissue disease. Rather the authors wrote, "we found no evidence of an association between silicone breast implants and . . . connective-tissue diseases . . . ." Sanchez-Guerrero et al., supra note 72, at 1670. [BACK]

78. . BNA Prod. Liab. Daily, Nov. 8, 1995. See also Frank B. Vasey & Noreen Aziz, Letter to the Editor, 333 New Eng. J. Med. 1423 (1995); Gary Taubes, Silicone in the System, Discover, Dec. 1995, at 65, 75 (describing the views of one pathologist who testifies for plaintiffs and compares his views to Galileo's). [BACK]

79. . Hall, at *14. The court's chemistry advisor expressed pointed reservations about the work of the chemist, Christopher Batich, and the proposed testimony of a biomaterials engineer, Harold Alexander, but he recognized that they were "at some level, employing reasonable scientific approaches." Id. at *20. The court chose "not [to] decide whether this evidence is admissible on the basis of valid scientific methodology because the evidence now does not 'fit' plaintiffs' case." Id.. [BACK]

80. . Casey v. Ohio Medical Prod., 877 F. Supp. 1380, 1386 (N.D. Cal. 1995). [BACK]

81. . Hall, at *21. [BACK]

82. . Cf. Cavallo v. Star Enter. 892 F. Supp. 756, 771 (E.D. Va. 1995), aff'd on this ground, rev'd on other grounds. 100 F.3d 1150 (4th Cir. 1996) ("the expert must 'rule in' the suspected cause as well as 'rule out' other possible causes."). [BACK]

83. . See D.H. Kaye, Science in Evidence 395-96 (1997). [BACK]

84. . "Relative risk" is the ratio of the risk in an exposed group to the risk in an unexposed group. "Risk" can be measured by the proportion of the group that manifests the disease. See, e.g., Abraham M. Lilienfeld & David E. Lilienfeld, Foundations of Epidemiology (2d ed. 1980). The use of a mechanical rule for relative risk is questioned in Diana B. Pettiti, Reference Guide on Epidemiology, 36 Jurimetrics J. 159, 168 (1996) (book review), and Stephen E. Fienberg et al., Understanding and Evaluating Statistical Evidence in Litigation, 36 Jurimetrics J. 1, 9 (1995), as well as in the report of the epidemiologist-advisor in Hall. [BACK]

85. . A relative risk of 1.0 means that there is no association between the disease and the exposure. [BACK]

86. . Charles H. Hennekens et al., Self-Reported Breast Implants and Connective-Tissue Diseases in Female Health Professionals: A Retrospective Cohort Study, 275 JAMA 616 (1996). [BACK]

87. . That study is a randomized, double-blind, placebo-controlled trial of low-dose aspirin, beta-carotene, and vitamin E in the primary prevention of cardiovascular disease and cancer among approximately 40,000 female health professionals, aged 45 years and older. J.E. Buring & C.H. Hennekens, The Women's Health Study: Summary of the Study Design, 4 J. Myocardial Ischemia 27 (1992). Letters of invitation and questionnaires were sent to 1.75 million women. Id. [BACK]

88. . Unlike the Nurses Health Study, the self-reports of connective-tissue disease in the Women's Health Study were not rigorously confirmed through an evaluation of the respondent's medical records. The questionnaires were completed between 1992 and 1995, when publicity regarding the potential adverse health effects of breast implants already had surfaced. Heightened awareness among clinicians and among women with breast implants who reported symptoms compatible with connective-tissue disease might have contributed to overreporting. Hennekens, supra note 85; Marcia Angell, Evaluating the Health Risks of Breast Implants: The Interplay of Medical Science, the Law, and Public Opinion, 334 New Eng. Med. J. 1513, 1515 (1996). Other biases are possible. Women with breast implants and connective-tissue disease might have been more likely to participate if they suspected that the study was investigating the potential health hazards of such implants. Or, such women may have been less likely to participate if they were involved in litigation. These possible sources of bias work in opposite directions. Id.

At this point, many observers would conclude that if silicone breast implants have some effect on the diseases studied, that effect is modest. See Sasha Nemecek, Augmenting Discord, Sci. Am., Apr. 1996, 36, 38 (quoting the leading author of the Women's Health Study as stating that "[c]onsidering all the available evidence, women with breast implants should be reassured that there is no large risk."); Angell, supra, at 1515 ("Even if the self-reports should prove to be accurate, the increase in risk is so small that women with breast implants should have found these latest results reassuring."). Whether future studies can measure the risk, if any, is open to question. See Angell, supra, at 1515 ("Future studies will certainly be increasingly plagued by the problem of reporting bias. Even attempts to validate self-reports by medical records will be subject to bias, because there are now a number of doctors whose patients are referred to them by plaintiffs' attorneys and who diagnose implant-related illness so often that their records would be highly suspect.") (notes omitted). [BACK]

89. . The statistical significance of these risks was near the conventional borderline of 0.05. Of course, in a study that examines many diseases, the individual P-values cannot be accepted at face value. See, e.g., Kaye & Freedman, supra note 36; Peter H. Westfall & Russell D. Wolfinger, Multiple Tests with Discrete Distributions, Am. Statistician, Feb. 1997, at 3. [BACK]

90. . Joe S. Cecil & Thomas E. Willging, Accepting Daubert's Invitation: Defining a Rule for Court-Appointed Experts in Assessing Scientific Validity, 43 Emory L.J. 995 (1994). [BACK]

updated 2/27/97

home page publications