Courtesy of Mike Imbroscio of Covington (who was on the winning team) today’s post is about a very satisfying Cymbalta win – not only did warning causation fail under the learned intermediary rule, but the warning for the drug was held adequate as a matter of law. The case is McDowell v. Eli Lilly & Co., ___ F. Supp.3d ___, 2014 WL 5801604 (S.D.N.Y. Nov. 7, 2014). It involved injuries (such as “brain zaps,” whatever those might be) allegedly suffered by the plaintiff when she was weaned from the drug too quickly.
We’ll briefly touch on the causation part first. Under New York’s learned intermediary rule, a prescriber’s “independent knowledge” of a supposedly unwarned-of risk breaks any causal connection between the warning’s inadequacy and the plaintiff’s injuries. 2014 WL 5801604, at *15. The prescriber (a nurse practitioner – we think McDowell is the first nurse-as-learned-intermediary decision under New York law) “testified that, based on her clinical experience and training, she was aware of the risk of discontinuation symptoms with abrupt cessation of [drug] treatment.” Id. (there’s a lot more than just that statement). Because the alleged understatement of the risk wouldn’t have changed the prescribing decision, plaintiff lost. Id. at *17-18. Plaintiff’s attempt to mislead the prescriber at her deposition with statements about the drug that had no basis in fact couldn’t create a disputed issue of fact. Id. at *18 (pointing out “the absence in the record of any support for the claim”). We also like the causation standard employed in McDowell:
[U]nder New York’s proximate cause standard, a plaintiff must demonstrate that had a different, more accurate warning been given, his physician would not have prescribed the drug in the same manner. Summary judgment is appropriate where a plaintiff fails to establish that a prescribing physician’s decision to prescribe a particular medication would have changed had a different warning been given
Id. at *16 (citations and quotation marks omitted). We’ve discussed as some length recently the muddle in New York law concerning the heeding presumption, with some federal courts ignoring state-court precedent that there’s no such animal. No such laxity appears in McDowell.
Warning causation rulings, however, are inherently case specific, since they depend on the facts of a particular prescriber’s reaction to a particular plaintiff’s inadequate warning claim. The finding in McDowell that the warning in question was adequate as a matter of law, however, has potentially much broader application. That’s where we want to spend most of our time today. The court in McDowell specifically held that the defendant’s “discontinuance warning is adequate.” 2014 WL 5801604, at *10. Why? Well, first the learned intermediary rule means that the duty to warn only runs to prescribers, not patients, so prescription medical product warnings are judged by what such learned intermediaries are expected to know and respond to. Id. Thus, “[a] warning is adequate as a matter of law if it provides specific detailed information on the risks of the drug.” Id.
The key question is how specific and how detailed? No matter how comprehensive a defendant’s treatment of a particular drug risk, any competent plaintiffs’ lawyer can dissect any warning and find something else that, in retrospect, could conceivably have made some sort of difference. As the court held, however, that’s not how warnings are properly viewed. “The warning should also be evaluated as a whole and not through the nitpicking prism of an interested legal advocate after the fact.” Id. at *11. In this particular case:
[T]he FDA-approved label for the medicine has included a three-paragraph warning on the risk of symptoms upon discontinuation of [its] therapy. That warning has included a statement about the occurrence of discontinuation symptoms in the [drug’s] clinical trial experience, including the important fact that the rate was significantly higher in [its] patients than patients on placebo; a recitation of the specific symptoms possible upon . . . discontinuation (including the symptoms alleged by the Plaintiff); and guidance on the appropriate protocol for safe discontinuation of the medicine.
McDowell, 2014 WL 5801604, at *11. That’s pretty tough to get around, since the “detailed catalog of symptoms possible upon discontinuation” includes “the symptoms the Plaintiff alleges that he experienced after stopping his [drug] treatment.” Id. at *12.
Time for the other side of the “v.” to get out its “nitpicking prism.” So what did they come up with in McDowell? Statistics, that’s what. “The Plaintiff contends that the [drug’s] discontinuation warning was both inadequate because it did not include the frequency of patients with discontinuation symptoms . . ., and misleading, because it utilized a threshold of ‘greater than or equal to 1%’ to identify individual discontinuation events.” Id. at *13. Plaintiff asserted this claim despite the defendant’s statistics being precisely what the FDA wanted:
During the period relevant here, the label included approximately a dozen symptoms occurring “at a rate greater than or equal to 1%” in placebo-controlled clinical trials. . . . This method of communicating information on individual symptoms appearing in clinical trials is consistent with the accepted practice of identifying such individual adverse events observed at or above a specified threshold and in accord with FDA regulations and guidance directing that the label “list the adverse reactions identified in clinical trials that occurred at or above a specified rate appropriate to the safety database.”
Id. at *12 (citations to FDA warning formatting guidance documents omitted). Plaintiff argued “I don’t want your stinkin’ ‘greater than 1%’ garbage, I don’t want “a significantly higher rate” than “placebo” (which the warnings also stated, id. at *13), I want exact numbers:
The Plaintiff contends that the [drug’s] discontinuation warning was both inadequate because it did not include the frequency of patients with discontinuation symptoms . . . and misleading, because it utilized a threshold of “greater than or equal to 1%” to identify individual discontinuation events.
Id. at *13 (quoting plaintiff’s brief).
That contention put at issue something we’ve never addressed in any detail on DDL Blog before – can the warnings for a prescription medical product be considered “inadequate” because exact risk statistics were not included? McDowell held that such statistics were not required. A product warning could be “accurate, clear, consistent on its face” and “portray” a drug risk “with sufficient intensity” to be adequate as a matter of law without including every statistic for every risk that ever showed up in a clinical study. Id. at *14.
[C]ourts have refused to graft onto the adequacy standard a requirement that a package insert must include specific adverse event frequencies. . . .[T]he warning warned about the injuries sustained by the plaintiff, despite lacking precise frequencies for each withdrawal event or for withdrawal symptoms as a category.
Id. Nor was anything defective about the FDA’s preferred more than “1%” statistical format. That was plenty, particularly as that the intended audience consisted of trained medical professionals:
The Plaintiff has contended that the [drug’s] label misleadingly suggests that discontinuation symptoms “occurred in only 1% of patients.” [The prescriber], an experienced medical professional, expressly rejected that interpretation. . . . Using a numerical threshold for the inclusion of adverse events in a label is an appropriate, standard methodology for identifying adverse events arising with sufficient frequency to warrant inclusion in the product label.
Id.
McDowell cited several previous decisions also holding that reams of risk statistics were not necessary for an adequacy finding. The first of these was Hurley v. Lederle Laboratories, 651 F. Supp. 993 (E.D. Tex. 1986), rev’d on other grounds, 863 F.2d 1173 (5th Cir. 1989) (preemption and pre-Buckman fraud on the FDA), involving the DTP vaccine. Such vaccine litigation (Bexis’ first exposure to a pharmaceutical mass tort) was ultimately adjudged so inimical to the public health that Congress prohibited it. Anyway, one ploy in DTP litigation was to demand inclusion of statistics for every adverse reaction. Hurley said “no”; holding that such allegations presented a slippery slope with no discernable stopping point:
The plaintiff cites no authority for the proposition that a drug manufacturer has a duty to warn prescribing physicians of the rate of adverse reactions. As a practical matter, this would be extremely difficult, perhaps impossible, with respect to a drug . . . which has many possible harmful side effects. Moreover, the warning . . . lists with specificity seven serious reactions, one of which inflicted the minor plaintiff. The warning also refers the physician to medical research material further explaining the possible consequences of administering the [product]. Based upon the foregoing, it is clear that the warning adequately warns of the severity and nature of adverse reactions as a matter of law.
651 F. Supp. at 1002. Hurley was actually quoting an earlier DTP decision that had involved the client Bexis represented, Smith v. Wyeth Laboratories, Inc., 1986 WL 720792, at *10 (S.D.W. Va. Aug. 21, 1986). Smith is a West Virginia case, so while it’s good law on the statistics point (and has a Westlaw citation that McDowell missed), it comes with the usual West Virginia learned intermediary caveat. McDowell also cited Percival v. American Cyanamid Co., 689 F. Supp. 1060 (W.D. Okla. 1987). That’s yet another DTP decision and really doesn’t add anything to Hurley and Smith, unless Oklahoma law is involved, since it quotes the same language. 689 F. Supp. at 1064.
There are other cases out there rejecting claims that warnings unencumbered by extensive statistical development are nonetheless adequate. In yet another DTP case, the Sixth Circuit, applying Ohio law, held that where the risk in question “was included in the list of possible adverse reactions in the warning insert,” the defendant was not required to go further and that risk’s specific “reported rate” or “incidence.” Ackley v. Wyeth Laboratories, Inc., 919 F.2d 397, 405 (6th Cir. 1990).
There are also non-DTP cases. The earliest of these – and the only other appellate decision − is Calabrese v. Trenton State College, 392 A.2d 600 (N.J. Super. App. Div. 1978), aff’d, 413 A.2d 315 (N.J. 1980) (per curiam). The plaintiff claimed that the incidence of rabies was so low that the risks of rabies vaccine were more serious than the “remote” likelihood of the disease. The product warning, according to plaintiff, was defective because it did not contain “statistical information” about the incidence of rabies. The claim was rejected:
Plaintiff’s central argument that a drug manufacturer’s warnings must, as a matter of law, include not only information concerning undesirable side effects of the drug it is marketing but, as well, statistical information concerning the incidence of the condition for which the drug can be used, is without merit.
Id. at 604 (citations omitted). That’s not 100% the same thing, since the statistics at issue didn’t involve product risks, but it’s the first statistics-in-warning case that we’re aware of.
More recent is Ames v. Apothecon, Inc., 431 F. Supp.2d 566 (D. Md. 2006), in which the plaintiff claimed that the defendant’s label was defective for failing to “quantify” the risk she suffered in addition to stating clearly what it was. The court rejected this claim.
The plaintiffs were unable to cite a single case in which a court has found a label to be defective because the incidence rate was not described. Moreover, warnings must be brief and focused to be effective. It would require an extended discussion, in the nature of a medical journal article, to lay out the debate [about incidence rate]. . . . One must also bear in mind that the warnings are intended to be read by learned intermediaries who are presumed to have considerable medical training as well as the ability to access the medical literature if they require additional information.
Id. at 573.
We aren’t denying that, where statistics are provided, they need to be accurate. Of course they do. E.g., Ackermann v. Wyeth Pharmaceuticals, 526 F.3d 203, 209 (5th Cir. 2008) (applying Texas law). But the question in McDowell and these other cases is different. It’s whether a plaintiff can demand quantification of the rate of incidence of risk where that risk has been accurately described without the use of such statistics. On that question, the law seems clear. There is no additional duty of quantification where a risk is otherwise adequately described in a defendant’s product labeling. “If a pharmaceutical manufacturer warns doctors that specific adverse side effects are associated with the use of a drug, then a causal relationship between use of the drug and development of potential side effects is implicit in the warning.” Ziliak v. AstraZeneca LP, 324 F.3d 518, 521 (7th Cir. 2003). Statistics don’t add anything material to this sort of risk discussion. To us, this is the most important point made in McDowell.