Giles v. Wyeth, Inc.

MEMORANDUM AND ORDER GILBERT, District Judge. I. Introduction Jeff Giles, a forty-six-year-old white male, committed suicide on October 30, 2002. Several days earlier, his family physician diagnosed him with depression and prescribed Effexor, an antidepressant. Jacquelyn Giles (Giles), Jeffs widow, contends that Effexor was a proximate cause of his suicide. In her complaint, Giles alleges that Jeff was a member of a “small vulnerable sub-population” in whom Effexor and other selective serotonin reuptake inhibitors (SSRIs) and selective seratonin norepi-nephrine reuptake inhibitors (SNRIs) increase the risk for violence and suicide. Giles brings this action against Wyeth 1 , the manufacturer of Effexor, alleging that it has known about this small vulnerable subpopulation for years, but “has failed to conduct any prospective tests to determine the frequency of this phenomenon or to develop means of identifying, screening, and protecting those patients who are in this risk group.” (Amend. Compl. at 3). Giles believes Wyeth should have warned doctors, pharmacists, and patients about this risk. To show Effexor caused her husband to commit suicide, Giles intends to rely on the opinions of two experts: Joseph Glenmul-len and Ronald Maris. After deposing Glenmullen and Maris and perusing their reports, Wyeth filed a motion to exclude their testimony as both irrelevant and unreliable under Federal Rule of Evidence 702 (Doc. 88). It contends “there is no evidence to ‘rule in’ [the experts’] theory of general causation and ... [that] then-views on specific causation do not ‘fit’ the facts of this case.” (Doc. 88 at 1). Giles has responded to Wyeth’s motion (Doc. 103) and Wyeth has replied to the response (Doc. 110). Having reviewed the motions and the record, and having heard the parties orally argue the motion, the Court is now prepared to rule. II. The Expert Reports A. Glenmullen Glenmullen 2 has concluded that Jeff “developed classic symptoms of antidepressant-induced decompensation leading to suicidality.” (Glen. Rep. at 1). He thinks the side effects of taking the drug— principally akathisia — caused Jeff to commit suicide. He based this opinion on his interviews with Giles and her son, his review of the record in this case, his clinical experience, and his reading of the relevant medical literature. i. General Causation Glenmullen’s opinion on general causation 3 — whether Effexor can cause suicidal *1050 ity as a general matter — is based on an amalgamation of different sources, including his experience in treating patients with “this side effect,” Food and Drug Administration (FDA) warnings on the risk of suicide, and academic journal articles. 4 Glenmullen first points to the warnings currently given with antidepressants, including Effexor, indicating they may cause “anxiety, agitation, panic attacks, insomnia, irritability, hostility, akathisia (severe restlessness), hypomania, and mania” and may cause or worsen suicidality. (Glen. Rep. at 5). Glenmullen focuses on akathisia, a condition which causes individuals to become anxious, agitated, panicky, sleepless, and look manic or hypomanic. He believes akathisia can cause patients to commit suicide and that it was a primary factor leading to Jeffs suicide. (Glen. Rep. at 7). Glenmullen relies on the research of Drs. Teicher and Cole, pioneers in the field, which suggests that antidepressant-induced suicidal thoughts “involve an intense violent suicidal preoccupation.” (Glen. Rep. at 8) (internal quotation marks omitted). Finally, Glenmullen relies on internal Eli Lilly documents from the early 1990’s suggesting the potential for a suicidal reaction to antidepressants. (Glen. Rep. at 9). ii. Specific Causation Glenmullen based his opinion on specific causation — whether Effexor caused Jeff to commit suicide — on his postmortem, differential diagnosis of Jeffs case. 5 Jeff took three Effexor pills in the two days before he killed himself. Though he was on Effe-xor for a short time, Glenmullen believes Jeff developed akathisia and other side effects because the night before he died, Jeff was agitated, restless, anxious, had trouble sleeping and acted out of character in a number of ways. (Glen. Rep. at 13). Glenmullen concluded that Jeffs suicide was not due to his underlying depression, an underlying anxiety disorder, an underlying psychotic disorder, being laid off work or financial hardship, previous surgery, alcoholism, substance abuse, a character disorder, another concurrent psychiatric condition, concurrent medical condition, or another prescription medication. Having ruled these causes out in the course of his differential diagnosis and because Jeff had no prior history of depression or suicide, but the traditional characteristics of antidepressant-induced akathisia, he believes Effexor caused the suicide. B. Maris Maris is a “suicidologist/suicide expert” for the Psychiatry and Family Medicine Departments within the University of South Carolina School of Medicine. As a suicidologist, Maris has investigated and studied thousands of suicides. (Maris Rep. at 2). Like Glenmullen’s, Maris’s report concludes that Effexor caused Giles’s suicide. *1051 i. General Causation Maris based his conclusions on general causation on his clinical experience and a number of studies that tie antidepressant-induced suicide to akathisia. Maris devotes considerable effort to addressing the faults of randomized clinical trials (RCTs) for purposes of studying suicide. For one thing, researchers exclude “seriously suicidal subjects” from these studies for ethical reasons. 6 (Maris Rep. at 9). Because of this, Maris thinks RCTs “are not studying the populations most at risk.” (Id. at 9). Two other concerns figure prominently in his critique. The first relates to clinical trials themselves. Maris believes the conditions of treatment in these trials differ greatly from those in a non-clinieal setting. Second, he observes that the RCTs relied on by drug manufacturers were not specifically designed to examine Giles’s hypothesis — antidepressants cause a small population of patients to become suicidal. Because they were not designed primarily to detect suicide, and RCTs are not structured to detect rare outcomes anyway, Maris believes they are too obtuse to be the “gold standard” on causation here. Maris’s problems with RCTs lead him to give more weight to small studies designed specifically to test the potential of antidepressant-induced akathisia and suicidality. Therefore, in coming to his conclusion, he puts greater emphasis on challenge/dechal-lenge/rechallenge studies and case reports. 7 In support of his reliance on these studies, he points to the prior testimony of an Eli Lilly expert, who said he considered these tests to be the preferred methodology for detecting antidepressant-induced suicide. ii. Specific Causation Maris based his opinion on specific causation on a “psychological autopsy,” which he defined as procedure “reconstructing an individual’s psychological life ... particularly the person’s lifestyle and those thoughts, feelings, and behaviors manifested during the weeks preceding death.” (Id. at 12). Maris found it important that Jeff had not been abusing alcohol or taking any other psychiatric medications before his death and that he developed “classic adverse suicidogenic reactions” after taking Effexor. Maris reached his ultimate conclusion somewhat differently than Glen-mullen, however, in that he found Jeffs other risk factors — his family history of depression, physical injuries, chronic pain, and employment status — were all suicido-genic as well. (Maris Rep. at 19). Because Jeff had had the other risk factors for many years without attempting suicide, and by all accounts, without suicidal thoughts, Maris sees his taking of Effexor as the “scale-tipper ... that pushed [him] over the edge.” (Id.). He admits that “Jeffs depressive disorder was one proximate cause of his suicide.” (Id. at 22). III. Standard In Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), the Supreme Court held that Federal Rule of Evidence 702 did *1052 not incorporate the “general acceptance” test set forth in Frye v. United, States, 293 F. 1013, 54 App. D.C. 46 (D.C.Cir.1923). Instead, the Court held that Rule 702 required district judges to be gatekeepers for proposed scientific evidence. Daubert, 509 U.S. at 589, 113 S.Ct. 2786; see also Joiner, 522 U.S. at 142, 118 S.Ct. 512. For scientific evidence to be admissible, the Court found, a district court must find it both relevant and reliable: it must be scientific knowledge grounded “in the methods and procedures of science” and consist of more than “subjective belief or unsupported speculation.” Daubert, 509 U.S. at 589-90, 113 S.Ct. 2786. When dealing with scientific evidence, the preliminary question is “whether the reasoning or methodology underlying the testimony is scientifically valid and ... whether that reasoning or methodology properly can be applied to the facts in issue.” Id. at 592-93, 113 S.Ct. 2786. Considerations pertinent to this inquiry include whether a theory or technique is capable of being or has been tested, whether it has been subjected to peer review and publication, its known or potential rate of error when applied, and whether it has gained general acceptance. Id. at 593-94, 113 S.Ct. 2786; accord United States v. Conn, 297 F.3d 548, 555 (7th Cir.2002). In 2000, Rule 702 was amended in response to Daubert. Conn, 297 F.3d at 555. In its current form, it reads as follows: If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise, if (1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the witness has applied the principles and methods reliably to the facts of the case. Fed.R.Evid. 702. Rule 702 and the directions from Daubert and its progeny govern the inquiry in this case. IV. Analysis A. Caraker In 2001, this Court decided Caraker v. Sandoz Pharm. Corp. It conducted a detailed analysis of the reliability of various evidence offered by the plaintiffs in support of their theory of causation, including epidemiological studies, ease reports, challenge/rechallenge/dechallenge reports, and medical texts. Ultimately, the Court concluded that the plaintiffs failed to meet their burden of demonstrating their experts’ testimony was reliable and that it “fit” the issues in the case. Caraker, 188 F.Supp.2d at 1029. Given the apparent similarity between this case and that, the parties rely extensively on that decision. Alisa Caraker and her husband sued Sandoz Pharmaceuticals Corporation after Alisa had a stroke allegedly caused by Sandoz’s drug Parlodel. Id. at 1028-29. To establish causation, the Carakers’ experts relied on a differential diagnosis methodology. Though it found differential diagnosis sound in the abstract, the Court concluded that “when it is used in the practice of science ... it must reliably ‘rule in’ a potential cause.” Id. (citing Glastetter v. Novartis Pharm. Corp., 252 F.3d 986, 989 (8th Cir.2001)). Speaking of epidemiology generally, the Court said the following: Epidemiology is the study of disease patterns and risks in human populations. “Epidemiology focuses on the question of general causation (i.e., is the agent capable of causing the disease?) rather than that of specific causation (i.e., did the agent cause the disease in this particular individual?).” *1053 In a typical epidemiologic study, an epidemiologist compares the health of people exposed to a substance to that of persons not so exposed to determine whether the exposure to the substance is associated with an increased rate of disease. There are essentially three types of study designs used by epidemiologists in attempting to determine whether there is an association between exposure to an agent and development of a disease: (1) randomized trial or randomized clinical trial, (2) cohort studies, and (3) case-control studies. In any epidemiological study, the first question is whether an association exists; the second is whether that association is actually a true association and not an association due to some error. The second step is important, because, while a causal relationship is one possible explanation for an association, an association does not necessarily mean that there is a cause-effect relationship. Epidemiology cannot objectively prove causation; rather, causation is a judgment for epidemiologists and others interpreting the epidemiological data. Epidemiologic studies typically provide an estimate of “relative risk” (“RR”), which is the strength of an association between exposure and disease. Relative risk is the ratio of the incidence of a disease in exposed individuals to the incidence in unexposed individuals. A relative risk of 1.0 means that the incidence in the groups is the same; that is, the exposure has no association with the disease. If the study is properly performed, a relative risk below 1.0 means that the exposure is associated with the absence of the disease (a negative association), whereas a relative risk significantly above 1.0 means that exposure is associated with an increased risk of the disease (positive association). If an epidemiologist finds any association (either positive or negative), he then scrutinizes the results in his particular study to determine whether that association indicates a causal relationship or is due to chance, error or bias. Similarly, a study that does not find an association may be erroneous on the same grounds. Id. at 1031-32 (internal citations and footnotes omitted). The Court concluded that the plaintiffs’ epidemiological evidence was not reliable. Id. at 1033. This did not, by itself, doom the plaintiffs’ case, because this Court and others do not impose an absolute epidemiology requirement, “or any other requirement, except reliability and relevance.” Id. (citing Glastetter, 252 F.3d at 992); see also Kennedy v. Collagen Corp., 161 F.3d 1226, 1229 (9th Cir.1998) (collecting cases). In any event, after reviewing the remaining scientific literature, the Court found flaws and excluded the experts’ opinions. B. General Causation i. Giles In her Daubert briefs, Giles primarily relies on seven or eight sources to support her experts’ contention that the link between SSRIs and suicidality, “particularly suicidality which is mediated by akathisia, is generally accepted.” (Doc. 103 at 12). The first of these sources is the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. In § 333.99, it states “Serotonin specific reup-take inhibitor antidepressant medications may produce akathisia.” It goes on to say that akathisia “may be associated with ... aggression, or suicide attempts.” American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, Text Revision, Appendix B (DSM-IY-TR) 801 (4th ed.2000). Wyeth rightly points out that the excerpt upon which Giles relies is in Appendix B to the DSM-IV-TR, not the DSM-IV-TR itself. Ap *1054 pendix B consists of “proposals for new categories and axes that were suggested for possible inclusion in DSM-IV.” These proposals were not included because researchers “determined that there was insufficient information to warrant inclusion ... as official categories or axes in the DSM-IV.” (Doc. 110, Ex. C). It is unclear what this information’s inclusion in Appendix B rather than within the body of the DSM-IV-TR means for purposes of this case. Giles also points to a passage in a textbook edited by one of Wyeth’s experts in this case, Dr. Schatzberg, which reads, “Akathisia in particular is associated with a poor clinical outcome, increased violence and even suicide.” (Doc. 103 Ex. 20 at 43-44). Like the material in Appendix B of the DSM-IV-TR, the material from Schatzberg’s textbook does not say that the association between akathisia and suicide rises to the level of causation. Giles, and many of the more recent articles Giles offers, relies on the work of Harvard researchers Martin Teicher and Jonathon Cole, who, in the early Nineties, achieved notoriety for their pioneering work in this area. Their 1993, peer-reviewed article concluded as follows: Although antidepressant drugs are of undeniable value and are the cornerstone of treatment of patients with moderate to severe depression, they are neither universally efficacious nor free from serious risk. Although these agents often diminish suicidal ideation in depressed patients, they do not do so in all patients, and some patients receiving these medications experience worsening or de novo emergence of suicidal ideation during treatment. Martin Teicher et al., Antidepressant Drugs and the Emergence of Suicidal Tendencies, 8(3) Drug Safety 207 (1993). Teicher and Cole recognized the association between antidepressants and suicidality, but did not venture further and explicitly attribute a causative effect. Another peer-reviewed article upon which Giles relies found a link between SSRIs and suicidality based on the administration of Prozac to patients who attempted suicide when they took the drug the first time. Anthony J. Rothschild & Carol A. Locke, Reexposure to Fluoxetine after Serious Suicide Attempts, the Role of Akathisia, 52:12 J. Clin. Psychiatry 491-93 (Dec.1991). Upon reexposure to the drug, the patients developed severe akathisia— one five days after treatment began. They said the akathisia made them feel suicidal and that this same feeling precipitated their prior suicide attempts. Once the researchers discontinued the administration of the drug, the akathisia ceased. Giles also relies on epidemiological data — primarily the work of researchers Donovan and Fergusson. Donovan’s cohort study compared patients taking tricyclic antidepressants with those taking SSRIs. The study concluded that “[significantly more [deliberate self harm] events occurred following the prescription of an SSRI than that of a [tricyclic antidepressant] (P<.001).” See Stuart Donovan et ah, Deliberate Self-harm and Antidepressant Drugs: Investigation of a Possible Link, 177 Brit.J. Psychiatry 551 (2000). At least one court has relied on this study in finding “a scientifically significant risk” that Prozac causes suicidal behavior in certain individuals. Cassidy v. Eli Lilly Co., No. 821, at 8 (W.D.Pa. June 1, 2002); see also Estates of Tobin v. Smithkline Beecham Pharm., 164 F.Supp.2d 1278, 1283 (D.Wyo.2001) (noting that plaintiffs expert relied on Donovan and ultimately concluding that the expert’s testimony was sufficiently reliable and relevant to present to the jury). But see Blanchard v. Eli Lilly & Co., 207 F.Supp.2d 308, 319 (D.Vt.2002) (finding Donovan unreliable, in part on the *1055 study’s admission that establishing a cause and effect as opposed to an associational relationship between SSRIs and deliberate self-harm or attempted suicide is “almost impossible”); Miller v. Pfizer Inc., Roerig Din, No. 99-CV-2326, 2001 WL 1793169, at *9-10 (D.Kan. Sept.4, 2001) (discussing confounding variables in the study). In Cassidy, Judge Ambrose, noting that the study had a low rate of error, concluded that “the incidence of deliberate self-harm of people on Prozac was several times more than that of people not on antidepressants.” Cassidy, No. 821 at 8. Importantly, in a subsection titled “Cause and Effect,” the authors of the study cautioned, It is difficult to attribute the cause of DSH behaviour to antidepressant treatment when such behaviour can also occur spontaneously during the course of depressive illness. Establishment of cause and effect for the different apparent risks of DSH associated with different antidepressants seen in this study is therefore almost impossible. Nevertheless, although non-fatal DSH is not a proxy for suicide because DSH is not always synonymous with failed suicide, it is a risk factor. Id. at 555. The authors went on to find that the study’s findings were consistent with previous studies, which indicated that “for whatever reason, the frequency of suicide, by any method, [is] greater in patients” who are prescribed SSRIs than patients who are prescribed TCAs. Id. at 556. Fergusson reviewed 702 clinical trials involving 87,650 patients and documented an association between suicide attempts and the use of SSRIs. Dean Fergusson et al., Association between Suicide Attempts and Selective Serotonin Reuptake Inhibitors: Systematic Review of Randomised Controlled Trials, 330 British Med. J. 653 (Mar.2005). Specifically, he found “A significant increase in the odds of suicide attempts (odds ratio 2.28, 95% confidence 1.14 to 4.55 ...; P=.02) for patients receiving SSRIs compared with placebo.” Id. With respect to non-fatal suicide attempts, he found that “a significant difference overall remained (2.70, 1.22 to 5.97; P=.01). In comparing fatal suicide attempts, [he] did not detect any differences between SSRIs and placebo (.95, .24 to 3.78).” Id. In another peer-reviewed article, Ivar Aursnes concluded, “Patients and doctors should be warned that the increased suicidal activity observed in children and adolescents taking certain antidepressant drugs may also be present in adults.” Ivar Aursnes, Suicide Attempts in Clinical Trial tuith Paroxetine Randomised against Placebo, 3:14 BioMed Central (2005). He found that previous “authors failed to convey the unanimous conclusion in the reviewed studies of an increased risk of suicide attempts.” Id. Though only one of the studies he reviewed “touched statistical significance,” he concluded, “The data strongly suggest that the use of SSRIs is connected with an increased intensity of suicide attempts per year.” Id. Other peer-reviewed literature supports the causal link as well. See David Healy, Lines of Evidence on the Risks of Suicide with Selective Serotonin Reuptake Inhibitors, 72 Psychotherapy & Psychosomatics 71-79 (2003). Giles also relies on cherry-picked data from the FDA’s 2006 study on antidepressant-induced suicide. The FDA based its analysis on data collected from 372 RCTs involving nearly 100,000 individuals. Marc Stone & M. Lisa Jones, Clinical Review: Relationship Between Antidepressant Drugs and Suicidality in Adults (Nov. 17, 2006) (hereinafter Clinical Review); Mark Levenson & Chris Holland, Statistical Evaluation of Suicidality in Adults Treated with Antidepressants, (Nov. 17, 2006) *1056 (hereinafter Statistical Review). The researchers concluded, “In contrast with the previous FDA review of pediatric studies, the pooled estimates of studies of the adult population support the null hypothesis of no treatment effect on suicidality.” (Doc. 89 Ex. 5 at 43). In short, the FDA studies found that antidepressants have a net protective effect for suicidality in adults 25-64 and that use in this age group is neutral to suicidal behavior. Despite the study’s overall conclusion, Giles grasps onto a subset of data that facially suggests that antidepressants cause suicidality in adults in the 45-54 age group. Specifically, she points to Table 18 in the first of the FDA reports, titled “Suicidal Behavior Risk for Active Drug Relative to Placebo — Preparation or Worse — Adults with Psychiatric Disorders — By Age.” Clinical Review, at 30. The odds ratio for antidepressants versus a placebo for Jeffs age group was 2.29, with a 95% confidence interval between .73-7.14, and a p-value .15. While Giles admits that a p-value of .15 is three times higher than what scientists generally consider statistically significant — that is, a p-value of .05 or lower — she maintains that this “represents 85% certainty, which meets any conceivable concept of preponderance of the evidence.” (Doc. 103 at 16). ii. Wyeth Wyeth attacks Giles’s theories of general causation on two primary grounds. First, it relies extensively on the FDA’s conclusions in its 2006 study. Emphasizing that RCTs are the gold standard in the industry, the depth of the study, the huge number of participants, and the reputations of the body and the researchers, it implicitly maintains that this study is the last word on the link between antidepressants and suicide in adults. Its first point dovetails into its second, which is that all the studies upon which Glenmullen and Maris rely are irrelevant because they deal with suicidal thoughts and attempts, not completed suicide. In its view, as the data do not support a causal relationship between completed suicide and antidepressants, the data on ideation, behavior, and attempts are irrelevant. In the course of making these broader arguments, Wyeth attacks each study relied upon by Giles as flawed in one or more important respects. With respect to the most recent FDA data, Wyeth makes several points. First, it notes that the FDA found no difference between the rates of completed suicide in antidepressant-treated and placebo-treated patients, no evidence “that, in patients treated with antidepressants for conditions (unlike depression) that did not themselves pose a risk for suicide or suicidality, antidepressants induced suicide related behaviors,” “a net protective effect of drug treatment among adults 26-64 for suicidality,” and that antidepressant use in the 25-64 age group was “neutral” as to “suicidal behavior.” (Doc. 110 at 2-3). Second, with regard to the data relied upon by Giles, it notes they are related to suicidality, not completed suicide. Third, Wyeth makes a number of observations about the statistical significance of the data. Because this data reflected a 15% possibility that the results were due to chance, and because the 95% confidence interval range includes 1.0 — meaning it failed to exclude the null hypothesis — it believes the results are completely unreliable. See generally Reference Manual at 361. Importantly, it also points to the Reference Manual on Scientific Evidence, which squarely rejects Giles’s attempts to equate statistical probabilities, especially p-values, to the eviden-tiary standard in civil actions. Id. at 358 n. 67 (“A common error made by lawyers, judges, and academics is to equate the level of alpha with the legal burden of proof. Thus, one will often see a statement that using an alpha of .05 for statistical significance imposes a burden of proof *1057 on the plaintiff far higher than the civil burden of a preponderance of the evidence.”). Wyeth takes on Donovan, Aursnes, and Fergusson as a group. It of course argues that their analyses are irrelevant because they deal almost entirely on patients who did not, in fact, complete the suicide. More important though, it faults these studies because they failed to age-stratify their results. As to Donovan specifically, Wyeth contends the data are irrelevant because he compared patients treated with different antidepressants to one another, not to untreated or placebo-treated patients. With respect to Fergusson, Wyeth notes that an FDA scientist observed that there “were serious limitations to this review, most important of which being a lack of any information on adverse events for 58% of the patients eligible for the analysis.” (Doc. 110 at 11). Citing the Reference Manual and this Court’s opinion in Caraker, Wyeth urges the Court to disregard all studies based upon case reports. See Caraker, 188 F.Supp2d at 1034-35; Reference Manual at 475 (“Causal attribution based on case studies must be regarded with caution.”). “[I]t is plainly wrong to rely on case reports to reach causal conclusions when a huge body of epidemiological evidence, including data from randomized, double-blind clinical trials that are the ‘gold standard’ of scientific evidence, fails to show an increased risk.” (Doc. 110 at 12). Wyeth argues that the Teicher and Cole and Rothschild studies are problematic because of the presence of a number of confounding factors, including prior suicide attempts and ideation or the concomitant use of other medications. See generally Reference Manual at 369-73. Because of these factors, Wyeth does not believe it is reasonable to draw any inferences on causation from these studies. Wyeth principally relies on an article by Gregory Simon for the proposition that “[S]uicide attempts and suicide death are quite distinct phenomena.” See Gregory Simon, Suicide Risk During Antidepressant Treatment, 163 Am. J. Psychiatry 41, 45 (2006). Simon “urged caution in using data regarding suicidal ideation or suicide attempts to make predictions regarding risk of suicide.” Id. at 46. It is not at all clear if he was speaking to clinicians regarding the treatment of their actual patients on an individual basis or more generally about inferring anything about the potential of antidepressant-induced suicide in a given case from data on antidepressant-induced suicidal ideation, behaviors, or attempts. During their depositions, Glenmullen and Maris appeared to express similar reservations about using ideation, behavior, and attempts as a basis for inference on completed suicides. Through selective quotation, however, Wyeth has overstated its case. During Glenmullen’s deposition, counsel for Wyeth asked him the following question. “As a matter of scientific methodology, in other words studying suicides, is using suicidal ideation as a surrogate for suicide, for identifying people who will suicide, is that a good methodology? Do you agree with that methodology?” (Glen. Dep. at 28). After clarifying that counsel was speaking in terms of studies as opposed to clinical practice, Glenmullen responded, “I would imagine it does not necessarily correlate.” (Id. at 29). Inferring from this response that Glenmullen does not believe it permissible to infer causation respecting completed suicides from data on suicide attempts and ideation, as Wyeth does, would expressly negate his entire expert report. Because of this, and considering that counsel’s question was unclear in at least four respects, it is not reasonable to infer that Glenmullen thinks *1058 data on ideation and attempts are irrelevant to completed suicide. Counsel asked Maris the following question. “Is it your belief that analysis of data regarding suicidal ideation or even behaviors is not sufficient data on — on which to base causal conclusions about suicide?” (Maris Dep. at 197). Maris responded, Well, sure. I mean it’s better — suicide — if you go back and look at my textbook, suicide is a — is a concept, or suicidology, over a panorama of indicators. Ideation, gestures, attempts, serious attempts, completions. Part of the problem is that 85 percent of the people in the United States who make a suicide attempt will die a natural death. So you’ve got to study completers. Now, does that mean that the ... FDA and Posner’s reclassification committee didn’t do any service at all by just looking at ideation in attempts? No. I mean, that’s — that’s helpful, but it doesn’t prove anything about complet-ers. (Id. at 198). To ascribe an intention to undermine his entire report on this rambling answer is unreasonable. The Court needs more than selective quotation to accept the proposition that data regarding suicidality generally have no bearing on whether an antidepressant can cause successful suicide attempts. The studies cited by Giles certainly differentiate data on completed suicides from data on attempts and ideation. Nevertheless, some of the authors of the studies implicitly rely on this data in finding an association between antidepressants and suicide. The distinction Wyeth wishes to draw is not apparent in the literature before the Court. 8 Suicide presents researchers seeking to study it with both ethical and practical difficulties. As a rare event, studying it for purposes of causation requires a huge number of participants. Simon, supra at 45 (“Given the risk observed in our sample, a new study comparing any single drug to placebo would require more than 300,000 to detect a twofold difference in risk of suicide death or serious suicide attempts.”). Perhaps more important, researchers have ethical qualms over treating “risky” patients with a placebo. Whatever the reason, suicidality itself has rarely if ever been specifically studied in large, randomised placebo-controlled double-blind epidemiological studies. The record is not clear on this point, but, at the very least, the trials upon which the FDA based its 2006 meta-analysis “were not designed to specifically detect suicidality.” Clinical Review at 43; see also Statistical Review at 48. The experts on both sides of this case have relied on epidemiology that goes both ways. Donovan and Fer-gusson’s relatively large scale epidemiological studies have shown an association suggestive of causation, at least according to Giles’s exceptionally well-credentialed experts. 9 On the other hand, the FDA’s large scale epidemiological study suggests otherwise. Wyeth’s exceptionally well-credentialed experts say that this study conclusively shows that antidepressants do not cause suicide. This is certainly enough for the Court to conclude that the causal link *1059 between antidepressants and suicide is not generally accepted. General acceptance, however, is no longer the linchpin of admissibility. Given the scope and importance of the FDA study, it is probably fair to say that the weight of the evidence goes against the Giles on epidemiology. Nevertheless, the Fergusson and Donovan studies are in peer-reviewed journals and have low rates of error. These studies have flaws, of course, but so do the studies going the other way. See S.M. v. J.K., 262 F.3d 914, 921 (9th Cir.2001) (“As commentators have repeatedly observed, ‘mental health professionals involved in everyday practice may disagree more than half the time even on major diagnostic categories such as schizophrenia and organic brain syndrome.’ ”). Large scale RCTs have not specifically tested the hypothesis that exposure to antidepressants causes patients to commit suicide. Thus, while epidemiology is probably the best way to determine causality generally — as Wyeth repeatedly notes, it is the “gold standard” — it may not be the best method to use regarding suicide, at least until the hypothesis is tested in one of these trials. 10 This figures into the weight the Court must give these studies, and necessarily requires attention to those studies that have specifically tested the hypothesis, like Rothschild’s and Teicher’s. Considering the flaws inherent to using epidemiology to study suicide and the fact that a large scale epidemiological study has not tested the specific hypothesis at issue, the Court cannot accept the contention that the FDA’s latest study is the last word. Even if Giles had no epidemiological evidence, which she does, this would not necessarily prove fatal. E.g., Kennedy, 161 F.3d at 1229; Benedi v. McNeil-P.P. C., Inc., 66 F.3d 1378, 1384 (4th Cir.1995). Other, non-epidemiological studies cited by Giles have repeatedly found an association between antidepressants and suicidality. At least in the minds of Giles’s experts, Rothschild, Teicher, the other case reports, and the adverse event reports strongly suggest causation. The authors of the studies cited by Giles (the epidemiological, case reports, ehal-lenge/dechallenge/rechallenge) are hesitant to conclude explicitly that antidepressants cause suicide. Glenmullen and Maris, recognized experts in the field, are willing to take the next step and infer causality from this data. Whether this is enough by itself the Court need not decide, because Maris and Glenmullen cite their extensive clinical experience to bolster their conclusions. See Heller v. Shaw Indus., Inc., 167 F.3d 146, 155-56 (3d Cir.1999); McCullock v. H.B. Fuller Co., 61 F.3d 1038, 1044 (2d Cir.1995). 11 Maris has investigated thou *1060 sands of suicides — patients that have had, in his opinion, antidepressant-induced suicide — and he is a trained epidemiologist. He concludes that the data are suggestive enough to infer causality. Glenmullen has treated patients that have had antidepressant-induced suicidality and draws on this experience and his reading of the medical literature to conclude that antidepressants cause suicide in a subset of patients. Though the point is close, having reviewed the myriad studies, the testimony of the experts, and the parties’ respective briefs, the Court concludes that Giles may present her theory of general causation to the jury. The proper method for attacking the evidence presented by Giles here is “[vjigorous cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof.” Daubert, 509 U.S. at 596, 113 S.Ct. 2786. C. Specific Causation Wyeth exerts considerably less effort in opposing Glenmullen and Maris’s testimony on specific causation. Its main argument is this: because Jeff had a number of suicide risk factors — depression, chronic pain, hopelessness, and unemployment among them — and because he only took three pills, the evidence is “connected to existing data only by the ipse dixit of the[ir] expert[s].” Joiner, 522 U.S. at 146, 118 S.Ct. 512. Joiner said the following on analytical gaps between data offered and an expert’s testimony: Respondent points to Daubert’s language that the focus, of course, must be solely on principles and methodology, not on the conclusions that they generate. He claims that because the District Court’s disagreement was with the conclusion that the experts drew from the studies, the District Court committed legal error and was properly reversed by the Court of Appeals. But conclusions and methodology are not entirely distinct from one another. Trained experts commonly extrapolate from existing data. But nothing 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. That is what the District Court did here, and we hold that it did not abuse its discretion in so doing. Id. (internal quotation marks and citations omitted). In Joiner, the plaintiff relied on a number of studies to support her contention that a chemical manufactured by the defendant caused cancer. She offered, among other things, animal studies and epidemiological studies. The district court found the animal studies unreliable be *1061 cause the mice in the studies received massive doses of the purported carcinogen directly into their stomachs at a much higher concentration than the plaintiff was exposed and developed alveolgenic adeno-mas, not cancer. Id. at 144, 118 S.Ct. 512. The district court similarly concluded that plaintiffs epidemiological studies were unreliable. The first was unreliable because it failed even to show an association between the purported carcinogen and cancer, the second because it was not statistically significant and its authors did not suggest a link between the carcinogen and cancer, and the third and fourth studies because the participants either had not been exposed to the purported carcinogen or were exposed to a great number of carcinogens. Id. The differences between this case and Joiner are stark. In this case, Giles has presented the Court with several statistically significant epidemiological studies that have demonstrated an association between antidepressants and suicidality. She has cited, in addition, ehallenge/rechal-lenge/dechallenge studies, case reports, and adverse event reports and relies on the combined experience of her experts, who have investigated thousands of suicidal patients. None of the studies in Joiner showed an association and none of the studies’ authors was even willing to opine on a link. Here, while the authors of most of the studies are hesitant to speculate on causation, the data in the studies are much more suggestive of causation. The ipse dixit concerns apparent in Joiner are not implicated in this case. Giles’s experts used differential diagnoses and a psychological autopsy to determine that Effexor caused Jeffs suicide. With respect to differential diagnosis, courts have held that “a medical opinion about causation, based upon a proper differential diagnosis, is sufficiently reliable to satisfy Daubert.” Glastetter, 252 F.3d at 989; see also Heller, 167 F.3d at 154-55; Westberry v. Gislaved Gummi AB, 178 F.3d 257, 262 (4th Cir.1999); McCulloch v. H.B. Fuller Co., 61 F.3d 1038, 1044 (2d Cir.1995); Glaser v. Thompson Med. Co., 32 F.3d 969, 978 (6th Cir.1994); Lennon v. Norfold & W. Ry. Co., 123 F.Supp.2d 1143, 1153 (N.D.Ind.2000). “[BJeeause a differential diagnosis is presumptively admissible, a district court may exercise its gatekeeping function to exclude only those diagnoses that are scientifically invalid.” Glastetter, 252 F.3d at 989. Perhaps more relevant here, a number of courts have held that a psychological autopsy is a generally accepted methodology for determining the cause of a suicide. Blanchard v. Eli Lilly & Co., 207 F.Supp.2d 308, 313 & n. 2 (D.Vt.2002) (citing Miller v. Pfizer, Inc., No. 99-2326, 1999 WL 1063046 (D.Kan. Nov.10, 1999)); Cloud v. Pfizer Inc., 198 F.Supp.2d 1118, 1135 (D.Ariz.2001); see Herrin v. Treon, 459 F.Supp.2d 525, 545 (N.D.Tex.2006); Yanco v. United States, 45 Fed. Cl. 782, 787 (Fed.Cl.2000). But see Halvorsen v. Plato Learning, Inc., 167 Fed.Appx. 524, 531 (6th Cir.2006) (unpublished decision) (affirming magistrate’s rejection of psychological autopsy where the doctor did not review the relevant medical records and had never performed such an autopsy before); Schad v. Schriro, 454 F.Supp.2d 897, 948 (D.Ariz.2006). After reviewing Jeffs medical records, interviewing his family, and the record in this case, both Maris and Glenmullen concluded that Effexor induced akathisia in Jeff almost immediately. His restlessness, his fidgeting, and other similar symptoms, in their minds — based on their treatment of patients with this specific side effect and their reading of the medical literature' — ■ made it clear that it was akathisia. Maris and Glenmullen have discovered, in their personal experience, that akathisia leads to suicide in some patients. They *1062 have read studies that have shown as much. Because Jeff had a history of coping with his problems — he had been in pain for years and had been out of work many times — and never exhibited any signs of suicidality previously, they both concluded that Effexor-induced suicide caused Jeff to kill himself. Glenmullen and Maris came to this conclusion in different ways. Giles argues that the difference is slight. Wyeth, on the other hand, thinks their diagnoses are fatally contradictory. After considering the risk factors and protective factors for suicide, Glenmullen ran through and dismissed (“ruled out” in differential diagnosis Daubert parlance) underlying depression, underlying anxiety disorder, underlying psychotic disorder, surgery, and several other possible causes as the cause of his suicide. Maris proceeded differently. For better or worse, he couched his conclusions in legal, rather than medical terms. Like Glenmullen, he ruled out the involvement of psychiatric drugs, psychotropic drugs, street drugs, and alcohol. He could not however, rule out depression and other risk factors entirely. He explained thus: I am not arguing that Effexor was the one and only cause of Jeff Giles’s suicide. Clearly his depressive disorder ..., family history of depression, physical injuries and resulting chronic pain, his September 12, 2002 disc surgery, his unemployment since July 14, 2002, and possible other risk factors ... were all suicidogenic, too. However, and this is important, Jeff lived with these stressors for many years without ever attempting suicide. It was only after starting Effexor treatment that Jeff became suicidal. I see Effexor as a “scale-tipper” or “just noticeable difference maker” that pushed Jeff over the edge into a violent suicide attempt. (Id. at 19). Citing Black’s Law Dictionary and other materials, he concluded that the adverse effects of Effexor “were a proximate cause or trigger” of Jeffs suicide. (Maris Rep. at 19) (emphasis added). Wyeth contends Giles’s failure to explain “how one of her experts, Dr. Glenmullen, could reliably ‘rule out’ every risk factor for suicide risk, while her other expert, Dr. Maris, rule each of them in” dooms her evidence on specific causation. Giles explains away this apparent contradiction by focusing on the fact that suicide is a “mul-tifactorial” phenomenon for which it is impossible “to ‘rule out’ other factors entirely.” (Doc. 103 at 29). While the Court sees some contradiction in the testimony of Giles’s two experts, the analytical leap that Wyeth harps on is not present. Because a combination of factors generally cause one to commit suicide, it is not surprising that Maris would refuse to admit that depression, financial hardship, and other factors played no role in Jeffs suicide. Glenmul-len’s testimony is no doubt different. 12 Viewed reasonably though, both experts saw Effexor as the decisive factor in Jeffs suicide. Whatever inconsistencies exist between the opinions of Giles’s experts will provide grist for cross-examination. Again, while close, the Court finds Giles’s evidence on specific causation sufficiently reliable and relevant to present to a jury. V. Conclusion Wyeth’s motion (Doc. 88) is DENIED. IT IS SO ORDERED. Footnotes [1] Giles sued both Wyeth, Inc. and Wyeth Pharmaceuticals. For the sake of simplicity, the Court will refer to the defendants collectively as Wyeth. [2] Wyeth does not dispute that Glenmullen and Maris are qualified to give testimony in this case. [3] As explained in Federal Judicial Center's Reference Manual on Scientific Evidence (2d ed.2000), “Epidemiology focuses on the question of general causation (i.e., is the agent capable of causing disease?) Rather than that of specific causation (i.e., did it cause disease in a particular individual).” Federal *1050 Judicial Center, Reference Manual on Scientific Evidence 336 (2d ed.2000) (hereinafter Reference Manual). [4] Wyeth does not argue that it is necessarily improper for Giles's experts to extrapolate from the data on other SSRIs. Courts recognize that "[t]rained experts commonly extrapolate from existing data.” Gen. Elec. Co. v. Joiner, 522 U.S. 136, 146, 118 S.Ct. 512, 139 L.Ed.2d 508 (1997). [5] Glenmullen explains differential diagnosis as follows: "When diagnosing conditions in medicine and psychiatry, one considers all the possible diagnoses that might account for the patient’s symptoms. In a process called the 'differential diagnosis,' one rules each of the diagnoses in or out to arrive at a final diagnosis.” (Glen. Rep. at 16). [6] There are, of course, serious ethical concerns attached to putting a seriously suicidal patient on a placebo. (Maris Rep. at 12). [7] As noted by this Court previously, ''Rechal-lenge occurs when a doctor re-exposes a patient to a drug believed to have caused an earlier adverse reaction; dechallenge removes that exposure.” Caraker v. Sandoz Pharm. Corp., 188 F.Supp.2d 1026, 1035 (S.D.Ill.2001). "[T]his type of data is substantially more valuable than run-of-the-mill case reports because a patient’s reactions are measured against his own prior reactions.” Id. [8] While not necessarily relevant, one of the patients in the Rothschild study attempted suicide on her first exposure to Prozac by jumping off the roof of a building. Because she hit a landing, she did not die, but sustained a subdural hematoma and compound fractures of her arms and legs. Whether she only intended to injure herself badly by landing on the landing, or she was saved by the grace of God, it seems that a suicide attempt like this is relevant to a completed suicide. [9] In his report, Maris stated that he has done basic epidemiological research on suicide since 1962. (Maris Rep. at 4). [10] Even the studies cited by Wyeth appear to bear this out. Simon said the following: Our data certainly do not exclude the possibility that antidepressant drugs may precipitate increased suicidal ideation or suicide attempts in a subgroup of vulnerable individuals. If it occurs, however, such a phenomenon must be infrequent enough to be hidden by the general decline in risk of suicide attempts after starting antidepressant treatment. Simon, supra at 45 (citation omitted). [11] This case is distinguishable from Caraker. There, plaintiffs' experts opined that Parlodel causes intracerebral hematomas (ICHs) because it causes arteries to constrict (i.e.vaso-constriction), vasoconstriction elevates blood pressure, and high blood pressure is a risk factor for ICHs. Caraker, 188 F.Supp2d at 1031. To support this series of conclusions, they relied on the kinds of evidence relied upon by Giles's experts here. Though the Carakers' experts concluded that the relevant epidemiological studies were fundamentally flawed, they relied on some of the data from these studies in support of their opinions on general causation. They used the fact that the studies were not statistically significant to attack the portions that were unfa *1060 vorable to their opinions, but failed to account for the lack of statistical significance when using the favorable data. Id. at 1032. For this and other reasons, the Court concluded that the epidemiological data upon which the experts relied was unreliable. Id. at 1033-34. The Court similarly found the other evidence confounded or unreliable. If Giles's experts relied upon the cherry-picked data from the FDA study alone, this would have put their epidemiological evidence in the category before the Court in Caraker. While they relied on this data in part, they also relied on statistically significant data from several other large, peer-reviewed epidemiological studies. Thus, the epidemiological evidence here is much stronger than that in Caraker. The case reports, adverse event reports, and challenge/rechallenge/dechallenge reports here apparently suffer from flaws similar to those in Caraker. These studies may well be unreliable because of confounding variables and power, like those in Caraker, but, given Giles’s epidemiological evidence and the extensive experience of Giles's experts in treating suicidality, this case is meaningfully distinguishable from Caraker. [12] Specifically, he ruled out other factors, stating, for example "in my opinion, based on a reasonable medical probability, Jeff’s suicide was not due to substance abuse.” (Glen. Rep. at 18) (italics omitted).