¶43. dissenting. The overarching issue in this appeal is whether plaintiff’s common-law claim for failure to warn conflicts with the FDA’s regulation of Phenergan, the drug *104responsible for plaintiffs injuries. I would conclude that the jury’s verdict in this case conflicts with federal law for two reasons.
¶ 44. First, it would be impossible for defendant Wyeth to comply with the requirements of both state and federal law. Specifically, the FDA approved IV administration of Phenergan and required that IV administration be listed on the Phenergan label. By contrast, plaintiffs theory of the case required Wyeth either to remove this approved use from the Phenergan label, add a warning that would directly contradict the label’s indication that IV administration was a safe and effective use, or, at a minimum, add a warning that only certain types of IV administration should be used. Thus, compliance with state law in this case would require Wyeth to eliminate uses of Phenergan approved by the FDA and required to be included in the Phenergan labeling.
¶ 45. Second, plaintiffs state-law claim conflicts with federal law in that it poses an obstacle to federal purposes and objectives. In short, by approving Phenergan for marketing and distribution, the FDA concluded that the drug — with its approved methods of administration and as labeled — was both safe and effective. See 21 U.S.C. § 355(d) (listing criteria for drug approval). In finding defendant liable for failure to warn, a Vermont jury concluded that the same drug — with its approved methods of administration and as labeled — was “unreasonably dangerous.” See Town of Bridport v. Sterling Clark Lurton Corp., 166 Vt. 304, 308, 693 A.2d 701, 704 (1997) (to succeed on failure-to-warn claim, plaintiff must show that “failure to warn made the product unreasonably dangerous and therefore defective”). These two conclusions are in direct conflict.
¶46. For both of these reasons I would conclude that the state-law cause of action is preempted. I respectfully dissent.
I. Impossibility of Compliance
¶ 47. As explained by the majority, because there is no clause in the FDCA expressly preempting state law, Wyeth must demonstrate that preemption is implied by showing either that federal law thoroughly occupies the regulatory field (a claim that Wyeth does not advance) or that there is an actual conflict between state and federal law. Actual conflict, in turn, can be demonstrated in one of two ways: by showing that it is impossible for the regulated party to comply with both state and federal law or that state law “stands as an obstacle to the accomplishment and *105execution of the full purposes and objectives of Congress.” Freightliner Corp. v. Myrick, 514 U.S. 280, 287 (1995) (quotations omitted).
¶ 48. The majority in essence concludes that it is not impossible for Wyeth to comply with both federal and state standards because Wyeth never sought FDA approval of a “stronger warning” of the type advocated by plaintiff. According to the majority, because the FDA was not presented with, and therefore did not explicitly reject, such strengthened language, there is no reason to presume that the FDA would disapprove. Therefore, the majority reasons, there is no actual conflict between state and federal law. See ante, ¶¶ 21-22. It is inaccurate, however, to characterize the requirements imposed by the jury verdict in this case as merely requiring a “stronger warning.” Rather, what plaintiff sought was an elimination of a use of Phenergan that had been approved by the FDA. Furthermore, the FDA’s rejection of Wyeth’s efforts to alter the language of the warning in 2000 supports Wyeth’s claim that the FDA had an affirmative preference for the language of the original warning.
A.
¶ 49. The crux of plaintiff’s claim was not based on the label warnings per se, but on the approved uses listed there. See, e.g., ante, ¶ 3 (“Plaintiffs experts testified that the label should not have allowed IV push as a means of administration . . . .”). A review of plaintiffs complaint and the evidence presented at trial makes clear that the standard plaintiff sought to establish (i.e., the change to the label that would be required in light of the jury’s finding of liability) was to remove IV administration — or at least certain types — as an approved use. For example, plaintiffs complaint asserted that the warnings on the label were inadequate and that:
[t]he Phenergan sold by defendant is . . . not reasonably safe for intravenous administration because the foreseeable risks of harm posed by intravenous administration of the drug are sufficiently great in relation to its foreseeable theraputic benefits that reasonable health care providers, knowing of such foreseeable risks and benefits, would not prescribe the drug intravenously for any class of patients.
*106(Emphasis added.) In her appellate brief, plaintiff characterizes the evidence as revealing “that Wyeth was aware of research indicating that direct IV administration of Phenergan was unsafe!’ (Emphasis added.) Plaintiff further refers to expert testimony “that the label should have restricted Phenergan to intramuscular injection as this method of administration presents no risk of inadvertent arterial injection; or, alternatively, that if IV administration is used, it must be by injecting the Phenergan into a hanging IV bag, not through a direct IV.” (Emphasis added.)
¶ 50. Here, the FDA clearly addressed the risks attending IV administration of the drug. The label approved IV administration generally, and specifically warned of the dangers of direct IV administration, including inadvertent arterial injection possibly resulting in amputation. In light of this, it cannot be argued that the FDA did not (1) assess the risk of IV administration, including direct IV administration and the associated risk of amputation due to inadvertent arterial injection; (2) conclude that the benefits of allowing IV administration with appropriate warnings outweighed the risk; and (3) reach a decision regarding precisely what warning language should be used. These assessments are, in fact, the very essence of the FDA’s approval and are in furtherance of the federal objective of advancing public health by balancing the risks and benefits of new drugs and facilitating their optimal use. See 21 U.S.C. § 355(d) (listing factors to be considered in approving or refusing new drug application); 21 U.S.C. § 393(b)(1), (b)(2)(B) (FDA is charged with promoting public health by acting promptly on new drug applications and protecting public health by ensuring that new drugs are both safe and effective).
¶ 51. The majority reconciles this manifest conflict by relying on 21 C.F.R. § 314.70(c), which allows a drug manufacturer to alter a label “[t]o add or strengthen a contraindication, warning, precaution, or adverse reaction” or “add or strengthen an instruction about dosage and administration” prior to FDA approval.7 On this basis, the majority concludes that Wyeth “was free under *107§ 314.70(c) to strengthen the warning without prior FDA approval.” Ante, ¶ 22. But, it is an overstatement to claim that manufacturers are “free” to change drug labels under § 314.70(c). To the contrary, a manufacturer may change a label only to add or strengthen a warning, not to eliminate an approved use, as plaintiff would require here. In other words, what plaintiff advocates is not a stronger warning but language that would directly contradict language approved and mandated by the FDA.
¶ 52. Further, the apparent purpose of § 314.70(c) is to allow manufacturers to address newly-discovered risks. See 44 Fed. Reg. 37434, 37447 (June 26, 1979) (allowing supplement to label “whenever possibly harmful adverse effects associated with the use of the drug are discovered”). Even courts that conclude that § 314.70(c) provides manufacturers broad latitude to add warnings to labels acknowledge that such supplements are aimed at previously unknown and unanalyzed risks. See McNellis v. Pfizer, Inc., 2005 WL 3752269, at *6 (D.N.J.) (concluding that § 314.70(c) “was promulgated precisely to allow drug manufacturers to quickly strengthen label warnings when evidence of new side effects [is] discovered”) (citing 30 Fed. Reg. 993 (Jan. 30, 1965)); Kurer v. Parke, Davis & Co., 2004 WI App 74, ¶ 18, 679 N.W.2d 867 (noting that, under § 314.70(c), “[d]rug manufacturers can strengthen warnings or petition for additional warnings when new risk information arises”). Another section of the regulation makes clear that any changes to a label that exceed the scope of § 314.70(c) are considered “major changes” that require prior approval before the drug may be distributed. § 314.70(b), (b)(2)(v). In short, the regulation does not allow manufacturers to simply reassess and draw different conclusions regarding the same risks and benefits already balanced by the FDA. Here, the FDA had already evaluated the risk of inadvertent arterial injection from direct IV administration of Phenergan, and had mandated warning language for the label to reflect that risk assessment.
¶ 53. In addition, any change accomplished under § 314.70(c) is subject to ultimate FDA review and approval. See § 314.70(c)(7) (providing that FDA may order manufacturer to cease distribution of drug if it disapproves supplemental application); see also Needleman v. Pfizer, Inc., 2004 WL 1773697, at *3 (N.D. Tex.) (noting that changes to label under § 314.70(c) are temporary and “must later be approved by the FDA”). Thus, any additional or different warnings must ultimately be supported by scientific *108research that meets the FDA’s standards. Neither a manufacturer, a state court, nor a state legislature can permanently substitute its judgment of the risk-benefit analysis for that of the FDA.
¶ 54. At its core, plaintiffs argument in this case was not that the warnings on the label were inadequate, but that an approved use (direct IV administration) was in fact unreasonably unsafe. Plaintiff did not seek to “add or strengthen” a warning or a dosage/administration instruction, but rather to eliminate an approved use of the drug. This is a disagreement that cannot be overcome by operation of § 314.70(c). Plaintiffs claim in this case — that a method of administration of the drug should be partially if not entirely eliminated from the labeling — represents a substantive disagreement with FDA policy that goes beyond labeling/warning issues alone. This disagreement creates opposing requirements, and a manufacturer could not satisfy both at once.
B.
¶ 55. Wyeth argues that even if § 314.70(c) theoretically allows a manufacturer to make unilateral changes to a drug label, in this case, the FDA actually rejected Wyeth’s attempts in 2000 to change the warning regarding intra-arterial injection and amputation. The trial court concluded that the FDA gave only “passing attention” to the risks of IV administration in 2000. Ante, ¶ 4. The majority similarly concludes that the record does not indicate “that the FDA wished to preserve the use of IV push as a method of administering Phenergan.” Ante, ¶ 23. I cannot agree with this assessment of the record.
¶ 56. Both the original label and Wyeth’s proposed alternative were titled “INADVERTENT INTRA-ARTERIAL INJECTION.” On the original label, the first two sentences of the warning read:
Due to the close proximity of arteries and veins in the areas most commonly used for intravenous injection, extreme care should be exercised to avoid perivascular extravasation or inadvertent intra-arterial injection. Reports compatible with inadvertent intra-arterial injection of [Phenergan], usually in conjunction with other drugs intended for intravenous use, suggest that pain, severe chemical irritation, severe spasm of distal vessels, and resultant gangrene requiring amputation are likely under such circumstances.
*109On the proposed label, the first sentence of the warning read: “There are reports of necrosis leading to gangrene, requiring amputation, following injection of [Phenergan], usually in conjunction with other drugs; the intravenous route was intended in these cases, but arterial or partial arterial placement of the needle is now suspect.” While the proposed change to the warning language may not reflect what plaintiff would require in a warning, it cannot be disputed that Wyeth’s proposed alternative warning (1) placed greater emphasis on the risk of necrosis and amputation by referencing it in the first sentence, and (2) gave the FDA the opportunity to consider the specific, alternative warning advanced, by Wyeth, as well as the adequacy of the warning in general. Despite this opportunity, the FDA mandated that Wyeth retain the language of the existing warning. The alleged extent of the FDA’s consideration of the issue is not relevant, in my view.
¶ 57. In 2000, the FDA confirmed its assessment that health care professionals should be permitted to choose IV administration in its various forms as a means of delivering the drug, where appropriate. Wyeth could not both list all forms of IV administration as an approved use, as required by the FDA, and exclude all or some forms of IV administration as unsafe, as required by the jury’s verdict in this case. It would be impossible to comply with both requirements.
II. Obstacle to Federal Purposes and Objectives
¶ 58. I would further conclude that Wyeth has demonstrated actual conflict preemption by showing that plaintiffs state-law failure-to-warn claim poses an obstacle to federal purposes and objectives. The majority does not address this issue, concluding that Wyeth does not have the option of proving this form of actual conflict preemption. The majority reaches this conclusion by relying on the following clause in the 1962 amendments to the FDCA:
Nothing in the Amendments made by this Act to the Federal Food, Drug, and Cosmetic Act shall be construed as invalidating any provision of State law . . . unless there is a direct and positive conflict between such amendments and such provision of State law.
Ante, ¶ 26 (quoting Drug Amendments of 1962 (Harris-Kefauver Act), Pub. L. No. 87-781, §202, 76 Stat. 780, 793 (1962)). Citing *110Southern Blasting Services, Inc. v. Wilkes County, 288 F.3d 584, 591 (4th Cir. 2002), the majority concludes that the provision “essentially removes from our consideration the question of whether common-law tort claims present an obstacle to the purposes and objectives of Congress,” because the 1962 provision “simply restates the principle that state law is superseded in cases of actual conflict with federal law such that compliance with both federal and state regulations is a physical impossibility.” Ante, ¶ 27 (internal quotations omitted). “In other words,” the majority explains, “under any circumstances where it is possible to comply with both state law and the FDCA, the state law in question is consistent with the purposes and objectives of Congress.” Id. Thus, the majority eliminates the possibility of proving actual conflict preemption independently through the “obstacle” prong of that standard.
¶ 59. But neither the passage in Southern Blasting on which the majority relies nor the United States Supreme Court decision8 cited as authority in that passage provide an explanation or even an affirmative statement that the phrase “direct and positive conflict” in the 1962 amendment eliminates the “obstacle” prong of the actual conflict preemption standard. Thus, the majority eliminates one of the two means by which Wyeth may show actual conflict based on a single, unclearly-reasoned Fourth Circuit decision that is itself lacking in case law support. There is no basis for eliminating this prong of the actual conflict standard, and I disagree with the majority’s conclusion to the contrary.9
¶ 60. Assuming, then, that Wyeth may demonstrate actual conflict preemption by showing that state law is an obstacle to *111federal regulatory purposes and objectives, I believe the facts here support the conclusion that the state tort-law verdict in this case is preempted. The United States Supreme Court’s decision in Geier v. American Honda Motor Co., 529 U.S. 861 (2000), is controlling on the question of when state law poses an obstacle to federal purposes and objectives. In that case, the Department of Transportation had issued a safety standard that required automobile manufacturers “to equip some but not all of their 1987 vehicles with passive restraints.” Id. at 864-65. Among the optional passive restraints were air bags. Honda was in compliance with this standard. Nonetheless, the plaintiff was seriously injured in a car accident while driving a 1987 Honda that was not equipped with an air bag, but with another form of passive restraint. The plaintiff brought suit, alleging Honda was negligent in failing to install a driver’s-side air bag in the car. Honda argued that the federal safety standard preempted the plaintiffs state-law negligence claim. The Supreme Court held that a lawsuit premising negligence on the failure to install an air bag conflicted with the objectives of the federal safety standard and was therefore preempted. Id. at 866.
¶ 61. In reaching this conclusion, the Court noted that the plaintiff and the dissenting opinion — like the majority in the instant case — viewed the federal regulation as setting a minimum safety standard that states were free to supplement or strengthen. Id. at 874. However, by examining the comments accompanying the regulation, the Court concluded that a safety standard allowing a choice of passive restraint systems while not mandating any particular system was a deliberate decision that reflected a balance of diverse policy concerns. See id. at 875 (noting that allowing mix of available safety devices available over time would “lower costs, overcome technical safety problems, encourage technological development, and win widespread consumer acceptance”). “In sum, . . . the 1984 version of [the safety standard] embodies the Secretary’s policy judgment that safety would best be promoted if manufacturers installed alternative protection systems in their fleets rather than one particular system in every car.” Id. at 881 (quotations omitted). Accordingly, the Court concluded that the tort action sought to impose a duty *112on manufacturers to impose air bags, rather than other types of passive restraint systems, and that this state-law requirement was an obstacle to the federal objective of allowing a mix of safety devices.
¶ 62. Application of the Supreme Court precedent in Geier dictates the same result in this case. As with the DOT in Geier, the FDA is primarily concerned with public safety. The conclusion of what is best for public safety is arrived at by considering various policy factors that are sometimes in tension with one another. For example, in developing the safety regulation at issue in Geier, the DOT considered not only which passive-restraint systems were safest on an absolute scale, but which were most cost-effective and which would gain consumer acceptance. Similarly, here the FDA balances its assessment of a drug’s safety against concerns for the drug’s efficacy, taking into account that a safer but less effective drug is not necessarily best for the public health overall. See 21 U.S.C. § 355(d) (FDA must consider safety and efficacy); 21 U.S.C. § 393(b)(1), (b)(2)(B) (FDA’s mission is to protect public from unsafe drugs and to promote public health by approving regulated products in timely manner). In the specific context of warnings on drug labels, the FDA considers not only what information to include, but also what to exclude. As the Eighth Circuit has noted in the medical device context, “[tjhere are ... a number of sound reasons why the FDA may prefer to limit warnings on product labels.” See Brooks v. Howmedica, Inc., 273 F.3d 785, 796 (8th Cir. 2001). For example, “[wjarnings about dangers with less basis in science or fewer hazards could take attention away from those that present confirmed, higher risks.” Id.
¶ 63. No drug is without risks. The FDA balances the risks of a drug against its benefits to maximize the availability of beneficial treatments. The FDA’s decision in approving a drug, its uses and labeling reflect consideration of these and other policy factors. While a state-court jury presumably shares the FDA’s concern that drugs on the market be reasonably safe, the jury does not assess reasonableness in the context of public health and the associated risk-benefit analysis. A jury does not engage in a measured and multi-faceted policy analysis. Rather, a jury views the safety of the drug through the lens of a single patient who has already been catastrophically injured. Such an approach is virtually guaranteed to provide different conclusions in different *113courts about what is “reasonably safe” than the balancing approach taken by the FDA. In act, different conclusions were reached in this case.
¶ 64. The jury in this case was instructed that “[a] prescription drug is unreasonably dangerous due to inadequate warnings or instructions if reasonable instructions regarding foreseeable risks of harm are not provided to the physician and other medical professionals who are in a position to reduce the risks of harm.” Faced with plaintiff’s tragic injuries, the jury concluded that allowing Phenergan to be delivered through IV administration was “unreasonably dangerous.” The jury’s verdict conflicts squarely with the FDA’s assessment of precisely the same issue: whether Phenergan is safe and effective when delivered through IV administration. The claim is preempted.
¶ 65. For the above reasons, I dissent.
This is also the approach employed by the numerous federal district court decisions cited by the majority. Ante, ¶ 14. Because I disagree with this analysis of the import of § 314.70(c), I do not find these decisions to be persuasive. Instead, I side with the minority view expressed in Needleman, which concludes that § 314.70(c) gives manufacturers very little latitude in unilaterally revising drug labels. Needleman v. Pfizer, Inc., 2004 WL 1773697, at *3 (N.D. Tex.).
See Hillsborough County v. Automated Med. Labs., Inc., 471 U.S. 707, 713 (1985). The cited passage in Hillsborough does not interpret the phrase “direct and positive conflict.” It merely cites the different forms of preemption, including the “obstacle” prong.
It is worth noting that the federal statute at issue in Geier v. American Honda Motor Co., 529 U.S. 861 (2000) (discussed below), contained an even broader savings clause than the 1962 amendment to the FDCA. The provision in Geier stated simply that the federal safety standard at issue did “not exempt any person from any liability under common law.” Id. at 868 Nonetheless, the Court concluded that all ordinary preemption principles — including actual conflict preemption and the obstacle prong of the standard — applied. The Court rejected the notion that Congress would so limit the effect of preemption as to allow an actual conflict with a federal objective: “Insofar as petitioners’ argument would permit common-law actions that ‘actually conflict’ with federal regulations, it would take from those who would enforce a federal law the very ability to achieve the law’s congression*111ally mandated objectives that the Constitution, through the operation of ordinary pre-emption principles, seeks to protect.” Id. at 872.