Alaska Elec. Pension Fund v. Pharmacia Corp. , 554 F.3d 342 ( 2009 )

554 F.3d 342 (2009)

ALASKA ELECTRICAL PENSION FUND; City of Sarasota Firefighters' Pension Fund; International Union of Operating Engineers Local 132 Pension Plan; New England Health Care Employees Pension Fund; Pace Industry Union-Management Pension Fund; Chemical Valley Pension Fund of West Virginia, as Class Representatives, on behalf of themselves and all other similarly situated
v.
PHARMACIA CORPORATION; Pfizer, Inc.; Fred Hassan; Dr. G. Steven Geis; Carrie Cox, Alaska Electrical Pension Fund; City of Sarasota Firefighters' Pension Fund; International Union of Operating Engineers Local 132 Pension Plan; New England Health Care Employees Pension Fund; Pace Industry Union-Management Pension Fund; Chemical Valley Pension Fund of West Virginia, Appellants in 07-4500
Pharmacia Corporation; Pfizer, Inc.; Fred Hassan; Dr. G. Steven Geis; Carrie Cox, Appellants in 07-4564.

Nos. 07-4500, 07-4564.

United States Court of Appeals, Third Circuit.

Argued: November 21, 2008. Opinion Filed: January 30, 2009.

Eric A. Isaacson, Esq. (Argued), Arthur C. Leahy, Esq., Matthew P. Montgomery, Esq., Coughlin, Stoia, Geller, Rudman & Robbins, San Diego, CA, Peter S. Pearlman, Esq., Cohn, Lifland, Pearlman, Herrmann & Knopf, Saddle Brook, NJ, for Appellants/Cross Appellees.

Gregg S. Levin, Esq., Joshua C. Littlejohn, Esq., Joseph F. Rice, Esq., Ann K. Ritter, Esq., Motley Rice, Mount Pleasant, SC, for Appellant/Cross-Appellee Pace Industry Union-Management Pension Fund.

Jonathan M. Hoff, Esq. (Argued), Jason M. Halper, Esq., Joshua R. Weiss, Esq., Cadwalader, Wickersham & Taft, New York, NY, William A. Dreier, Esq., Steven A. Karg, I, Esq., Robert Mahoney, Esq., Norris, McLaughlin & Marcus, Bridgewater, NJ, for Appellees/Cross Appellants.

Before: BARRY, CHAGARES, Circuit Judges, and RESTANI,^[*] Judge.

OPINION OF THE COURT

BARRY, Circuit Judge.

In this securities fraud class action, plaintiffs allege that defendants violated §§ 10(b) and 20(a) of the Securities Exchange Act of 1934 by making, with scienter, materially false statements about a clinical study of Celebrex, a popular anti-inflammatory medication. In particular, defendants are alleged to have misled investors by distorting the study's results with the intent to show that Celebrex had a better safety profile than similar medications. Finding the action to be untimely, the District Court granted summary judgment to defendants. This appeal and cross-appeal followed.

I. Factual Background

Celecoxib, known and marketed commercially as Celebrex, is an anti-inflammatory prescription drug sold by defendant Pharmacia Corporation. Substantially more expensive than many other nonsteroidal anti-inflammatory drugs ("NSAIDs"), Celebrex's promise was rooted in the hope that it would cause fewer gastrointestinal ("GI") side-effects than the less costly NSAIDs.^[1] To help the hope become a reality, defendants^[2] commissioned a long-term clinical study of Celebrex's effect on the GI system, the Celecoxib Long-term Arthritis Safety Study ("CLASS study"). This litigation focuses on the aftermath of that study.

According to the complaint, the results of the CLASS study were a disappointment to defendants: Celebrex did not show the desired reduction in GI side-effects as compared to the other drugs studied. Fearing a decrease in sales and stock price, defendants allegedly undertook to distort the results of the study so that it would appear that Celebrex possessed a better GI safety profile than, in fact, it did. Towards this end, in April 2000, defendants released only the results from the first six months of the CLASS study; those results, divorced from the entire set of data, were capable of positive construction.

Defendants released the truncated results of the CLASS study with great fanfare, declaring that the study "shows that Celebrex has a truly exceptional safety profile," and that "the long-term outcome data paints a clear and compelling picture of Celebrex's safety versus NSAIDs." (Joint App. ("JA") 59, 64.) Some documents issued by defendants noted that the CLASS study lasted a full thirteen months, but the reason for excising the last seven months from the analysis was not revealed.

Scientists affiliated with defendants then drafted an article based on the truncated results and submitted it for publication to the Journal of the American Medical Association ("JAMA"). As would become known only later, however, neither defendants nor the article's authors informed JAMA that the data in the article was incomplete. In September 2000, JAMA published the article, which reached the following conclusion: "The overall incidence of GI symptoms experienced by patients taking [Celebrex] was significantly lower than by those taking NSAIDs, as was the rate of withdrawal [from the study] due to GI intolerability." (Id. at 45.)

Defendants hoped to convince the FDA to allow Celebrex to be marketed without the standard GI warning label required for other NSAIDs, and so submitted the complete data from the CLASS study to the FDA in June 2000. FDA staff members, in preparation for hearings on the warning label issue, reviewed the data. On February 6, 2001, the reports of those staff members were published on the FDA's website, alongside defendants' own report. Defendants' report defended the decision to use only the truncated data, asserting that data after the six-month point was biased in favor of the comparator drugs:

The GI safety data presented are for the six-month treatment timepoint based on the analysis of risk factors prespecified in the protocol. In brief, a disproportionate withdrawal of patients at high risk of an ulcer complication from the entire study was observed after six months (depletion of susceptibles). Additionally, a significantly greater withdrawal of patients on diclofenac for GI intolerance occurred during the initial six months of the study. The withdrawal of patients for GI intolerance prematurely removed a group at high risk for ulcer complications and symptomatic ulcers from the diclofenac treatment arm (informative censoring).

(Id. at 381.)

The FDA staff reports, stating in part as follows, disagreed with defendants' reliance on the truncated data:

• A rheumatologist's report stated that it was "unclear" that the rationale put forth by defendants "represented a significant bias in assessment of the outcome." (Id. at 631.)

• A gastroenterologist's report stated that defendants' "rationale for analyzing the first 6 months as a meaningful endpoint independent of the success at the study completion is not convincing." (Id. at 472.) However, this report also stated that "[t]he six-month analysis will be reviewed only as a potentially supportive analysis." (Id. at 482.)

• A statistician's report rejected the six-month analysis as "not valid," and asserted that there was "no reason to include information only in the first 6 months." (Id. at 666.)

These reports were prepared to assist the FDA's Arthritis Advisory Committee ("Advisory Committee") in deciding whether to recommend the label change sought by defendants. The day after the publication of the reports, the Advisory Committee held hearings on the issue, and ultimately declined to recommend the label change. The staff reports and the Advisory Committee's recommendation received substantial media attention. The market also took note of the disappointing outcome: between February 6-8, 2001, the price of Pharmacia's stock dropped approximately 9.0%.

After these events, defendants issued a series of positive statements about Celebrex's GI safety profile as well as about the chances for a label change. Defendants claimed, for example, that the CLASS study data presented a "compelling case" for a label change, and that because the CLASS study was "an extremely rigorous and complex trial," it was "difficult for the [Advisory Committee] to analyze." (Id. at 171, 1519.)

Financial analysts also continued to rate Pharmacia's stock positively. Even while noting the reduced chances for a label change and the disagreement over the results of the CLASS study, JP Morgan, Merrill Lynch, Lehman Brothers, and Bank of America all continued to rate Pharmacia's stock as a "buy" or "strong buy." Several analysts noted the challenge to defendants' use of truncated data: JP Morgan wrote that the staff reports called the six-month analysis "unjustified and invalid," and Bloomberg News wrote that defendants were only able to show a benefit "by looking at selected parts of the data — a practice discouraged by the [FDA]." (Id. at 714, 719.)

Months later, on August 5, 2001, the Washington Post reported that defendants had withheld the full CLASS study data from JAMA. In the article, JAMA's editor described herself as "disheartened" and stated that "a level of trust ... was, perhaps, broken." (Id. at 203.) Additionally, a scientist who wrote an editorial published in conjunction with the JAMA article stated that he was "flabbergasted" when he saw the complete data; another scientist "said he complained to JAMA after noticing differences between the published [JAMA] report and the data presented to the FDA." (Id. at 203-04.)

After the Washington Post article raised the red flag of impropriety, other sources began to question defendants' good faith. For example, an article from The Sunday Times noted that the scandal involving the CLASS study had inspired medical journals to "stop drug firms from `cheating' on medical studies." (Id. at 1360.) On June 1, 2002, an article in the British Medical Journal called the "explanations for [the] serious irregularities [in the JAMA article]... inadequate." (Id. at 757.) The article also stated that "[p]ublishing and distributing overoptimistic short term data using post hoc changes to the protocol, while omitting disappointing long term data of two trials ... is misleading." (Id.) Following the publication of this article, the price of Pharmacia's stock dropped 7% in three days.

II. Procedural History

This action was initiated on April 7, 2003, when the first securities fraud class action complaint was filed. Related actions were shortly thereafter consolidated into it. The District Court denied defendants' motion to dismiss and granted plaintiffs' motion for class certification, but shortened the class period by more than a year, finding that investors could not have reasonably relied on defendants' alleged misrepresentations after February 6, 2001, the date on which the FDA staff reports were published.

Defendants subsequently moved for summary judgment on statute of limitations grounds, asserting that if reliance was unreasonable after February 6, 2001, plaintiffs must necessarily have been on inquiry notice of their claims at that time. Because the first securities fraud suit was filed on April 7, 2003, and the statute of limitations for § 10(b) claims is two years, see 28 U.S.C. § 1658(b), defendants asserted that plaintiffs' claims were untimely. The District Court agreed, and granted defendants' motion for summary judgment.

The cross-appeals of the parties are now before us for our consideration.

III. Jurisdiction and Standard of Review

The District Court had jurisdiction pursuant to 15 U.S.C. § 78aa. We have jurisdiction pursuant to 28 U.S.C. § 1291. We exercise plenary review over the Court's decision to grant defendants' motion for summary judgment and the decision to deny their motion to dismiss. See, e.g., Mest v. Cabot Corp., 449 F.3d 502, 510 n. 7 (3d Cir.2006); Farber v. City of Paterson, 440 F.3d 131, 134 (3d Cir.2006). The decision to certify a class action and the determination of the class period are reviewed for abuse of discretion. Holmes v. Pension Plan of Bethlehem Steel Corp., 213 F.3d 124, 136 (3d Cir.2000).

IV. Legal Analysis

There are a number of issues before us on these appeals. Plaintiffs challenge, first and foremost, the District Court's grant of summary judgment on statute of limitations grounds. Plaintiffs also dispute the Court's determination of the class period. On cross-appeal, defendants assert that the Court erred in denying their motion to dismiss, and in granting plaintiffs' motion for class certification. We address each of these issues in turn.

A. Inquiry Notice and the Statute of Limitations

The District Court determined that investors were on inquiry notice of possible securities fraud as of February 2001. The statute of limitations will begin to run when the plaintiff is on inquiry notice. "Whether the plaintiffs, in the exercise of reasonable diligence, should have known of the basis for their claims depends on whether they had sufficient information of possible wrongdoing to place them on inquiry notice or to excite storm warnings of culpable activity."^[3]Benak ex rel. Alliance Premier Growth Fund v. Alliance Capital Mgmt., L.P., 435 F.3d 396, 400 (3d Cir. 2006) (internal citations and quotations omitted). The inquiry notice determination requires a "totally objective" analysis that pinpoints the time at which "a reasonable investor of ordinary intelligence would have discovered the information [demonstrating possible liability] and recognized it as a storm warning." Mathews v. Kidder, Peabody & Co., Inc., 260 F.3d 239, 252 (3d Cir.2001).

In line with this objective analysis, plaintiffs are "presumed to have read prospectuses, quarterly reports, and other information relating to their investments." Id. However, the hypothetical reasonable investor need not be a scientific expert; to the contrary, the relevant inquiry is whether a reasonable investor of "ordinary intelligence" would have recognized the available information as indicative of possible fraud. Id.

Inquiry notice seeks to deter putative plaintiffs from sitting on their hands, awaiting the discovery of the elusive smoking gun. Inquiry notice will thus be triggered when plaintiffs "should have discovered the general fraudulent scheme," In re NAHC, Inc. Secs. Litig., 306 F.3d 1314, 1326 (3d Cir.2002) (internal citations and quotations omitted), and "cannot avoid the time bar simply by claiming they lacked knowledge of the details or narrow aspects of the alleged fraud." Benak, 435 F.3d at 400 (internal citations and quotations omitted).

Our recent decision in In re Merck & Co., Inc. Securities, Derivative, & `ERISA' Litigation, 543 F.3d 150 (3d Cir.2008) (hereinafter "Merck") informs our decision here. In Merck, we were faced with a similar set of factual circumstances: after a long-term clinical study of Merck's own blockbuster drug, Vioxx, the company published a questionable interpretation of the study's results, allegedly in order to boost Vioxx's sales and Merck's stock price. The questionable interpretation advanced by Merck attempted to explain why patients taking Vioxx experienced a higher rate of negative cardiovascular events than patients taking the study's comparator drug, naproxen. This so-called naproxen hypothesis emphasized the possibility that naproxen had a positive impact on the cardiovascular system, and discounted the possibility that Vioxx had a negative impact.

The science behind this explanation was debatable, and, consequently, the FDA scolded Merck for its repeated promotion of the naproxen hypothesis. In a public warning letter issued to Merck, the FDA called the marketing campaign for Vioxx "false, lacking in fair balance, or otherwise misleading .... and [minimizing of] the potentially serious cardiovascular findings." Id. at 156 (internal citations and quotations omitted). The FDA ordered Merck to send letters to doctors in order "to correct false or misleading impressions and information." Id. at 157. Despite the public nature of these strong words, as well as, inter alia, news reports questioning the naproxen hypothesis and consumer lawsuits alleging negative cardiovascular effects from Vioxx, we found that investors were not on inquiry notice of securities fraud.

Most importantly for our purposes here, Merck found that inquiry notice, in securities fraud suits, requires storm warnings indicating that defendants acted with scienter. "Thus, to trigger storm warnings of culpable activity, in the context of a claim alleging falsely-held opinions or beliefs, investors must have sufficient information to suspect that the defendants engaged in culpable activity, i.e., that they did not hold those opinions or beliefs in earnest." Id. at 166 (emphasis added) (internal citations and quotations omitted).

Accordingly, for investors to be on inquiry notice of § 10(b) claims, there must be some indication that defendants did not, in fact, hold the views expressed. Inquiry notice requires storm warnings of "culpable activity." See Benak, 435 F.3d at 400. Under § 10(b), a corporation does not engage in culpable activity unless it acted with scienter.^[4] Scienter is not incidental to § 10(b), it is elemental.^[5]

In support of a finding of inquiry notice in February 2001, defendants point to the drop in the price of Pharmacia's stock, the FDA staff reports, the Advisory Committee meeting, analyst reports discussing the events at the FDA, and the fact that the full length of the CLASS study had long been publicly known. Whatever else those facts may have indicated, they did not provide storm warnings of possible fraud.

For one thing, the drop in stock price following the events of February 6-8, 2001 did not indicate fraud or even the possibility of fraud. Rather, the drop in price is easily explained by the fact that the market had been expecting that the FDA would approve a label change. When the Advisory Committee issued a negative recommendation, the market reacted accordingly. But mere investor disappointment does not ipso facto imply fraud.

Neither, in our view, does the content of the FDA staff reports suggest fraud. Defendants rely on a handful of words in those reports: to wit, "not valid," "unclear," and "not convincing." (JA 666, 631, 472.) But the staff reports span over 250 pages of highly complex scientific and statistical analysis. These few words and phrases, lacking in accusatory intent and buried like needles in a haystack, could not give rise to storm warnings of fraud.^[6]

The Advisory Committee meeting and concomitant negative recommendation also could not give rise to storm warnings of fraud; indeed, the transcript of the Advisory Committee meeting explicitly supports a finding that the experts believed that the dispute between defendants and the FDA was a good faith, legitimate scientific dispute. We note, for example, the following statements made at the Advisory Committee meeting:

[FDA Representative]: Just to add a couple of other comments to that, ... I don't think that there were differences between us and the company that were meaningful in terms of the findings of the analyses that were done.

We spent more time describing certain analyses and the company spent more time describing other analyses, but I don't think there is any dispute that we have with what the company presented, and I think that the company understands where we were coming from with our analyses, and I don't think that they are off target either in terms of how the company sees them.

(Id. at 1189.)

[Advisory Committee Consultant]: The challenge here for me is that it seems to me everybody is speaking truth. I agree with everyone who speaks. I agree with the sponsor and their emphasis, I agree with the FDA in their description.... It depends on which piece of this you pick out.

(Id. at 1154 (emphasis added).)

As should be evident by these statements, the Advisory Committee simply did not believe that anything untoward had occurred, and we obviously will not expect more of the reasonable investor than we would expect of experts on the FDA's Advisory Committee.

Arguably the most troublesome pieces of evidence came in the form of two analyst reports discussing the events at the FDA: a JP Morgan report stated that the use of the truncated data was "unjustified and invalid," and Bloomberg News stated that defendants could only show a benefit "by looking at selected parts of the data — a practice discouraged by the [FDA]." (Id. at 714, 719.) Yet even in conjunction with the other evidence to which defendants point, we simply cannot conclude that these statements would alert a reasonable investor that fraud, as opposed to a mere disagreement over the best method of scientific analysis, had occurred.

Defendants also argue that storm warnings existed by February 2001 because no new information was revealed at that time or later: it was always known that the CLASS study lasted longer than six months. But while defendants had acknowledged in April 2000 that the study lasted 13 months, there was no indication that they deliberately withheld data from JAMA, or improperly massaged the data, until the Washington Post article in August 2001. The mere fact that the study lasted thirteen months, and that there was a technical dispute between scientists about whether to use the full data or only a portion of the data, would not have provided storm warnings of fraud to the reasonable investor.^[7]

Finally, defendants' own reassuring statements after the Advisory Committee meeting foreclose their argument here. In reaction to the Advisory Committee's negative recommendation, defendants defended the results of the CLASS study, and their use of the truncated data. Defendants stated that they had a "compelling case" for a label change, and that it had been "difficult for the [Advisory Committee] to analyze" the study because it was "an extremely rigorous and complex trial." (Id. at 171, 1519.)

These reassuring statements operate as a sort of antidote to any storm warnings that may have existed. See, e.g., Merck, 543 F.3d at 167 n. 14 ("We have recognized that reassurances can dissipate apparent storm warnings if an investor of ordinary intelligence would reasonably rely on them to allay the investor's concerns.") (internal citations and quotations omitted). Just as we require investors to act upon public information indicating fraud, so, too, do we allow them to rely upon corporate statements discounting the possibility of malfeasance.^[8]

The totality of the evidence in the public realm as of February 2001 did not indicate a possibility of fraud or even hint at any malfeasance or intentional impropriety; rather, the evidence only supported the view that there existed a legitimate dispute over scientific and statistical models.^[9] But for inquiry notice of § 10(b) claims, we require some reason to suspect that defendants did not genuinely believe the accuracy of their statements. No such evidence surfaced until the publication of the Washington Post article which stated that defendants withheld data from JAMA.

We conclude that investors are not put on inquiry notice of fraud when, in the context of this case, an apparently legitimate scientific dispute arises between the FDA and a pharmaceutical company, and note that the Private Securities Litigation Reform Act ("PSLRA"), 15 U.S.C. § 78u-4, itself suggests this conclusion. A rule that would place investors on inquiry notice of fraud the moment that the FDA questions the seemingly good faith scientific analysis of a pharmaceutical company would encourage putative plaintiffs to file premature securities suits. In imposing heightened pleading requirements, Congress evinced an intent to discourage such suits; our inquiry notice jurisprudence reflects this intent.

For inquiry notice to take hold, there must be some indicia of potential malfeasance. Because no such indicia existed here, we will vacate the District Court's grant of summary judgment.^[10]

B. Length of the Class Period

Largely for the reasons discussed above, we also find that the District Court erred in terminating the class period in February 2001. Particularly in light of defendants' repeated defense of the CLASS study and their optimism regarding a potential label change, it was reasonable for plaintiffs to rely upon defendants' statements until the publication of the Washington Post article on August 5, 2001.^[11]Cf. Basic, Inc. v. Levinson, 485 U.S. 224, 108 S. Ct. 978, 99 L. Ed. 2d 194 (1988) (outlining presumption of reliance in fraud-on-the-market securities suits). Accordingly, we will reverse the Court's order limiting the class period to February 5, 2001.^[12]

C. Cross-Appeal

1. Motion to Dismiss

The District Court correctly denied defendants' motion to dismiss. In that motion, defendants asserted that plaintiffs failed to meet the pleading requirements of the PSLRA, focusing on the admittedly thin two paragraphs of the Amended Complaint labeled "Scienter Allegations." (JA 270-71.) When examined as a whole, however, the Amended Complaint is replete with allegations that defendants acted with the requisite scienter. See Tellabs, Inc. v. Makor Issues & Rights, Ltd., 551 U.S. 308, 127 S. Ct. 2499, 2509, 168 L. Ed. 2d 179 (2007) (requiring courts to "consider the complaint in its entirety"). In particular, the Amended Complaint documents the alleged scheme to trick JAMA, and thus the investment community, by using only the truncated data.

While it is true that a legitimate disagreement over scientific data does not give rise to a securities fraud claim, plaintiffs alleged something quite different: a bad faith misrepresentation of scientific data. Those allegations are sufficient to withstand the "[e]xacting pleading requirements" of the PSLRA, which require that the allegations give rise to a strong inference of scienter. See, e.g., Tellabs, Inc., 127 S.Ct. at 2504-10. The District Court correctly denied defendants' motion to dismiss.

2. Class Certification

Defendants also argue that the District Court's decision to grant plaintiffs' motion for class certification was in error. The gravamen of this argument is that the results of the CLASS study were immaterial as a matter of law in light of the lack of movement in stock price following the initial release of those results in April 2000. We disagree.

Plaintiffs' own expert acknowledges that the announcement of the results of the CLASS study "had little measurable effect on [Pharmacia's] stock price." (JA 1298.) But that fact does not negate a finding of materiality when the market was expecting that the results of the study would be positive, and plaintiffs have presented evidence indicating precisely that. (See id. (citing Morgan Stanley report written the day after the CLASS study results were released that states, "we are making no change to our forecasts, as we had anticipated the study to corroborate the strong safety profile of the product").) And, of course, the materiality of the alleged misrepresentations is self-evident when we look at the market's negative reaction—to the tune of a nine-percent drop in stock price in three days—when defendants' analysis of the CLASS study was questioned in February 2001.^[13]

V. Conclusion

For the foregoing reasons, we will vacate the judgment of the District Court, and remand for proceedings consistent with this Opinion.

NOTES

[*] Honorable Jane A. Restani, Chief Judge, United States Court of International Trade, sitting by designation.

[1] Unlike traditional NSAIDs, which inhibit both the COX-1 and COX-2 enzymes, Celebrex is a selective COX-2 inhibitor. Because the inhibition of COX-1 enzymes often leads to negative GI side-effects, it was hoped that selective COX-2 inhibitors would not possess the harmful side-effects associated with traditional NSAIDs.

[2] Defendants include Pharmacia Corp., Pfizer, Inc., and individual defendants Fred Hassan, Steven Geis, and Carrie Cox.

[3] Plaintiffs do not contend that they "exercised reasonable due diligence and yet were unable to discover their injuries." Mathews v. Kidder, Peabody, & Co., Inc., 260 F.3d 239, 252 (3d Cir.2001). Thus, if we find storm warnings of fraud, plaintiffs' claims are untimely.

[4] Thus, as defamation is not assault, so is § 10(b) not § 11. Section 11 does not require a plaintiff to plead or to prove scienter; § 10(b) does. This is a distinction with a difference, both in terms of what a plaintiff must show for recovery and in terms of what information must be available for inquiry notice to take hold. We did not find it necessary in Merck to discuss § 11, nor do we find it necessary to do so here. Were this is a § 11 case, which it is not, the evidence in the public realm as of February 2001 might well have given rise to storm warnings of misstatements, and thus triggered the second step of the inquiry notice analysis—the duty to investigate potential claims.

[5] "To state a valid claim under Rule 10b-5, a plaintiff must show that the defendant (1) made a misstatement or an omission of a material fact (2) with scienter (3) in connection with the purchase or the sale of a security (4) upon which the plaintiff reasonably relied and (5) that the plaintiff's reliance was the proximate cause of his or her injury." Semerenko v. Cendant Corp., 223 F.3d 165, 174 (3d Cir.2000) (emphasis added); see also Ernst & Ernst v. Hochfelder, 425 U.S. 185, 96 S. Ct. 1375, 47 L. Ed. 2d 668 (1976) (requiring more than evidence of negligence in § 10(b) cases).

[6] Additionally, one of the staff reports described the disputed six-month model as a "potentially supportive analysis." (JA 482.)

[7] In a different context in this litigation, even defendants' counsel acknowledged as much: "Well, what does that show? That scientists can disagree on how you interpret the data. That doesn't make fraud. That means people have different interpretations." (JA 2028 (argument in opposition to class certification).)

[8] Corporations are, of course, unlikely to admit to culpable conduct, and we do not purport to state a rule that precludes a finding of inquiry notice until they do so. But, under circumstances such as these, where the evidence of wrongdoing is entirely speculative, reassuring statements are a relevant consideration.

[9] Indeed, after the events of February 2001, reports portrayed Celebrex's outlook positively, and took seriously defendants' truncated analysis. (See, e.g., JA 716 (JP Morgan analysis stating there is some "support [] [for Pharmacia's] contention that there was a [justified reason to limit the analysis to six months], but the FDA statisticians dispute this"); id. at 1707-08 (calling "the scientific evidence on this question ... unclear").)

Additionally, numerous financial analysts retained Pharmacia's stock at a "buy" or "strong buy" rating. See Merck, 543 F.3d at 157-58 (noting that, during the relevant period, "securities analysts were of one voice in their projections for Merck and Vioxx; ... all maintained their ratings for Merck stock at `buy' or `hold' and/or continued to project increased future revenues for Vioxx").

[10] We note that inquiry notice — in securities fraud suits and otherwise — is alive and well in this Court. Neither in Merck nor here have we replaced inquiry notice with an actual notice standard. Here, we merely conclude that, in the absence of any indication that defendants did not believe the truth of their own statements, investors were not on inquiry notice of § 10(b) claims.

[11] The appropriate date is August 5, 2001, and not the later date, May 31, 2002, of the British Medical Journal article. As of August 5, 2001, investors should have known that there was a possibility that defendants' claims were false. Any new information found in the British Medical Journal article is different in only degree, and not in kind.

[12] Plaintiffs also challenge the District Court's decision to seal the record and presumably our decision to continue the seal pending appeal. We will leave the question of continued sealing to the District Court. We note, however, that we are unpersuaded by the Court's reliance on the so-called "self-critical analysis privilege" as a basis for sealing. The self-critical analysis privilege has never been recognized by this Court and we see no reason to recognize it now. Cf. Union Pac. R.R. Co. v. Mower, 219 F.3d 1069, 1076 n. 7 (9th Cir.2000) (calling the privilege "novel," and noting that the Ninth Circuit has not recognized the privilege).

[13] We also reject, without further discussion, defendants' argument that the District Court erred in denying their motion to strike certain documents.