Full opinion text
FINDINGS OF FACT AND CONCLUSIONS OF LAW LEE, District Judge. Introduction This pharmaceutical products liability action was originally filed by the plaintiff, Lisa A. Soldo, in the United States District Court for the District of New Jersey which transferred the action to this Court because the plaintiff is a resident of Pennsylvania and also because Pennsylvania is the situs where she allegedly suffered an in-tracerebral hemorrhage as a result of her ingestion of Parlodel®, a drug manufactured and marketed by the defendant and also where she received most of her medical treatment. The Court has jurisdiction based on diversity of citizenship and the amount in controversy pursuant to 28 U.S.C. § 1332. The plaintiff is a citizen of the Commonwealth of Pennsylvania, residing at 101 West Lake Road, Transfer, Pennsylvania 16154. The defendant, Sandoz Pharmaceuticals Corporation, now Novartis Pharmaceutical Corporation (“NPC”), is organized and existing under the laws of the State of Delaware, with its principal place of business located at 59 Route 10, East Hanover, New Jersey 07936. Procedural Background Before the Court for disposition is the defendant’s Motion for Summary Judgment on Issues of Medical Causation (Document No. 77), to which plaintiff responded in Plaintiffs Memorandum of Law in Opposition to Defendant’s Motion for Summary Judgment on Issues of Medical Causation (Document No. 84). NPC moves the Court to enter judgment in its favor as a matter of law on the basis that plaintiffs evidence of general and specific causation fails to meet the test of scientific reliability set out in Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993) and followed by the Court of Appeals for the Third Circuit in In re Paoli Railroad Yard PCB Litig., 35 F.3d 717 (3d Cir.1994) and Heller v. Shaw Indus., Inc., 167 F.3d 146 (3d Cir.1999). Pursuant to NPC’s Motion for Eviden-tiary Hearing Regarding NPC’s Motion for Summary Judgment on Issues of Medical Causation (Document No. 63), the Court conducted a Daubert hearing during which medical expert witnesses testified on behalf of the parties and exhibits were introduced into the record. At various other times, on motions of the parties, other extensive exhibits, including medical treatises, were also introduced into the record. Following the Daubert hearing, with the assistance of the Duke University School of Law Registry of Independent Scientific and Technical Advisors, the Court appointed three medical experts who were directed to opine as to whether the methodology or technique employed by the plaintiffs medical experts in opining that Parlodel ® can cause stroke and did cause plaintiffs intracerebral hemorrhage is scientifically reliable. Those three experts are: (i) David A. Savitz, Ph.D. — Epidemiology (ii) William J. Powers, M.D. — Neurology/Radiology (in) David A. Flockhart, M.D., Ph.D.— Pharmacology Additionally, both before and after the Daubert hearing, the parties submitted proposed findings of fact and conclusions of law, and, after receipt of the reports of the court-appointed experts, the parties were invited to file and did file supplemental proposed findings of fact and conclusions of law. Based on the record before it, the Court enters the following Findings of Fact and Conclusions of Law. Findings of Fact A. Findings of Fact Regarding the History of Parlodel ® 1. Parlodel ® is a prescription drug formulated and sold by Novartis Pharmaceutical Corporation f/k/a Sandoz Pharmaceuticals Corporation (“NPC”) since 1978. The active ingredient of Parlodel ® is bro-mocriptine mesylate (“bromocriptine”). 2. In November 1976, NPC submitted a New Drug Application (“NDA”) for Par-lodel ® for treatment of amenorrhea/galac-torrhea. [Summary for Basis of Approval: Amenorrhea/Galactorrhea] (Att.61). 3. Parlodel ® has been approved by the Food and Drug Administration (“FDA”) since 1977 for treatment of amenor-rhea/galactorrhea associated with hyper-prolactinemia. [Summary for Basis of Approval: Amenorrhea/Galactorrhea] (Att.61). 4. In 1980, after reviewing extensive submissions from NPC’s predecessor San-doz Pharmaceuticals Corporation (“SPC”), the FDA approved Parlodel ® for the indication prevention of physiological lactation (“PPL”). Parlodel ® was found to be “both effective and safe” for the prevention of lactation. [Summary for Basis for Approval of Parlodel ®: Prevention of Physiological Lactation, at 9] (Att.62). 5. The FDA approved the use of Parlo-del ® to treat individuals with Parkinson’s Disease and also to treat infertility associated with hyperprolactinemia in 1981. [Summary for Basis for Approval of Parlo-del ®: Parkinson’s Disease] (Att.63); [Summary for Basis for Approval of Parlo-del ®: Agromegaly] (Att.64). 6. The FDA approved Parlodel® for the treatment of acromegaly in 1984. [Summary for Basis for Approval of Parlo-del ®: Female Infertility] (Att.65). ■ 7. The FDA approved Parlodel® for the treatment of Prolactin-Secreting Ade-nomas. [Summary Basis of Approval of Parlodel ®: Prolactin-Secreting Adeno-mas] (Att.66). 8. In 1990, an approved indication for Parlodel® was the PPL.1990 PDR, (Att.68). 9. At all times relevant to this case, Parlodel ® was FDA-approved for the indication PPL. [Summary for Basis for Approval of Parlodel ®: Prevention of Physiological Lactation] (Att.63). 10. In its 1984 FDA Drug Bulletin, FDA noted that though the labeling of Parlodel ® was being revised to reflect reports of adverse reactions, “[a] cause and effect relationship has not been established.” FDA Drug Bulletin, April, 1984 (Ex. 19). The 1984 Drug Bulletin expressly referenced dechallenge and rechallenge data. 11. The 1988 FDA Advisory Committee concluded that there was insufficient “evidence to indicate a causal relationship between the use of Parlodel ® and postpartum stroke/seizure.” See 1988 Summary Minutes (Ex. 20). 12. The 1989 FDA Advisory Committee concluded that there was no “need” for pharmaceutical treatment of postpartum breast engorgement, but did not present or review any new data on safety, did not review any new data on efficacy, and did not vote on the safety and efficacy of Par-lodel® for the PPL. See 1989 Summary Minutes (Ex. 21). 13. Subsequent to the 1989 Advisory Committee meeting, Dr. Solomon Sobel prepared an internal memorandum to the Commissioner of the FDA concerning the Advisory Committee’s recommendation that notes, inter alia, that “Ms. Ann Witt in the General Counsel’s office reports that we have a case for a NOOH [Notice of Opportunity for a Hearing] based on updated perceptions of efficacy and safety, ‘but it won’t be easy’ since we can raise doubts about safety but we cannot prove that risks exist.” See Memorandum from Solomon Sobel to The Commissioner, June 27, 1989 at 4 (Ex. 22) (emphasis added.) 14. SPC voluntarily withdrew the Par-lodel® indication for PPL on August 18, 1994. [Letter from Thomas Koestler to Solomon Sobel, 8/18/94] (Att.89). 15. FDA’s August 1994 Notice of Opportunity for Hearing (“NOOH”) — which was a proposal to withdraw the indication PPA — did not conclude that there was a causal connection between Parlodel® and stroke in general, or ICH in particular. See 59 Fed.Reg. 43347 (August 23, 1994). 16. FDA’s August 1994 NOOH states only that the information on adverse events raises safety questions, and seeks consideration of those issues. See 59 Fed. Reg. 43347, 43351 (August 23,1994). 17. The FDA Notice of Opportunity for Hearing was based on FDA’s receipt of reports of adverse experiences, and the Notice articulated the FDA’s perception that no pharmaceutical intervention was needed, though it confirmed that FDA could not prove that Parlodel® was not both “effective and safe,” as it had determined in 1980. The Notice, in this regard, also confirmed the FDA’s internal assessment in 1989 (when the FDA requested voluntary withdrawal of all lactation prevention drugs) that FDA could “raise doubts about safety but [FDA] cannot prove that risks exist.” Memorandum from Solomon Sobel to The Commissioner, June 27,1989, at 4 (Att.90). 18. SPC’s voluntary withdrawal of the indication PPL from Parlodel ® mooted the administrative hearing process, and thus no hearing or formal proceeding was held. 19. Notwithstanding SPC’s withdrawal of the indication PPL from Parlodel ®, on January 17, 1995, the FDA formally withdrew the indication PPL from Parlodel ®. 60 Fed.Reg. 3404-03 (January 17, 1995), (Att.94). 20. At least 10,000,000 (ten million) women in the United States are estimated to have used Parlodel ® for PPL between 1980 and 1994. lily/Revels Dep. at 58 (Att.lC); Iffy Dep. at 137 (Att.lA). 21. Parlodel® remains FDA approved today for the treatment of Parkinson’s Disease, amenorrhea and galactorrhea, and pituitary and Prolactin disorders, such as acromegaly. B. Use of FDA Proceedings in Assessing the Effects of Parlodel ® Use in Postpartum Women 22. The current FDA-approved labeling for Parlodel® states that “a causal relationship between Parlodel ® (bromo-criptine mesylate) administration and hypertension, seizures, strokes, and myocardial infarction in postpartum women has not been established.” Physicians’ Desk Reference, Aug. 1, 1998 (bold emphasis in original), Ex. RB. 23. The WARNINGS section of the current package labeling for Parlodel ® states that a causal relationship between Parlodel ® and the adverse events of stroke, seizure, and hypertension has not been established: Symptomatic hypotension can occur in patients treated with Parlodel ® (bromocriptine mesylate) for any indication. In postpartum studies with Parlodel® (bromocriptine mesylate), decreases in supine systolic and diastolic pressures of greater than 20 mm and 10 mm Hg, respectively, have been observed in almost 30% of patients receiving Parlodel ® (bromocriptine mesy-late). On occasion, the drop in supine systolic pressure was as much as 50-59 mm of Hg. While hypotension during the start of therapy with Parlodel ® (bromocriptine mesylate) occurs in some patients, in postmarketing experience in the U.S. in postpartum patients 89 cases of hypertension have been reported, sometimes at the initiation of therapy, but often developing in the second week of therapy; seizures have been reported in 72 cases (including 4 cases of status ep-ilepticus), both with and without the prior development of hypertension; 30 cases of stroke have been reported mostly in postpartum patients whose prenatal and obstetric courses had been uncomplicated. Many of these patients experiencing seizures and/or strokes reported developing a constant and often progressively severe headache hours to days prior to the acute event. Some cases of strokes and seizures were also preceded by visual disturbances (blurred vision, and transient cortical blindness). Nine cases of acute myocardial infarction have been reported. Although a causal relationship between Parlodel ® (bromocriptine me-sylate) administration and hypertension, seizures, strokes, and myocardial infarction in postpartum women has not been established, use of the drug for prevention of physiological lactation, or in patients with uncontrolled hypertension is not recommended. Physicians’ Desk Reference, Aug. 1, 1998 (bold emphasis in original), Ex. RB. 24. In the sub-section entitled “Adverse Events Observed in Other Conditions, Postpartum Patients ” of the ADVERSE REACTIONS section, the current package labeling further states: In postmarketing experience in the U.S. serious adverse reactions reported include 72 cases of seizures (including 4 cases of status epilepticus), 30 cases of stroke, and 9 cases of myocardial infarction among postpartum patients. Seizure cases were not necessarily accompanied by the development of hypertension. An unremitting and often progressively severe headache, sometimes accompanied by visual disturbance, often preceded by hours to days many cases of seizure and/or stroke. Most patients had shown no evidence of any of the hypertensive disorders of pregnancy including eclampsia, preec-lampsia or pregnancy induced hypertension. ... The relationship of these adverse reactions to Parlodel® (bro-mocriptine mesylate) administration has not been established. Physicians’ Desk Reference, Aug. 1, 1998, Ex. RB (emphasis added). 25. This language appeared on the Parlodel® label in March 1995, just two months after FDA published in the Federal Register its notice of the withdrawal of the prevention of PPL. See Ex. RP 26. At the Daubert hearing in Railey v. Sandoz Pharmaceuticals Corporation, Dr. Kenneth William Kulig agreed with Judge McDade that FDA, being a prudent agency, would err on the side of caution if there were even a possibility that an adverse effect outweighed the benefit of a drug. 11/9 Tr. at 124-25; see also Ex. SR (Ku-lig/Railey Tr. at 118). 27. In his testimony at the Daubert hearing in Railey v. Sandoz Pharmaceuticals Corp., Dr. Kulig admitted that the December 1994 FDA Federal Register notice regarding Parlodel® was not proof that Parlodel® causes strokes. Ex. SR (Kulig/Railey Tr. at 119). C. Findings of Fact Regarding the Pharmacology of Parlodel ® 28. Like dozens of other drugs, bromo-criptine is derived from ergot, a naturally-occurring substance. The drugs deriving from ergot are known as “ergot alkaloids.” Berde and Strumer, “Introduction to the Pharmacology of Ergot Alkaloids and Related Compounds as a Bases of Their Therapeutic Application,” Ergot Alkaloids and Related Compounds (hereinafter “Berde”), (Att.23). 29. Bromocriptine differs physically from the other ergot alkaloids in several respects, the most notable of which is that a bromine atom has been added. Clark, et al, “Actions on the Heart and Circulation,” Ergot Alkaloids and Related Compounds 321 (1978), (Att.67). 30. Slight differences in molecular structure can cause seemingly similar compounds to have radically different biological effects. Berde p. 2, (Att.23). 31. For example, bromocriptine inhibits uterotonic activity, whereas methlyer-gotamine has potent uterotonic activity in the rabbit. Berde p. 4, (Att.23). 32. Bromocriptine acts on dopamine receptors in the brain and elsewhere to produce its clinically useful effects. “Par-lodel ®,” Physicians’ Desk Reference, (Medical Economics Data 1990) (hereinafter “1990 PDR”), (Att.68). Most of these effects occur due to the drug’s action on dopamine receptors in the pituitary gland, a midbrain structure that controls many hormonal functions. Id. Bromocriptine blocks the secretion of the hormone Pro-lactin, which acts on the breasts to induce secretion of milk. Bromocriptine thus prevents lactation from occurring by blocking the hormone that causes it. Id. Because it prevents the secretion of Prolactin, bromo-criptine has traditionally been used-(and is still used today) for a number of disorders characterized by hyperprolactinemia, or excess prolactin secretion: amenorrhea, galactorrhea, some types of female infertility, hypogonadism, and Prolactin-secreting adenoma. Id. In addition, bromocriptine is used for acromegaly and Parkinson’s disease. S3. For PPL, Parlodel® is typically taken for 14 days, but hyperprolactinemia, acromegaly, and Parkinson’s patients may take the drug every day for years. Id. D. Findings of Fact Regarding the Medical History Giving Rise to This Lawsuit 34. Plaintiff was admitted to the hospital and delivered her second child on December 26, 1990. 12/26/90, Labor and Delivery Summary, Sentara Norfolk General Hospital, (Att.69). 35. Plaintiff was normotensive (that is, did not have elevated blood pressure) before or during her pregnancy or during or immediately after her delivery. 8/3/89 and 4/5/90, Office Notes, Dr. Shawne R. Bryant, (Att.70); 6/4/90 to 12/26/90, Prenatal Flow Sheet, Dr. Gad E. Broseh, (Att.71); 12/26/90, Labor Record, Sentara Norfolk General Hospital, (Att.72); 12/26/90, Anesthesia Record, Sentara Norfolk General Hospital, (Att.73); 12/26/90, Recovery Room Record, Sentara Norfolk General Hospital, (Att.74); 12/26/90 to 12/27/90, Postpartum Nurses’ Records, (Att.75). 36. Plaintiff elected not to breast feed. 12/26/90, Assessment Screening Room Form, Sentara Norfolk General Hospital, (Att.76). 37. On December 26, 1990, plaintiffs treating OB-GYN, Dr. Gad E. Broseh, dictated a 15-day order for Parlodel ®, 5 mg/ day, to be taken in two 2.5 mg doses per day. 12/26/90, Physician’s Post Partum Orders Form, Sentara Norfolk General Hospital, (Att.77). 38. The hospital medication administration records reflect Parlodel® was not administered while plaintiff was in the hospital. 12/26/90, Medication Administration Record, Sentara Norfolk General Hospital, (Att.78). 39. Plaintiff was discharged from the hospital on December 27, 1990. 12/27/90, Physician’s Order Form, Sentara Norfolk General Hospital, (Att.79). 40. There are no records of any prescription for Parlodel® being filled after plaintiff left the hospital on December 27, 1990. 41. Plaintiff does not recall when she started taking Parlodel®. Deposition of Lisa Soldo (“Soldo Dep.”), p. 121, (Att.8). 42. Plaintiff does not remember how often she took Parlodel®. Soldo Dep. p. 121-22, (Att.8). 43. Plaintiff does not remember how many pills of Parlodel ® per day she took. Soldo Dep. p. 121-22, (Att.8). 44. Plaintiff testified that she took Par-lodel ® while she was visiting her parents in Transfer, Pennsylvania. Soldo Dep. p. 129, (Att.8). 45. Plaintiff recalls that she discarded her empty Parlodel ® bottle “about one or two days” before her stroke. Soldo Dep. p. 128, (Att.8). 46. Plaintiffs experts credit plaintiffs deposition testimony that she did not follow her prescription and instead completed her Parlodel ® regimen “one or two days” before her intracerebral hemorrhage (“ICH”). (11/8 Tr. at 73); 11/15 Tr. at 38, 57. 47. Plaintiffs experts cannot state, given plaintiffs deposition testimony, when plaintiff took her last dose of Parlodel®. 11/15 Tr. at 57. 48. There is no evidence other than plaintiffs deposition testimony regarding the frequency and duration of plaintiffs Parlodel ® usage. 49. There is no scientific method to determine when plaintiff took her last dose of Parlodel ®. 11/15 Tr. at 62. 50. A 15-day prescription for Parlo-del ®, started on December 27, 1990, would have been completed on or around January 9,1991. 51. On January 18, 1991, 23 days after her discharge, while still in Pennsylvania, plaintiff complained of a very severe headache, and laid down in a room at her mother’s house. Soldo Dep., pp. 132-33, (Att.8). 52. When awakened several hours later, plaintiff was unresponsive. 53. Plaintiff was taken to Sharon General Hospital, where a CT-scan of the brain revealed an ICH. 1/18/91, Head CT-scan Report, Sharon General Hospital, (Att.80). 54. Plaintiffs Emergency Room admission form, completed with information provided by her family, lists possible aspirin use, but not Parlodel ®. 1/18/91, Emergency Room Record, Sharon General Hospital, (Att.52). 55. Shortly after admission to Sharon General Hospital, at 11:12 p.m., a urine sample was collected for a toxicology screen. The drug screen results indicated the presence of salicylate (aspirin) and “large amount present” of amphetamines. 1/18/91, Laboratory Report, Sharon General Hospital, (Att.81). 56. On January 19, 1991, plaintiff was transferred to Saint Elizabeth Hospital Medical Center for further treatment. While there, she was given a four-vessel cerebral arteriogram to help diagnose her condition. 1/19/91, Arteriogram Report, Saint Elizabeth Hospital Medical Center, (Att.82). Plaintiff also underwent a cranio-tomy to evacuate a large hematoma that had built up as the result of her cerebral bleed. 1/20/91 Operation Report, Saint Elizabeth Hospital Medical Center (incorrectly dated 1/21/91), (Att.83). Fragments of the hematoma were examined and found consistent with an acute hemorrhage. 1/20/91, Pathology Report, Saint Elizabeth Hospital Medical Center, (Att.84). 57. Plaintiffs highest recorded blood pressure in the Sharon General emergency room was 130/70. 1/18/91, Emergency Room Record, Sharon General Hospital, (Att.52); 1/18/91 — 1/19/91, frequent Vital Signs Form, Sharon General Hospital, (Att.91). Plaintiffs highest recorded blood pressure at Saint Elizabeth Hospital Medical Center, prior to her craniotomy was a single reading of 150/90. Most readings at Saint Elizabeth Hospital Medical Center averaged in the 110/80 range. 1/19/91, Critical Care 24 Hour Flowsheet, Saint Elizabeth Hospital Medical Center, (Att.85). 58. On January 20, 1991, plaintiff was again screened for substances in her blood. The results were negative for most substances, including salicylates, acetaminophen, and amphetamine. The test also found “substance present consistent with sympathomimetic amine.” 1/20/91, Laboratory Report, Saint Elizabeth Hospital Medical Center, (Att.86). 59. There were no objective measurements made of the amount of Parlodel ®, if any, in plaintiffs blood or tissue at the time of plaintiffs ICH. 11/8 Tr. at 91-92. 60. Plaintiffs experts testified that the half-life of Parlodel ® in the blood may be as short as three hours and may be a high as 100 hours. 11/15 Tr. at 58. 61. Plaintiffs experts assume that plaintiff had “a substantial amount” of Par-lodel® in her system at the time of her ICH. 11/15 Tr. at 20. 62. If plaintiff took her last dose of Parlodel® only one day (24 hours) before her ICH, based on a three-hour serum half life the Parlodel ® in plaintiffs system would have gone through eight half-lives prior to the event. After eight half-lives, only l/256th of the amount of Parlodel ® initially in plaintiffs blood stream would be present. 11/15 Tr. at 60-61. 63. If plaintiff took her last dose of Parlodel ® only two days (48 hours) before her ICH, based on a three-hour serum half life only 1/65,000th of the amount of Parlo-del® initially in plaintiffs blood stream would have been present at the time of the event. 11/15 Tr. at 61. 64. If plaintiff last took Parlodel® as much as 60 hours prior to her stroke, a possibility recognized by plaintiffs experts (11/15 Tr. at 57), based on a three-hour serum half-life her blood levels of bromo-criptine would be reduced by a factor of 2 20 from their initial, therapeutic level, ie., would be less than one two-millionth of their starting levels at the time of her event. 65. If plaintiff took her last dose of Parlodel® only one day (24 hours) before her stroke, a possibility recognized by plaintiffs experts (11/15 Tr. at 58), based on a 100-hour half-life, her blood levels at the time of the stroke would be only barely lower than they were when last at therapeutic levels. 66. Given the uncertainties in the timing of plaintiffs last dose and uncertainties with respect to the half-life of bromocrip-tine articulated by plaintiffs experts, it is unknown whether the level of bromocrip-tine in plaintiffs blood stood at the therapeutic level at one extreme or 1/2,000,000 of the therapeutic level at the other extreme or somewhere in between. 67. Plaintiff has not demonstrated a scientifically valid basis to conclude that she was taking Parlodel ® one or two days before her ICH. 68. Plaintiff has not demonstrated that she had a substantial amount of Parlodel ® in her system at the time of her ICH. 69. Plaintiff has not demonstrated that she had an amount of Parlodel® in her system at the time of her ICH sufficient to cause any biological effect. 70. At the time of her stroke, plaintiff smoked between half a pack and a pack of cigarettes per day. 1/19/91, History and Physical, Saint Elizabeth Hospital Medical Center, (Att.52). 71. At her deposition, plaintiff could not remember the date of her stroke. Sol-do Dep. p. 129, (Att.8). 72. Plaintiff has testified that her stroke keeps her from remembering facts about her Parlodel ® usage. Soldo Dep. p. 122, (Att.8). E. Findings of Fact Regarding Epidemiology 73. Stroke is a relatively common and widespread disease in the United States; there are 700,000 new stroke cases a year in the United States, and it is the third leading cause of death in the United States. 11/16 Tr. at 69. 74. A background risk for stroke exists in all age groups. 11/17 Tr. at 56. 75. Dr. Kulig conceded that he does not know the annual incidence of stroke in the United States. 11/8 Tr. at 169. 76. Dr. Kulig conceded that he does not know whether stroke is more common than breast cancer in the United States. 11/8 Tr. at 171. 77. Dr. Kulig conceded that he does not know whether stroke is the third leading cause of death in the United States after diseases of the heart and all cancers combined. 11/8 Tr. at 170-71. 78. Dr. Kulig conceded that he does not know what percentage of stroke victims in the United States is persons under age 65. 11/8 Tr. at 170. 79. Differential diagnosis alone cannot establish causation to a degree of medical certainty in a case involving a disease as common as stroke. 11/16 Tr. at 99; see also Ex. SQ (In re New York State Silicone Breast Implant Litigation, Brusch v. Cooper Companies, No. 128115/93, Tr. (9/29/97)) at 859 (“a cause and effect relationship” cannot be shown with a disease as common as breast cancer in humans “by a process of differential diagnosis ”) (discussed in 11/8 Tr. at 179-80). 80. Roughly one-third of all strokes, despite careful evaluation, go undiagnosed as to their cause. 11/15 Tr. at 174. 81. Strokes exist for which a particular cause cannot be ascertained, even after extensive investigation. 11/10 Tr. at 212. 82. There are strokes in persons of any age for which we do not have a mechanism to explain their causality. 11/10 Tr. at 214. 83. “In the absence of an understanding of the biological and pathological mechanisms by which disease develops, epidemiological evidence is the most valid type of scientific evidence of toxic causation.” Federal Judiciary Center, Reference Manual on Scientific Evidence (“Ref. Man. Sci. Evid.”) at 126. 84. Regardless of whether the mechanism is known, given the existence of a background risk of stroke, the scientific way to determine whether bromocriptine increases the risk of stroke in humans is through a proper controlled clinical or epi-demiologic study. 11/15 Tr. at 181-82; 11/17 Tr. at 56; Ex. SQ at 859 (Dr. Kulig agrees controlled study is required to establish a cause and effect relationship between a substance and a disease as common as breast cancer). 85. For example, because of the background risk of birth defects, it was necessary to conduct epidemiologic studies to determine whether Bendectin use raises the risk of developing birth defects. Ultimately, epidemiology demonstrated that there was a negative association between Bendictin and an increased risk of birth defects, or, put another way, Bendectin use did not raise the odds of having a child with a birth defect. 11/17 Tr. at 57-58. 86. A particular epidemiologic study’s measurement of relative risk has no meaning by itself but must be interpreted in conjunction with its statistical degree of confidence. Ref. Man. Sci. Evid. at 152-55. Relative risk is always expressed with “confidence intervals” that indicate a range of relative risk values in which the “true” relative risk is very likely to fall. Id. at 154-55. A confidence interval that includes 1.0 means that the relative risk estimate in a particular study is not statistically significant. Id. at 154-55. See generally Ref. Man. Sci. Evid. at 147-49, 154-55. Relative risk is the ratio of the incidence of disease in exposed individuals to the incidence in unexposed individuals. A relative risk of 1.0 means that the incidence in each group is the same, ie., the exposure has no association with the disease. A relative risk significantly below 1.0 means that the exposure is associated with the absence of the disease, whereas a relative risk significantly above 1.0 means that exposure is associated with an increased risk of the disease. 87. During the postpartum period, women are at an increased risk of many types of cerebrovascular accidents, including cerebral infarction, ICH, and sub-arachnoid hemorrhage. See The Kittner Study; see also 11/15 Tr. at 168-71. 88. Indeed, postpartum stroke is a common serious complication of pregnancy. 11/15 Tr. at 168; 11/16 Tr. at 24; see also Kulig/Roberts Tr. at 44-45 (pregnancy and delivery are risk factors for stroke; probably increased incidence of postpartum stroke;) see generally Lanska and Kryscio, Peripartum Stroke and Intracra-nial Venous Thrombosis in the National Hospital Discharge Survey, 89 Obstetrics & Gynecology 412 (1997), Ex. GT; see also 11/16 Tr. at 73-74. 89. “There are a number of physiological changes that occur in the transformation from pregnancy back to the non-pregnant state. These take place in what’s known as the postpartum period, which is defined as the first six weeks post-delivery. During that time there’s a major decrease in blood volume; there are hormonal changes, as the woman shifts from the hormonal state of pregnancy to non-pregnancy; there are changes in coagulation of the blood that are thought to create a hypercoagulable state, that is a state in which blood clots more easily in some women in this period. Those are some of the mechanisms that have been put forth to account for the rise in stroke in the postpartum period.” 11/15 Tr. at 170. 90. Data on pregnancy and the postpartum period gathered for the past five decades reflect that postpartum stroke is a common serious complication of pregnancy. Douglas J. Lanska, M.D., M.S., M.P.H., and Richard J. Kryscio, Ph.D., “Stroke and intracranial venous thrombosis during pregnancy and puerperium,” 51 Neurology 1622, 1627 (1998) (table 3 citing epidemiologic studies of stroke), Ex. GU; see also 11/15 Tr. at 168-70. 91. Plaintiff concedes that no epidemiology exists that demonstates that a woman taking Parlodel® postpartum is more than twice as likely to have a stroke than a woman who has not taken Parlo-del®, i.e., statistically-significant epidemiology demonstrating a relative risk greater than 2.0. 11/8 Tr. at 9,10. 92. Plaintiff cannot present any statistically significant study demonstrating an association between any ergot alkaloid and stroke in human beings. 11/9 Tr. at 132. 93. Plaintiff cannot cite any study showing that the rate of postpartum stroke increased significantly starting in 1980 when Parlodel® was introduced for the prevention of postpartum lactation in the United States. 11/15 Tr. at 70. 94. Plaintiff cannot cite any article that indicates that the risk of postpartum stroke significantly decreased after 1994 when the Parlodel® PPL indication was withdrawn in the United States. 11/15 Tr. at 71. 95. There is no prospective, double-blind, randomized, placebo-controlled study — published or unpublished — that shows that bromocriptine causes stroke. 11/10 Tr. at 187. 96. Epidemiology has methods and standards and, as such, is by its very nature “testable.” Epidemiologists express study results in terms of a relative risk. 97. Plaintiffs experts, Drs. Kenneth Kulig and Dennis Petro, disregard the express conclusion of the studies that ec-lampsia is not a sufficient explanation for the increased risk of postpartum stroke. E.g., id.; Kulig¡Hollander Dep. 117 (Att.2A); Petro Dep. 225, 227 (Att.3E); Petro/Rider Dep. 242-43 (Att.3A). 98. Drs. Kulig and Petro both concede that there is no statistically-significant epi-demiologic study showing that Parlodel® increases the risk of stroke. See Pe-tro/B/G/Q Dep: 290 (Att.3C); Iffy/NJC Dep. 46-52, 143 (Att.lA); Kulig/Hollander Dep. 108-09 (Att.2A). 99. At his deposition in Brasher v. Sandoz Pharmaceuticals Corporation, 160 F.Supp.2d 1291 (N.D.Ala. 2001) Dr. Petro, plaintiffs expert, acknowledged that the postpartum period itself is a risk factor for stroke. Petro/Brasher Dep. at 322. 100. Notwithstanding the existence of compelling evidence of an elevated risk of stroke in the postpartum period, Dr. Petro offered no basis to rule out the postpartum period in performing his differential diagnosis for plaintiffs stroke. E.g., 11/10 Tr. at 105 (“there’s no reason to believe that just having a child three weeks prior will in fact make that person susceptible to stroke”). 101. Plaintiff designated, but declined to call to testify, Dr. George Macones, an expert epidemiologist. 102. Dr. Macones rejects Dr. Kulig’s hypothesis. Dr. Macones previously testified that the epidemiology clearly showed an increased risk of stroke in the postpartum period, even excluding preeclampsia and eclampsia. In an unrelated Parlodel ® case, Dr. Macones testified regarding the Kittner Study: Q. So postpartum stroke can clearly occur in women who up to that point have had normal pregnancies and are deemed healthy. Correct? A. Yes. Q. Now, focusing on figures one and two, if we excluded all strokes associated with preeclampsia and eclampsia, we could apply the formula that we discussed earlier to make a relative risk estimation for the postpartum period compared to the balance of pregnancy. Correct? A. Yes, we could use your formula. Q. And that would yield a relative risk estimate of 11.9, roughly? A. 11.9, good job. Q. So using that estimate that would indicate that if one excludes preeclamp-sia and eclampsia, there still seems to be substantial increased risk of stroke in the postpartum period compared to the balance of pregnancy. Correct? A.... [Ajgain, using person weeks is one way of doing it. And if you look at it in terms of weeks like that and weeks at risk, then your 11 relative risk is right. I think another legitimate way to look at it is just to look' at pregnancy and postpartum and not count the number of weeks.... [T]he relative risk would be whatever, 1.8, 1.9, something like that. Q.... [E]ven if we did it your way, ... one still finds roughly twice as many postpartum strokes as strokes during pregnancy. Correct? A. Yeah, that’s absolutely what they found. Q. Even if you exclude preeclampsia and eclampsia? A. Yeah, that’s correct. Macones/B/G/Q Dep. 94-95 (Att.4A). Thus, Dr. Macones testified that he had no basis to disagree with the conclusion of Kittner that “[a] causal role for preeclamp-sia and eclampsia does not fully explain the much stronger associations in stroke found for the postpartum state than for pregnancy itself.” Macones/B/G/Q Dep. 101 (citing Kittner) (Att.4A). 103. Dr. Macones admits that the epi-demiologic data do not support the conclusion that Parlodel ® increases the risk for postpartum stroke: “Based on the epidemiological data that I have reviewed, not having reviewed anything else, the answer would be that I can’t say either way.” Macones/NJC Dep. 41^42 (Att.43). 104. Dr. Macones has testified that it is unknown whether there is a positive or a negative association between Parlodel ® and stroke. Macones/Hernandez Dep. 65-66 (Att.9). 105. Plaintiffs experts are similarly unable to point to any clinical trial for any indication of Parlodel ® in which there was a statistically-significant increased risk of stroke. PetroIB/G/Q Dep. 311 (Att.3C). Nor can plaintiffs experts point to any treatises or textbooks stating that bromo-criptine causes stroke. Petro/B/G/Q Dep. 337 (Att.3C); Iffy/NJC Dep. 181-83 (Att.lA). 106. Plaintiffs experts do not rely on any clinical trial that demonstrated stroke associated with any use of Parlodel ®. 11/9 Tr. at 74. (i) Study 60 107. Dr. Kulig testified that the Sandoz Study 60 shows that “at least one case of hypertension was caused by the drug [Par-lodel ®] using the drug company’s own causation assessment.” 11/9 Tr. at 78. 108. The investigators/authors of San-doz Study 60 do not state anywhere in the report that hypertension was demonstrated in any participant in the study. Ex. LG (Study 60). 109. Indeed, the authors of Sandoz Study 60 stated that “Parlodel ® was safe and relatively well tolerated, although a blood pressure lowering effect was noted.” Ex. LG at 6. 110. Dr. Kulig does not recall whether he reviewed the actual blood pressure data from any of the patients in Sandoz Study 60 to see whether the data supported his assertion that at least one case of hypertension during the clinical trial was caused by Parlodel ®. 11/9 Tr. at 83. 111. In Sandoz Study 60, one trial participant-Patient 62 — exhibited a single diastolic hypertensive blood pressure reading during a second 24-week phase of a three-phase clinical trial. 11/16 Tr. at 165-66. 112. As plaintiffs expert Dr. Petro testified, a single reading of elevated blood pressure is insufficient to support a finding of hypertension. 11/10 Tr. at 195-97. 113. In any event, Patient 62 in the Sandoz Study 60 was hypertensive prior to participating in the Parlodel ® clinical trial. 11/16 Tr. at 171. 114. After her 72-week involvement in the Sandoz Study 60, Patient 62’s measured blood pressure was significantly lower than it had been before her participation in Study 60. 11/16 Tr. at 171-72. 115. The Sandoz Study 60 did not demonstrate that Parlodel® treatment causes hypertension or elevated blood pressure. 116. ■ There is no evidence that the San-doz Study 60 raw data was “sanitized” in any way, at any time. 11/17 Tr. at 14-15. 117. Plaintiff presented no factual evidence that Sandoz Study 60 was terminated prematurely or “sanitized” in any way. 118. Plaintiff has not demonstrated that the Sandoz Study 60 supports her hypothesis that Parlodel ® taken in therapeutic doses causes cerebral vasoconstriction or vasospasm. 119. Plaintiff has not demonstrated that the Sandoz Study 60 raw data supports her hypothesis that Parlodel ® taken in therapeutic doses causes cerebral vaso-constriction or vasospasm. 120. Plaintiff has not demonstrated that the Sandoz Study 60 or its raw data supports her hypothesis that Parlodel® taken in therapeutic doses causes ICH. (ii) Hand Vein Study 121. Dr. Kulig testified that the Sandoz “hand vein study” demonstrates that “Par-lodel ®, like the other ergot alkaloids, is a vasoconstrictor, and in this case the blood vessel that was examined was the hand veins [sic] of human beings.” 11/8 Tr. at 145-46. 122. At his deposition in Siharath v. Sandoz Pharmaceuticals Corp., Dr. Kulig stated that he does not know whether the hand vein study results can be extrapolated to cerebral veins. 11/9 Tr. at 114; see also Kulig/Siharath Dep. at 199 (Att.10). 123. Dr. Kulig conceded that he also does not know whether the results of the hand vein study can be extrapolated to cerebral arteries. 11/9 Tr. at 114-19. 124. Dr. Kulig did not attempt to compare the doses and blood levels of bromo-criptine in Sandoz’ experiment against those seen in women receiving oral doses of Parlodel ®. 11/9 Tr. at 119. 125. A woman would have to take 5,000 Parlodel ® 2.5 mg tablets in a single dose to place the same amount of bromocriptine in her bloodstream as was used in the “hand vein study.” 11/16 Tr. at 154-55. 126. The hand vein study is a dose response study in which no effect was noted except at the highest of the test infusion doses, which was many times the dose and blood level of bromocriptine ingested under prescription for the Parlodel ® PPL indication. 127. The hand vein study does not demonstrate that any person taking Parlo-del® at therapeutic doses would develop any of the outcomes which Dr. Kulig asserts based on his interpretation of the hand vein study. 128. The hand vein study does not demonstrate that any person taking Parlo-del® at therapeutic doses would develop cerebral vasoconstriction. 129. Extrapolation from the massive Parlodel® doses given in the hand vein study to postpartum women taking Parlo-del ® does not comport with the fundamental principle of dose response. Cf. 11/17 Tr. at 42^3. 130. Plaintiffs experts, Drs. Kulig and Petro, do not use a scientifically valid methodology in relying on the results of the hand vein study as support for their opinion that Parlodel® can cause ICH in postpartum women when taken at therapeutic doses. (iii) Epidemiological Studies re: Par-lodel ® and Stroke 131. Among the epidemiologic studies concerning Parlodel® and stroke are the ERI Study, the HCIA Study, the Kittner Study and the Witlin-Sibai Study. In the first study, investigators reviewed hospital databases with information about 280,096 women delivering babies. Kenneth Roth-man, An Epidemiologic Evaluation of the Possible Relation Between Bromocriptine, Puerperal Seizures and Strokes, (Sept. 30, 1988) (“ERI Study”) (Att.14). (The case-control model of epidemiologic studies is explained in detail in the Ref. Man. Sci. Evid. at 136-38.) Out of a total of 10 postpartum strokes in this population, only one occurred in a woman who had taken Parlodel ®. The resulting relative risk calculation (8.4) was not statistically-significant, and the study was deemed “not informative.” ERI Study (Att.14) at 2. 132. Dr. Rothman found that, at the 90% confidence level, the lower confidence interval for the risk of stroke due to Parlo-del ® use was only 0.40, consistent with a negative association. Id. 133. Plaintiffs experts state that this single occurrence of a stroke among more than 280,000 women is evidence of general causation, though they nonetheless agree that it lacks statistical significance. Pe-tro/B/G/Q Dep. 409 (“the sample size was inadequate to appropriately address the question [whether Parlodel ® causes stroke]”) (Att.3C); Iffy/N/C Dep. 48 (ERI study did not reach statistical significance) (Att.lA); Kulig/NJC Dep. 83 (“I don’t believe it’s a very rehable study....”) (Att.2B); Kulig/Daubert Hearing Transcript in Nussel (Railey v. Novartis Pharms. Corp., Case No. 94-1440 (C.D.Ill., Peoria Div.)), Apr. 6, 1999, Vol. I, at 79-80 (“I’m not claiming that [the ERI] study shows that the drug Parlodel ® causes stroke”) (Att.2C); Kulig/O’Connor Dep. 35-39 (admission that he is bound by investigator’s statement that study is inconclusive) (Att. 2D). 134. Dr. Kulig testified that the ERI study is the only epidemiologic study on which he relies as support for his opinion that Parlodel ® causes ICH in the postpartum period. 11/8 Tr. at 206; see Ex. KW. 135. Dr. Kulig concedes that the confidence interval for the stroke data in the ERI study crossed the number one and therefore could not exclude the possibility that the calculated relative risk of stroke in women using Parlodel ® was due to chance. 11/8 Tr. at 207, 212-13. 136. The results of the ERI study concerning stroke are negative in terms of the hypothesis that Parlodel® causes stroke. 11/15 Tr. at 183. 137. In his deposition in O’Connor v. Sandoz Pharmaceuticals Corp., Dr. Kulig testified: “[The ERI study] doesn’t prove anything basically if you want to use proof in a very scientific sense of the word, it doesn’t prove that Parlodel ® causes strokes or seizures, it’s suggested that it does, but it doesn’t prove it, and I think we need to prove it one way or the other in order to call this drug safe or effective.” Kulig/O’Connor Dep. at 38 (Att. 7). 138. In an affidavit submitted in the case Railey v. Sandoz Pharmaceuticals Corporation, Dr. Kulig wrote, “This [ERI] study is inherently unreliable and is not relied upon....” 11/8 Tr. 191; see also Ex. SP CRailey Affidavit). 139. Dr. Kulig concedes that the basis for his opinion in Railey v. Sandoz Pharmaceuticals Corporation is the same as the basis for his opinion in this case. 11/8 Tr. at 209. 140. Dr. Kulig testified that he relies on the ERI study as support for his opinion that Parlodel ® causes ergotism. 11/9 Tr. at 8-9. 141. As Dr. Kulig concedes, the ERI study nowhere concludes or states that Parlodel ® causes ergotism. 11/9 Tr. at 10. 142. Indeed, the ERI study does not make findings about a link between Parlo-del ® and ergotism. 11/15 Tr. at 184. 143. Dr. Petro testified that he is not relying on the ERI study for any portions of his opinion. 11/10 Tr. at 99. 144. Dr. Petro nevertheless cites the ERI study as evidence that Parlodel® used in the postpartum period was a significant risk factor for stroke. 11/15 Tr. at 71. 145. Dr. Macones admits that the ERI study on Parlodel ® and postpartum stroke, upon which plaintiffs other experts rely, is “uninformative” on that issue and does not even begin to address the question. Macones/Hernandez Dep. at 65 (Att.9). 146. Dr. Macones admits that, if additional stroke cases had been found in the ERI study, it is entirely speculative as to whether such stroke cases would have been women who used Parlodel ® or women who did not. Maeones/B/G/Q Dep. at 78-80 (Att.42). Similarly, Dr. Macones admits that, if additional stroke eases had been found, additional controls would have been selected and it is entirely speculative as to whether such controls would have been women who used Parlodel ® or women who did not. Id. 147. The ERI study stroke results are not statistically significant and may not be used in a scientifically valid manner to support an expert’s opinion that bromo-criptine causes stroke. 148. The Witlin-Sibai study, “Postpartum Stroke: A Twenty-Year Experience,” examined the incidence of stroke in postpartum women. Ex. OE. 149. When the underlying study data were examined for the possible role of Parlodel® use in postpartum stroke, the Witlin-Sibai study results supported the hypothesis that bromocriptine use in the postpartum period was protective of stroke, or, to put it another way, the study showed that women taking bromocriptine were eight times less likely than women not taking bromocriptine to develop stroke in the postpartum period (Odds Ratio 0.12). This result is statistically significant. 11/17 Tr. at 67-68. 150. Dr. Sibai reliably obtained the 40,-000 Parlodel ® user figure used in the Wit-lin-Sibai study by asking Roberta Rogers, a Pharm.D., to review the hospital pharmacy records to determine how many Par-lodel® prescriptions were written over a two-year period. This figure was then extrapolated and applied over the entire period when Parlodel ® was used for postpartum lactation at Dr. Sibai’s hospital. 11/17 Tr. at 77-78. 151. Even if the number of bromocrip-tine users in the Witlin-Sibai study were overstated by 33% (of which there is no evidence), the results of the study would not fundamentally change; the study results would still reflect that women taking bromocriptine were five times less likely than women not taking bromocriptine to develop stroke. This result would still be statistically significant. 11/17 Tr. at 70. 152. The Witlin-Sibai study was peer-reviewed and initially accepted for publication. 11/17 Tr. at 73. 153. After plaintiffs counsel contacted the journal editor by telephone and in writing, the journal editor “knuckled under” and declined to publish the study. 11/17 Tr. at 73. 154. Dr. Laura Carolyn Green relied upon Dr. Sibai’s affidavit regarding the Witlin-Sibai study, an affidavit with a higher degree of reliability than the kinds of explanatory information she would normally have access to in assessing the scientific validity of a study. 11/17 Tr. at 65. 155. Although Dr. Kulig characterizes the Witlin-Sibai study as “litigation science,” Dr. Witlin was not an expert witness for NPC when the manuscript was written and Dr. Sibai was not an expert witness for NPC when the data on which the manuscript is based was collected. 11/9 Tr. at 21. 156. A third epidemiologic study analyzed 533,816 delivery records from 128 hospitals and tracked postpartum complications, correlating these complications with Parlodel® use. HCIA, Postpartum Complications and Parlodel ® (October 1995), Ex. DZ. This study estimated a relative risk for stroke associated with bro-mocriptine use of 1.088 with a confidence interval (“Cl”) from 0.448 to 2.643. Because the Cl included 1, this result was not statistically-significant. Id.; see also 11/9 Tr. at 14-15 (dismissing results from HCIA study); KuligINJC Dep. 78 (“ ... overall I think the [HCIA] study is not reliable in answering the questions that need to be answered.”) (Att.8); Ma-cones/Hernandez Dep. 76-77 (“the confidence intervals are extremely wide which suggest ... huge amounts of uncertainty in the data.”) (Att.9). 157. The HCIA study does not support plaintiffs hypothesis that bromocriptine use increases the risk of stroke in postpartum women. 158. The Kittner study determined that the risk of ICH during the postpartum period is 28.3 times higher than in similarly aged women who are not postpartum. 11/15 Tr. at 173. 159. Plaintiffs ICH falls in the postpartum time frame identified by the Kitt-ner study as a period of significantly increased risk for stroke. 11/15 Tr. at 176— 77. 160. The results of the Kittner study are consistent with the long-standing literature and studies that support the hypothesis that the postpartum period is a risk factor for stroke. 11/17 Tr. at 137-38; 11/16 Tr. at 73-76. 161. Because of the different baseline risk for stroke between European and African-American women, and the differing baseline risks of stroke depending on age, the Kittner study was age and race adjusted to minimize these possible biases in the study data. 11/16 Tr. at 16. 162. Although the Kittner study population included both European and African-American women, there is no reason to believe that the elevated relative risk for stroke in the postpartum period is different for white women and black women, even though white women and black women have different baseline risks for stroke. 11/16 Tr. at 15,18. 163. Plaintiffs experts suggest that Parlodel® perhaps accounted for the significant increased risk documented in the Kittner study. The suggestion is based on at least two critical assumptions for which no evidence was presented: —• that Parlodel ® was in fact in regular use at the hospitals involved in the Kittner study during the two years of that study; —■ that some or all of the women identified in the Kittner study with postpartum stroke had been (a) bottle-feeding, and (b) using a drug to suppress lactation. 164. In Dr. Kulig’s own hospital, Parlo-del ® was taken off of the preprinted standing orders in the mid-1980’s, i.e., before the time frame of the Kittner study. 11/8 Tr. at 32. 165. After the Kittner study was published, Dr. Kittner engaged in a case-control study examining the potential risk factors for ischemic stroke in the same geographic area. 11/15 Tr. at 179-80; Ex. GB. The case-control study did seek information concerning drug use within one month of an incident stroke, and none of the seven postpartum women who had a stroke in that study indicated usage of Parlodel ®. These facts were set forth in a letter from Dr. Kittner published in the New England Journal of Medicine. 11/15 Tr. at 178-80; Ex. GB. 166. The facts support an inference that Parlodel® may not have been available in the hospitals covered by the Kitt-ner study. In any event, there was no evidence whatsoever presented to support plaintiffs experts supposition that Parlo-del ® may have played a role in the Kittner study. 167. Parlodel® is not a scientifically probable confounder for the increased risk of stroke in postpartum women reported in the Kittner study. 11/15 Tr. at 180-81. 168. The Kittner Study specifically evaluates the role of eclampsia and concludes that eclampsia does not account for the findings of significant increased risk of stroke (for example, the 28-times increased risk of ICH). Ex. GA at 773. 169. In still another epidemiologic study, investigators compared hospital admissions and drug use to identify women who experienced ischemic heart disease, hypertension, or cerebrovascular events (such as stroke) before, during, and after Parlodel® for PPL. No women were admitted to hospitals for these conditions during the presumed exposure period or in the two months following. Herings and Strieker, Bromocriptine and Suppression of Postpartum Lactation, 17 Pharmacy World & Sci, 133-37 (1995), Ex. EA. 170. The Herings and Strieker study does not support plaintiffs hypothesis that bromocriptine use increases the risk of stroke in postpartum women. 171. The only two patients documented in the Herings and Strieker study to have had cerebrovascular disease, which includes stroke, were not users of Parlodel ®. Macones/Colangelo at 66-67. F. Scientific Method 172. The definition of science is being able to test a hypothesis in a manner which is valid — that is, controlled, unbiased, blinded whenever possible, significant in its conclusions by statistically valid techniques, and where the conclusions are supported by the data. 11/8 Tr. at 182; see also 11/10 Tr. at 182. 173. The scientific method is the naming of a hypothesis, the careful testing of that hypothesis, and the use of scientific judgment to evaluate the results of those tests. 11/17 Tr. at 35. 174. The hallmark of the scientific method is the generation of testable hypotheses which are then subjected to the real world crucible of experimentation, validation, and replication. Daubert v. Merrell Dow Pharm., 509 U.S. 579, 593, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993) (citing K. Popper, Conjectures and Refutations: The Growth of Scientific Knowledge, at 37 (5th ed.1989) (“the criterion of the scientific status of a theory is its ... testability”)). The Daubert Court went on to note that “‘scientific methodology is based on generating hypotheses and testing them to see if they can be falsified; indeed, this methodology is what distinguishes science from other fields of human inquiry.’ ” Id. (citations omitted). 175. To “falsify” a hypothesis in this context means to prove that the “null hypothesis” — that Parlodel ® has no effect on the risk of postpartum stroke — is false, i.e., that Parlodel® in fact significantly increases the risk of postpartum stroke. The failure of plaintiffs experts to show any study proving that the null hypothesis has been falsified demonstrates that their causal hypothesis has not been tested or verified by the means of science. 176. Plaintiffs experts acknowledge that epidemiologic studies are the best evidence of medical causation. See Ku-ligZNussel Hearing Transcript, Apr. 6, 1999, Vol. II, at 168-70 (Att.2C) (well performed epidemiologic study generally strongest evidence of causation); Iffy/Glo-betti Dep. 89-90 (case reports are “much less suitable” than epidemiology for proving medical causation) (Att.lB). 177. In the following dialogue, which occurred between Dr. Kulig and Chief Judge McDade in an evidentiary Daubert hearing, Dr. Kulig conceded that epidemio-logic studies are the best evidence of causation: THE COURT: If you had a choice between that type of study [epidemiologic study] and adverse event reporting sheet, which would you choose? THE WITNESS: Well, if it was the only choice? THE COURT: Yes, if that was the only choice. THE WITNESS: And the epidemio-logic study was a good one. I would obviously choose that. THE COURT: You would choose it in every case when it’s matched against something else, wouldn’t you? THE WITNESS: If it was well performed. THE COURT: Yes. THE WITNESS: Yes. Kulig/A%sseZ Hearing Transcript, Apr. 6, 1999, Vol. II at 170 (Att.2C). 178. Dr. Kulig testified that he uses “exactly the same” scientific methodology in assessing whether a substance causes a potential adverse event in both his Parlo-del ® litigation work on behalf of plaintiffs and his breast implant litigation work on behalf of defendants. 11/8 Tr. at 36-37 (Kulig). He testified to his scientific methodology in the breast implant litigation as follows: Q. Doctor, on a more general level, can a cause and effect relationship be established with a disease as common as breast cancer in humans without first showing an association through a controlled study? A. No. Q. Can it be shown with case reports? A. No. Q. Can it be shown with case series, multiple case reports? A. No. Q. Can it be shown by a process of differential diagnosis? A. No. Ex. SQ (In re New York State Silicone Breast Implant Litigation, Brusca v. Cooper Companies, No. 128115/93, Tr. (9/29/97)) at 859 (discussed at 11/8 Tr. at 172-81) (emphasis added). 179. In assessing medical causation, the scientific method requires valid scientific proof first that a drug can cause the effect in question and then valid scientific proof that the drug did cause the effect in a particular individual. For example: Dr. Kulig agrees that he would not offer an expert opinion as to causation in a specific case with one patient unless he thought as a matter of science that both general causation and specific causation had been established in a scientifically reliable way. 11/9 Tr. at 140. Dr. Petro testified that he must know whether bromocriptine can cause ICH before being able to state that a particular individual suffered an ICH caused by bromocriptine. 11/10 Tr. at 181-82; Pe-tro/Brasher Dep. at 107 (Att.2). G. Toxicologic Principles of Dose Response and Threshold 180. The principle of dose response is fundamental to the scientific method, the toxicological method, and the medical method. 11/17 Tr. at 42; 11/10 Tr. at 158. 181. The principle of dose response states that the possibility of an effect increases as the amount of substance to which a living being is exposed is increased. 11/10 Tr. at 158. 182. The principle of threshold is fundamental to toxicology. 11/10 Tr. at 158. 183. The principle of threshold states that no effects are seen in a living being until they are exposed to a certain' — ie., threshold — level of a substance. 11/10 Tr. at 158-59. 184. Bromocriptine, the parent compound, does not accumulate in the human body even after multiple doses. 11/15 Tr. at 133. 185. In this Court’s judgment, plaintiffs experts abandon the scientific method — as they themselves define it — in this case. For example, Dr. Petro acknowledged that the scientific method requires the formulation and testing of hypotheses. Petro/B/G/Q Dep. 348 (Att.3C). To test the hypothesis that a particular drug causes a particular adverse event, Dr. Pe-tro admitted that the scientific method would require one to (1) conduct a prospective, double-blind, randomized, placebo-controlled study, id. at 351; (2) utilize a single patient trial design, id. at 356-57; or (3) establish through epidemiology that an overwhelming number of people experience the adverse event when given the drug compared to those who experience the event in its absence, id. at 368-69. However, when asked whether such studies had ever been conducted showing that bromocriptine causes stroke, Dr. Petro admitted that they had not. Id. at 351-52 (no prospective, double-blind, randomized, placebo-controlled study); id. at 360 (no single patient trial design); id. at 369 (no epidemiology). 186. Dr. Petro admitted that one could not show general causation using scientific methodology in the absence of such studies: Q. In the absence of such studies, is there a particular methodology that tests the hypothesis that substance A causes effect B? A. Well, again, the observation of the effect in an uncontrolled manner does not meet the standard you are raising. Q. And when you use the term, sir, weight of evidence, that is not a scientific methodology, is it? A. Well, in certain situations, you can’t do any of these other tests, so you make a judgment. Again, it’s more subjective than scientific methodology. Q. All right. A. But again, I would suggest that that has a certain merit in scientific research in the absence of the other type of study designs, but it’s not conclusive, et cetera. I mean I — it does not rise to the standard you are suggesting. Id. at 369-70. Thus, plaintiffs experts’ methodology in this case is subjective — -in the words of her own expert — as opposed to scientific. 187. Similarly, Dr. Leslie Iffy described the scientific method as requiring “controlled studies [that] ... show ... significant evidence for [a] certain effect. ..”. Iffy/Revels Dep. 75-76 (Att.lC); see also Wy/Globetti Dep. 58 (causation established through epidemiology or “getting up controlled and blinded investigations in orde