CALABRESI, Circuit Judge.

The United States District Court for the Southern District of New York (Kaplan, J.) granted Defendants' motion to dismiss for failure to state a claim under Fed.R.Civ.P. 12(b)(6), and Plaintiffs appeal. Because Plaintiffs' suit involves claims of damages that were caused directly by Defendants' alleged fraud, we reverse and remand for further proceedings.

I.

A. The Parties

Defendant Warner-Lambert is a pharmaceutical company incorporated in Delaware with headquarters in New Jersey. Parke-Davis Company is an unincorporated division of Warner-Lambert, and also has its headquarters in New Jersey. From February 1997 through March 2000, Defendants marketed and sold Rezulin, an oral drug that was approved in January 1997 by the FDA to treat Type II (adult onset) diabetes, a disease from which more than 15 million Americans suffer.

Plaintiff Louisiana Health Service Indemnity Company ("Blue Cross") is a Blue Cross/Blue Shield health benefit provider ("HBP") with headquarters in Baton Rouge, Louisiana. Blue Cross filed a class action suit in the United States District Court for the Eastern District of Louisiana on August 29, 2000; the suit was subsequently transferred to the Southern District of New York by the judicial panel on multidistrict litigation.

Plaintiff Eastern States Health and Welfare Fund ("Eastern"), an ERISA plan that provides benefits to members of the Needletrades, Industrial, and Textile Employees Union, is based in New York. Eastern filed suit in the Southern District of New York in March 2001. The two suits were consolidated by the district court in a pre-trial order in April 2001, and the Plaintiffs filed a consolidated complaint the same month.

B. Allegations in Plaintiffs' complaint

According to Plaintiffs' complaint, the allegations of which must be accepted as true for the purposes of a motion to dismiss, see Conley v. Gibson, 355 U.S. 41, 45-46, 78 S.Ct. 99, 2 L.Ed.2d 80 (1957), Rezulin first gained prominence in June 1996, when the National Institutes of Health (NIH) announced a $150 million clinical study to test the drug's effectiveness. The NIH's decision to include Rezulin in the study was heavily influenced by Dr. Jerrold Olefsky, who held prominent positions in the NIH, and by Dr. Richard Eastman, the NIH's senior diabetes researcher who had overall responsibility for the study. Both doctors had financial interests in promoting the drug: Dr. Olefsky is listed as inventor or co-inventor on three patents regarding Rezulin's use for preventing diabetes and was co-chair of a Warner-Lambert-created group that promoted the drug; Dr. Eastman was a board member of two organizations financed by Warner-Lambert and the drug's developer. And, as it turned out, Dr. Eastman received more than $78,000 in undisclosed compensation from Warner-Lambert while he was overseeing the study. When these payments were brought to light in December 1998 — prompting a conflict-of-interest investigation by the Inspector General's office — Dr. Eastman was forced to resign from the NIH.

Warner-Lambert submitted a New Drug Application ("NDA") for Rezulin to the FDA on July 1, 1996, and the FDA agreed to give the drug a "fast track" (six month) review. The FDA initially assigned Dr. John L. Gueriguian as the chief medical officer to oversee the review. After concluding that the drug posed serious health and safety concerns, Dr. Gueriguian recommended against approval in a detailed written evaluation dated October 9, 1996. Specifically, Dr. Gueriguian found "very worrisome liver toxicity," and concluded that Rezulin "offered very little therapeutic advantage" over existing diabetes medications. He noted that Warner-Lambert's clinical trial data showed that Rezulin users developed liver problems at four times the rate of control patients who received a placebo, and also expressed concern about potential cardiovascular damage from the drug. Dr. Gueriguian met with Warner-Lambert representatives in September 1996 and suggested the drug's prospects were not good. Warner-Lambert then met with Dr. Gueriguian's superiors at the FDA who, in November 1996, removed Dr. Gueriguian from the Rezulin review.

Dr. Gueriguian was replaced by Dr. G. Alexander Fleming, whose medical review noted that Warner-Lambert's clinical trials had "identified significant safety issues" and "suggested unpredictable damage associated with Rezulin." Nevertheless, Dr. Fleming's presentation to the FDA's Endocrinologic and Metabolic Drugs Advisory Committee made no mention of either Dr. Gueriguian's reservations about the drug, or of the problematic results of Warner-Lambert's clinical trials. Three advisory committee members later told the press that, if they had been informed of the adverse liver results, they would have required liver-function monitoring as a condition of approval. The advisory committee voted unanimously (8-0) to recommend approval of Rezulin and, on January 29, 1997, the FDA approved the drug, but only for use in combination with insulin or with metformin or sulfonylurea drugs. In August 1997, this approval was broadened to allow the use of Rezulin as a "stand-alone" therapy.

Warner-Lambert marketed Rezulin aggressively, and priced it at nearly three times the cost of appropriate treatments by other drugs. It touted Rezulin as "the first anti-diabetes drug designed to target insulin resistence." That statement prompted the FDA to accuse Warner-Lambert of making "false and misleading" claims, and to recommend that the company "immediately discontinue" circulating new releases containing the claim. More significant to this litigation, Warner-Lambert published two full-page color advertisements — one in the May 1, 1997 issue of The New England Journal of Medicine, and one in the June 19, 1997 Washington Post — describing Rezulin as a drug with breakthrough effectiveness, and "Side Effects Comparable to Placebo." The company allegedly made this statement while knowing that its own clinical trial data showed Rezulin users were three to six times more likely to suffer liver injury than patients taking the placebo.

By July 1997, seven people receiving Rezulin had died from the same side effects that Warner-Lambert had observed in its pre-market Rezulin tests. By the fall of 1997, the FDA began to receive reports of Rezulin patients suffering serious liver injuries, including death following liver failure. The FDA appointed a Dr. Misbin to evaluate the adverse event reports. He estimated that more than 12,000 Rezulin users would experience some liver injury, and that 2,000 of those patients might die unless their liver functions were closely monitored.

Warner-Lambert sent a letter to doctors in October 1997, reporting that thirty-five Rezulin users had experienced "idiosyncratic hepatocellular injury" ranging from mild effects to "one liver transplant and one death." The letter said these reports were "RARE," but nevertheless recommended that liver function be tested within the first month or two of Rezulin use, every three months for the first year thereafter, and then periodically.

On December 1, 1997, Warner-Lambert sent a second letter to healthcare professionals, announcing a label change on the drug which recommended more frequent liver-function monitoring for patients using Rezulin. The letter suggested that the FDA remained an advocate of Rezulin and a partner with Warner-Lambert in assuring safety to the drug's users.

On that same day, December 1, the sale of Rezulin's active ingredient was banned in the United Kingdom, based on a conclusion by the company marketing the drug that "there was no way to predict which Rezulin patients would be harmed, and the pace of liver injuries and deaths was by that point `unacceptably high.'" Warner-Lambert reacted by issuing a press release on December 7 stating it was "disappointed with the mischaracterization of its actions and intentions regarding the development and marketing of Rezulin," and that it was "particularly disturbed" by media reports that focused on the drug's risks while ignoring its significant benefits. Moreover, Warner-Lambert reaffirmed that the FDA supported the drug, and noted that the injured patients "may not have been monitored as recommended in the labeling."

Throughout this period, the NIH study of Rezulin continued. In May 1998, Audrey Jones, a healthy, non-diabetic participant in the study who had taken Rezulin, died, following a liver transplant made necessary by the irreversible liver failure she had developed. Mrs. Jones had undergone the monitoring tests that Warner-Lambert recommended. Warner-Lambert issued a press release stating that Mrs. Jones had died "apparently due to complications unrelated" to her Rezulin use. This report, however, was contradicted by NIH physicians, who concluded that the patient's death had "probably" been caused by Rezulin. The following month, citing concerns about the safety of the remaining 580 participants in the study, Rezulin was withdrawn from the NIH study.

In July 1998, Warner-Lambert sent a third letter to health care professionals announcing another "label change" for Rezulin. The letter noted that a "few new cases of hepatocellular events" since 1997 had occurred in "patients who may not have been monitored as recommended in the labeling." The statement allegedly was made despite the company's knowledge that Mrs. Jones had been tested and carefully monitored, in accordance with Warner Lambert's recommendation, by the NIH. Under the new label change, liver testing was required for all Rezulin patients monthly for the first eight months, every two months for the remainder of the first year, and periodically after that. Despite the new label calling for additional monitoring, the reports of patients suffering liver failure continued to rise. In November 1998, two patients who had been monitored as recommended by the safety labels then in effect developed liver failure within two weeks of the tests, and died.

The complaint alleges that Warner-Lambert did not stop misdescribing the drug as safe. In a March 1999 article in the Los Angeles Times, a Warner-Lambert vice president asserted that he was unaware of any deaths related to Rezulin after the July 1998 label change. But, in fact, Warner-Lambert had itself reported to the FDA the deaths of thirty-one Rezulin users between July 28 and December 17, 1998.

On March 26, 1999, the FDA convened a second advisory committee meeting to evaluate Rezulin. At that meeting, a senior FDA epidemiologist verified that Rezulin had been implicated by the FDA in forty-three reported cases of liver failure, and that the drug could be expected to cause liver failure in one out of every 1,800 patients. The committee recommended that the FDA withdraw its approval of Rezulin as a stand-alone therapy. On that same day, Warner-Lambert sent out a series of press releases claiming, inter alia, that the new labeling had substantially reduced the reported rate of liver events.

In June 1999, the FDA announced that it was allowing Rezulin to remain on the market, but was requiring another label change. Under the new label, Rezulin could no longer be used as a stand-alone therapy, and the required liver-function monitoring was again increased. Patients using the drug were required to undergo liver-function testing before beginning treatment, monthly for the first 12 months, and quarterly during the second year.

Despite these changes, mortality statistics of Rezulin users continued to climb, and Warner-Lambert allegedly continued to lie about them. In February 2000, the company released a statement claiming that it believed there were no liver-failure deaths attributable to Rezulin after the June 1999 label change. The day after this statement was made, the FDA discredited it, saying the agency had been informed of six cases of liver failure with onset after July 1999, of which at least three resulted in death.

In March 2000, doctors at the FDA and elsewhere became even more concerned about Rezulin, with one doctor writing to others that "at each juncture in the management of Rezulin's liver failure risk, hindsight shows that [the monitoring] had little or no effect and that Warner-Lambert's assertions that the liver failure problem was solved were proved false." Another doctor, who had formerly been a principal researcher in Warner-Lambert 1994 Echo study, wrote to Senator Edward Kennedy that she "believe[d] that the company . . . deliberately omitted reports of liver toxicity and misrepresented serious adverse events experienced by patients in their clinical studies." Moreover, an article in the March 10, 2000 Los Angeles Times reported that the FDA had linked Rezulin to eighty-nine liver failures, including sixty-one deaths.

On March 21, 2000, Warner-Lambert withdrew Rezulin from the U.S. market. The withdrawal was at the request of the FDA, which had concluded that "continued use of Rezulin now poses an unacceptable risk to patients." Since then, the FDA has acknowledged reports of liver damage occurring after patients were taken off the drug, which indicates that the risk of harm does not disappear with the withdrawal of the drug.

C. Plaintiffs' claims and relief sought

In their April 2001 consolidated complaint, Plaintiffs sought class relief on behalf of all health benefit providers that paid for Rezulin during the period between February 1997 and April 2001. According to the complaint, the vast majority of the purchase price of prescription drugs, including Rezulin, is paid directly by health benefit providers, pursuant to contracts that exist between HBPs and virtually every retail and mail-order pharmacy in the United States. Plaintiffs assert that members of the class paid approximately $1.4 billion to purchase the drug. This amount includes only the portion of the prescription paid by the HBPs and excludes the part paid by the patients, in the form of a "co-pay." A one-month prescription of Rezulin cost approximately $150, of which the typical HBP paid about $135 (with the patient paying the rest). Prior to the introduction of Rezulin, the most commonly prescribed oral drug therapy for Type II diabetes was metformin, which had a prescription cost of approximately $55 per month, of which the typical HBP paid about $50.

In their complaint, Plaintiffs sought to recover the moneys they spent purchasing Rezulin. Asserting that the suit is governed by New Jersey law, the complaint alleges a) knowing concealment, suppression or omission of material facts, in violation of section 56:8-2 of the New Jersey Consumer Fraud Act (count 1); b) breach of express warranties and of the implied warranty of merchantability, in violation of New Jersey's Commercial Code sections 2-313 and 2-314 (counts 2 and 3); and c) Unjust Enrichment (count 4).

D. The Proceedings Below

On October 22, 2001, the district court (Lewis A. Kaplan, J.) granted Defendant's 12(b)(6) motion to dismiss. The court rested its holding squarely on its analysis of two decisions, Holmes v. Securities Investor Protection Corp., 503 U.S. 258, 112 S.Ct. 1311, 117 L.Ed.2d 532 (1992), and Laborers Local 17 Health and Benefit Fund v. Philip Morris, Inc., 191 F.3d 229 (2d Cir. 1999). It concluded that, under those cases, Plaintiffs could not establish that Defendants' actions were the proximate cause of Plaintiffs' damages.

In Laborers Local 17, plaintiff HBPs sued several major tobacco companies, alleging that the companies had conspired to deceive the public and the plaintiffs concerning the health risks of smoking. Plaintiffs sought to recover the money they had spent treating their insureds' smoking-related illnesses. 191 F.3d at 232, 233. A panel of this court first noted that the plaintiffs' RICO claims required a showing of "proximate cause," and stated that, in making the determination of whether proximate cause existed, courts should apply "common-law principles," id. at 234 (citing Holmes, 503 U.S. at 267-68, 112 S.Ct. 1311), which, it found, required the elements of (1) direct injury, and (2) foreseeability. Id. at 235-36. As to the first element, the court said that "the critical question posed by the direct injury test is whether the damages a plaintiff sustains are derivative of an injury to a third party. If so, then the injury is indirect; if not, it is direct." Id. at 238-39. Applying that test to the facts before it, the court concluded that the plaintiffs' injuries arose only from the injuries sustained by its insureds:

Without injury to the individual smokers, the [plaintiffs] would not have incurred any increased costs in the form of the payment of benefits, nor would they have experienced the difficulties of cost prediction and control that constituted the crux of their infrastructure harms. Being purely contingent on harm to third parties, these injuries are indirect. Consequently, because defendants' alleged misconduct did not proximately cause the injuries alleged, plaintiffs lack standing to bring RICO claims against defendants.

Id. at 239. The Laborers Local 17 court also determined that the plaintiffs' fraud and special duty claims under New York common law, like its RICO claims, were barred by their inability to show proximate cause. Id. at 243.

The district court in the instant case described the claims in Laborers Local 17 as "closely analogous" to the claims brought by the HBPs:

Putting aside the question whether the HBPs "bought" Rezulin, the crux of their complaint is that Warner-Lambert marketed a defective product for use by their insureds, that it failed to disclose information undermining implicit or explicit representations as to safety, and that the demand for the product — and thus the payments the HBPs made to provide the drug to or to reimburse its insureds for it — were higher than they would have been absent the alleged torts. Thus, absent the alleged misconduct vis-a-vis the insureds and their health care providers, the HBPs would not have been injured. Accordingly, Laborers Local 17 controls here, and the complaint must be dismissed.

In Holmes, the plaintiff was the Securities Investor Protection Corporation ("SPIC"), a private non-profit corporation whose members included most broker-dealers registered under the Securities and Exchange Act, Holmes, 503 U.S. at 261, 112 S.Ct. 1311, and which insures investors against losses when a member organization is unable to meet its obligations to its investors. When an SIPC member was the victim of a fraudulent scheme by Holmes and, as a result, went into default, the SIPC took over and spent about $13 million to compensate the affected investors. Id. at 262-63, 112 S.Ct. 1311. The SIPC then, as subrogee of the investors, sued Holmes for violations of SEC rules, mail and wire fraud statutes, and the RICO Act. Id. The Ninth Circuit allowed the case to go forward, concluding that the investor's losses were proximately caused by Holmes's fraud. Id. at 264, 112 S.Ct. 1311.

The Supreme Court reversed. Id. at 265, 112 S.Ct. 1311. The Court first analyzed the RICO statute and concluded that it required not only "but for" causation, but also proximate causation. Id. at 268, 112 S.Ct. 1311. The Court (still concerned with the requirements of RICO) stated that proximate cause demands "some direct relation between the injury asserted and the injurious conduct alleged." Id. It then provided three policy rationales for this requirement of "directness":

First, the less direct an injury is, the more difficult it becomes to ascertain the amount of a plaintiff's damages attributable to the violation, as distinct from other, independent, factors. Second, quite apart from problems of proving factual causation, recognizing claims of the indirectly injured would force courts to adopt complicated rules apportioning damages among plaintiffs removed at different levels of injury from the violative acts, to obviate the risk of multiple recoveries. And, finally, the need to grapple with these problems is simply unjustified by the general interest in deterring injurious conduct, since directly injured victims can generally be counted on to vindicate the law as private attorneys general, without any of the problems attendant upon suits by plaintiffs injured more remotely.

Id. at 269-70, 112 S.Ct. 1311 (citations omitted).

The district court in the instant case used Holmes as an alternate rationale for its decision, noting that "even if Laborers Local 17 were not controlling," Plaintiffs would have failed to establish proximate cause under the three factors articulated in Holmes.

As to the first factor, the court assumed that "it appears to be common ground that [Rezulin] did not harm, and indeed benefitted, a great many" and, given this assumption, asserted that "[c]ertainly the HBPs have no claim to recover the cost of providing the drug to patients who were not harmed." It concluded that the extent of Plaintiffs' injuries therefore "depends upon whether each insured for whose Rezulin they paid benefitted from or was harmed by Rezulin, a matter that will turn on the individual proof as to each patient for whom it was prescribed." On the second Holmes factor, the district court characterized the HBPs as "essentially financial intermediaries which," the court said, "doubtless passed on all or much of the costs they incurred to employers and perhaps even to insureds themselves." For that reason, the court concluded that "the apportionment problems that would be created by allowing the HBPs to sue here would be monumental." Finally, stating that "[t]here simply is no reason why the individuals who allegedly were injured by Rezulin cannot recover the costs of the drug and any subsequent diagnostic testing or monitoring," the court held that the third Holmes factor also weighed against the Plaintiffs.

II.

We review de novo a district court's grant of a motion to dismiss pursuant to Rule 12(b)(6). Todd v. Exxon Corp., 275 F.3d 191, 197 (2d Cir. 2001). "On a motion to dismiss for failure to state a claim, we construe the complaint in the light most favorable to the plaintiff, accepting the complaint's allegations as true." Id. "A complaint should not be dismissed for failure to state a claim `unless it appears beyond doubt that the plaintiff can prove no set of facts in support of his claim which would entitle him to relief.'" Id. at 197-98 (quoting Conley v. Gibson, 355 U.S. 41, 45-46, 78 S.Ct. 99, 2 L.Ed.2d 80 (1957)). "Thus, the issue is not whether a plaintiff will ultimately prevail but whether the claimant is entitled to offer evidence to support the claims." Id. at 198 (citation and quotation marks omitted).

A.

We note as a threshold matter that the district court seems to have acted on the assumption that the relevant legal standard to be applied in this case is the law of proximate cause as stated by the Supreme Court in Holmes, and by us in Laborers Local 17. We do not agree with that view. Both of those cases arose under RICO, and therefore the applicable standard of causation in those cases was the standard that was established by that federal statute. See Abrahams v. Young Rubicam Inc., 79 F.3d 234, 237 (2d Cir. 1996) (noting that, in a suit seeking recovery granted by a statute, liability does not "depend on whether there is proximate causation as that term is used at common law," because "[w]ith statutory claims, the issue is, instead, one of statutory intent").

In Holmes, the Supreme Court interpreted the RICO statute as not only requiring "but-for" causation, but also as incorporating what it termed "common-law principles of proximate causation." Holmes, 503 U.S. at 267, 112 S.Ct. 1311. It then proceeded to define — among many possible versions of common-law proximate cause — what the proximity requirements of RICO were. Id. at 268-69, 112 S.Ct. 1311. We, of course, applied the Supreme Court's interpretation in Laborers Local 17, since it also involved a claim under RICO. Because the issue was one of statutory interpretation, the proximity requirements could have been broader or narrower than those applied under the common law of any particular state. See Abrahams, 79 F.3d at 237 n. 3 (noting that, while the proximity requirements for statutory liability are usually narrower than most common-law rules governing proximate cause, a legislature is free to "go beyond the common law and create rights of recovery for plaintiffs who are not foreseeable and who are injured by defendants' wrongdoing," and concluding that "[w]ere such a statute in issue, substantial problems could arise from the continued use of `proximate cause' language to define when plaintiffs are meant by the legislature to be given a cause of action"). In fact, the proximate cause requirements of RICO were more stringent than those of most states. See Moore v. PaineWebber, Inc., 189 F.3d 165, 178, 179 (2d Cir. 1999) (Calabresi, J., concurring) (explaining that "a RICO plaintiff's burden to show that his case meets the common-law requirement of proximate causation derives from the legislature's intent to impose that causation requirement, and not directly from the common law itself," and noting that "[i]n practice, our cases have held RICO plaintiffs to a more stringent showing of proximate cause than would be required at common law").

Unlike Holmes and Laborers Local 17, the legal standard of proximate cause that is relevant to the case before us is not the law of RICO; it is, rather, the law of New Jersey. And Plaintiffs strenuously argue that they meet all the requirements of proximate cause under New Jersey's common law. Defendants, on the other hand, do not address this point at any length. Our task is, therefore, to determine what New Jersey's common law of proximate cause requires. In cases involving state law, if we believe the issue to be uncertain, we have the option to certify the question to the state's highest court so that it can tell us definitively what the law is. Certification is not available in this case, however, because the Supreme Court of New Jersey allows certification only from the Third Circuit. See N.J. Court Rules, R.2:12A-3 (2002).

Plaintiffs contend that the law of New Jersey on proximate cause does not have the relatively narrow directness requirements applied by the federal courts under RICO in Holmes and Laborers Local 17. Defendants' position, charitably read, is that New Jersey's common-law proximate cause requirements mirror those of Holmes and Laborers Local 17. On this issue, we are much inclined to agree with Plaintiffs. But even assuming arguendo that Defendants are correct, their argument is to no avail, for the situations in Holmes and Laborers Local 17 are significantly different from the instant case.

B.

In Laborers Local 17, the tobacco companies' alleged tort directly harmed only the smokers, who suffered both a health injury (smoking-related illness) and an economic injury (the purchase price of the fraudulently marketed cigarettes). The smokers' health injuries, in turn, caused economic losses to the insurance companies, who had to reimburse patients for the cost of their smoking-related illnesses. That case was therefore clearly one in which the plaintiffs' damages were entirely derivative of the injuries to their insured. For, as we there noted, "[w]ithout injury to the individual smokers, the [plaintiffs] would not have incurred any increased costs." Laborers Local 17, 191 F.3d at 239.

In the instant case, instead, Plaintiffs allege an injury directly to themselves; an injury, moreover, that is unaffected by whether any given patient who ingested Rezulin became ill. Plaintiffs' claim is that the Defendants' wrongful action was their misrepresentation of Rezulin's safety, and that this fraud directly caused economic loss to them as purchasers, since they would not have bought Defendants' product, rather than available cheaper alternatives, had they not been misled by Defendants' misrepresentations. Thus the damages — the excess money Plaintiffs paid Defendants for the Rezulin that they claim they would not have purchased "but for" Defendants' fraud — were in no way "derivative of damage to a third party."

Defendants argue, however, that "[i]f Rezulin had been effective in all diabetic patients without any side effects, plaintiffs would have no basis for a claim." But it is easy to see how Defendants' reasoning is flawed. Consider, for example, a hypothetical in which a defendant drug company markets a "new," much more expensive drug claiming it is a great advancement (safer, more effective, etc. than metformin — the standard diabetes drug) when in fact the company is simply replicating the metformin formula and putting a new label on it. In other words, the only difference between metformin and the "new" drug is the new name and the higher prescription price (paid almost entirely by the insurance company). In that case, the "new" drug would be exactly as safe and effective as metformin, and thus there could be no injury to any of the insurance company's insured. Nevertheless, the insurance companies would be able to claim — precisely as they do here — that the defendants engaged in a scheme to defraud it, and that the company suffered direct economic losses as a result.

C.

Nor does Holmes help Defendants' argument. First, since Plaintiffs' claims do not rely on injuries sustained by Rezulin users, the first prong of the Holmes test for liability is satisfied. As explained above, Plaintiffs allege (1) that Warner-Lambert misrepresented its product to Plaintiffs; and (2) that, because of these misrepresentations, the HBPs overpaid to purchase Rezulin. No consequential injury is, therefore, at issue. Second, Plaintiffs' suit raises no apportionment difficulties because each HBP and its patient co-payer has its own, segregable, claim for economic harm, to the extent of their respective co-pay. Finally, as to the third prong in Holmes, Plaintiffs correctly note that the HBPs, as 90 percent co-payers, are the parties with the largest direct economic injury, and are the most motivated parties to pursue recovery from the $1.4 billion they paid for Rezulin.

D.

The district court characterized Plaintiffs as "essentially financial intermediaries," rather than direct purchasers of Rezulin. Although this court has not to date held that insurance companies are, in all instances, the "purchasers" of the drugs for which they reimburse pharmacies, we, like several other courts, have indicated that in a variety of contexts they are the buyers. See, e.g., Med. Arts Pharmacy of Stamford, Inc. v. Blue Cross Blue Shield of Conn., Inc., 675 F.2d 502, 505 (2d Cir. 1982) ("[T]he Pharmacy Agreements . . . are merely arrangements for the purchase of goods and services by Blue Shield.") (quoting Group Life Health Ins. Co. v. Royal Drug Co., 440 U.S. 205, 214, 99 S.Ct. 1067, 59 L.Ed.2d 261 (1979)); Blue Cross Blue Shield United of Wis. v. Marshfield Clinic, 65 F.3d 1406, 1414 (7th Cir. 1995) (holding that insurance companies had standing, as the "direct purchaser[s]," to maintain an antitrust suit). Moreover, and more directly relevant to this case, perhaps, Plaintiffs point out that this and other courts have long recognized the right of HBPs to recover from drug companies amounts that were overpaid due to illegal or deceptive marketing practices. See, e.g., Hartford Hosp. v. Chas. Pfizer Co., 52 F.R.D. 131, 133 (S.D.N.Y. 1971) (approving $10 million class action settlement of antitrust claims brought by insurance plans against drug companies). In any event, on a motion to dismiss pursuant to Fed.R.Civ.P. 12(b)(6), the court is obliged to accept as true Plaintiffs' assertion that they were, in fact, the purchasers of the drug. It follows that the district court erred in its characterization.

III. CONCLUSION

In dismissing Plaintiffs' claims on a 12(b)(6) motion, the district court erred by treating Plaintiffs' complaint as one analogous to the insurance companies' claims in Laborers Local 17. Instead, the court should have considered the complaint as it was presented. Under this characterization, the insurance companies were the direct victims of Defendants' fraudulent marketing. As such, their complaint must surely survive a 12(b)(6) motion. The district court's order dismissing Plaintiffs' claims is therefore VACATED, and the case is REMANDED for further proceedings consistent with this opinion.