In re Mirena Ius Levonorgestrel-Related Prods. Liab. Litig.

Paul A. Engelmayer, United States District Judge

This multi-district litigation involves products liability claims regarding a contraceptive product: the Mirenaintrauterine device developed, manufactured, and distributed by defendants Bayer HealthCare Pharmaceuticals, Inc., Bayer Pharma AG, and Bayer Oy (together, "Bayer"). The Mirena IUD functions by releasing a synthetic steroid hormone known as levonorgestrel ("LNG"). Plaintiffs claim that the hormonal component of Mirena caused them to suffer from a disease known as idiopathic intracranial hypertension ("IIH"), also known as pseudotumor cerebri ("PTC"). IIH is an uncommon disease marked by increased cerebrospinal fluid ("CSF") pressure in the skull. If untreated, IIH can cause headaches and vision problems, including, in extreme cases, blindness.

Currently before the Court are motions by each side, pursuant to Daubert v. Merrell Dow Pharmaceuticals, Inc. , 509 U.S. 579 , 113 S.Ct. 2786 , 125 L.Ed.2d 469 (1993), to exclude the other's expert testimony with respect to the issue of general causation-that is, whether Mirena's release of its hormonal component, LNG, is capable of causing IIH. In overseeing this MDL, this Court, heeding the guidance of the United States Judicial Panel on Multi-District Litigation ("JPML"), prioritized discovery and Daubert motions and briefing with respect to this issue.

Plaintiffs have put forward seven expert witnesses on general causation. Each opines that use of Mirena can cause IIH. These are: two obstetrician-gynecologists ("OB/GYNs"), an ophthalmologist, a neuroscientist, a pediatric neurologist, an epidemiologist, and a pharmacologist-toxicologist. In response, Bayer has put forward 12 expert witnesses. Each opines that the available scientific evidence does not reliably permit the conclusion that using Mirena can cause IIH. These are: three epidemiologists, a pharmaco-kineticist, five neuro-ophthalmologists, and three OB/GYNs.

Each of these 19 expert witnesses submitted an expert report, was deposed, and was the subject of separate Daubert briefing. On April 10-11, 2018, the Court held a *218 Daubert hearing and heard argument as to the admissibility of each expert's testimony as to general causation.

For the reasons that follow, the Court grants Bayer's motions under Daubert to preclude the testimony of plaintiffs' general causation experts. In light of the possible implications of this ruling for this litigation, the Court denies as potentially moot plaintiffs' motions to preclude the testimony of Bayer's general causation experts. This ruling is without prejudice to plaintiffs' right to renew their motions to preclude Bayer's general causation experts should the litigation proceed past summary judgment.

I. Procedural Background

A. Brief History of This Litigation

1. The JPML's Centralization of Mirena IIH Cases Before This Court

On April 6, 2017, the JPML centralized in this District pretrial proceedings in the 113 cases then pending across 17 districts nationwide in which plaintiffs had alleged IIH injuries caused by the hormonal component of the Mirena IUD. Most of the Mirena/IIH cases were at a relatively early stage of discovery or at the pleading stage, although fact and expert discovery had closed in the 10 longest pending actions. See generally Dkt. 1, at 2-3 (JPML transfer order).

The JPML had previously, in July 2014, denied a motion to centralize the Mirena/IIH actions, at a time when nine such actions, spanning six districts, were pending. Explaining its 2017 decision to centralize the pending cases, the JPML emphasized several factors that made centralized proceedings more efficient. Two are relevant here.

First, the JPML noted the heightened difficulty coordinating discovery and other pretrial proceedings given the increased number and dispersal of pending actions and of participating law firms. Id. at 2. "The record," the JPML stated, "demonstrates that centralization is necessary to facilitate the efficient conduct of common discovery." Id. at 3.

Second, the JPML noted, general causation had emerged as an important issue common to all proceedings. "[T]he records in the many actions filed since [2014] demonstrate that discovery and pretrial motions concerning the issue of general causation have been, or will be, at the center of all actions-that is, whether the hormonal component in Mirena is capable of causing intracranial hypertension." Dkt. 1, at 3; see also id. at 4 ("Issues concerning general causation [and] the background science ... will be common to all actions."). 1

2. Organization of This MDL

In overseeing this action, this Court has given priority to the matters that led the JPML to centralize the Mirena/IIH actions.

*219 Specifically, on June 21, 2017, after appointing plaintiffs' leadership team and reviewing written submissions and eliciting input at an initial conference as to the proper sequencing of proceedings, see generally Dkt. 51 (transcript of June 13, 2017 hearing), the Court issued an order stating that priority would be given to: (1) "the process of providing common fact discovery to plaintiff from Bayer," and (2) "resolving whether plaintiffs have admissible evidence sufficient to establish general causation" by Mirena of IIH. Dkt. 40, at 1 (June 21, 2017). The Court has done so as follows.

Outgoing discovery from Bayer : On July 27, 2017, after receiving submissions delineating the 14 discovery disputes identified by the parties and inquiring about them at a hearing, the Court resolved these disputes in a series of bench rulings. See generally Dkt. 51, at 10-59 (transcript of July 27, 2017 hearing). The Court ordered that Bayer broadly produce written discovery on all common issues, including electronic records from more than 50 Bayer custodians, and including broad production from Bayer's adverse-events database. Both as to custodians and as to search parameters, the common discovery ordered from Bayer extended well beyond the parameters theretofore utilized in the individual cases comprising the MDL. It also substantially exceeded the discovery that Bayer had produced in an immediately prior MDL also relating to the Mirena IUD. See MDL No. 2434 (the "Perforation MDL"). In that case, overseen by the Hon. Cathy Seibel of this District, the plaintiffs had alleged different injuries: that the hormone-release feature of Mirena had caused the IUD to migrate within the uterus after its insertion, leading to uterine perforation and related migration injuries.

General causation : The Court directed that the issue of general causation be litigated in the MDL as a threshold issue. To facilitate that issue's prompt resolution, the Court ordered that all fact discovery relating to general causation-including all document and deposition discovery-be completed by December 8, 2017. 2 Dkt. 62. Alerted by counsel that each side expected to make Daubert challenges to the admissibility of each other's experts on general causation, 3 the Court set deadlines spanning late December 2017 through late March 2018 for the submission of all expert reports on general causation, depositions of general causation experts, and reciprocal Daubert briefing. The Court set for the week of April 9, 2018 a "Science Day" tutorial for the Court on the background scientific issues in the case, followed by a Daubert hearing as to expert testimony on general causation. Id.

In prioritizing general causation, the Court was informed by, in addition to the guidance of the JPML, the representations of counsel in this case that the issue of general causation would be common and *220 identical to all potential Mirena/IIH plaintiffs so as make it an appropriate issue for this transferee court to resolve at the threshold. See, e.g. , Dkt. 51, at 29. The Court was also informed by the experience of the Perforation MDL. There, Judge Seibel held that plaintiffs' proposed expert testimony as to the general causation proposition at issue-that the Mirena IUD's release of the hormone LNG was capable of causing the Mirena IUD to migrate after insertion and cause uterine perforation-was not reliable under Daubert . See In re Mirena IUD Prods. Liab. Litig. , 169 F.Supp.3d 396 , 427-461 (S.D.N.Y. 2016) (" Mirena Perforation/Daubert "). Based on that ruling, Judge Seibel thereafter granted summary judgment for Bayer on all claims, holding that, without any admissible expert testimony, the remaining evidence was insufficient to establish general causation. See In re Mirena IUD Products Liability Litigation , 202 F.Supp.3d 304 , 310-28 (S.D.N.Y. 2016) (" Mirena Perforation/SJ ", aff'd , 713 Fed. Appx. 11 (2d Cir. 2017). The possibility of a similar outcome here, this Court determined, counseled deferring other time- and cost-intensive phases of this litigation pending resolution of the anticipated Daubert motions.

For that reason, the Court deferred almost all plaintiff-side discovery until after the anticipated Daubert motions were resolved. The MDL currently encompasses more than 850 plaintiffs, a result of the fact that, since creation of this MDL, nearly 750 plaintiffs have filed new lawsuits that have been received into it. The Court required only that each existing or new plaintiff file a detailed fact sheet ( e.g. , as to her Mirena usage, medical history, and symptoms). See Dkt. 62; see also Dkt. 78 (August 30, 2017) (order approving plaintiff fact sheet). The requirement of filing such fact sheets was intended to enable the litigation to move forward expeditiously in the event that admissible evidence sufficient to establish general causation were to be found, including by expediting selection of plaintiffs for individualized discovery in anticipation of bellwether trials.

B. The Expert Witnesses at Issue

The seven expert witnesses whom plaintiffs propose to call as to general causation are: (1) Dr. Lemuel A. Moyé, an epidemiologist; (2) Dr. Laura M. Plunkett, a pharmacologist and toxicologist; (3) Dr. James M. Wheeler, an OB/GYN; (4) Dr. Frederick W. Fraunfelder, an ophthalmologist; (5) Dr. Philip Darney, an OB/GYN; (6) Dr. Conrad E. Johanson, a neuroscientist; and (7) Dr. Vincent Salpietro, a pediatric neurologist.

The 12 expert witnesses whom Bayer would call are: (1) Dr. Robert Langer, an epidemiologist; (2) Dr. Kurt T. Barnhardt, an epidemiologist; (3) Dr. Todd A. Lee, an epidemiologist; (4) Dr. William Jusko, a pharmaco-kineticist; (5)-(9) Drs. Nancy J. Newman, Gregory Van Stavern, Joseph F. Rizzo, Dean M. Cestari, and Marc J. Dinkin, each a neuro-opthalmologist; and (10)-(12) Drs. Vanessa Dalton, Geri D. Hewitt, and Dana R. Gossett, each an OB/GYN.

Each expert has authored a report, was deposed, and was the subject of individualized Daubert briefing and oral argument. 4

*221 II. Factual Background

This section sets out background relevant to all Daubert motions. It addresses: (1) the Mirena product; (2) Mirena's hormonal component, the progestin LNG; (3) the disease IIH and its history and characteristics; and (4) the state of scientific research, prior to the instant Daubert proceedings, regarding both the causes of IIH in general and, more specifically, as to whether contraceptives (including Mirena ) that use LNG as a hormonal component can cause IIH. 5

A. Mirena

Mirena is a commonly used type of long-acting reversible contraceptive ("LARC"). There are several different types. Some LARCs are copper intrauterine devices ("IUDs"). 6 Others are subdermal, hormone-releasing implants. Others, of which Mirena is by far the most predominant, are hormone-releasing IUDs. Of these types of LARCs, hormone-releasing implants are generally considered the most effective at preventing conception, followed by hormone-releasing IUDs, followed by copper IUDs. Statistically, all LARCs are far more effective in preventing conception than traditional contraceptive methods such as the Pill or the male condom.

Mirena is made of a T-shaped polyethylene frame with a silicon-based steroid reservoir around the vertical stem. It is inserted into the uterus vaginally. See Mirena Prescribing Information and Label § 11.1 (Dkt. 135-6) ("Mirena Label"). In the uterus, Mirena releases the synthetic steroid hormone LNG, a progestin, at an initial rate of 20 µg LNG/day, decreasing over its five-year life-span. See Mirena Label §§ 2, 11, 12.3.

The FDA approved Mirena on December 6, 2000, following studies for safety and efficacy in two large clinical trials in Finland and Sweden. It is currently approved in the United States for up to five years of use. See id. §§ 14.1-14.2. As of May 2017, an estimated 45.5 million Mirenas had been inserted worldwide amounting to close to 142 million woman-years of use since its introduction. See Bayer Response to Questions from German Federal Institute for Drugs and Medical Devices ("BfArM") (Oct. 2017) (Dkt. 167-71) ("Bayer BfArM Response"), at 44-45.

In contrast to combined oral contraceptives, which contain both progestin and estrogen and whose effectiveness among obese women has been questioned, Mirena is believed to be effective regardless of the user's weight. Also commending its use among obese women, studies indicate that Mirena does not increase the risk of weight loss and blood clots and does not expose the user to potential risks associated with estrogen-containing contraceptives. See, e.g. , Alison Edelman & Bliss Kaneshiro, Contraception counseling for obese women , UptoDate.com (Jan. 25, 2017) at 2 (Dkt. 167-26) ("Edelman & Kaneshiro") (terming efficacy of oral contraceptives "suboptimal" for obese women; reasons hypothesized include "that the inherent effectiveness of oral contraceptives may be diminished in obese women because obesity increases metabolic rate, increases *222 clearance of hepatically metabolized drugs, increases circulating blood volume, and increases absorption of contraceptive steroids by adipose tissue"); id. ("[T]he pharmacokinetics of steroid hormones appear to be altered in obese oral contraceptives users compared with normal weight users."); Mary L. Marnach, et al., Current issues in contraception , 88 Mayo Clin Proc. 295, 297 (2013) (Dkt. 167-43) ("Progestogen-only and non-hormonal contraceptives are preferred methods of contraception for women who are obese."); see generally Centers for Disease Control and Prevention, US medical eligibility criteria for contraceptive use, 2016 , Morbidity and Mortality Weekly Report (July 29, 2016) (Dkt. 167-77); Darney Dep. at 99-100.

As a result, Mirena is widely believed to be preferentially, i.e. , disproportionately, prescribed to overweight and obese women. One recent study found that 63% of Mirena users were overweight or obese, compared to 48% of the general population of reproductive-age women. See Bayer Omnibus Br. at 3 (collecting studies); see also Edelman & Kaneshiro, supra , at 3 (terming such IUDs "the best contraceptive option for obese women who have no contraindications to use of this method" and stating that [LNG] implants "appear to be highly effective in overweight and obese women"); id. at 6 (recommending such IUDs for this population); Cestari Rpt. at 14 ("According to evidence-based guidelines, Mirena is a preferred contraceptive for obese women."); Wheeler Rpt. at 12 ("[O]ral contraceptives are less effective in obese women."); Wheeler Dep. at 90 ("[H]ormonal IUDs are preferentially prescribed to obese women.").

B. Levonorgestrel

LNG, the hormone Mirena releases, is a synthetic progestin compound derived from testosterone. It has "been used as an active ingredient in contraceptives since the 1980's." Plunkett Rpt. at 9.

LNG mimics the effects of the naturally occurring sex hormone progesterone, which is involved in pregnancy and menstruation. LNG is widely used in gynecology, including in hormonal replacement therapy. Its primary use is in numerous contraceptives. These include (1) LNG-releasing intrauterine devices such as Mirena ; (2) LNG-releasing subdermal implants such as Norplant, which from 1991 to 2002 was marketed in the United States, and Jadelle, which is currently marketed in Europe; (3) in the single-dose hormone contraceptive known as Plan B; and (4) as the progestin component of numerous combined oral contraceptives (which usually include both LNG and an estrogen compound). Id. at 8-9.

The exact mechanism by which LNG prevents pregnancy is unknown. See Mirena Label § 12.1 ("The local mechanism by which continuously released LNG enhances contraceptive effectiveness of Mirena has not been conclusively demonstrated."). However, it is generally understood that LNG, and Mirena by extension, inhibits contraception by bringing about various systemic effects, including blocking ovulation, thickening cervical mucous (decreasing sperm penetration), and altering the endometrium, the lining of the uterus (impairing the implantation of fertilized eggs). See id. ("Studies of Mirena and similar LNG IU[D] prototypes have suggested several mechanisms that prevent pregnancy: thickening of cervical mucus preventing passage of sperm into the uterus, inhibition of sperm capacitation or survival, and alteration of the endometrium."); see also Plunkett Rpt. at 9-10.

The Court notes here two chemical characteristics of LNG relevant to the pending motions. These characteristics are germane *223 to the analyses of the plaintiff experts, who theorize scientific mechanism(s) by which Mirena can cause IIH.

First, as noted, LNG is a progestin. It is not an androgen, a male sex hormone like testosterone. However, progestins can have androgenic effects. LNG, for example, is known to have a low but substantial affinity to bond with androgen receptors and to cause systemic androgenic side effects, such as acne. See Philip D. Darney, The Androgenicity of Progestins , 98 Am. J. Med. (1995) at 1A-105S (Dkt. 199-8); Christina Björklun, Expert report on the toxico-pharmacological documentation to the application for drug marketing approval of Mirena (2003), at 8 (Dkt. 199-9) ("Björklun") ("LNG is not a pure progesterone agonist. It has a low, but substantial affinity to the androgen receptor and the mineralocorticoid receptor, and to some transport proteins."); Hofmann 30(b)(6) Dep. at 167-68. Androgen receptors are found in the choroid plexus ("CP"), the area of the brain that produces cerebrospinal fluid ("CSF"). As discussed below, IIH arises from the excessive buildup of CSF.

Second, LNG is known to bond with mineralocorticoid receptors ("MRs"). See Björklun, supra , at 8. It is a point of dispute in this litigation whether LNG is an MR agonist (meaning it binds to a receptor and activates it) or antagonist (meaning it binds to a receptor and blocks it). MRs are also found in the choroid plexus. See Brian E. McGeeney & Deborah I. Friedman, Pseudotumor Cerebri Pathophysiology , 54 Headache 445, 452 (2014) (Dkt. 196-14) ("McGeeney & Friedman").

C. Idiopathic Intracranial Hypertension

Idiopathic intracranial hypertension, or IIH, "is the clinical syndrome of raised intracranial pressure, in the absence of space-occupying lesions or vascular lesions, without enlargement of the cerebral ventricles, for which no causative factor can be identified." Alex K. Ball & Carl E. Clarke, Idiopathic intracranial hypertension , 5 Lancet Neurology 433, 433 (2006) (Dkt. 167-17) ("Ball & Clark"); see also McGeeney & Friedman, supra , at 445. IIH is alternatively referred to in the scientific literature as "pseudotumor cerebri syndrome" ("PTC" or "PTCS"). See, e.g. , Deborah Friedman, et al., Revised diagnostic criteria for the pseudotumor cerebri syndrome in adults and children , 81 Neurology 1159 (2013); Fraunfelder Rpt. at 7.

By whichever name, these terms connote a syndrome marked by heightened intracranial pressure that derives not from a tumor or other lesion but from the excessive buildup of cerebral spinal fluid, or CSF. Because the parties have most often used the term IIH in this litigation, and because the term "idiopathic" usefully captures the heretofore scientifically indeterminate causes of the applications of the syndrome at issue here, the Court will use the term IIH in this decision, save when directly quoting sources that use other terms. 7

Although the immediate cause of IIH is the excessive buildup of CSF, IIH's pathogenesis-the biological mechanism that brings it about-is poorly understood. See McGeeney & Friedman, supra , at 445 *224 ("[C]erebrospinal fluid dynamics and homeostatis in PTCS are complex and incompletely understood."); Ball & Clark, supra , at 435 ("The pathophysiology underlying the raised intracranial pressure is unclear."); Cestari Rpt. at 9 (canvassing alternative theories and noting that "[t]he mechanism by which excess weight and weight gain increase the risk for IIH is [also] poorly understood"); see also Bayer Omnibus Br. at 2 n.1 (quoting scholarly literature sources cited by plaintiffs' experts to the effect, inter alia , that IIH's pathogenesis "is currently unknown," that its "pathogenesis remains a mystery," that its "underlying pathogenesis is uncertain," and that "IIH is a disorder of unclear pathophysiology") (citations omitted).

Some scientists have theorized that IIH is caused by an increase in the production of CSF. See Johanson Rpt. at 8-9. Others have theorized that IIH is caused by impaired CSF absorption; this latter theory appears to have a greater number of proponents. See Vincenzo Salpietro, et al., Recent insights on pediatric pseudotumor cerebri syndrome pathophysiology: From the "Unifying neuroendocrine perspective" to the "Integrated bioenergetic-hormonal mechanism ," 13 J. Pediatric Neurology 11, 12 (2015) (Dkt. 167-57) (noting that "the paraphysiology of [PTCS] is still poorly understood," but describing "hampered outflow of CSF into the venous system" as "a more generally accepted hypothesis" of the cause of IIH); McGeeney & Friedman, supra , at 447 (reviewing competing theories, and noting that "[m]ost of the focus in PTCS has been on resistance to CSF absorption"); Ball & Clark, supra , at 435 (observing that the "more popular hypothesis is that [IIH] is a syndrome of reduced CSF absorption").

Symptoms of IIH vary dramatically. The most common is a headache, which occurs in almost all (92-94% of) cases. Many patients (64-87%) also experience a whooshing sound in their ears called pulsatile tinnitus. Other symptoms include papilledema, photophobia, phonophobia, nausea, transient visual obscurations, binocular diplopia, and blurred vision. Papilledema, the swelling of the optic nerves due to intracranial pressure, is sometimes called the hallmark symptom of IIH. However, a patient can still be diagnosed with IIH without this symptom, and these symptoms can result from other causes. A papilledema is detected with an ophthalmoscopic or funduscopic examination. Most patients show a degree of visual loss. And, in extreme cases, if untreated, papilledema can cause serious vision loss and even blindness. See Ball & Clark, supra , at 433, 436-438; Vincent Giuseffi, et al., Symptoms and disease associations in idiopathic intracranial hypertension (pseudotumor cerebri ): A case-control study , 41 Neurology 239, 239-41 (1991) (Dkt. 167-33) ("Giuseffi"); Cestari Rpt. at 4-7; Darney Rpt. at 21; Fraunfelder Rpt. at 4; Moyé Rpt. at 26; Plunkett Rpt. at 24; Salpietro Rpt. at 12-13.

Ultimately, IIH is a diagnosis of exclusion. While doctors use two largely overlapping sets of diagnostic criteria-the "modified Dandy criteria" and the "Friedman criteria"-in their diagnoses, IIH is ultimately diagnosed by excluding alternative causes of these symptoms. Fraunfelder Rpt. at 4. Papilledema, for example, can also be caused by brain tumors, meningitis, or intracranial thrombosis, id. , and headaches can arise from numerous causes. To exclude alternative causes of IIH, doctors use a combination of neural imaging, neurologic examination, and a spinal tap or lumbar puncture, which directly measures CSF pressure. See Bayer BfArM Resp. at 9-11; Cestari Rpt. at 2-3; Fraunfelder Rpt. at 4; Moyé Rpt. at 25-27.

IIH is an extremely rare disease. It occurs with a frequency of about one case *225 per year out of a population of 100,000. See, e.g. , Giuseffi, supra , at 239; Ball & Clark, supra , at 433 (canvassing studies reporting incidences of IIH of between 0.03 per 100,000 and 2.2 per 100,000); Cestari Rpt. at 3 (0.9 cases per 100,000 population); Fraunfelder Rpt. at 5. As reviewed infra , IIH differentially affects certain demographic groups. Relevant here, IIH occurs by far most commonly-and its incidence is far greater-among women of child-bearing age, and in particular among overweight and obese such women.

IIH is treated in a variety of ways. It is typically recommended that overweight patients with IIH lose weight; medications are often prescribed to assist weight loss. Doctors also often prescribe acetazolamide, a diuretic, to prevent CSF production. Lumbar punctures, which are used to diagnose IIH, can also be used to drain excess CSF. Optic nerve sheath fenestration-the "cutting into the sheath, or covering, of the optic nerve"-"may [also] be used to decrease pressure on the optic disc, thereby preserving vision." Fraunfelder Rpt. at 7. Finally, shunts can be placed in the brain or lumbar spine to drain excess CSF into the abdominal cavity. See Ball & Clark, supra , at 439-40; Cestari Rpt. at 7-8; Moyé Rpt at 28.

D. Incidence of, and Risk Factors for IIH

Symptoms consistent with intracranial hypertension were first observed centuries ago by Eskimos, who came to associate these symptoms with excess ingestion of polar bear liver. John Chen & Michael Wall, Epidemiology and risk factors for idiopathic intracranial hypertension , 54 Int'l Ophthalmology Clinics 1, 6 (2014) (Dkt. 196-10) ("Chen & Wall"). 8 More recently, however, the cause or causes of the elevated CSF pressure in the skull that marks IIH have proven elusive to scientists. Their inquiries have been complicated by factors including the rarity of the condition and, as discussed infra , the practical inability to study causation using randomized, prospective trials involving control groups.

Scientists have, however, identified a number of risk factors associated with IIH. These include: (1) being a woman of child-bearing age; (2) being overweight, obese, or having experienced recent weight gain; (3) using certain drugs, such as vitamin A and retinoids; and (4) experiencing endocrine disturbances caused by diseases such as Addison's Disease and steroid withdrawal. See generally Ball & Clark, supra , at 434-35; Chen & Wall, supra , at 5-6; Giuseffi, supra , at 239-40; McGeeney & Friedman, supra , at 450-53; see also Fraunfelder Rpt. at 5; Moyé Rpt. at 27-28; Salpietro Rpt. at 12-13.

The first two risk factors are particularly common, as IIH "predominantly affects obese women of childbearing age." Cestari Rpt. at 2; see also Fraunfelder Rpt. at 5.

More than 90% of patients afflicted with IIH are women, see Chen & Wall, supra , at 5. And the incidence of IIH among women of child-bearing age is approximately 3.3 to 3.5 cases per 100,000, more than three times that of the population as a whole. Id. at 1; Cestari Rpt. at 3.

IIH's incidence among obese women, meanwhile, has been calculated as close to 20 times that of women of normal weight; and its incidence among overweight women *226 has been calculated as some 6.5 times that of women of normal weight. Calculations on this point vary among the scientific literature, but in general, researchers describe IIH's incidence among obese women as between approximately 15 and 22 cases per 100,000 population, and as particularly affecting women who have recently gained weight. This literature also describes IIH's incidence as rising over time in the United States as the nation's population has become more obese. See, e.g. , Anthony B. Daniels, et al., Profiles of obesity, weight gain and quality of life in idiopathic intracranial hypertension (pseudotumor cerebri ) , 143 Am. J. Ophthalmology 635, 637 & n.1 (2007) (Dkt. 167-22) (noting that IIH "tends to occur in young women of childbearing age"; "the strongest evidence for association exists for obesity and weight gain"); see also Giuseffi, supra , at 239; Ball & Clark, supra , at 433, 440 (reporting 15 to 19 IIH cases per 100,000 women who are 20% or more above their normal body weight; IIH's "association with female sex and obesity is striking"); McGeeney & Friedman, supra , at 445 (reporting 11.9 IIH cases per 100,000 obese women; "studies have confirmed that the vast majority of patients with IIH are obese women of child-bearing age"); Chen & Wall, supra , at 2 (reporting 19 IIH cases per 100,000 obese women); see also Bayer BfArM response, supra (reporting more than 20 IIH cases per 100,000 obese women); Cestari Rpt. at 3 (reporting 19.3 IIH cases per 100,000 obese women); Fraunfelder Rpt. at 5 (reporting 11.9 to 21.4 IIH cases per 100,000 population); Moyé Rpt. at 27 (reporting 22 IIH cases per 100,000 obese women).

As addressed later, the heightened incidence of IIH among overweight and obese women of reproductive age complicates the study of the relationship between Mirena and IIH. That is because Mirena, by definition, is prescribed only to women of reproductive age, and because Mirena is widely understood to be disproportionately prescribed to overweight or obese women.

E. Studies and Data, Outside of this Litigation, Bearing on Whether Use of Mirena Causes IIH

The Court next reviews the state of research, outside of the instant litigation, as to the question whether use of Mirena is a cause of IIH.

In brief, although plaintiffs' experts in this litigation have now so opined, outside of this litigation, no medical organization, regulatory agency, article in peer-reviewed scientific literature, or other research has found that use of Mirena is a cause of IIH.

However, writings in several categories exist that bear on this question. First, two published epidemiological studies have addressed the possibility of a causal connection between use of Mirena and IIH. One was principally authored by Dr. Mahyar Etminan; the other, by Reuben M. Valenzuela.

In addition, epidemiological studies have been conducted as to certain other contraceptive products containing LNG. None have found that the use of these products causes IIH.

Finally, published case reports involving patients with IIH symptoms, where the patient had used Mirena or another LNG-based contraceptive, have raised the question of a connection between the use of these products and IIH.

The above-summarized pre-litigation writings figure prominently in the reports of the experts in this case and in the parties' Daubert briefing. In particular, to varying degrees, plaintiffs' seven experts draw on these materials to support their claim that using Mirena can cause IIH.

*227 Various of these experts also draw upon case reports made to Bayer of "adverse events." The subset of Bayer's "adverse events" database relating to Mirena and IIH was made available in discovery during this MDL.

Because these materials are foundational sources for the experts who are the subjects of the instant Daubert motions, the Court-before considering any individual expert-reviews the preexisting scholarship and data. The Court first discusses conceptually the categories of studies and other evidence upon which experts may draw in exploring whether a drug can cause a given disease. The Court then reviews the publications that exist, outside of this litigation, regarding Mirena and IIH. Lastly, the Court reviews the publications regarding other ( i.e. , non-Mirena) LNG-based contraceptive products and IIH.

1. Categories of Studies Generally

In general, three broad categories of human studies can be used to explore a possible causal connection between a drug and a disease or the efficacy of particular modes of treatment of a disease: randomized control trials, epidemiological studies, and studies analyzing anecdotal evidence ( i.e. , case reports).

Randomized control trials : A randomized control study is widely considered the gold standard of human studies. Such a study consists of a "true clinical experiment in which an intervention is compared with a standard treatment, no treatment, or a placebo, with allocation to treatment by chance." Leon Speroff & Philip D. Darney, A Clinical Guide for Contraception 429 (5th ed. 2011) (Dkt. 167-62) ("Speroff & Darney"). Such experiments are often used to test possible treatments for a disease. The hallmarks of a classic clinical trial include a contemporary control group, the allocation of participants so as to assure that participant characteristics are distributed evenly across the treatment and placebo groups, and blinded as to both participants and investigators. These features enable researchers, inter alia , to control for known and unknown confounding factors. However, in the context of testing the capacity of a drug to cause a disease, randomized control studies are often unattainable in practice. Constraints presented by norms of medical ethics may preclude certain tests. And, in the context of rare ailments, such a study might have to be impracticably large to yield statistically significant results, particularly where confounding factors are present. See, e.g. , Moyé Rpt. at 12, 15-16.

Epidemiological studies : Epidemiology is the study and analysis of the distribution and determinants of health and disease conditions in defined populations. The two most common types of epidemiological studies are cohort observational studies and case-controlled observational studies.

Cohort studies involve sorting subjects into separate groups based on their exposure (or lack of exposure) to the drug in question. Such studies may be prospective or retrospective. In prospective cohort studies, patients are followed over time to see which persons develop the disease; in retrospective studies, patients are interviewed to determine whether they have been diagnosed with the disease and, if so, under what circumstances. See Speroff & Darney, supra , at 429-30. A case-control study, in contrast, involves sorting patients into separate groups based on their having been diagnosed with having (or not having) the disease in question, and comparing these groups. Id. at 430.

Relative to randomized control tests, however, both types of epidemiological studies, when used to test a thesis of general causation, present inherent challenges *228 controlling for potential confounding factors. Addressing such factors calls for intelligent study design and/or rigorous statistical analysis of results. See id. at 431.

Anecdotal evidence/case reports : Anecdotal evidence in cases involving claims that a drug caused a disease usually takes the form of "case reports." A case report is a report of an individual incident or episode, generally keyed to a specific user of a drug, "that serves to bring attention to a possible problem or condition." Id. at 430.

Individual manufacturers of drugs, such as Bayer, maintain their own adverse events databases. Another database-relevant here because one of the two epidemiological studies of Mirena and IIH, that by Dr. Etminan, drew upon it-is maintained by the Food and Drug Administration (FDA): the FDA Adverse Event Reporting System ("FAERS"). It contains approximately 5 million adverse event reports, medication error reports, and product quality complaints resulting in adverse events that were submitted to the FDA. The database is designed, inter alia , to support the FDA's post-marketing safety surveillance program for FDA-approved biologic products. Reports are submitted to FAERS by healthcare professionals, consumers, and manufacturers. Reports from healthcare professionals ( e.g. , physicians, pharmacists, and nurses) and consumers ( e.g. , patients, family members, and lawyers) are voluntary. If a manufacturer receives a report from a healthcare professional or consumer, however, regulations require it to send the report to the FDA.

Consistent with the case law reviewed infra , scholars, including plaintiffs' experts in this litigation, agree that while case reports often have utility in generating hypotheses about the possible relationship between a drug and a medical condition, such anecdotal accounts-except in extremely rare circumstances-cannot, without more, demonstrate the causation by a drug of such a condition. 9 See, e.g. , Darney Dep. at 206 ("[C]ase reports cannot establish causation."); id. at 144 (noting that the "principal use" of case reports is "to generate a hypothesis for epidemiologic ... investigation"); Moyé Rpt. at 21 ("[I]t is the rare case report that demonstrates causality in and of itself"; case reports are "essential to the causation process" because they "demonstrate what is possible"); Moyé Dep. at 131 ("[C]ase reports are not even sufficient to show association because there is no comparison group."); Plunkett Dep. at 184 (noting case reports "show[ ] associations, not general cause"); Wheeler Dep. at 206 ("[A]dverse event data cannot be used to establish causation" and "the proper use of adverse event report data is ... hypothesis generating ...."); cf. Fraunfelder Rpt. at 8 (observing that case reports may serve as "signals," defined as "reported information on a possible causal relationship between an adverse event and a drug").

Animal studies : Animal studies can sometimes be used to test a causal connection between a drug and a disease. Advantages of animal studies include the fact that, because fewer ethical constraints apply *229 to such studies, these can be "conducted as true experiments." Michael D. Green, et al., Reference Guide on Epidemiology, in Reference Manual on Scientific Evidence 549, 563 (Fed. Jud. Ctr., 3d ed. 2000). But animal studies have several disadvantages. One is that it may be difficult to extrapolate from such studies to the context of humans. Another "difficulty with inferring human causation from animal studies is that the high doses customarily used in animal studies require consideration of the dose-response relationship and whether a threshold no-effect dose exists." Id. at 345-46. Animal studies are of limited relevance in this case, in that no such study has assessed whether LNG (however administered) causes IIH. The animal studies cited by the parties (involving other hormones) instead relate to discrete steps in longer biological chains of causation posited by individual experts who opine as to possible mechanisms by which Mirena might cause IIH.

2. Studies and Case Reports Regarding Mirena

a. Randomized Control Trials

There have not been randomized control trials addressing whether Mirena can cause IIH. The parties appear to agree that such a trial would not realistically be attainable, as a result of medical ethics and/or prohibitive cost. The latter is because IIH's rarity would likely make it prohibitive to design a statistically significant study. See, e.g. , Moyé Rpt. at 16.

b. Epidemiological Studies

Outside of this litigation, two epidemiological studies have been published addressing the relationship between Mirena and IIH: the Etminan and Valenzuela studies. Because these two studies figure centrally in the Daubert litigation, the Court reviews each in detail. As to the Etminan study, the Court further reviews the unusual series of events that culminated in lead author Etminan's repudiation of a significant portion of his study.

(i) The Etminan study

2015-Etminan's initial study : Etminan's study was published in 2015 in the Journal of Therapeutic Advances in Drug Safety. See Mahyar Etminan, et al., Risk of intracranial hypertension with intrauterine levonorgestrel , 6 Therapeutic Advances in Drug Safety 110 (2015) (Dkt. 167-27) ("Etminan" or the "Etminan study"). At the time that he published the study, Dr. Etminan was serving as a retained expert for plaintiffs in the pre-MDL Mirena/IIH litigation. The Etminan study did not disclose this relationship. The study declared in its "conflict of interest statement" that the authors did not have any conflict of interest to declare. Id. at 113.

The Etminan study contained two analyses designed to supplement one another. The outcomes of the analyses pointed in opposite directions. The first was a "disproportionality analysis" (DPA) of adverse event reports in the FDA's FAERS database. Id. The DPA analysis found that the "reporting odds ratio" as to IIH, and as to search terms associated with IIH, was higher for Mirena than for a comparison group consisting of users of all other drugs in the FAERS database, which captured some 5 million reported adverse events. Id. at 112. 10 Etminan opined that this finding was statistically significant. Id. at 112.

*230 The second analysis consisted of a retrospective cohort study. It compared the risk of intracranial hypertension between, on the one hand, LNG-releasing IUDs, and, on the other hand, two combination oral contraceptives that did not contain LNG: ethinyl estradiol (EE) and norethindrone and EE-norgestimate. Id. at 111. The study drew upon a large health claims database, IMS LifeLink, which contained more than 102 healthcare plans. This comparison was undertaken, Etminan stated, because "[o]ral and intramuscular contraceptives including progestins have been linked to [intracranial hypertension ]." Id. at 112. This second analysis did not find any difference in the risk of IIH between users of EE-norgestimate and Mirena. Id. It found, in fact, a lower risk of IIH for users of EE-norethindrone relative to users of Mirena, although it concluded that that result was not statistically significant. Id.

Etminan's study, which described itself as "the first large epidemiologic study that has examined the risk of [IIH] with Mirena," id. at 112, did not definitively conclude that Mirena causes IIH. And, the study noted, the authors "did not have information on all risk factors for [IIH]." Id. at 113. However, the authors urged that "the risk of [IIH] with Mirena must be clearly conveyed to young women who are planning to use them." They further opined that "the small risk of [IIH] may outweigh the risk of unintended pregnancies." Id. at 113.

2016-Dr. Friedman's critique : In 2016, the Etminan study came under attack. Dr. Deborah Friedman, the author of various studies relating to IIH, published, in the same journal as the Etminan study, a letter critiquing on multiple grounds the Etminan study's methodology. Centrally, she noted that Etminan's disproportionality analysis-the part of his study that had pointed to an increased IIH risk for Mirena uses-had failed to adjust for age and gender, thus comparing Mirena patients (reproductive-age females) to populations like older men who almost never get IIH. Friedman termed the Etminan study's conclusions "erroneous and misleading." See Deborah Friedman, Risk of intracranial hypertension with intrauterine levonorgestrel , 7 Therapeutic Advances in Drug Safety 23, 23 (2016) (Dkt. 213-7) ("Friedman Letter").

As to Etminan's DPA analysis, Friedman wrote, the search terms that Etminan had used to isolate IIH-related symptoms within the FAERS database had been overbroad and had included terms not demonstrably related to that condition. Id. As to Etminan's second analysis-the cohort study comparing Mirena to two oral contraceptives in the IMS LifeLink database-Friedman wrote, Etminan's data set and methodology were problematic in multiple respects. 11 Etminan's cohort study *231 was also faulty, Friedman wrote, because it did not include a control group of women who were not using the hormonal contraceptives at issue, or any such contraceptives at all. Id. at 24. The study, she stated, had not found a statistically significant difference in the development of IIH between users of the two oral contraceptives and Mirena. Further, it had failed to assess whether the rate among users of Mirena"is higher than would be expected in their population sample in general." Id. She noted that no published evidence had shown a causal relationship between IIH and hormonal contraception. Id. 12

2016-Dr. Etminan's letter response : In 2016, Dr. Etminan published a brief (1.5 page) letter response to Friedman's letter. That letter, Dr. Etminan acknowledged, had raised "some questions." See Mahyar Etminan, Risk of intracranial hypertension with intrauterine levonorgestrel : reply , 7 Therapeutic Advances in Drug Safety (2016) (Dkt. 167-28) ("Etminan Letter"). 13

As to his DPA analysis, Dr. Etminan's 2016 letter stated that, even limiting the search terms used within the FAERS database to exclude conditions such as "cerebral edema" not associated with IIH, there were still a disproportionate number of adverse events associated IH for "women with Mirena." Id. at 1. As to his cohort study, Dr. Etminan clarified and/or defended his methodology. 14 Finally, Dr. Etminan disputed "that there are no epidemiological studies linking hormonal contraceptives to intracranial hypertension." He noted two studies (from 1990 and 1993) that he acknowledged "lacked statistical power," and a 2015 case-control study (a study by Rai, et al. , that previewed the Valenzuela study addressed infra ) that, Etminan stated, had found a statistically significant increase in IIH among women with intrauterine LNG use compared to nonusers, but which, Dr. Etminan acknowledged, did "not seem to have adjusted for body mass index (BMI), a major confounder for this question." Id.

*232 2016-Dr. Etminan's affidavit repudiating his findings : Later in 2016, Dr. Etminan changed course. He repudiated much of his study's analysis. On December 17, 2016, after being served with a notice of a deposition in a lawsuit against Bayer that is now part of this MDL, Dr. Etminan furnished to counsel for Bayer a sworn affidavit retracting many of his study's findings. See Dkt. 167-29 ("Etminan Affidavit").

Addressing his study's disproportionality analysis, Dr. Etminan clarified that the study had calculated "reporting odds ratios" (RORs) for Mirena "versus all other products in the FAERS database"; "the comparator group," he now acknowledged, "had not been limited to oral contraceptive pills or any other specific product." Id. ¶¶ 3-4. Further compromising the results, Dr. Etminan admitted, the analyst whom he had used to extract data from that database

did not limit the groups compared in the FAERS DPA to reproductive age females. Because the background incidence of intracranial hypertension is higher in reproductive age females than in other demographic groups, failing to limit the comparator group to reproductive age females can artificially inflate the ROR for Mirena, as almost all Mirena users are reproductive age females. Therefore, a proper analysis would be limited to women of reproductive age.

Id. ¶ 6. And, Dr. Etminan stated, the study's DPA analysis might separately be flawed in that:

I do not know if my analyst used unique cases, or instead used unique reports, for the FAERS DPA. Because the same individual case can have multiple unique reports filed in the FAERS database, using unique reports can result in individual cases being counted more than once in the analysis. The proper analysis should use unique cases, not unique reports.

Id. ¶ 7. In sum, Dr. Etminan stated, as to his study's DPA analysis:

I have no basis to say that the results of the FAERS DPA would have been statistically significant, or that the point estimate would have been greater than 2.0, had the control group been properly limited to reproductive age females and had the analysis properly used unique cases rather than unique reports. Indeed, rerunning DPA analysis of the FAERS data using OpenVigil 2.1 software and properly limiting it to unique cases and women of reproductive age results in no elevated ROR for Mirena, suggesting that intracranial hypertension and Mirena use are "likely not related."

Id. ¶ 8 (emphasis added).

As for his retrospective cohort study, Dr. Etminan's affidavit noted that it had not found any statistically significant difference in risk between Mirena and either of the combination oral contraceptive comparators. Id. ¶ 9. Dr. Etminan further explained, "For neither the FAERS DPA nor the retrospective cohort analysis did I have weight data (BMI or recent weight gain) that would have allowed me to control for weight as a potential confounding variable as the FAERS data do not generally provide BMI information for each case." Id. ¶ 10.

Dr. Etminan summed up the implications of his revelations as follows.

Based on the above, as the lead author of this article, I acknowledge that neither of the analyses in the article provide evidence that Mirena use increases the risk for intracranial hypertension . Therefore, there is no basis to say, based on these analyses, that the risk of intracranial hypertension with Mirena *233 use outweighs the risk of unplanned pregnancies.

Id. ¶ 11 (emphasis added). 15

2017-Dr. Etminan's letter repudiating his findings : In April 2017, Dr. Etminan submitted another letter to the editor. See Mahyar Etminan, Revised disproportionality analysis of Mirena and benign intracranial hypertension , 8 Therapeutic Advances in Drug Safety 299, 299-300 (2017) (Dkt. 167-30) ("Etminan Second Letter"). The letter largely reprised the points Dr. Etminan had made in his affidavit. It acknowledged that both of his 2015 analyses had failed to "control[ ] for reproductive age[,] which has a strong correlation with [IIH]." Id. at 1. In this 2017 letter, Dr. Etminan added: "[A]n unstratified analysis by reproductive age can artificially overestimate the risk of [IIH] with Mirena." Id. at 1 (illustrating the point with data). And he further conceded, "Based on the age-restricted analysis a signal is not detected with Mirena with respect to [IIH] when child bearing age is accounted for." Id. (emphasis added).

(ii) The Valenzuela study

The Valenzuela study was published in 2017. See Reuben M. Valenzuela, et al., An estimation of the risk of pseudotumor cerebri among users of the levonorgestrelintrauterine device , 41 Neuro-Ophthalmology 192 (2017) (Dkt. 167-64) ("Valenzuela" or the "Valenzuela study"). A retrospective case-control study, it addressed the risk of IIH among certain patients in Utah and Denmark. 16

In particular, the study examined whether IIH patients in these populations were using LNG-releasing IUDs (principally Mirena ), 17 whether the use of such IUDs was associated with an increased risk of IIH, and whether IIH patients who used such IUDs had signs or symptoms different from those observed in IIH patients who did not use such IUDs. Id. at 1-2. Towards this end, the study compared the incidence of IIH among reproductive age women who used LNG-releasing IUDs with the incidence of IIH among reproductive age women who were not using LNG-releasing IUDs. Id. at 2.

The Valenzuela study found a statistically significant correlation between a patient's use of an LNG-releasing IUD and the patient's having IIH. However, the authors emphasized that they had not found causation of IIH by use of an IUD, but merely a correlation between the two:

*234 Our investigation does not indicate that an LNG-IU[D] [such as Mirena ] can cause PTC, and the number of women with an LNG-IU[D] was too small to determine if an LNG-IU[D] is an independent risk factor for PTC. Although use of an LNGIU[D] seems [to] be associated with an increased risk of PTC, it is possible that this observation occurred because use of an LNG-IU[D] is also associated with other established risk factors that are known to be associated with PTC (e.g., obesity and recent weight gain). This analysis was also limited by the lack of temporal data to confirm that exposure to LNG-IU[D] occurred prior to PTC symptom onset or diagnosis.

Id. at 4 (emphasis in original).

The Valenzuela study noted two possible explanations for the correlation between IIH and use of LNG-based IUDs. Id. at 5. One is that LNG causes increased intracranial pressure, through an as-yet undetermined biological mechanism. Id. at 5. 18 The other

is that LNG does not cause increased intracranial hypertension, but that the PTC is more likely to occur in the same population of women who are more likely to have an LNG-IU[D] recommended to them by their physician. LNG-IU[D] is often, although not exclusively, recommended for women who may have difficulty with other forms of contraception. For instance, women with obesity, headache, and/or polycystic ovarian syndrome are more likely to be intolerant to oral contraceptives. For this group of women, LNG-IU[D] may be better tolerated as a form of contraception. This same group of women, with obesity, headache, and polycystic ovarian syndrome, are also more likely to develop PTC. When interpreting the findings presented here, it is also important to consider that the risk analysis does not account for potential confounders.

Id. (footnotes omitted).

The Valenzuela study further noted that future research "may or may not be able to distinguish between these two possibilities." Id. As to possible future areas of research, the study noted that there are currently "no reliable animal models of PTC," and that a prospective trial of LNG-IUDs in a population of women at risk for PTC "would likely be too costly and would not settle the question of whether an LNG actually causes [IIH]." Id. The authors added that "[l]imitations of our study include the retrospective nature of the investigation and the relatively small number of PTC subjects with an LNG-IU[D] at the time of diagnosis." Id. The authors emphasized that "[o]ur findings are preliminary, and caution should be exercised in applying this information to clinical practice." Id.

The Valenzuela study concluded: "We do not recommend the removal of LNG-IU[D]s from women with PTC, as the benefit of effective contraception for these women likely outweighs the risk. Likewise, if a woman with PTC or at risk for PTC needs contraception, an LNG-IU[D]

*235 should still be considered as an effective form of contraception." Id. at 6. 19

c. Case and Adverse Event Reports

Outside of this litigation, two published case reports have addressed IIH in the context of usage of LNG-based IUDs.

One 2010 report from a doctor discussed the case of a 45-year-old woman and noted a possible connection between an LNG-based IUD and IIH. See H. Martinez, et al., Atypical pseudotumour cerebri , 34 Neuro-Ophthalmology 255 (2010) (abstract); but see Cestari Rpt. at 14 (questioning whether this woman, who did not have headaches or papilledema, met the Dandy criteria for IIH and noting that the case report did not state whether the woman was overweight or had had recent weight gain or whether her IUD use predated the IIH symptoms).

A second case report, published in 2017, described an overweight woman who had developed IIH approximately two years after starting use of Mirena. It noted that the woman, soon before developing symptoms, had been prescribed a medicine, minocycline, that is associated with IIH, and that the IIH symptoms had abated shortly after the minocycline use (although not the woman's Mirena use) had ceased. See F.J. Ros Forteza, et al., Minocycline-induced intracranial hypertension in a patient with a levonorgestrelintrauterine device , Neurologia (2017) (letter to the editor) (in Spanish).

As to adverse event reports, Bayer has periodically conducted "signal analyses" regarding Mirena and IIH, utilizing the adverse events reports in its pharmacovigilance database.

The most recent, and therefore the most comprehensive, was in October 2017 during discovery in this case. Bayer summarized the findings of this analysis in a submission to a German regulator, BfArM. Bayer identified a total of 315 reported cases involving symptoms indicative of possible IIH symptoms among users of Mirena, and an additional four cases indicative of possible IIH symptoms involving Bayer's more recent Skyla IUD product, which is also LNG-based. See Bayer BfArM Response at 44-45. Of those patients for whom weight data is available, Bayer reported, most were obese or overweight. See Cestari Rpt. at 20. More than 60% of the 315 reported cases were in the form of lawsuits filed after December 2013. See Bayer BfArM Response at 17, 45-46. As developed later, various of plaintiffs' experts draw upon case reports within this *236 set in support of their conclusions. See, e.g. , Darney Rpt. at 24; Fraunfelder Rpt. at 17-24; Moyé Rpt. at 34-37; Plunkett Rpt. at 28, 34; Salpietro Rpt. at 30; Wheeler Rpt. at 39.

3. Studies of Other LNG-Based Contraceptive Devices

The Court next summarizes the existing studies and data bearing on the relationship between other LNG-based contraceptives and IIH.

a. Combined Oral Contraceptives

Between 1984 and 1993, five epidemiological studies into IIH were conducted that considered, among other issues, whether there was a link between combined oral contraceptives (those containing both estrogen and progestin) and IIH. 20 LNG-containing oral contraceptives result in a substantially higher rate of LNG circulating in the user's blood stream than does Mirena. See Plunkett Rpt. at 22 and Plunkett Dep. at 94 (noting that LNG-containing oral contraceptives have LNG levels between 3,000-6,000 pg/ml, some 20-30 times more than in Mirena).

All five studies failed to find a causal link. 21 Two of plaintiffs' experts, Drs. Fraunfelder and Darney, in fact, have acknowledged that science has largely disproven a link between IIH and combined oral contraceptives containing LNG. See Fraunfelder Dep. at 16 (agreeing that "an association between oral contraceptives and IIH has been largely disproven"); Darney Dep. at 43 ("My opinion is that [oral contraceptives] do not [cause IIH]."). Bayer's experts similarly assess the state of such research. See, e.g. , Cestari Rpt. at 11 ("This lack of an association has been consistently confirmed ....")

*237 b. Norplant

Norplant, a contraceptive product manufactured by American Home Products and its subsidiary, Wyeth-Ayerst Laboratories, was approved by the FDA in 1990 and introduced in 1991. Norplant released LNG from an implant placed in the woman's arm. Norplant marketing was discontinued in the United States in 2002 and globally in 2008, although a different implant that releases LNG, Jadelle, is currently marketed by Bayer outside the United States. Norplant contained substantially more LNG than Mirena and produced substantially higher total blood concentrations of LNG. Compare 1997 Norplant Label, Dkt. 167-69, at 00055905 (216 mg) with Mirena Label at § 11.1 (52 mg); compare 1997 Norplant Label, at 00055898 (327 ± 119 pg/mL after one year) with Mirena Label at § 12.3 (180 ± 66 pg/mL after one year).

As various plaintiffs' experts observe, see, e.g. , Darney Rpt. at 22; Fraunfelder Rpt. at 16; Plunkett Rpt. at 34, Salpietro Rpt. at 28-29, during the period Norplant was in use, several case studies noted an association or possible association between Norplant and IIH. See John B. Alder, et al., Levonorgestrel implants and intracranial hypertension , 332 New England J. Med. 1720, 1720-21 (1995) (Dkt. 167-16) ("Alder") (letter to editor, reporting the development of IIH in two women following the insertion of Norplant and identifying 56 additional possible cases of IIH; author, however, notes existence of confounding factors and adds, "[LNG] may have contributed to the onset of [IIH], or it may have had nothing to do with it"); A.J. Sunku, et al., Benign intracranial hypertension associated with levonorgestrel implants , 34 Annals Neurology 299, 299 (1993) (Dkt. 167-63) ("Sunku") (poster presentation, reporting two women who developed IIH while using Norplant; author concludes that further study is needed to determine whether relationship between Norplant and IIH is "etiological or coincidental"); see generally Diane E. Wysowski & Lanh Green, Serious adverse events in Norplant users reported to the Food and Drug Administration's MedWatch Spontaneous Reporting System , 85 Obstet Gynecology 538, 540 (1995) (Dkt. 167-67) ("Wysowski & Green").

Another study drawn from the FDA's adverse events database described 39 reports of women with symptoms associated with IIH that developed during Norplant use between February 2001 and December 2003. See Wysowski & Green, supra , at 540. Of the 39 women, the study noted, all for whom weight data was available were obese (16 women) or overweight (2 women). The study noted that the duration of time between the insert of the Norplant and the onset of IIH symptoms had ranged from one to 17 months; that four patients had continued using Norplant yet the symptoms resolved; that 16 patients had the Norplant removed and the symptoms resolved; and that six had the Norplant removed but their IIH symptoms continued. Id. The authors found the assembled data inconclusive. They stated, "[I]t is not possible to determine whether Norplant, obesity or weight gain, or both factors are related to the occurrence of [IIH] .... Epidemiologic research (case-control or cohort studies) would be required to determine if a causal association between Norplant and ... [PTC] exists." Id. at 541. No such studies of Norplant, however, were ever conducted. See Valenzuela, supra , at 4 ("[T]hese previous [Norplant] reports did not imply a causative role in the pathogenesis of PTC.").

Notwithstanding the absence of epidemiological studies, as a result of early adverse event reports of IIH, Norplant's *238 manufacturer in 1993 voluntarily placed the following language on Norplant's label:

There have been reports of idiopathic intracranial hypertension in NORPLANT SYSTEM users.... Patients with these symptoms, particularly obese patients or those with recent weight gain, should be screened for papilledema and, if present, the patient should be referred to a neurologist fur further diagnoses and care. NORPLANT SYSTEM should be removed from patients experiencing this disorder.

1997 Norplant Label (Dkt. 167-69) at 00055901. A similar warning as to IIH has been-and remains-included on the label of Bayer's Jadelle product, an LNG-based implant that is not currently marketed in the United States. See Fraunfelder Rpt. at 15; Plunkett Rpt. at 9 n.7.

c. Other Contraceptive Devices Containing LNG

Finally, apart from the contraceptives reviewed above-Mirena, combined oral contraceptives, and Norplant and Jadelle-other contraceptive devices have had LNG as their hormonal component. Those referenced in the expert reports include: (1) Kyleena and Skyla, both of which are IUDs manufactured by Bayer (Skyla, approved in 2013, has an average release rate of 6 µg LNG/day for up to three years, while Kyleena, approved in 2016, has an average in vivo release rate of 9 µg LNG/day for up to 5 years); and (2) Allergan's Liletta, an IUD releasing 19.5 µg LNG/day for up to three years. Darney Rpt. at 7; Plunkett Rpt. at 9 n.8, 23-24; Wheeler Rpt. at 2. The parties have not drawn to the Court's attention studies or case reports bearing on a possible connection between these products and IIH.

III. Bayer's Challenges to Plaintiffs' Expert Witnesses

The state of research outside of this litigation as to the general causation proposition here-that using the Mirena IUD can cause a woman to develop IIH-presents a challenge for an expert witness here who would so testify.

As the above review reflects, to date, no prospective experiments have been undertaken that sought to address that question. Two epidemiological studies have examined that question but neither has found such causation. One such study, Etminan, has been retracted to the extent that it-based on its DPA analysis of reports in the FAERS database-had initially found an increased IIH risk among Mirena users. And the surviving half of the Etminan study (which compared Mirena with oral contraceptives) did not find any such increased risk. The other epidemiological study (Valenzuela) found a correlation between Mirena use and IIH. But it found only that. In language that warned against conflating correlation with causation, the Valenzuela study emphasized that its finding of such a correlation "does not indicate" that an LNG-based IUD such as Mirena is an "independent risk factor" for IIH. Rather, as Valenzuela recognized, alternative explanations for the correlation between Mirena and IIH are apparent-notably, the confounding factors of overweightness and obesity among reproductive-age women. As to the other contraceptive products using LNG, five studies of combined oral contraceptives have affirmatively found that these products, which contain notably higher amounts of LNG than Mirena, do not cause IIH. And no study has established a causal link between IIH and Norplant, which also contained substantially more LNG than Mirena.

In the face of this assembled historical record, with no medical organization or regulator or peer-reviewed scientific literature *239 having found that Mirena or any contraceptive product using LNG is a cause of IIH, an expert witness who would so opine as to Mirena necessarily would break new ground in this litigation.

Each of plaintiffs' seven expert witnesses reach this conclusion in their reports. None has done so through an experiment, laboratory work, or a new epidemiological study of his or her own. Four of plaintiffs' expert witnesses (Drs. Moyé, Plunkett, Wheeler, and Fraunfelder) arrive at this result largely by drawing upon existing sources. These include the Valenzuela study plus, depending on the witness, some or all of the following: the repudiated portion of the Etminan study; case reports regarding Mirena; case reports regarding Norplant and other subdural implants; and Norplant's warning label. These experts also draw upon a newly available source of data: the case reports regarding Mirena first added in Bayer's 2017 signal investigation, which was made available to plaintiffs in discovery. To varying degrees, each of these four experts also articulates a theory as to a biological mechanism by which Mirena might cause IIH.

Plaintiffs' remaining three expert witnesses (Drs. Darney, Johanson, and Salpietro) are predominantly "mechanism" experts. Their reports each develop a thesis as to how, biologically, use of Mirena may cause IIH.

Bayer, for its part, argues that plaintiffs' experts' proposed testimony is unreliable, measured against the requirements for admission of expert testimony. As to the four experts who rely largely on existing studies and data, Bayer contends, inter alia , that the alchemy is elusive by which a witness can reliably find such causation where the studies that comprise the expert's central source material have not so found. As to the three "mechanism" experts, Bayer contends that their untested theories are based on conjecture, and often on mishandling the scholarship on which they postulate causal mechanisms. Overall, Bayer contends that, rather than reflecting rigorous application to scientific methodologies, the conclusions of each of the seven witnesses amounts to a non-scientific ipse dixit .

The Court's analysis proceeds as follows.

The Court first reviews the legal standards governing receipt of expert testimony as set out in Daubert and ensuing decisions.

The Court then summarizes each witness' proposed testimony and assesses it against these standards. The Court first considers the four witnesses who reach the conclusion that Mirena is a cause of IIH based largely on existing studies and data. These witnesses describe the multi-factor methodologies they use in assessing this evidence as, alternatively, a "Bradford Hill" and/or a "totality of the circumstances" methodology.

The Court then considers the remaining three witnesses, whose reports, as noted, largely articulate theories about biological mechanisms by which Mirena might cause IIH.

A. Applicable Legal Standards

Trial courts serve as "gatekeep[ers]," responsible for "ensuring that an expert's testimony both rests on a reliable foundation and is relevant to the task at hand." Daubert , 509 U.S. at 597 , 113 S.Ct. 2786 ; see also Wills v. Amerada Hess Corp. , 379 F.3d 32 , 48 (2d Cir. 2004). Pursuant to Federal Rule of Evidence 702, the party seeking to admit expert testimony must show by a preponderance of the evidence that: (i) the expert is qualified, (ii) the testimony is based on sufficient data, *240 (iii) the testimony is the product of reliable methods reliably applied, and (iii) the testimony is relevant and will assist the jury.

In Daubert , the Supreme Court set out a list of non-exclusive factors that courts should consider in determining whether an expert's methodology is reliable. These are: (1) whether the expert's technique or theory can be or has been tested; (2) whether it has been subjected to peer review and publication; (3) whether there is a high error rate for the expert's technique, and whether there are "standards controlling the technique's operation"; and (4) whether the expert's technique or theory is generally accepted by the relevant scientific community. Daubert , 509 U.S. at 592 -94 , 113 S.Ct. 2786 ; accord Nimely v. City of New York , 414 F.3d 381 , 396 (2d Cir. 2005). Courts also consider whether the proffered expert opinions were developed for the purposes of litigation. In re Rezulin Prods. Liab. Litig. , 369 F.Supp.2d 398 , 420 (S.D.N.Y. 2005) (" In re Rezulin II ").

A proffered opinion may fail all four Daubert reliability factors and still be admitted. Before admitting proposed testimony in those circumstances, however, a court must "carefully scrutinize," pause, and take a "hard look" at the expert's methodology. Mirena Perforation/Daubert , 169 F.Supp.3d at 430 , 449 ; In re Methyl Tertiary Butyl Ether (MTBE) Prods. Liab. Litig. , 593 F.Supp.2d 549 , 564 (S.D.N.Y. 2008). "The flexible Daubert inquiry gives the district court the discretion needed to ensure that the courtroom door remains closed to junk science while admitting reliable expert testimony that will assist the trier of fact." Amorgianos v. Nat'l R.R. Passenger Corp. , 303 F.3d 256 , 267 (2d Cir. 2002).

Experts can fail to meet Rule 702 and Daubert 's standards for various reasons, relating to the expert's qualifications and/or methodology:

Qualifications : The witness simply may not be qualified to address the area in question. See Fed. R. Evid. 702 (providing that in order to testify as an expert under Rule 702, a witness must be "qualified as an expert by knowledge, skill, experience, training, or education."). "To determine whether a witness qualifies as an expert, courts compare the area in which the witness has superior knowledge, education, experience, or skill with the subject matter of the proffered testimony." United States v. Tin Yat Chin , 371 F.3d 31 , 40 (2d Cir. 2004).

In determining whether the witness has the relevant experience, courts consider factors including the degree to which that experience was developed for the litigation. See, e.g. , Mancuso v. Consol. Edison Co. of N.Y. , 967 F.Supp. 1437 , 1443 (S.D.N.Y. 1997) ("We cannot help but conclude that [the plaintiffs' expert] was not in fact an expert ... when he was hired by plaintiffs, but that he subsequently attempted, with dubious success, to qualify himself as such by a selective review of the relevant literature."); Prohaska v. Sofamor, S.N.C. , 138 F.Supp.2d 422 , 437 (W.D.N.Y. 2001) (criticizing "litigation-driven expertise" where expert "relied upon the plaintiff's attorney to provide him with the relevant scientific literature"). Although extensive experience can make up for an absence in specialized training, 22 if the witness does not possess superior knowledge, education, experience, or skill *241 in the relevant area, the Court must exclude his or her testimony. See, e.g., Mirena Perforation/Daubert , 169 F.Supp.3d at 439 (excluding witness with experience in engineering and biomaterials but without experience in hormonal contraception).

Methodology : A witness may also be excluded if his or her proposed methodology is not sufficiently rigorous. See Nimely , 414 F.3d at 396 ("[R]eliability within the meaning of Rule 702 requires a sufficiently rigorous analytical connection between that methodology and the expert's conclusions."). The following are among the principles that guide a court's assessment of reliability.

"To warrant admissibility, ... it is critical that an expert's analysis be reliable at every step." Amorgianos , 303 F.3d at 267 . "[N]othing in either Daubert or the Federal Rules of Evidence requires a district court to admit opinion evidence that is connected to existing data only by the ipse dixit of the expert." Gen. Elec. Co. v. Joiner , 522 U.S. 136 , 146, 118 S.Ct. 512 , 139 L.Ed.2d 508 (1997). "A court may conclude that there is simply too great an analytical gap between the data and the opinion proffered" to permit admission. Id.

Courts have found analytical gaps to be too great, for example, when a critical step in a prospective expert's reasoning is based on a highly dubious analogy. See, e.g., Mirena Perforation/Daubert , 169 F.Supp.3d at 439 ("Such a subjective comparison of muscle of a pig heart to a female uterus creates simply too great an analytical gap between the data and the opinion proffered to pass muster under Rule 702 and Daubert ." (quotation marks omitted) ); Shatkin v. McDonnell Douglas Corp. , 727 F.2d 202 , 208 (2d Cir. 1984) (rejecting expert methodology based on an "apples and oranges" comparison).

An expert is, further, expected to "employ[ ] in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field." Kumho Tire Co. v. Carmichael , 526 U.S. 137 , 152, 119 S.Ct. 1167 , 143 L.Ed.2d 238 (1999). "Expert testimony should be excluded if it is speculative or conjectural," Boucher v. U.S. Suzuki Motor Corp. , 73 F.3d 18 , 21 (2d Cir. 1996), or where the proffered opinion is "based on data, a methodology, or studies that are simply inadequate to support the conclusions reached," Amorgianos , 303 F.3d at 266 .

"[W]hen an expert relies on the studies of others, he must not exceed the limitations the authors themselves place on the study." In re Accutane Prods. Liab. , No. 8:04-MD-2523-T-30TBM, 2009 WL 2496444 , at *2 (M.D. Fla. Aug. 11, 2009), aff'd , 378 F. App'x 929 (11th Cir. 2010) ; Mirena Perforation/Daubert , 169 F.Supp.3d at 452 (same). 23

Opinions that assume a conclusion and "reverse-engineer[ ] a theory" to fit that conclusion are, similarly, inadmissible. Mirena Perforation/Daubert , 169 F.Supp.3d at 430 ; In re Gen. Motors LLC Ignition Switch Litig. , No. 14-CV-5810, 2017 WL 6729295 , at *8 (S.D.N.Y. Dec. 28, 2017) ; see also Faulkner v. Arista Records LLC , 46 F.Supp.3d 365 , 381 (S.D.N.Y. 2014) ("[M]ethodology ... aimed at achieving one result ... is unreliable, and ... must be excluded."); In re Zoloft (Sertraline Hydrochloride) Prods. Liab. Litig. , 858 F.3d 787 , 796-800 (3d Cir. 2017) (" In re Zoloft ") (affirming exclusion of "conclusion-driven" analysis).

*242 Finally, "an expert may not 'pick and choose' from the scientific landscape and present the Court with what he believes the final picture looks like." In re Rezulin Prod. Liab. Litig. , 309 F.Supp.2d 531 , 563 (S.D.N.Y. 2004) (" In re Rezulin I "). Where an expert ignores evidence that is highly relevant to his conclusion, contrary to his own stated methodology, exclusion of the expert's testimony is warranted. See id.

B. Dr. Lemuel Moyé

Dr. Moyé is a trained medical doctor with a Ph.D. in Community Sciences Biostatistics. He bases his opinion that Mirena is a cause of IIH on what he describes as an application of the Bradford Hill criteria.

1. Qualifications

Dr. Moyé works as a tenured professor of biostatistics within the Department of Biostatistics and Data Sciences at the University of Texas School of Public Health in Houston, Texas. He became a licensed physician in Texas in 1984 and worked as a general practitioner for eight years. See Moyé Rpt. at 2. Dr. Moyé does not, however, hold himself out as an expert in pharmacokinetics or pharmacodynamics, specialties associated with expertise in developing biological mechanisms for diseases. Moyé Dep. at 85-86.

Dr. Moyé has extensive experience in conducting large clinical trials and analyzing their results statistically. He has been a Principal Investigator in a large cardiovascular study, a Coordinating Center Principal Investigator in a study regarding the treatment of strokes, and is currently the Coordinating Center Principal Investigator for the Cardiovascular Cell Therapy Research Network. He has published numerous peer-reviewed articles about the studies he has conducted, as well as several books on the use of statistics in medicine. See Moyé Rpt. at 2-3.

Dr. Moyé has also been hired as an expert witness on numerous occasions. In the last four years, he has been deposed as an expert witness 14 times. Before this litigation, he did not have any experience related to Mirena, and he had limited exposure to IIH. Moyé Dep. at 85. Outside of this litigation, he has never conducted research on LNG. Id. at 85-88.

2. The Bradford Hill Criteria

The Bradford Hill criteria derive from a 1965 lecture by a British epidemiologist and statistician, Sir Austin Bradford Hill. See David E. Bernstein, The Admissibility of Scientific Evidence After Daubert v. Merrell Dow Pharmaceuticals, Inc. , 15 Cardozo L. Rev. 2139 , 2167 (1994) ("In a celebrated lecture in 1965, Sir Austin Bradford Hill proposed nine criteria to aid scientists in deciding whether a reported association in an epidemiological study is causal."). "The Bradford Hill criteria are metrics that epidemiologists use to distinguish a causal connection from a mere association." In re Zoloft , 858 F.3d at 795 . These criteria "start with an association demonstrated by epidemiology and then apply" eight or nine criteria to determine whether that association is causal. In re Breast Implant Litig. , 11 F.Supp.2d 1217 , 1234 (D. Colo. 1998).

The nine Bradford Hill criteria are as follows.

Statistical Association [alternatively referred to as "Strength of Association"]. There must be some degree of statistical association between a cause and its effect. A strong association (large in magnitude) is more likely to represent causation than a weak association (small in magnitude).

Temporality. A cause must precede its effect. Strength in temporality, such as when a cause immediately precedes its *243 effect, supports an inference of causation.

Biological Plausibility ... A cause and effect relationship between exposure and disease should be biologically plausible ... with other information about the disease or harm.

Coherence. A cause and effect relationship between exposure and disease should be ... consistent with other information about the disease or harm. 24

Dose-Response Effect. Causation is more likely if greater amounts of the putative cause are associated with corresponding increases in the occurrence of disease or harm.

Consistency. When similar findings are generated by several epidemiological studies involving various investigators, causation tends to be supported.

Analogy. Substantiation of relationships similar to the putative causal relationship increases the likelihood of causation.

Experimental Evidence. Causation is more likely if removing the exposure in a population results in a decrease in the occurrence of disease or harm.

Specificity. When there is but a single putative cause for the disease or harm, causation is supported.

Thompson, supra , at 268; see also In re Zoloft , 858 F.3d at 795 ; Moyé Rpt. at 16-20.

3. Proposed Testimony

Dr. Moyé principally relies on the following materials in applying the Bradford Hill criteria: (1) the Valenzuela study, (2) 36 case reports, drawn from among the 315 adverse event reports as to Mirena and IIH noted in Bayer's 2017 signal investigation, and Bayer's summary of that investigation in its BfArM submission; (3) a single published article involving Norplant, and (4) a review of literature based upon which Dr. Moyé develops a theory as to a "biologic[ally] plausib[le]" mechanism. Moyé Rpt. at 39; see id. at 29-43. He applies the nine Bradford Hill criteria, or factors, to that evidence as follows:

Strength of association : Dr. Moyé bases his finding of a "positive" strength of association, Moyé Rpt. at 34, on the Valenzuela cohort study. See id. at 29-34. As noted, that study reviewed populations in Utah and Denmark. Dr. Moyé concludes that that, as to both populations, the Valenzuela data revealed an association between Mirena and IIH. The relative risk ratio (a measure of the likelihood of IIH occurring in patients using Mirena compared to those not using Mirena ) for both populations was, he states, statistically significant. See id. at 31 (relative risk ratio for LNG-IUD group in Utah population was 7.69); id. at 32 (relative risk ratio for LNG-IUD group in Denmark population was 3.90). Dr. Moyé concedes that those ratios, being derived from Valenzuela, did not control for the confounding factors of age or weight. He terms that shortcoming "regrettable." Id. at 31. Nevertheless, Dr. Moyé concludes, because the relative risk ratios (particularly as to Utah) were substantial, "it would be unreasonable to assume that the large odds ratios observed in the Utah cohort would be completely adumbrated by adjustment for obesity and age." Id. at 31-32. The association between Mirena and IIH, he states, was "not likely to be overshadowed by any adjustment for confounders." Id. at 32.

In discussing this factor, Dr. Moyé also responds to Bayer's argument that the *244 Valenzuela study, independent of its not having controlled for confounding factors, was compromised by selection-bias flaws in its choice of populations. 25 Bayer has argued that the Valenzuela study ascertained Mirena usage differently between its IIH (case) groups and its non-IIH (control) groups, causing the study to undercount Mirena use in the control groups and to make Mirena use appear comparatively more common among IIH patients. Bayer Omnibus Br. at 16-17. 26 In his response as to the Utah population, Dr. Moyé does not address whether it was inappropriate for Valenzuela to include certain types of patients in the case but not the control group. Instead, he states that "there is no perfect process by which one can ascertain exposure status." Moyé Rpt. at 31. 27 In his response as to the Denmark population, Dr. Moyé does not address whether (as Bayer has claimed) it had been wrong not to consider the amount of time between the start of the patient's Mirena usage and the patient's manifestation of IIH symptoms. Instead, he responds to Bayer's separate argument that the correct proportion of women using Mirena in Denmark at the time is 10%, and not the 3.5% used by Valenzuela. Dr. Moyé questions the study on which Bayer's 10% calculation was based. 28 Id. at 33. Dr. Moyé ultimately concedes that, lacking the underlying data the studies used, "it is impossible to determine the relationship between the two and reconcile the apparent contradiction between the two papers ...." Id. 29 In the *245 end, Dr. Moyé states, "selection and misclassification biases would need to be overwhelming to reverse" his finding of an association between Mirena and IIH in the two populations. Id. at 34. He observes that "the consistency of the finding from two different populations is striking, even after taking into account that the results are provided by the same investigators." Id.

Temporality : Dr. Moyé finds the temporality factor met, based on case reports in Bayer's 2017 adverse events database. See Moyé Rpt. at 34-37. "Temporality requires that exposure occur before the disease," Dr. Moyé notes, and "[c]ases in the Bayer database demonstrate the occurrence of [IIH] after the insertion of the Mirena device." Id. at 34. Dr. Moyé identifies 13 cases in that database that he states showed that Mirena had likely caused the patient's IIH symptoms. Id. at 34-36. Each, he contends, reflected "a reasonable time relationship between Mirena exposure" and the patient's IIH; "it is unlikely," he states, "that the [IIH] would be attributed to other factors." Id. at 37. Dr. Moyé draws these 13 examples from the approximately 36 case reports that he reviewed from the Bayer database. Moyé Dep. at 231. It is unclear how the 36 cases were selected. Dr. Moyé appears to have reviewed them either in the order they were provided to him by counsel and/or by reviewing a chart of all case reports and deciding (based on criteria that are unclear) which to examine. Dr. Moyé did not review all of Bayer's adverse event reports regarding IIH. Id. at 229, 231.

Biological gradient : Dr. Moyé finds that "there is a biological gradient"- i.e. , that IIH occurs more frequently in Mirena users whose LNG levels are highest. See Moyé Rpt. at 39; id. at 37-39. He bases this finding on three sources: (1) the Valenzuela study, (2) Bayer's 2015 signal investigation, and (3) Bayer's 2017 BfArM response. 30 He explains that a "biologic gradient involves the presence or absence of a relationship between the exposure to the potential hazard and the occurrence of the disease." Id. at 37. Dr. Moyé finds such a dose-response effect because (1) in the Valenzuela study, the average duration of exposure prior to symptom onset was 22 months; (2) in Bayer's 2015 signal investigation and its 2017 response to BfArM, respectively, 77.8% and 62.9% of the cases of IIH (with time-to-onset information) occurred within the first two years of patients' Mirena exposure; and (3) during the first two years of usage, LNG levels are the highest and most variable. Id. at 39. However, Dr. Moyé concedes that he did not have the full underlying data for the time-to-onset discussed in the Valenzuela paper nor the median time to onset. Id. He therefore infers, based on the fact that the mean exposure was less than two years, that the majority of IIH cases addressed *246 there had occurred within the first two years of the patient's Mirena use. Id.

Biological plausibility : Dr. Moyé proposes a biological mechanism as to how LNG might plausibly cause IIH. He embraces an "androgen" theory: He posits that LNG mimics "the action of the mineralocorticoid aldosterone, thereby increasing sodium ion and water flow into the central spinal fluid, and subsequently increasing CSF pressure." Moyé Rpt. at 40. Dr. Moyé does not develop this theory. Instead, he recites nine propositions for which he stated there are biomedical sources. These, he states, when viewed in combination, support such a causal mechanism for Mirena. Id. Dr. Moyé includes a blanket citation to 40 academic articles. Id. at 40-41.

Coherence: Dr. Moyé finds this factor met, because the biological mechanism he endorses "relies on known science ... and is therefore in concordance with modern molecular biology." Id. at 41.

Consistency : Dr. Moyé finds this factor met, because the Valenzuela study had involved two different populations and had used different protocols to select patients (health data in Utah; census data in Denmark). Id.

Specificity : Dr. Moyé concedes that IIH has multiple causes. But, he states, alternative causes to Mirena"can either be excluded, statistically adjusted out of consideration, or accounted for in a multi-factorial causation model." Id. He does not state that this exercise has ever been performed anywhere.

Experimental evidence : In finding this factor met, id. at 41-42, Dr. Moyé relies on five case reports drawn from the 2017 Bayer database. He terms these "representative" examples of a "challenge-dechallenge"-that is, a situation in which a patient developed IIH while using Mirena, and then, after the Mirena had been removed, "either recovered or improved." Id. at 41. Dr. Moyé's examples, however, are not representative of the cases in Bayer's adverse events database, as described in Bayer's BfArM report. That report states that, of the 68 patients whose Mirena had been removed and for whom further information was available, 31 (or 46%) "did not recover," 23 (or 34%) had "improved," and 14 (or 20%) had "recovered." Bayer BfArM Response at 49. For an additional 128 patients, "no information on the further course of the disease was provided. Id. In contrast, of Dr. Moyé's five case reports, four patients (80%) had recovered, one (20%) had "improved," and none "did not recover." See Moyé Rpt. at 41-42. Pressed in his deposition on his claim that his five-case sample was "representative," Dr. Moyé stated that he had viewed his sample as representative not of the Bayer database, but of the subset of patients where the "phenomenon of dechallenge was observed"-that is, where the IIH symptoms had disappeared after a Mirena was removed. Moyé Dep. at 241.

Analogy : In finding this factor met, Moyé Rpt. at 42-43, Dr. Moyé analogizes to one published study of Norplant, by Wysowski and Green. He opines that it supported that Mirena causes IIH. However, as noted earlier, that study is not an epidemiological study. It discusses a collection of adverse event reports involving Norplant. Wysowski and Green concluded that, based on the evidence before them, it was "not possible" to determine whether "a causal association exists" between Norplant and IIH. Wysowski and Green, supra , at 541.

4. Analysis Under Daubert

Either explicitly or implicitly, Dr. Moyé finds that each of the nine Bradford Hill factors is met here. On that basis, he *247 opines that use of the Mirena IUD is a cause of IIH.

Of the four plaintiffs' experts who reached this conclusion by applying the Bradford Hill or on a similar totality-of-the-circumstances approach, Dr. Moyé's assessment of the constituent factors is, in the Court's assessment, the most thorough and substantial. However, on inspection, his analysis is flawed by serious methodological deficiencies. These include an unweighted and unmoored application of the nine Bradford Hill factors, a failure to consider known contrary evidence, a contravention of principles which Dr. Moyé has acknowledged should guide an epidemiologist's inquiry, a selective use of case report data, a lack of qualification to opine on biological mechanisms by which Mirena might cause IIH, and the citation of the Valenzuela study for propositions that it did not find. Under Daubert principles, these flaws, reviewed below, make Dr. Moyé's proposed testimony unreliable and inadmissible.

To begin, as appears undisputed, Dr. Moyé's opinion does not satisfy any of the four reliability factors identified in Daubert . He has not tested his theory. He has not subjected it to peer review or had it published. He has not identified an error rate for his application of the nine Bradford Hill factors. And vetting of a multi-factor inquiry to yield a numeric error rate appears realistically impossible, as there are no "standards controlling the technique's operation." Daubert , 509 U.S. at 594 , 113 S.Ct. 2786 . Finally, the theory he advances in this litigation has not been "generally accepted by the relevant scientific community." Id. at 584 , 113 S.Ct. 2786 . Quite the contrary: Outside of this litigation, there is a complete absence of scholarship opining that Mirena, or for that matter any LNG-based contraceptive, is a cause of IIH.

The Court therefore must take a "hard look" at Dr. Moyé's methodology. See Mirena Perforation/Daubert , 169 F.Supp.3d at 430 , 449 .

Such scrutiny is particularly warranted given Dr. Moyé's choice of methodology. As courts have recognized, it is imperative that experts who apply multi-criteria methodologies such as Bradford Hill or the "weight of the evidence" rigorously explain how they have weighted the criteria. Otherwise, such methodologies are virtually standardless and their applications to a particular problem can prove unacceptably manipulable. Rather than advancing the search for truth, these flexible methodologies may serve as vehicles to support a desired conclusion.

As the Third Circuit has put the point: "To ensure that the Bradford Hill/weight of the evidence criteria is truly a methodology, rather than a mere conclusion-oriented selection process ... there must be a scientific method of weighting that is used and explained." In re Zoloft , 858 F.3d at 796 (quotation marks omitted); Magistrini v. One Hour Martinizing Dry Cleaning , 180 F.Supp.2d 584 , 607 (D.N.J. 2002) (same), aff'd , 68 F. App'x 356 (3d Cir. 2003). And as the First Circuit has required, while the expert's bottom-line conclusion need not be independently supported by each of the nine Bradford Hill factors, 31 in analyzing the factors, separately and together, the expert must employ "the 'same level of intellectual rigor' that he employs in his academic work."

*248 Milward v. Acuity Specialty Prods. Grp., Inc. , 639 F.3d 11 , 26 (1st Cir. 2011) (quoting Kumho Tire , 526 U.S. at 152 , 119 S.Ct. 1167 ).

Accordingly, where experts have claimed to apply Bradford Hill, courts have insisted on a clear explication of the weighting assigned to the different criteria. They have also demanded that the expert's application of the individual criteria be performed with proper rigor. "[T]he specific techniques by which the weight of the evidence/Bradford Hill methodology is conducted must themselves be reliable according to the principles articulated in Daubert ." In re Zoloft , 858 F.3d at 796 ; see id. ("An expert can theoretically assign the most weight to only a few factors, or draw conclusions about one factor based on a particular combination of evidence. The specific way an expert conducts such an analysis must be reliable; 'all of the relevant evidence must be gathered, and the assessment or weighing of that evidence must not be arbitrary, but must itself be based on methods of science.' " (quoting Magistrini , 180 F.Supp.2d at 602 ) ). Insistence upon such rigor guards against the pitfall of which Judge Seibel warned in excluding expert testimony in the earlier Mirena litigation: "reverse-engineering a theory to fit the desired outcome." Mirena Perforation/Daubert , 169 F.Supp.3d at 430 ; see also Faulkner , 46 F.Supp.3d at 381 .

Measured against these standards, Dr. Moyé's report falls short.

At the most general level, his report does not explain the weight that he attaches to any of the Bradford Hill criteria or address the relationship among them. Instead, he finds that all nine criteria support a finding that Mirena causes IIH. Indeed, he nowhere concedes that any criterion even is only weakly supportive of a finding of causation. By leaving obscure the weight that he attaches to each of the nine Bradford Hill factors and the relationship among them, Dr. Moyé's approach effectively disables a finder of fact from critically evaluating his work.

To illustrate the point: A finder of fact might, for example, take issue with Dr. Moyé's assessment of the first factor-the gating factor of strength of association. Dr. Moyé finds a "strong association" based principally on the Utah component of the Valenzuela study. Moyé Rpt. at 31. Dr. Moyé's fellow expert, Dr. Wheeler, however, reaches a different conclusion-he concludes that, based on the Valenzuela and Etminan studies, "there is no association demonstrated." Wheeler Dep. 200. And the Valenzuela study on which Dr. Moyé bases his strength-of-association finding stopped well short of finding a strong association. The authors pointedly cautioned that "[a]lthough use of an LNG-IU[D] seems [to] be associated with an increased risk of PTC, it is possible that this observation occurred because use of an LNG-IU[D] is also associated with other risk factors that are known to be associated with PTC (e.g., obesity and recent weight gain)." Valenzuela, at 4; see also id. at 5 ("[I]t is also important to consider that the risk analysis does not account for potential confounders."). A finder of fact might therefore come to a conclusion other than that of Dr. Moyé as to strength of association, perhaps finding, for example, in between Drs. Moyé and Wheeler, that there is a statistical association, but only a weak one, between Mirena and IIH. Such a finder would then have to consider whether the other Bradford Hill factors, if now viewed in the context of only a weak statistical association, support a finding of causation. Dr. Moyé's failure to weigh or explain the relationship among the factors in his analysis, however, would disable such an inquiry.

*249 Alternatively, a finder of fact might take issue with Dr. Moyé's conclusions that other Bradford Hill factors support Mirena's causation of IIH. A finder might so conclude based on concerns about proper methodology (including along the lines discussed below). Or a finder might substantively disagree with Dr. Moyé as to particular factors. Such a finder would then need to assess whether to find general causation in the absence of certain factors. Dr. Moyé's failure to weight factors or explain the relationship among them would preclude this assessment, too. Does Dr. Moyé's conclusion of causation still stand if the Bradford Hill factor of analogy is not found to favor that result? Or the factor of specificity? Or the factor of consistency? Or all of the above-or yet other factors? Dr. Moyé's unscientific "black box" approach to Bradford Hill review almost entirely prevents the finder of fact, or other experts seeking to validate or check his work, from conducting a meaningful and informed review.

As to his assessment of individual Bradford Hill factors, Dr. Moyé's mode of analysis, too, departs repeatedly from reliable methodology. Four examples, among others, illustrate the point.

As to the Bradford Hill factor of analogy, this factor requires "substantiation of relationships similar to the putative causal relationship ...." Thompson, supra , at 268. Dr. Moyé elsewhere has acknowledged this requirement. 32 In this case, Dr. Moyé finds that this factor supports concluding that Mirena causes IIH based on an analogy he makes to Norplant. But this analogy is founded on an unestablished hypothesis about Norplant. As noted, a cause-and-effect relationship between Norplant and IIH has never been substantiated. No epidemiological study of the relationship, if any, between Norplant and IIH has ever been conducted. Dr. Moyé bases his conclusion that Norplant causes IIH, and hence can be the basis for an analogy, solely on case reports discussing Norplant as reviewed in the study by Wysowski and Green. But these Norplant case reports, as Dr. Moyé admitted in his deposition, were all subject to confounding factors such as obesity and weight gain. Moyé Dep. at 270-72. And the Wysowski and Green study, on which Dr. Moyé relies, clearly stated that this evidence falls well short of establishing a causal relationship between Norplant and IIH. See Wysowski and Green, supra , at 541; see also Reference Manual on Scientific Evidence at 218 (Dkt. 167-86) ("Anecdotal evidence usually amounts to reports that events of one kind are followed by events of another kind. Typically, the reports are not even sufficient to show association, because there is no comparison group."); McClain v. Metabolife Int'l, Inc. , 401 F.3d 1233 , 1254 (11th Cir. 2005) ("[C]ase reports raise questions; they do not answer them."). In his deposition, Dr. Moyé admitted that he is unable to opine that Norplant causes IIH, and that it is an open question whether Norplant does so. Moyé Dep. at 56. Dr. Moyé's analytic approach as to the analogy factor, based as it is on "the ipse dixit of the expert," Gen. Elec. Co. , 522 U.S. at 146 , 118 S.Ct. 512 -on an inaccurate account of the scholarship as to whether the analog product, Norplant, causes IIH-is therefore foundationally unsound.

As to the Bradford Hill factor of specificity, this factor inquires into the number of causes of a disease. As Dr. Moyé explained: "The greater the number of causes of a disease ( i.e. , the more multifactorial *250 the risk factors causing the disease), the more nonspecific the disease is, and the more difficult it is to demonstrate a new causal agent is involved in the production of the disease." Moyé Rpt. at 19. In finding the specificity factor satisfied, Dr. Moyé devoted two sentences to his discussion. In the first, he admits that "[IIH] has multiple causes." Id. at 41. In the second, he states that the alternative causes ( i.e. , all but Mirena) "can either be excluded, statistically adjusted out of consideration, or accounted for in a multi-factorial causation model." Id. Dr. Moyé's reliance on this conclusory statement as a basis for finding the specificity factor met, too, departs from rigorous methodology. Dr. Moyé does not cite any study in which any such statistical adjustment has actually been performed so as to isolate Mirena as a cause of IIH, as distinct from other factors. His conceptual point about the potential to exclude alternative causes is, at best, theoretical. And the one epidemiological study on which Dr. Moyé relies in his Bradford Hill analysis, Valenzuela, did not undertake any such statistical adjustment. On the contrary, it pointedly disclaimed any finding that Mirena causes IIH, on account of confounding risk factors prevalent among the dominant population of Mirena users. See Valenzuela at 5 ("When interpreting the findings presented here, it is also important to consider that the risk analysis does not account for potential confounders.").

As to the Bradford Hill factor of consistency, it tends to require "similar findings ... generated by several epidemiological studies involving various investigators." See Thompson, supra , at 268; see also Reference Manual on Scientific Evidence at 604 ("It is important that a study be replicated in different populations and by different investigators before a causal relationship is accepted by epidemiologists and other scientists."). Dr. Moyé opines that the Valenzuela study "satisfies the consistency metric," because Valenzuela considered two separate populations. Moyé Rpt. at 41. These studies, however, were conducted by the same investigators. 33 More problematic as a matter of methodology, neither cohort study in Valenzuela found what was "important"-"a causal relationship." Reference Manual on Scientific Evidence at 604. Even accepting Valenzuela as containing "similar findings" from multiple epidemiological studies, Dr. Moyé's finding that the consistency metric was satisfied ignores the very limited point that the studies addressed and on which they were therefore consistent. The studies were consistent only as to the fact of a correlation-subject, as Valenzuela repeatedly pointed out, to potential confounders-between Mirena and IIH. They explicitly disclaimed a finding of causality.

Finally, as to the Bradford Hill factor of biologic plausibility, Dr. Moyé bases his finding that this factor was satisfied on his conclusion that a mineralocorticoid (MR) theory by which Mirena causes IIH is plausible. He posits that LNG mimics "the action of the mineralocorticoid aldosterone, thereby increasing sodium ion and water flow into the cerebral spinal fluid, and subsequently increasing CSF pressure." Moyé Rpt. at 40; see Moyé Dep. 249 ("Q: As I understand what you have written in your report you rely on the MR theory; is that correct?" A: Correct. Q: Not the androgen receptor theory? A: [T]hat is correct.").

*251 But, as Bayer rightly argues, Dr. Moyé-a medical doctor who doubles as a professor of biostatistics-is not qualified to opine about a biological mechanism by which Mirena might cause IIH. Dr. Moyé is not educated in the relevant disciplines: pharmacokinetics and pharmacodynamics. Moyé Dep. 85-86 ("Q: Do you hold yourself out as a pharmacokinetics expert? A: No. Q: Do you hold yourself out as a pharmacodynamics expert? A: No."). He is not trained in gynecology, either. And plaintiffs have not come forward with evidence that Dr. Moyé has acquired sufficient practical experience in these areas to make up for his lack of relevant education or training. Before this litigation, Dr. Moyé had not conducted research on LNG generally or on its hormonal impact on MR receptors. Id. at 85, 262. Dr. Moyé, instead, provides a blanket citation to 40 academic articles, none of which address Mirena, to support his finding of biological plausibility based on an MR mechanism.

Simply stated, by any measure, Dr. Moyé is unqualified to give an expert opinion as to a biological mechanism of causation of IIH. His lack of qualifications to so opine makes unreliable not only Dr. Moyé's assessment as to biological plausibility-that the MR mechanism is a biologically plausible one. It also compromises his assessment of the companion Bradford Hill factor of coherence, because Dr. Moyé finds that factor met explicitly based on his assessment that the mechanism he had endorsed is a coherent one. See Moyé Rpt. at 41 ("The mechanism of 1) increased egress of Na+ ion and water flow into the CSF from the choroid plexus producing CSF, and 2) decreased flow [ ]either by decreased activity of carbonic anhydrase, or reversal of the effect of steroid hormones that mimic aldosterone in the MR can reduce the symptoms of [IIH] does not contradict known science.")

Common to Dr. Moyé's misapplications of the above Bradford Hill factors is this: Each of Dr. Moyé's departures from settled and rigorous methodology favors the same outcome. Each enables him to find that the Bradford Hill factor at issue supports concluding that Mirena is a cause of IIH. At bottom, as the Supreme Court instructed in Daubert , expert testimony "must be supported by appropriate validation- i.e. , good grounds, based on what is known." Daubert , 509 U.S. at 590 , 113 S.Ct. 2786 (internal quotation marks omitted). Dr. Moyé's unidirectional misapplication of a series of Bradford Hill criteria is concerning-it is a red flag. Rather than suggesting a scholar's considered neutral engagement with the general causation question at hand, it suggests motivated, result-driven, reasoning. See Faulkner , 46 F.Supp.3d at 381 ("[M]ethodology ... aimed at achieving one result ... is unreliable.").

Yet other methodological lapses also preclude, as unreliable under Daubert , Dr. Moyé's proposed testimony.

For one, Dr. Moyé ignores scientific standards that he has conceded govern inquiries into general causation. In his deposition, Dr. Moyé agreed, for example, with certain widely accepted principles of medical research. One is that case reports do no more than raise the question of a causal connection between a drug and a disease; they cannot establish causation. Moyé Dep. at 272. A related principle that he acknowledged is that, to prove general causation, observational studies, such as a case control study or a cohort study, generally are required. See id. at 125 (agreeing that a single observational study "[r]arely, if ever ... persuasively demonstrate[s] a cause-effect relationship"); id. (agreeing that "it is important that an observational study be replicated in different populations and by different investigators *252 before a causal relationship is accepted").

Dr. Moyé's report is unfaithful to these precepts. Dr. Moyé finds causation of IIH by Mirena in the absence of any of the sorts of studies that he concedes generally are required. As to studies, he relies solely on the Valenzuela study, a retrospective epidemiological study which stopped well short of finding causation, recognizing that it had not controlled for major confounding variables such as obesity and weight gain that are prevalent both among IIH patients and the population of Mirena users. In his deposition, Dr. Moyé admitted Valenzuela's limitations. See, e.g. , Moyé Dep. at 215 ("Q: Would you conclude causation with just the results of Valenzuela standing alone? A: No, sir, I would not."); id. at 216 ("I think an appropriate statement would be to say that it's possible that LNG-IU[D] causes PTC, but this study doesn't show it."). And Dr. Moyé also appears to rely on case reports for more than merely raising a question about whether a causal relationship between Mirena and IIH existed. In his report, Dr. Moyé uses case reports (and a study aggregating such reports) almost exclusively to establish three of the nine Bradford Hill factors: experimental evidence, temporality, and analogy. See Amorgianos , 303 F.3d at 269 (affirming exclusion of proffered expert who "fail[ed] to apply his stated methodology reliably to the facts of the case") (internal quotation marks omitted); Soldo v. Sandoz Pharm. Corp. , 244 F.Supp.2d 434 , 561 (W.D. Pa. 2003) (excluding experts "[b]ecause consistency is a hallmark of the scientific method [and] plaintiff's experts must be required to satisfy their own standards of reliability"); cf. Kumho Tire Co. , 526 U.S. at 152 , 119 S.Ct. 1167 (expert is expected to "employ[ ] in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field").

Dr. Moyé also chooses not to consider evidence that undercuts his opinion-potentially very materially. This is a separate methodological failing. It is most strikingly apparent in Dr. Moyé's consideration of the analogy factor. Dr. Moyé elects to analogize to Norplant. As to Norplant, he draws on isolated case reports canvassed in a study (Wysowski and Green) that disclaimed a finding of causation. He also analogizes to case reports in which, he states, a relationship was indicated between IIH and "elevated testosterone levels." Moyé at 42-43. As to this factor, he thus is willing to range outside the area of LNG-containing intra-uterine devices, considering case reports as to other products and other hormones.

In so doing, however, Dr. Moyé, tellingly, does not consider an analogy between Mirena and other LNG-containing contraceptive drugs: combined oral contraceptives containing LNG. Consideration of those products stood to seriously undermine Dr. Moyé's conclusion. As noted, five epidemiological studies have uniformly failed to demonstrate a link between such oral contraceptives and IIH, even though LNG-containing oral contraceptives generally contain far higher doses of LNG than Mirena, and IIH. Whatever the scientific distinctions that might have been made between oral and intra-uterine delivery systems for LNG, these studies demanded Dr. Moyé's consideration. Yet Dr. Moyé does not review them. Moyé Dep. at 51-52.

Dr. Moyé's decision to brush aside these epidemiological studies, while analogizing to a different medical device and a different hormone based exclusively on case reports, departs from reliable, rigorous methodology. "[A]n expert may not pick and choose from the scientific landscape." In re Rezulin Prod. Liab. Litig. , 309 F.Supp.2d at 563 (internal quotation marks omitted); see also *253 MTX Commc'ns Corp. v. LDDS/WorldCom, Inc. , 132 F.Supp.2d 289 , 293 (S.D.N.Y. 2001) (excluding expert who omitted "major variables" from his analysis); In re Bextra & Celebrex Mktg. Sales Practices & Prod. Liab. Litig. , 524 F.Supp.2d 1166 , 1176 (N.D. Cal. 2007) (" In re Bextra ") ("[R]ejecting or ignoring" unfavorable evidence "is not 'good science.' "). 34

Finally, Dr. Moyé's sampling of case reports-which played a central role in his report at various points-is performed in a manner not bespeaking scientific rigor or neutrality. Rather than selecting case reports through random sampling or by a representative cross section, Dr. Moyé appears to draw the case reports that he considers from those presented to him by counsel or from a spreadsheet whose design is elusive. This skewed methodology has obvious potential to yield a skewed sample and result. Relatedly, Dr. Moyé's five chosen "dechallenge" cases are unrepresentative of the much larger sample covered by the BfArM report; and the examples he chooses tended to favor his bottom-line causation conclusion. 35

Dr. Moyé's failure to apply neutral selection methodology as among such anecdotal evidence is unscientific. Viewed in combination with the other deficiencies noted, this failing, too, supports excluding his testimony. See U.S. Info. Sys., Inc. v. Int'l Bhd. of Elec. Workers Local Union No. 3, AFL-CIO , 313 F.Supp.2d 213 , 235 (S.D.N.Y. 2004) (excluding expert who relied on biased sample of data); Rowe Entm't, Inc. v. William Morris Agency, Inc. , No. 98 Civ. 8272 (RPP), 2003 WL 22124991 , at *3 (S.D.N.Y. Sept. 15, 2003) (same); but see Auto. Ins. Co. of Hartford, Conn. v. Electrolux Home Prods., Inc. , No. 10-CV-0011 CS, 2012 WL 6629238 , at *2 (S.D.N.Y. Dec. 20, 2012) ("Potential sample bias is a subject for cross-examination, and goes to the weight, not the admissibility, of the expert testimony.")

In conclusion, Dr. Moyé's proposed testimony is compromised by a range of serious methodological flaws. It fails to meet the standard for reliability required by Daubert . The Court must exclude it.

B. Dr. Laura Plunkett

1. Qualifications

Dr. Plunkett is a pharmacologist and toxicologist with more than 20 years' experience *254 in those fields. She received her B.S. in 1980 from the University of Georgia and a Ph.D. in pharmacology in 1984 from the University of Georgia, College of Pharmacy. Plunkett Rpt. at 1. She has worked as a Pharmacology Research Associate Training (PRAT) fellow at the National Institute of General Medical Sciences (approximately two years), an Assistant Professor of Pharmacology and Toxicology at the University of Arkansas for Medical Sciences (almost three years), and a consultant for ENVIRON Corporation (almost eight years). Id. at 1-2. Dr. Plunkett has also served as a consultant for Integrative Biostrategies and at Plunkett & Associates. Id. Her professional career has consisted predominantly of working as a paid consultant, including substantial service as a compensated expert. See Plunkett Dep. at 351 (expert work made up about 55% of her income in 2016). In the last five years she has testified approximately 65 times. See Plunkett Rpt. at App'x C.

Dr. Plunkett has experience "examining the risks associated with exposure to hormones and the risks associated with altered hormonal status in women." Id. at 2. However, she is not a medical doctor. She has neither published on IIH nor spoken about the disease to doctors who treat it. Plunkett Dep. at 111-12.

2. Proposed Testimony

Dr. Plunkett's proposed testimony can be divided into two parts: (1) a discussion of the pharmacokinetics of LNG-releasing contraceptives, and (2) an application of the Bradford Hill methodology, which she alternatively describes as a "weight of the evidence" analysis, with the conclusion that Mirena causes IIH.

a. The Pharmacokinetics of LNG

Dr. Plunkett's discussion of the pharmacokinetics of LNG informs her application of the Bradford Hill methodology. Because Bayer either concedes or does not substantially contest most of this discussion, the Court highlights only Dr. Plunkett's major points.

First, Dr. Plunkett discusses the difference between "free" LNG and total LNG. She asserts that free LNG levels should be used to access a patient's exposure and response to the hormone. The vast majority of LNG in the blood stream is not "free" floating. Instead, she states, upwards of 98% of LNG is bound to plasma proteins (specifically albumin and a carrier protein called steroid hormone binding globulin). However, she states, LNG's effects are not generally created by such bound LNG. Rather, they are created by "free" LNG activating nuclear hormone receptors. Thus: "It is the free or unbound LNG levels that correlate best with the biological responses observed in humans, not the total level of LNG in blood." Plunkett Rpt. at 7.

Dr. Plunkett also states that, although free LNG levels for Mirena are lower than the levels associated with other LNG-releasing contraceptive devices (such as Norplant), the LNG levels in the blood of patients using such devices overlap with the levels observed in Mirena patients. This overlap, Dr. Plunkett asserts, justifies analogizing Mirena to Norplant and other LNG-containing contraceptive devices. Id.

Dr. Plunkett notes that LNG is a progestin, which mimics the effects of progesterone. She observes that, relative to other progestins, LNG is androgenic, meaning that it can bind with androgen receptors. However, Dr. Plunkett notes, discerning a mechanism to explain the pharmacological and toxicological effects of progestins like LNG is extremely complicated. Id. at 10-12.

b. Application of Bradford Hill

Dr. Plunkett applies the Bradford Hill factors, as follows.

*255 Strength of association : Dr. Plunkett frames her discussion of this threshold factor by quoting Sir Bradford Hill himself, to the effect that a scholar "must not be too ready to dismiss a cause and effect hypothesis merely on the grounds that the observed association appears to be slight," id. at 26 (citation omitted), and that "[n]o formal tests of significance can answer" whether there is a "cause and effect relationship," id. Dr. Plunkett then opines that three studies in the scientific literature "show[ ] that exposure to LNG from LNG intrauterine drug products has been associated with an increased risk of intracranial hypertension." Id. at 27. These were: (1) Etminan, (2) Valenzuela, and (3) Rai et al., The Relationship Between the Levonorgestrel-Releasing Intrauterine System and Idiopathic Intracranial Hypertension , ARVO (2015) ("Rai poster board presentation"). Beyond citing these studies, Dr. Plunkett does not discuss any of them in more than a sentence. As to the Etminan study, Dr. Plunkett acknowledges that, after Etminan's 2017 letter to the editor that factored in age-adjusted results, the results of his retrospective cohort study "were no longer statistically significant." Id. Nevertheless, she writes of the three studies:

These human studies are an important piece of the weight-of-the-evidence for LNG exposure with Mirena and PTC, as they have shown a statistically significant risk of PTC when women are exposed to LNG drug products, including intrauterine devices such as Mirena. I used these human studies as part of my weight-of-the-evidence assessment.

Id. Dr. Plunkett's report does not mention or contend with the fact that the Valenzuela study had explicitly disclaimed a finding of a causal connection between IIH and Mirena. Nor does Dr. Plunkett's report note that the Rai poster presentation was a preview of the Valenzuela data presented prior to its 2017 publication and thus was substantively the same study as Valenzuela. In her deposition, Dr. Plunkett conceded those facts. Plunkett Dep. at 136-37, 165-66.

Temporality : In finding this factor satisfied, Dr. Plunkett states that she relies on three studies which, she stated, had "addressed" temporality (by Valenzuela, Rai, and Alder) and on three case reports drawn from the Bayer database. Plunkett Rpt. at 28. Beyond identifying these materials, however, Dr. Plunkett does not address their contents. The Valenzuela authors, for instance, specifically noted in their publication that they lacked temporal data. See Valenzuela, supra , at 4 ("The analysis was also limited by the lack of temporal data to confirm that exposure to LNG-IU[D] occurred prior to PTC symptom onset or diagnosis."). And Rai, as noted, was a precursor to Valenzuela and based on the same data. The third publication she cites, by Alder, was a letter to the editor that discussed two cases of IIH which arose in Norplant users after the Norplant's implantation. The Alder authors disclaimed any causal findings: "[Norplant] may have contributed to the onset of [IIH], or it may have had nothing to do with it." Alder, supra , at 1721. Dr. Plunkett's report does not note this disclaimer. As to the three case reports Dr. Plunkett cites, two involved Mirena and one involved Skyla. 36 While the Court assumes arguendo that the patients described in these reports developed IIH symptoms after they began use of the LNG-based contraceptive product, Dr. *256 Plunkett's report does not so state. Id. at 28.

Biological gradient : Dr. Plunkett quotes Sir Bradford Hill in framing her discussion of this factor, in recognizing the "difficulty [of] secur[ing] some satisfactory" data to permit assessment of a dose-response effect. Id. In finding a possible such effect with respect to Mirena and IIH, Dr. Plunkett states that she relies on three sources: (1) a 1992 article (V. Brache, et al., Free Levonorgestrel Index and Its Relationship with Luteal Activity During Long-Term Use of Norplant Implants , 8 Soc'y for the Adv. of Contraception 319 (1992) ("Branche"); (2) Bayer's 2015 signal investigation; and (3) the Mirena 2017 signal assessment. The Branche article, according to Dr. Plunkett, had shown that in the first two years of LNG exposure, free LNG levels are higher than they otherwise would be. Plunkett Rpt. at 29. Dr. Plunkett finds that probative of a biological gradient, because approximately 75-80% of the IIH adverse events in Bayer's 2015 signal investigation had been reported during the first two years of Mirena use. Id. at 29. As for Bayer's 2017 signal assessment, Dr. Plunkett notes, it had compared the reporting frequency for IIH in Mirena and Skyla, which releases substantially less LNG than Mirena. The reporting frequency of IIH for each product was the same, which might seem to suggest, Dr. Plunkett acknowledges, the lack of a biological gradient. Id. at 30. However, Dr. Plunkett rejects that conclusion. She notes that there is "variability in LNG pharmacokinetics with both products, which would affect the levels of LNG in the blood achieved from patient to patient, and confound any attempt to identify a dose-response for LNG exposure and [IIH] in humans." Id. She also ventures that adverse events as to these drugs may have been significantly under-reported, a phenomenon particularly likely as to "drugs that may have been on the market for many years." Id. She ultimately concludes that "the available dose-response data provide support for the cause and effect assessment in terms of showing that increased levels of LNG in blood are more likely to be associated with adverse effects generally." Id. at 31. However, she acknowledges, "[u]nfortunately, no dose or blood level threshold that is associated with an increased risk of [IIH] has been identified to date." Id.

Biological plausibility : Dr. Plunkett embraces the "androgen theory" of how LNG might cause IIH. 37 She does not, however, articulate a proposed mechanism to support that theory, as had Dr. Moyé. Instead, she cites literature that she contends found a link between IIH and androgens. See id. at 31-32; Plunkett Dep. at 311-16. Although acknowledging that LNG is a progestin, not an androgen, she terms it an "androgenic progestin." Plunkett Rpt. at 32.

Coherence : Dr. Plunkett does not cite any sources as to this factor, which she addresses in a short paragraph. Instead, she refers to her discussion of the biological plausibility that Mirena causes IIH, in that IIH "has been linked with androgenic activity." Id. at 33. She also cites "human epidemiological data" as to Mirena and IIH, apparently although not explicitly referring to the Valenzuela and Etminan studies. "[T]he data reported" in those studies, she states, "are consistent with what is known about the etiology of [IIH]." Id.

*257 Consistency : Dr. Plunkett describes the consistency factor as turning on "whether the association being investigated has been seen repeatedly by different people, in different circumstances, and/or at different times." Id. at 27. She concludes that this factor is met, citing, although not further discussing, (1) the Valenzuela and Etminan studies, (2) several Norplant studies reviewed earlier (Wysowski & Green, supra ; Sunku, supra ; Alder et al., supra ; and R.A. Tang, et al., Pseudotumor Cerebri Associated with the Norplant Contraceptive Device , Investigative Ophthalmology & Visual Science 36 (1995), and (3) an analogy to case reports of IIH experienced by patients using Norplant. Id. In her deposition, Dr. Plunkett acknowledged that such case reports could not alone demonstrate causation. Plunkett Dep. 181-82 ("Absolutely. You need more than the case studies.").

Specificity : Dr. Plunkett states that this factor considers "whether the association is limited to specific types of activities or injuries." Id. at 27. Dr. Plunkett does not explicitly find this factor met, stating more obliquely that "considering specificity of the relationship has been part of my overall assessment." Id. at 28. She acknowledges that the scientific literature addressing the relationship between Mirena and IIH described "confounding factors ... such as obesity [and] female gender." Id. at 27-28.

Experimental evidence : Dr. Plunkett casts this factor as "relat[ing] to the ability to collect data in order to analyze the cause and effect relationship." Id. at 33. In finding it met, Dr. Plunkett relies on publications above regarding Norplant, and on Bayer's 2015 signal assessment data, the latter of which, she stated, contained examples of "dechallenge." She concedes, however, that animal or cell experiments are the evidence typically used in experiments used to find causation, and that none have been conducted regarding LNG and IIH. Plunkett Rpt. at 35.

Analogy : Dr. Plunkett analogizes Mirena to Norplant. She relies on the publications cited above regarding Norplant. As noted, however, these were not epidemiological studies-they instead discussed case reports-and none found causation by Norplant of IIH. Dr. Plunkett also cites several studies which discuss IIH's etiology and list LNG as associated with IIH. None, however, had found that LNG is causally linked to IIH. See id. at 34-35.

3. Analysis Under Daubert

Bayer argues that Dr. Plunkett's methodology in applying Bradford Hill is unreliable. Bayer is correct. The application of Bradford Hill in Dr. Plunkett's report has a number of methodological flaws. Some are common with those noted in connection with Dr. Moyé. Dr. Plunkett's report is also validly called out for independent lapses, as reflected in the Court's assessment, below, of her applications of certain Bradford Hill factors.

To begin, in common with Dr. Moyé, Dr. Plunkett's report does not meet any of the four Daubert reliability factors. She has not tested her theory. She has not subjected it to peer review or had it published. She has not identified an error rate for her technique, and, as with Dr. Moyé's application of the Bradford Hill methodology, there are no standards controlling its operation. Finally, Dr. Plunkett's theory that Mirena is a cause of IIH, far from achieving general acceptance, has not been accepted by any part of the scientific community outside of this litigation. A "hard look" is therefore warranted as to her analysis, too. Compounding these problems, Dr. Plunkett, in common with Dr. Moyé, does not explain the weights she places on the various Bradford Hill factors. Instead, she reviews each in isolation, and opines or *258 (where unable or willing expressly to so state) implies that each has been satisfied. And Dr. Plunkett's handling of individual factors is subject to grave methodological deficiencies. For all these reasons her application of Bradford Hill does not survive Daubert scrutiny.

The Court begins with Dr. Plunkett's reliance on the discredited portion of the Etminan study. That study is a centerpiece of her application of the first Bradford Hill factor. That factor requires a statistical, or strong, association between the cause under review and its asserted effect. 38 This factor is a necessary, or gating, factor for any Bradford Hill analysis to proceed, such that, if such a statistical association is not found, there is no charter to undertake a Bradford Hill analysis at all. See Federal Judicial Center, Reference Manual on Scientific Evidence, 599 n.141 (3d. ed. 2011) ("In a number of cases, experts attempted to use these guidelines to support the existence of causation in the absence of any epidemiologic studies finding an association.... There may be some logic to that effort, but it does not reflect accepted epidemiologic methodology."); Restatement (Third) Torts § 28 cmt. (c)(3) ("Even when epidemiologic studies find an association between a substance and a disease, further analysis is necessary before a causal conclusion can be drawn.... [I]f an association is found, epidemiologists use a number of factors (commonly known as the 'Hill guidelines') for evaluating whether that association is causal or spurious."); see also In re Lipitor (Atorvastatin Calcium) Mktg., Sales Practices & Prods. Liab. Litig. (No II) MDL 2502 , 892 F.3d 624 , 640 (4th Cir. 2018) (" In re Lipitor ") (noting that the Reference Manual on Scientific Evidence and case law require a demonstrated epidemiological association). 39

As reviewed earlier, Dr. Etminan retracted as unsound the disproportionality analysis in his study. He did so expressly in his 2016 affidavit and again functionally in his 2017 letter to the editor. As he explained in those writings, once the control group in the FAERS database was limited to reproductive age females, no elevated reporting odds ratio was observed for Mirena (statistically significant or otherwise). This, Dr. Etminan stated, "suggest[ed] that intracranial hypertension and Mirena use are 'likely not related.' " Etminan Affidavit ¶ 8 (emphasis added).

Notwithstanding these serial retractions, Dr. Plunkett relies on the Etminan study as a whole, including its DPA analysis, as a basis for her finding of a significant statistical association between Mirena usage and IIH. 40 Her report acknowledges Dr. Etminan's letter to the editor stating that the data he reviewed in fact was not statistically significant as to a relationship between Mirena and IIH. Plunkett Rpt. at 27. Her report, however, then proceeds as if Dr. Etminan's repudiation had not occurred. Pressed at her deposition to justify relying on Dr. Etminan's discarded findings, Plunkett vaguely stated that the Etminan study's earlier analysis could be considered as "part" of a "weight of the evidence"

*259 analysis and that it was "supportive of [her] overall opinions." See Plunkett Dep. 146-52; id. at 152. ("Q: So now having seen the Etminan affidavit, you're still relying on Etminan 2015 as support for your causation opinion, correct? A: I am relying on that paper as part of the weight of the evidence. Certainly it by itself is not sufficient to prove causation." (objection omitted) ); id. at 148 ("Q: Dr. Etminan states, 'Based on the above, as a lead author of this article, I acknowledge that neither of the analyses in the article provide evidence that Mirena use increases the risk for [IIH].' Do you disagree with that statement? A: Well, that's his opinion, so I can't agree with-disagree with his opinion. I do believe that there is data within this that is informative to weight of the evidence." (objection omitted) ); id at 443 ("It is part of my overall weight of the evidence."); id. at 445 ("I think it is supportive of my overall opinions."). But, beyond stating that Dr. Etminan's initial findings could be considered as part of an overall mix, she did not explain on the merits why the repudiated findings warranted rehabilitation. 41 In the Court's assessment, Dr. Plunkett's embrace of these author-repudiated findings is presumptively unsound methodology. See Moyé Dep. at 283-84 ("Q: Now, as a matter of epidemiology, would it be appropriate to rely on any part of a paper that was functionally retracted? A: I don't believe so."). It is reasonable to expect of an expert who relies upon a study repudiated by its own author to justify coherently this improbable approach.

Dr. Plunkett did not do so. Defending this component of her analysis in her deposition, Dr. Plunkett embraced anew the results of Etminan's study. She testified that the 1.85 reporting odds ratio used in Dr. Etminan's 2015 disproportionality analysis survived and remained "a reliable piece of evidence to be used within the weight of the evidence for causation." See Plunkett Dep. at 138-39. That premise is wrong. In his 2016 affidavit, Dr. Etminan explicitly repudiated the 1.85 reporting odds ratio. He implicitly did so again in his 2017 letter to the editor. There, he made clear that once the subject's reproductive age was taken into account, the reporting odds ratio would change. 42 Dr. Plunkett's uncritical and unwarranted reliance on the Etminan 2015 study in the face of the ultimate red flag-the study author's serial repudiations of it-is not *260 consistent with rigorous dispassionate thinking. It suggests the sort of "conclusion-driven" analysis, In re Zoloft , 858 F.3d at 798 , that Daubert does not permit.

A similar methodological lack of rigor is reflected in Dr. Plunkett's application of Bradford Hill's analogy factor. In finding that the analogy factor supported finding that Mirena causes IIH, Dr. Plunkett, like Dr. Moyé, centrally analogizes Mirena to Norplant. Plunkett Rpt. at 34-35. But, as Dr. Plunkett conceded in her deposition, she has not concluded that Norplant causes IIH. See Plunkett Dep. at 33-34. Nor, as noted earlier, has any existing scholarship reached that conclusion, as various of plaintiffs' other experts conceded. See, e.g. , Darney Dep. at 138. Dr. Plunkett's treatment of the analogy factor, like Dr. Moyé's, is therefore unsound, insofar as the premise of its analogy is an assumed, not an established, fact. For both of these experts, there is too great an analytic gap between the available data and the conclusion they draw.

Dr. Plunkett's approach as to the Bradford Hill factor that assesses experimental evidence is also problematic. Like Dr. Moyé, Dr. Plunkett does not, and given the non-existence of any experiment testing whether Mirena tended to cause IIH, could not point to any such experiment.

To find the experimental evidence factor satisfied, Dr. Plunkett instead, like Dr. Moyé, turns to examples of purported "dechallenges." Dr. Plunkett does not, however, cite, let alone analyze, any concrete such example. Instead, she cites without comment Bayer's 2015 signal assessment and publications addressing Norplant. Even on Dr. Plunkett's own terms, however, these sources were flawed as a basis for that conclusion, because the "dechallenge" examples addressed therein either explicitly were, or potentially were, subject to confounding factors. Bayer's 2015 signal assessment, for example, states that "[n]o unambiguous case of positive dechallenge could be identified." Bayer BfArM Resp. at 49. The writings that Dr. Plunkett cites regarding Norplant are much the same. Sunku's poster presentation recounted two anecdotal examples of dechallenge; in each, the patient was treated with diuretics, supplying an obvious alternative to the removal of the Norplant as an explanation for the resolution of the patient's IIH symptoms. See Sunku, et al., supra . And Wysowski's study is silent about whether the Norplant removal of the patients that he described coincided with other treatment. See Wysowski & Green, supra , at 540. In any event, all the patients addressed by Wysowski for whom weight data was available were overweight, and most were obese. Id.

Like Dr. Moyé's, Dr. Plunkett's report also departs at points from standards that she acknowledges govern her work. This is exemplified by her discussion of the Bradford Hill factor of temporality. As background, Dr. Plunkett is not a medical doctor. She is not qualified to diagnose IIH or to apply a differential diagnosis to determine what caused a patient's IIH. And in her deposition, she therefore acknowledged that she cannot offer an opinion on why a patient's IIH resolved. Plunkett Dep. at 111, 124, 127-29, 131-35. Rather, she acknowledged that, when considering case reports, she was limited to considering "what is described by the doctor." Id. at 126. Dr. Plunkett, however, does not thus confine the analysis in her report. Drawing on Bayer's adverse events database in the course of discussing the Bradford Hill factor of temporality, her report cites three case reports. See Dkt, 167-73 ("Case Report A"); id. Ex. 74 ("Case Report B"); id. Ex. 75 ("Case Report C"). Yet in each, the patient's physician specifically reported that *261 it was unlikely that the patient's apparent "IIH" had been caused by the Mirena (or Skyla). See Case Report A at MIR_JSEU_01057385 ("Reporter's comment: The event is unlikely related to study medication."); Case Report B at MIR_PIEU_157046 ("original reporter's clinical assessment" was of "no relationship" between the patient's IIH and Mirena); Case Report C at MIR_INDNDA_00473903 ("The investigator considered the event as not related to study drug and as not related to study conduct."). In fact, in Case Report A, the physician did not even specifically diagnose IIH. See Case Report A at MIR_JSEU_01057384-85. And in Case Report B, the physician stated that the diagnosis was "very unsure." See Case Report B at MIR_PIEU_157046. Dr. Plunkett, however, exceeds the boundaries of this source material. In finding the temporality factor met, she goes beyond what the percipient physicians had concluded, notwithstanding her own lack of medical training. This, too, is unsound methodology.

Dr. Plunkett also fails to consider evidence that did not support her opinion. In her discussion of Bradford Hill's biological gradient (dose-response effect) factor, for example, Dr. Plunkett does not consider any data regarding combined oral contraceptives that contain LNG. These contraceptives have total LNG serum levels that are some 20-30 times higher than Mirena. To be sure, Dr. Plunkett asserts in her report and stated in her deposition that the more relevant measure of LNG exposure is the patient's free LNG level, not total LNG serum levels. But Dr. Plunkett does not take the next logical step: to explore the free LNG levels of combined oral contraceptives, which would have permitted her to make a useful comparison between Mirena and these contraceptives, which, as noted, have been found in five epidemiological studies not to cause IIH. See Plunkett Dep. 22, 28, 30, 95-96. She instead writes off these products as irrelevant. Dr. Plunkett's decision to liken Mirena to Norplant (as to which no epidemiological studies regarding IIH exist) while ignoring the entire category of LNG-based combined oral contraceptives (which have been studied extensively) is suggestive, too, of an outcome-driven approach, not a search for truth. See In re Rezulin , 309 F.Supp.2d at 563 ("[A]n expert may not pick and choose from the scientific landscape ...."). 43

Dr. Plunkett's handling of the Valenzuela study is similarly problematic. Her report cites that study (and its preview in Rai) as support for finding no fewer than three of the nine Bradford Hill criteria: strength of association, Plunkett Rpt. at *262 27; consistency, id. ; and temporality, id. at 28. But the report's citations of that study contain no discussion or analysis of that study. Instead, Dr. Plunkett effectively treats the study, ipse dixit , as affirmatively establishing those criteria. But her report nowhere acknowledges the vital limitation emphasized by the Valenzuela authors: that the study had not controlled for important confounding factors (obesity and recent weight gain) and therefore could not be read as finding causality. Nor did Dr. Plunkett otherwise contend with the fact of these alternative explanations for the IIH symptoms of the patients covered in the Valenzuela study. See Plunkett Dep. at 372-75 (admitting she did not know the background rate of IIH among obese women or their relative risk); cf. In re Abilify (Aripiprazole) Prods. Liab. Litig. , 299 F.Supp.3d 1291 , 1353 (N.D. Fla. 2018) ("Dr. Glenmullen also considered alternative explanations for the association between Abilify, impulsive gambling, and other impulse control disorders.").

This lapse bespeaks a larger methodological deficiency characterizing Dr. Plunkett's proposed testimony. For an expert's testimony to be reliable, she "must demonstrate that [she] has adequately accounted for obvious alternative explanations." U.S. Info. Sys., Inc. , 313 F.Supp.2d at 238 ; see also Reference Manual on Scientific Evidence ("[R]esearchers first look for alternative explanations for the association, such as bias or confounding factors.... Once this process is completed, researchers consider how guidelines for inferring causation from an association apply to the available evidence."); In re Lipitor (Atorvastatin Calcium) Marketing, Sales Practices and Products Liability Litigation , 174 F.Supp.3d 911 , 916 (D.S.C. 2016) (same). Although his report proved methodologically deficient on other grounds, Dr. Moyé at least attempts to engage with Valenzuela's caveat about confounding factors, opining that because of the large reporting ratio observed in Valenzuela's Utah population, the correlation between Mirena and IIH was not likely to be confounded. Dr. Plunkett's report makes no such effort. Nor does it advance any scientific argument for why the IIH of the patients in Valenzuela could not equally plausibly be attributed to the non-Mirena factors of obesity and recent weight gain.

Finally, Dr. Plunkett's embrace of the "androgen theory" of the biological mechanism by which Mirena might cause IIH suffers from severe analytical gaps. As addressed infra in connection with Dr. Darney's and Dr. Johanson's proposed testimony, it is not tenable for an expert simply to assume that LNG, a progestin, causes IIH by the same mechanism as androgens ( e.g. , testosterone ). See Brumbaugh v. Sandoz Pharm. Corp. , 77 F.Supp.2d 1153 , 1157 (D. Mont. 1999) ("Testimony extending general conclusions about similar drugs does not meet Daubert's requirement of reliability."); Dunn v. Sandoz Pharm. Corp. , 275 F.Supp.2d 672 , 681 (M.D.N.C. 2003) ("Dr. Kulig's assertion that because bromocriptine is an ergot alkaloid and may behave like other ergot alkaloids and cause vasoconstriction simply does not support the proposition that Parlodel causes stroke in postpartum women."); Mirena Perforation/Daubert , 169 F.Supp.3d at 433 ("To conclude that Mirena would cause the same effect as Depo-Provera because they both contain progestin ... is to impermissibly draw grossly 'overreaching conclusions,' which are connected solely to the data by [the witness's] say-so." (internal citation omitted) ). Yet Dr. Plunkett did so here.

For all the above reasons, Dr. Plunkett's proposed testimony is beset by methodological *263 deficiencies. It falls far short of satisfying Daubert 's standard of reliability. Her testimony, too, must be excluded.

C. Dr. James Wheeler

1. Qualifications

Dr. Wheeler is an OB/GYN who works in private practice. He is a graduate of Harvard College and Baylor College of Medicine, where he received his medical degree and concentrated in reproductive medicine. Wheeler Rpt. at 3-4. He completed a residency in obstetrics and gynecology at Baylor, and a post-residency subspecialty in reproductive endocrinology and infertility at Yale University School of Medicine. Id. at 4. While at Yale he was a member of Yale's Robert Wood Johnson Clinical Scholars Program, which teaches clinical epidemiology; 44 he also earned a masters in public health (M.P.H.) from the Yale University School of Public Health, majoring in Biostatistics with a minor in Maternal/Child Health. Id. at 5. He worked as a faculty member at Baylor between 1988 and 1994. He also earned a J.D. from the University of Houston Law Center, where he concentrated in health law, and earned a Legal Nurse Consulting diploma in 2016. Id. at 6.

Since 1994, Dr. Wheeler has been in private practice. Id. at 4-5. He has provided contraceptive counseling and advice, including as it relates to IUDs, to thousands of patients. Id. He has personally placed or supervised the placement of many IUDs, including Mirenas. Id. at 5. However, Dr. Wheeler has not conducted research on Mirena or LNG outside of this litigation. Nor has Dr. Wheeler ever diagnosed a case of IIH. Wheeler Dep. at 39-40, 63, 329-30. Dr. Wheeler frequently serves as an expert in litigation. During the last five years, he estimates, he has spent between 15 and 20 percent of his professional time on legal matters. Id. at 46.

2. Proposed Testimony

Dr. Wheeler's testimony that Mirena causes IIH, like that of Drs. Moyé and Plunkett, is primarily based on an application of the Bradford Hill factors. See Wheeler Rpt. at 31, 34-39. 45 As the source material for his Bradford Hill analysis, Dr. Wheeler relies mostly on (1) a dataset of 115 case reports drawn from Bayer's 2015 signal investigation in which patients were diagnosed with IIH after having had a Mirena inserted, and (2) the Valenzuela study (and its Rai precursor). Dr. Wheeler also cites Norplant case reports. He also notes the fact that the labelling on both Norplant and Jadelle warns of a possible, *264 although by no means established, connection between LNG and IIH.

He applies the Bradford Hill factors as follows.

Strength of association : In finding this factor "sufficiently demonstrated," Dr. Wheeler notes that within Bayer's 2015 dataset, there were "115 cases of [IIH] in women with a Mirena in situ." Wheeler Rpt. at 35. Although he acknowledges "the potential for selection bias and accrual bias operating due to the manner in which these cases were collected," he states that "this is nonetheless a significant number of cases collected." Id. Dr. Wheeler also cites the Rai/Valenzuela study as "medical literature reporting on possible association of [IHH] with intrauterine use of LNG." Id.

During his deposition, however, Dr. Wheeler reversed course. He conceded that the Rai/Valenzuela study does not demonstrate an association between Mirena and IIH:

Q: And the-so as you sit here today, you would agree there is no epidemiological evidence establishing even an association between Mirena and [IIH], true?

A: Based on these papers and what we've reviewed, there's no demonstration of an association. The question is raised, but there is no association demonstrated.

Wheeler Dep. 200 (objection omitted).

Consistency of association : Dr. Wheeler opines that this factor is met because of a "sufficient degree of heterogeneity" in his dataset of 115 cases that he drew from Bayer's 2015 signal investigation, because "the association of [IIH] and LNG was sufficiently consistent with other LNG-containing devices to be included in the Product Information brochure of LNG implants Norplant and Jadelle," and because "the Rai/Valenzuela group found a statistically significant increased risk of [IIH] in a large case-control observational study of two different groups." Wheeler Rpt. at 35.

Specificity : Dr. Wheeler opines that this factor is met based on the 115 cases he draws from Bayer's data set, the Rai/Valenzuela study, and the Norplant and Jadelle labels. Although Dr. Wheeler's report elsewhere notes that IIH has associated risk factors, id. at 44, his analysis of specificity does not address these factors, id. at 36.

Temporality : Because almost every case in the data set that Dr. Wheeler extracts from Bayer's 2015 signal investigation arose after a Mirena had been inserted, Dr. Wheeler finds that the "[a]nalysis of Temporality favors a causal relationship." Id.

Biological gradient : Dr. Wheeler notes that within Bayer's 2015 signal investigation, one case had been included involving Bayer's lower-dose LNG, Skyla. While Dr. Wheeler states that he does not know why only one Skyla case was included in that data set, he postulates that, "if it is the only known case, or one of only a few," in which IIH was associated with Skyla, given the larger number of IIH cases associated with Mirena, "a dose-response between LNG and [IIH] may be proposed." Id. at 36-37. Dr. Wheeler also notes that women with an embedded Mirena have higher systemic levels of LNG and that at least two patients in Bayer's 2015 dataset had an embedded Mirena. "This," he states, "also lends support to a biologic gradient." Id.

Biological plausibility : Addressing this factor, Dr. Wheeler embraces the androgen theory of IIH causation. It is "more likely than not," he states, that IIH is "caused by sex hormones including androgens." Id. at 37. He does not articulate a mechanism by which this occurs.

*265 Coherence : Dr. Wheeler states that "based on the medical literature analysis" and the IIH warning that appears on "other LNG-containing devices," "I conclude Coherence is satisfied in proposing a causal relationship between LNG-IU[D] and [IIH]." Id. at 38.

Experimental evidence : Dr. Wheeler acknowledges that, in epidemiology, " 'experiment' usually means a randomized clinical trial." Id. at 38. But, he states, a "randomized placebo-controlled trial involving contraceptive choices" is likely "not practical, and potentially not ethical due to the differential likely effects on pregnancy rates and complications within a placebo-controlled group, of a comparison otherwise-treated group." Id. at 38. Dr. Wheeler states that, in lieu of experiments, he would look to " 'challenge-dechallenge-rechallenge' " evidence. Id. He states that the 11 "dechallenge" cases he had noted in Bayer's dataset, of which nine patients "recovered" from their symptoms and signs of IIH, "may also inform this criterion." Id. at 25, 38. Dr. Wheeler admits, however, that "the sample to which this applies is relatively small," id. at 38, and that he had been "unable to discover any dechallenged, then rechallenged, patients" within the dataset. Id. at 25.

Analogy : Dr. Wheeler states that "[IIH] is associated with other hyperandrogenic symptoms including Polycystic Ovary Syndrome" and that "[t]his clinical observation satisfies Analogy to a reasonable degree." Id. at 38.

3. Analysis Under Daubert

For the reasons reviewed below, methodologically, Dr. Wheeler's application of Bradford Hill is flawed in multiple respects. Some flaws echo those noted about the Bradford Hill analyses conducted by Drs. Moyé and Plunkett. Other flaws are specific to Dr. Wheeler's analysis.

Specific to Dr. Wheeler, his use of Bradford Hill is fatally compromised, at the threshold, by a concession he made at his deposition. He testified that there is no statistical association between Mirena and IIH. As explained earlier, the Bradford Hill criteria are, at bottom, a methodology for evaluating whether a demonstrated epidemiological association is, or is not, causal. It follows that, absent such an association, there is no basis to apply the Bradford Hill criteria. See Federal Judicial Center, Reference Manual on Scientific Evidence 599, n.141 (3d. ed. 2011) ("In a number of cases, experts attempted to use these guidelines to support the existence of causation in the absence of any epidemiologic studies finding an association.... There may be some logic to that effort, but it does not reflect accepted epidemiologic methodology."); see also Soldo , 244 F.Supp.2d at 569 ("The Bradford-Hill criteria start with an association demonstrated by epidemiology and then apply such criteria as the temporal sequence of events, the strength of the association, the consistency of the observed association, the dose-response relationship, and the biologic plausibility of the observed association."); Dunn , 275 F.Supp.2d at 679 ("The first step in the causation analysis pursuant to Bradford Hill is an epidemiological study that has identified an association between two variables."); cf. In re Fosamax Prods. Liab. Litig. , 645 F.Supp.2d 164 , 187 (S.D.N.Y. 2009) ("[W]here there is a positive association between the exposure and the disease," "epidemiologists often apply a set of considerations described by Sir Austin Bradford Hill in a famous 1965 lecture"). 46

*266 Here, in contrast to Drs. Moyé and Plunkett, Dr. Wheeler admitted in his deposition that the one study in which an epidemiological relationship between LNG and IIH was demonstrated and not later withdrawn-the Valenzuela study-does not provide evidence of association, given the presence of obvious confounding factors as noted by the study's authors. See Wheeler Dep. at 200 ("[T]here's no demonstration of an association. The question is raised, but there is no association demonstrated."). Although these proposed experts disagree amongst themselves on whether the Valenzuela study supplies evidence of an association between Mirena and IIH-and although Dr. Wheeler is perhaps to be commended for his candor under oath in disavowing any such association-once Dr. Wheeler conceded the lack of a statistical association, there was no longer a logical charter for him to undertake a Bradford Hill inquiry at all. See Federal Judicial Center, Reference Manual on Scientific Evidence 599, n.141 (3d ed. 2011).

The only other evidence which Dr. Wheeler's report cites in finding Bradford Hill's first factor ("strength of association") satisfied are the 115 adverse event reports that he drew from Bayer's 2015 signal assessment. See Wheeler Rpt. at 39 ("Additional evidence that Mirena causes or substantially contributes to [IIH] is derived from the collection of 115 cases of women with [IIH] who have a Mirena in place."); id. at 35 ("This is a sizable series of [IIH] patients, especially with any knowledge as to their contraceptive practices."). But these case reports-on which Dr. Wheeler relies in his report to satisfy fully five of the nine Bradford Hill factors, see id. at 34-39-are inadequate as a basis upon which to find Bradford Hill's vital first factor, an association demonstrated by epidemiology. As virtually all experts in this case (including Dr. Wheeler) acknowledge, adverse event data primarily serves instead a more limited function-as a tool to generate hypotheses. 47

And the set of case reports on which Dr. Wheeler relies, without more, certainly do not reveal the required association. An association exists when the rate of a reported condition is greater among the group taking a drug than among the portion of the population not taking that drug. The adverse event reports which Dr. Wheeler cites, however, by definition cannot reveal any such association, because they are silent as to the comparison group: They do not reveal the reporting rate of IIH among the portion of the population not taking Mirena. They do not provide a basis on which Dr. Wheeler can opine as to whether there is an elevated rate at which Mirena patients experience IIH relative to persons without a Mirena inserted- i.e. , whether there is an association between Mirena and IIH. Dr. Wheeler has not undertaken this inquiry. See Wheeler Dep. at 221 ("Q: So you can't tell us whether there is an elevated reporting rate of intracranial hypertension with Mirena users compared to any other group of patients, true? A: True."); see also id. at 220-23.

The 115 adverse event reports, which Dr. Wheeler draws from a spreadsheet given to him, in fact supply an unusually good illustration of why the case law, including *267 in the earlier round of Mirena litigation before Judge Seibel, has hesitated to base findings of causation or even an epidemiological association on adverse event data alone. See, e.g., Mirena Perforation/SJ , 202 F.Supp.3d at 304 ("Case reports are not reliable evidence of causation."); see also McClain , 401 F.3d at 1250 (noting that "reports reflect complaints called in by product consumers without any medical controls or scientific assessment"); In re Accutane Prods. Liab. , 511 F.Supp.2d 1288 , 1298 (M.D. Fla. 2007) ("[Adverse event] reports are unreliable as proof of causation because, in general, the events were not observed in such a way as to rule out coincidence or other potential causes."); Cloud v. Pfizer, Inc. , 198 F.Supp.2d 1118 , 1138 (D. Ariz. 2001) ("[I]ndividual case reports and retrospective medical articles summarizing individual case reports are not an adequate basis from which a jury could conclude that Zoloft causes suicide."). The adverse event reports made to Bayer reporting an IIH diagnosis were not necessarily filed by a medical professional, as Dr. Wheeler acknowledges. Instead, they were filed by "someone (clinician, lawyer, nurse, medical student, relative, or consumer themselves)." Wheeler Rpt. at 21. Of the 115 adverse event reports that Dr. Wheeler cites, 32 were generated by the filing of a lawsuit against Bayer, a fact of which Dr. Wheeler was unaware at the time of his report. 48 (These 32 appear to have been plaintiffs in the IIH lawsuits later consolidated in this MDL. Today, with more than 850 plaintiffs having joined this litigation since the formation of the MDL and the Court's appointment of lead and liaison counsel for plaintiffs, commensurately more adverse event reports presumably have been recorded by Bayer.) The filing of such lawsuits-and Bayer's required, ministerial act of logging such lawsuits as adverse event reports-is not evidence of causation or association.

Further, even assuming that all 115 case reports on which Dr. Wheeler relied could be credited as true to the extent they reported that the subject both had used Mirena and had experienced IIH symptoms, the observation that the number 115 is "sizable," Wheeler Rpt. at 35, falls short of supporting an epidemiological association between these two facts. More than this absolute number would be needed for this correlation to have such meaning. Such information would include how the 115 reports (as among the broader universe of Mirena users) compared with the incidence of such symptoms in the population generally. It would also include whether the subject cases were subject to confounding factors ( e.g. , obesity, recent weight gain) that could supply obvious alternative explanations for the IIH symptoms. In his deposition, Dr. Wheeler acknowledged that Mirena is preferentially prescribed to the precise population with the highest reporting of IIH-obese women of reproductive age. 49

Unsurprisingly, then, at his deposition, Dr. Wheeler retracted his report's claim of a statistical association between Mirena and IIH. That claim had been based on *268 two data points-the Valenzuela study and the 115 adverse event reports-neither of which could sustain finding such an association. However, with such an association disclaimed, the vital precondition for embarking on a Bradford Hill analysis was eliminated. This alone warrants exclusion of Dr. Wheeler's testimony under Daubert .

Independent of his inability to find this gating criterion, Dr. Wheeler's proposed expert testimony report, like that of Dr. Moyé and Dr. Plunkett, does not satisfy any of Daubert 's four core reliability factors: It is untested; it has not been subject to peer review; there is neither an error rate nor are there standards controlling its operation; and his conclusion lacks any acceptance, let alone general acceptance, in the scientific community outside of this litigation. 50 This requires the Court to carefully scrutinize his methodology, too. Dr. Wheeler's ensuing analysis does not withstand this scrutiny. Rather, as with the two prior experts, Dr. Wheeler's application of the Bradford Hill factors to find that Mirena causes IIH is subject to a series of methodological flaws that independently preclude its admission.

These include:

Lack of weighting or discussion of relationship of factors : Like Drs. Moyé and Plunkett, Dr. Wheeler applies the Bradford Hill factors without revealing the weight he attaches to individual factors or addressing the relationship among them. See Wheeler Rpt. at 35-39. As discussed in connection with Dr. Moyé, this mall

This text is long and has been trimmed here. Open the source document for the complete record.