Oral argument — In Re Lipitor Antitrust Litigation

Appellate Numbers 24-2184, 2185, 2189, 2194, 2202, 2203, 2256, and 2257, counsel, we'll take your orders. Okay. Good afternoon, your honors. May it please the court. My name is Barry Refson. I represent the plaintiff appellants, and I'm going to argue first on the issue of the summary judgment issue. My colleagues are going to deal with the class issue to the extent that the panel is interested. This is the second time that this 14-year-old case has been before this court. The first time, it was when the district court granted a motion to dismiss alleging that the plaintiffs had not properly pled that there was a large reverse payment that violated the antitrust laws. This time, the district court granted summary judgment on the issue of causation. The case involves the drug Lipitor, a cholesterol drug that was at one time the largest pharmaceutical product in the world. At the time of the agreements that are being challenged in this case, Lipitor was making a billion dollars in sales a month, which is... The FDA was very enthusiastic about trying to get folks out there to manufacture the product, right? That's exactly why I made this point to begin with. The FDA knew it was the largest blockbuster product ever and was very focused on the date that the parties had agreed to in their agreement, the date that Pfizer paid Rambazzi to adopt. That was... What facts, if viewed in your favor, would show the FDA would have, not may have, not could have, using Wellbutrin's language, approved the end of before November 30, 2011, the target date? I don't think that's quite the issue for the jury, whether they would have. The issue is... I'm asking you whether Wellbutrin requires that, showing which it does, because that's the language of the case. I'm asking you what facts, viewed in your favor, would have showed the FDA would have, not may have, or could have. I will answer that question, but I don't quite agree with that framing of what Wellbutrin said. Wellbutrin was an issue about causation, but Wellbutrin was a case where there was no evidence from which a jury could draw an inference about what would happen. Because they had a blocking patent. Because there was a third-party blocking patent. There was a very complicated set of agreements involving six different parties. Very complicated regulatory process here, too, that serves as a potential third-party block, which is why Judge Quartz is asking... Yeah. ...except the Wellbutrin standard. Yeah. Well, I... Let's get to that. Okay. I will get to that. I disagree quite, like I said. I don't agree quite with the framing of Wellbutrin. I think Wellbutrin was a different issue. But there are a lot of facts here from which a jury could draw inferences... But it was more likely than not about what would have happened in the but-for world. I mean, here we have that the FDA constantly recognized that the regulatory process was not an absolute block. It said from the time it adopted the... It issued the AIP, or said that Graham Baxley was subject to the AIP, it recognized that exceptions could be drawn, it could be made to that policy in the interest of public health. And it recognized that making Lipitor available to the public was a great issue of public health. And it continually targeted the date that the parties had agreed to, the date that Pfizer paid Graham Baxley to adopt. The FDA continually targeted that date, that November 30th date. You brought up the AIP. The AIP is issued in February 2009. A couple months after the settlement agreement in June. August of 2009, Graham Baxley tells the FDA. So now that's several, many months later, seven, eight months later. And not until May of 2011 does the FDA give the exception. That's right, Your Honor. And doesn't that fact hurt you? No. It demonstrates that the FDA wasn't focused on a date, but was focused on making sure an exception should be granted to the AIP. Yeah. I don't think that's what the facts show. And I skipped over the most important fact here. And it relates to that timeline that you just set out, Judge Schwartz. The AIP was implemented in 2009. And the settlement agreement was three and a half years before the FDA ultimately granted approval. The date that was in the settlement agreement was November 30, 2011. Three and a half years later, the FDA granted approval on November 30, 2011, exactly the date that the parties agreed to in 2008. The district judge said that was just a cosmic coincidence, or it was more likely than not that that was a cosmic coincidence that the FDA agreed to the exact date that Pfizer paid Graham Baxley to adopt. Now, that's not an inference that a jury would be obligated to reach. I mean, the jury could say, look, this date that Pfizer paid Graham Baxley for in 2008, that the FDA looked at that date and made every effort to approve this drug by November 30, 2011. And they weren't going to make a decision that was unsafe for the public, but they were going to move to do everything possible to approve this drug in order to make it available for the public. And it actually was able to do that by that date. How do you deal with the fact that the day before, November 29, Graham Baxley is calling the FDA and saying, we know we've got a problematic manufacturing facility. How about we just delete that from their application? And the response from the FDA was, we still can't guarantee we're going to make it. Doesn't that fact really undermine your position? That's a fact that a jury is entitled to hear, and the defendants are entitled to make their argument. But the standard for causation is not a guarantee. I mean, this is an issue that's been dealt with by the Supreme Court in Truett Payne. I mean, this is an issue where it hasn't been challenged that Pfizer and Graham Baxley were wrongdoers, that they violated the antitrust laws. And the Supreme Court has said, how do you deal with causation when you have a wrongdoer like Graham Baxley? And the Supreme Court said in Truett Payne that when you have a wrongdoer, that the wrongdoer is not entitled to insist that causation be proven with, is not entitled to demand specific and certain evidence. And that's exactly what the defendants are coming here and arguing in this case. They're saying it has to be specific and concrete. It seems to me what they're saying is there's a limit to the amount of speculation that the jury can be permitted to engage in in order to return a verdict in your favor, right? I mean, think of it, think of it this way. We all suffered through evidence, right? And we know that it's a summer day and someone comes into the courtroom and they take an umbrella and they shake it out and they brush some water off of it. We're allowed to make the inference that it's raining out, right? And I understand that to be your argument, the jury was allowed to infer that the FDA had the settlement date in mind, that this was not some hypothetical goal. This is what they were thinking about. But it seems like what you want is the further speculation that had it not been raining out, I would have gone to the beach instead of coming to work today. And there's a million things that could have or should have or might have happened. But that second level of speculation seems well outside the jury's province. So I get your first point. Get me the second conclusion why that should be there for the jury as well, which is, but for this, the FDA would have done something different even a day sooner regarding Rand Baxley's application or I suppose one of the other generics. I think it's clear, Your Honor, that if the date had not been November 30, 2011, and had in June of 2008 had the parties agreed to a settlement date of November 29 or maybe June 28, 2011, whatever that date was, the FDA would have been looking at that date for three years. And when those other generics started complaining at the end of 2010 and 2011 when they were raising issues that it looks like Rand Baxley's first-to-file exclusivity might be a blocking exclusivity, their letters specifically refer to that November 30, 2011 date. Had there been another date that they were looking at, had the parties agreed to a different date because Pfizer did not pay Rand Baxley to accept that November 30 date, everybody would have been looking at that earlier date. Those other generics would have started writing to the FDA, it looks like Rand Baxley's not going to be able to enter on November 29 or June 28 of 2011. Whatever that date was, they would have been complaining about the failure to enter on that date instead of November 2011. The problem wasn't the date for the other generics. It was the conclusion by the FDA that there was substantially complete or substantially fulsome, I can't remember the second word, but substantial compliance with their ANDA and therefore they had exclusivity and they weren't backing off from that finding. No, that didn't turn out to be the issue. The issue was for whether Rand Baxley would lose its first-to-file exclusivity, the issue was whether its ANDA was substantially complete. Right, and they found it was substantially complete. So therefore it had first-to-file exclusivity, but that's not what controlled when the FDA made its determination. I understand that. What should we look at to say a jury could rely on this fact to conclude the FDA, more likely than not, would have approved that ANDA before November, whatever the target date was, before November 30th? What evidence do we have from which a reasonable jury, following up on Judge Meady's question, could make that inference? Well, the issue is that they, there's this date, November 30th, that was agreed to three and a half years before the FDA approved the ANDA. There wasn't anything special about November 30, 2011, other than the fact that it was in the settlement agreement. These parties agreed on that date, then the FDA looked at that date throughout its approval process, it said, everybody recognizes the significance of this date. It's the date when the parties agreed that this largest drug was not subject to any further restraints from the patent. It was when the parties agreed that it could come to market. One of those communications also said there was no guarantee that the FDA, by the FDA, saying we can't guarantee we're going to make this date. Right, there is no guarantee, and in fact, even though there was no guarantee, the FDA approved this ANDA on November 30, 2011. That's the inference that a jury is entitled to make, that even if there was no guarantee in the but-for world, if they were targeting another date, they would have met that date too. And in fact... The evidence I'm hearing you bring to our attention is there was a date in the agreement that the FDA was targeting, and from that the jury should infer if a different date was in there, they would have done it sooner. I think that's a big part of it. That's not the extent of the evidence. I think if you also look at what was happening during the ANDA process, all of the main disciplines, the FDA has three main disciplines that they have to look at before they approve it. There's labeling, there's bioequivalence, and there's manufacturing. I mean, the district court didn't allow us to take discovery to show exactly what the reformed agreement would have looked like, what economically the parties would have agreed to. The focus here is on what the FDA would have done, not what the parties would have done. No, no, but the predicate is, I mean, for our full causation argument, had the district court allowed us to do full discovery instead of phasing it to this narrow issue, the predicate issue that we would have had an economist that would have said, you know, had the parties not used a reverse payment, it would have been profitable for them to enter into a different agreement without a reverse payment, and without the reverse payment, that entry date would have been earlier as a matter of economics, because it would have made economic sense for them to have agreed to an earlier date. And then that earlier date is the date that the FDA would have been looking at from the time of 2008 forward. And then, I mean, but the district court didn't let us do that. It said we had to look at this narrow issue first, and the issue became whether it would be just one day earlier than November 30, 2011. And if you're just talking about that one day earlier, the issue here is all of those three main disciplines were complete by October 25, 2011. The only thing that remained on October 25, 2011, was an inspection of the Toansa facility in India. And that was scheduled on November 14, and it was scheduled for November 21 to November 25. So there was clearly time to approve it just one day earlier. And the FDA ultimately approved it, even though it said on November 29 that it wasn't guaranteed, it ultimately approved it on November 30. And it didn't even wait to do the entire review, the issue, the establishment inspection report. That report, the final report didn't issue until April 12 of 2012. But the FDA recognized the importance of approving this, getting this drug on the market for consumers so they could have lower priced generic Lipitor, and they got it on the market on November 30, 2011. I mean, that's an inference that a jury is entitled to make. Jury doesn't have to agree with our district judge that it was just a cosmic coincidence that the FDA approved this drug on November 30, the day that the parties agreed to and that Pfizer paid Rambaxi to accept. Thank you. Oh, you didn't reserve any time. I did. I spent four minutes. I didn't hear it. You have to just let me know. I'm sorry. Four minutes, we've been reserved. Thank you. That's fantastic. Non-pro-top. Do you want to hear the class issues next, or do you want to do the summary? We're going to stay with the appellants. Good afternoon, and may it please the Court. Deepak Gupta, appearing on behalf of the end-payer plaintiffs. I'd like to try to reserve the minimum amount, a minute for rebuttal. So I'm going to turn to the class certification issues now. And we do believe that the class certification decision reflects an impermissibly heightened understanding of this Court's ascertainability precedence, but there's a threshold question. What did you think when you walked out and you got that opinion and said, wow, this wasn't enough, I think the district court was asking us to do, tell me what you think the district court wanted you to do, and tell us why that's, from your point of view, not consistent with our precedence. And I do want to just emphasize, I think there's a threshold question about why the district court reached this opinion at all after having granted summary judgment for the defendant. So I want to make sure I return to that. But first, I think the district court was demanding a number of things that this Court's precedence don't demand. It was demanding that the plaintiffs generate a list at class certification. It was demanding a rule that there's no individualized inquiry. I think it's clear after this Court's cases like Hargrove and City Select that that's not the law anymore. It was demanding that the plaintiffs produce the data. That's an opinion at page 48. And it was demanding time cost estimates of how all of this would go down at a later stage. None of that is demanded. One of the plaintiff's concerns with respect to all of those things was the PBM data only covered about 40 percent of the market share, and it didn't think that that was going to be sufficient for ascertainability. How do you respond to that? Yeah, I think that the district court was just operating on a fundamental misunderstanding of what the methodology that was being proposed here. In NIOSPAN, it's true that the plaintiffs were attempting to take this PBM data in the aggregate and abstract from that a class list. I think they were laboring under a misunderstanding, a more heightened demanding understanding of what ascertainability requires. But here, the methodology was much more straightforward and much more like a routine class action, which is that, as the expert Laura Kraft explained, people would come forward, whether they're in the consumer class or whether they're in the third-party payer class, they would have to file a claim, file an affidavit, and then produce the records. And those records are standardized records that are kept for every pharmaceutical transaction in this country. They're required by law under HIPAA, and there are multiple sources for that data. But I think the district court got confused by the complexity of the way the market works, and that obscured the fact that this is actually a very straightforward methodology. So are you saying, if I'm misunderstanding you, please tell me, that the PBM data might have only covered 40 percent of the market share, but you are to proffer through your experts and the other materials from the other data sources would have covered the market? Yeah, but more fundamentally, that it's not about accumulating data sources and then making an abstract inference from that data. It's instead envisioning a claims process where people come forward. I didn't mean to insinuate that. What I understood the representations to be from the materials from the district court was that there were many data sets from which the classes could be identified. That's right. But I think, again, it's not about a top-down set of inferences that's drawn from a set of aggregated data. It's instead envisioning a claims process where anyone who claims to be a class member has to come forward. They have to attest that they are a class member. They have to attest that they purchased Lipitor within the period, and then they have to provide the records that show that. And, you know, when you walk into a CVS or a Walgreens and you hand over your Lipitor prescription, the pharmacy instantly checks your plan's information and is able to compute within a minute whether or not, you know, who's paying for this drug, is the drug covered, how much does the plan pay, how much does the patient pay. All of that is in standardized records that are required by HIPAA that are obtainable from multiple sources. So this is actually like a highly regulated market where it's actually quite a bit easier than it is in the typical consumer class action. So I think the district court went fundamentally awry under this court's ascertainability precedence, but it also went awry in deciding this question at all. Because recall that, and I think the sequencing here is critical. I'm just going to, because you're over time, I'm just going to see if either of my colleagues have questions on that subject, because the briefing was quite sufficient on that point. I'm sorry to interrupt you, but do you have any? Yeah, I think your decision in city select, Your Honor, approves the use of affidavits and records. And I think actually that was a tougher case where there were gaps in the records and the court explained that affidavits can be used to fill those gaps. You didn't have complete logs there of the junk faxes. This case is much more straightforward. We're not even saying that we need to use the affidavits to fill gaps. We're saying that any person that comes forward, whether they're in the consumer class or the third-party payer class, can attest that they are in the class and use the available records, which are complete, to show whether they purchased Lipitor. Okay, thank you, Counselor. I'll have you on the line. Thank you. Good afternoon. My name is David Sorensen. I only have two minutes, but I'd like to reserve one, if possible. Of course, because everybody gets a minute. Thank you, Your Honors. Your Honors, Rule 23A1 asks whether, quote, the class is so numerous the joinder of all members is impracticable. The district court erred below by ignoring critical and undisputed evidence showing the joinder of all 63 class members here is not practicable. What is that evidence? In five similar antitrust cases before this one, five, involving many of the same class members, after class certification was denied on the basis of numerosity, there were joinder actions. Many class members, of course, all five cases did not join. I actually was counsel in all five, one of the class counsel, only relying on public information here. But in every case, numerous class members did not join. You were feeling lonely. I'm sorry? You were feeling lonely, that people weren't coming along to your party, and you're assuming that they wouldn't come again. I'm not assuming. What I'm saying is that there's an overlap in the classes, and it's a repeated conduct pattern. The district court recently, just a couple of weeks ago, in Amatiza, looked at the same evidence certifying a class saying, quote, this pattern, quote, constitutes a valid predictor of whether it would be impracticable for class members here to pursue joinder. That's the Imre Amatiza antitrust case. Just to finish the site, 2025, Westlaw, 2690871 at Star 15, District of Massachusetts, September 19th, 2025. That's the first court to actually look at this pattern that the district court here ignored and say what common sense tells you. It's been happening again and again and again. If ultimately, if this summary judgment is reversed and we go back and classes denied, it will happen again here. By the time this is up, I will. Can I have one question for you? On a slightly different point. One of the considerations talks about expenses, and there's a representation in the briefs to us that the class members, if required to join as a new party, would be $30,000-ish. Was that information presented to the district court? That quote from $30,000 is the cost of the? I think the answer is no. I think the answer is no. There was different cost information presented that actually was higher. We used a different number of $3.4 million, $3.7 million, which was the out-of-pocket in a similar case. You divide that by 63, you get $59,000 per class member. But the answer to your question is no. Thank you, counsel. Thank you. Good morning, Your Honors, and may it please the Court. Judge Schwartz, I want to pick up with where you were having a dialogue with counsel about Will Butrin and the standard. Because as the Court, I think, identified, the appellants to withstand the summary judgment motion have to produce evidence that it is more likely than not that, in that case, Anshin would have obtained a license, in this case, that the FDA would have approved an earlier ANDA. And what the Court said is, while it may be better than speculative that Anshin and Andrix would have had an incentive to talk, it's pure speculation they would have reached an agreement. And that's exactly what we have here. I don't think anyone is necessarily disputing that the FDA was aware and indeed targeting November 30th. But targeting something is a far cry from achieving it. And what the Court looked carefully at the evidence, and then on page 59 of the order and then again at page 71 of the order, said there is no evidence that either of the two things that could have happened would have happened. One is that they could have started the review process earlier. If there were evidence that the AIP exception would have been granted earlier, then maybe you could draw a reasonable evidentiary-based inference that they could have done the approval on the same timeframe and reached an earlier date. The Court said there is no evidence of that, and I'll address the other generic point in a minute. The second thing the Court said is there is no evidence they could have compressed the timeframe and done it faster. I can run an eight-minute mile, and I can target, well, I used to. I can run a ten-minute mile, and I can train to run a ten-minute mile, and I can succeed. I could set as a target to run a nine-and-a-half-minute mile, and that could be my objective, and I can target that. But the fact that I ran the ten-minute mile doesn't mean I'm going to run it shorter in a but-for world. Maybe I could, but that's not evidence that I would. What would evidence be? Evidence might be that I did a couple practice runs at nine-and-a-half. Now, you asked a question to counsel, well, what was the evidence that you had, and counsel didn't give you the evidence that they had. I want to tell you the evidence that they could have potentially brought, which, of course, they didn't. They could have shown evidence of the time that it took the FDA to schedule facility inspections after completion of the core review, like bioequivalence, and could have shown that that was faster than it, as a general rule, than it took here. They could have shown that they could have done the TANSA inspection earlier, the first week of November, not the third week of November when they still then had all sorts of questions that they had to resolve, and indeed, as you pointed out, on the 29th they still said, far from certain that we can grant the approval. They could have had evidence that FDA has an inspection team on the ground in India available to inspect TANSA on a day's notice. They didn't show that. They could have shown evidence that most ANDAs reviewed by the FDA were approved on or before the party's settlement date as agreed to, and yet what the evidence actually showed was with lots of other blockbuster drugs. Diavan, FDA was 20 months late. EpiPen, big, big drug. The FDA cared a great deal about EpiPen. There was that whole issue with Myelin. Thirty-eight months they were too late. Nexium, which is an enormous blockbuster drug. The FDA was six months late in approving it. Their point is that all of this is stuff for the jury, right? I mean, I get your point. But their argument is all of this should be considered by the jury, right, that the jury could look at all of this and say, sure, but they were targeting the date of the settlement. Of that, no one disagrees, or at least there's evidence from which one could conclude that. Why can't they also reasonably conclude that they would have done it quicker? Well, Buterin answers that exact question where the court said, the fact that juries may predict the outcome of hypothetical scenarios says nothing about the type or amount of evidence that is needed for a plaintiff to withstand summary judgment on a claim involving a counterfactual world. That's really the heart of the issue. Yes, there is reasonable evidence to suggest they would have targeted it, but that's not what the question is. The question is, could they have done it in a compressed timeframe? And by the way, well, Buterin has been applied by a lot of courts since this court decided it. In Seroquel, which is within this circuit, there was sufficient evidence. What was the sufficient evidence? That the parties had actually entered into a supply agreement. And then the court said, well, that would be sufficient evidence to go to a jury to see whether they would have entered into the same supply agreement. In Ziram, in Leitaderm, where there was evidence, for example, that the parties had made plans to do an at-risk launch, that would have been sufficient, perhaps, to get past. But here, all we have is what this court in Robutrin called pure speculation. And again, while the court said in that case, while it may be better than speculative, that Anshin and Andrix would have had an incentive to talk, substitute that the FDA might have targeted the date, it's pure speculation they would have reached the agreement. And that's really the essence of the argument here. And their expert opinion, what does their expert say? Their expert said, well, had there been a need for the FDA to inspect Tonsa earlier, they could have certainly done so. That's just speculation. He actually admitted that launch dates for the FDA are, quote, aspirational. That's at JA2879 and 80. And, quote, there's no guarantee the FDA will meet a launch date because, and what he said was, FDA is not going to cut corners or approve a product that they haven't determined is safe or effective. And then finally, their expert at JA2873 said, well, where there's a will, there's a way at the FDA. And if they need to get something done, they'll find the resources. Again, that's not evidence. That is speculation. And it's completely inferential. How do you respond to your colleague's argument concerning inability to conduct additional discovery on, I believe he was focused on the activities of the parties, not the FDA. That's precisely the point. What they said they wanted discovery on was they wanted to be able to show that the parties had an incentive to reach an earlier date, and they wanted to have expert evidence to show what the economics would have been of an earlier date. And we are stipulating for the purposes of this summary judgment question, that's what the court said, that's why the court limited it to just one day. We're stipulating that in a different world, there might well have been an earlier entry date. Obviously, if we had to try the case, we would show evidence as to why we don't believe that would have happened. But for the purpose of this, they didn't need discovery for that. The issue is, what would the FDA have done? And they had full access to the FDA. They had access to take discovery of the FDA. Maybe they chose not to take some discovery because they didn't want to know what the answers were, but nothing impeded them from that. And they got all of our communications with the FDA, and all of the FDA's communications with us, and even with some third parties. The fact of the matter is, what we have here, Your Honor, is simple speculation as to a result, and that's what Wellbutrin says is simply not sufficient. If there are no questions, I will let my colleague speak to class certification. Thank you, Your Honors. Good afternoon, Your Honors. Devorah Allen for Mbaxi. I'll start with the EPP motion for class certification. Whatever the EPPs want to call their methodology, they don't dispute that the district court correctly concluded that the final step of that methodology is always going to have to be confirming that the entity that's named in the data is actually the end payer that's in the class. And the only reliable way to do that is extensive individualized review of the plan documents. Now, that determination is not just entitled to deference under clear error. It is compelled by this Court's precedent, because NIOSPAN already found that it was reasonable, and obviously not clear error, for a district court to conclude that consideration of, quote, the individual contractual relationships underlying each transaction, same plan documents. NIOSPAN, though, is different, because NIOSPAN only had the KBM data. You have to concede there's a lot more information that has been identified, and isn't that what's required? You still identify records that would have the information. You have to identify the records exist, and that those records would allow you to identify the class members. That's all that's required under RealPage. And I want to just run through a couple of other teachings from our court, and you tell us why the district court acted in compliance with that. I've just reviewed RealPage. Burt said it's permissible to have to do a cross-check to verify that the class member is, in fact, an eligible person to be compensated as a member of the class. And at this stage, all you need to show is that the members can be identified. And finally, that there has to be, using Hargrove, a method, and you can use data and affidavits to determine these facts. I'd like you to explain to me how the district court followed those teachings and its conclusions about the ascertainability issue here. Absolutely, Your Honor. The district court specifically held that affidavits were permitted. There's obviously no categorical ban on using affidavits. CitiSelect makes that clear. But the availability of affidavits depends on whether you can corroborate what's in the affidavits with objective evidence that is not, in and of itself, individualized. So we know that we don't just have to rely on affidavits. Carrera tells us we don't have to rely on the claim and say so. And so the question becomes, if we want to challenge the reliability of what's in the affidavits, how do we do that? If that process is itself individually, it requires extensive individualized inquiry, it fails ascertainability. Wouldn't that be the case for any product that someone wants to buy? I mean, from your point of view, you'd have to show an individual person made a purchase of the product that's the subject of the class action. No class could be ascertained under your explanation. No, no, no. I don't agree, Your Honor. Individualized inquiry is okay. Matching is okay. Looking at the name of the entity in the data and saying, did they make a purchase, that's individualized to be sure, but it's not extensive. What's extensive is looking at a plan document. It's not an up-down yes-no like in a bird. Is it a household member or not? All you have to do is check the address. Or I'll give you another example. Members of the court are excluded from the class, so the plaintiffs have to generate a list of court staff, and then you have to check, is a class member on that list or not on that list? That matching process across data sets is absolutely permitted by this Court's precedent. What is not permitted, and this is exactly what NISBAN looked at, is looking at a 144-page plan document and reaching arguable legal conclusions about what is the contractual relationship between the end payer and the PBM, because that's not clear from the face of the document, and that's exactly what the district court found. The district court found that ultimately, no matter where the data came from, it was going to reduce to review of those plan documents. Okay, so I'll change my comment to general products to products that are subject to insurance plans. You're suggesting that no class action could be brought that happened to involve plaintiffs who happened to have insurance that covered the product that they purchased. No. I think that what the district court wanted was a record-based, manageable process to figure out who is in and out of the class. Now, I'm not a PBM, and I'm not an insurer, so I don't know what records they have and don't have, but I don't think it's the case that there can never be a class certified. This PBM data, as the NISBAN court found, is indeterminate, and I did just want to get back to Your Honor's point about, well, this case is different than NISBAN because there's more data here. The district court looked at that evidence. It looked at the evidence that was submitted where the plaintiff said, no, no, no, it's not going to come from the PBMs. We're going to submit it through the individual claimants. And what the district court concluded, based on the concessions by the EPPs, is it was the same PBM data that's in the record joint appendix 10144 and 10475. The only difference is, instead of being collected by subpoena, straight from the PBM and submitted for all class members, each end-payer gets the data, doctors it themselves so it only lists their name, and then submits a couple of columns from the data into the claims administrator. I think it should go without saying that the same inadequate, ambiguous data doesn't somehow magically confirm end-payer status. In this language, doctor, are you suggesting there was not accurate information conveyed to the district court? That was your language. You used doctor. Did you really mean doctor? Did you mean redacted? No, I think the EPPs acknowledge that they're doing this. I'm not implying there was any sort of – What do you mean by doctor? What I mean was – That's a pretty heavy word, and I want to make sure that – Yeah, let me explain exactly what I mean. So Ms. Craft at the evidentiary hearing walked through data from NMUFCW, one of the end-payers. That data that she walked through, she said it's very clear in the data. You can see these entries only list NMUFCW as the end-payer. When you compare the data that she walked through at the evidentiary hearing to the data that NMUFCW actually produced, they got it from the PDM, right? So we know where the data came from. It didn't come from the end-payer. They requested it. They got it from the PDM. That data has, I think, eight more data fields than the data that Ms. Craft presented that came through the end-payer. And so, again, I'm not accusing the plaintiffs of hiding the ball. I think they're saying the quiet part out loud. I think they're being very transparent. They say the problem in NISBAN is there was too much data. And so, essentially, we're going to filter it. We're going to take away some columns and add some columns, and then the claimant's going to submit it, and it's only going to list them as the end-payer. But the district court concluded, and this is explicit at JA190, that even if the entity submits the data along with the claim form, quote, the answer will be verified through the PDM data. It ultimately reduces to the PDM data because it comes from the PDM data. It's no different. It's just going through another step where there's actually more individualized inquiry that's happening at the front end before it ever gets to the claims administrator. So let me ask you this. Assume the panel were to affirm the district court's order granting summary judgments around vaccine. Could the court conclude that the named plaintiffs aren't adequate representatives, and that's why class certification should be denied? Because there's no named plaintiff that has a viable cause of action? I ask you this because you're the class certification person, so I'm not asking about summary judgment. I'm asking as the class cert person. I do think summary judgment should be affirmed, if you're asking my opinion on that. I think that if the court affirms summary judgment, certainly at that point it could essentially dismiss the class certification appeals as moot. It doesn't have to reach them. We can't do that under our case law, can we really? The case called Gale, which requires that the class certification motion was filed when the case was still alive. It doesn't necessarily moot it. Acknowledging that we have to consider how Gale might interact with our obligations, my question for you is, is that a proper answer? Do we have to go through the exercise of all the numerosity and the ascertainability issues? I think at that point, so I completely agree, the court has article-free case and controversy. It has the jurisdiction to decide the issues. It is not divested of its jurisdiction, just like the district court was not divested of its jurisdiction. That is clear. I think as a prudential matter, this court has the discretion to weigh judicial economy and its resources and decide whether it wants to engage in resolving the class cert issues and give guidance to the lower court, or if it wants to at that point, just say we're not going to address it anymore and we're just going to affirm and not actually go through the class cert analysis. Is there an adequate representative if we granted summary judgment were affirmed? So I think that the answer is that where it is affirmed based on a legal right, this is under the Roper decision, page 333, the cause of action is not mooted until the time to appeal has run. And so I would say the technical matter, until the time to appeal has run, the claim is still alive and the court still has jurisdiction to address it, just like the district court. I'm asking whether or not the plaintiffs remain as adequate class representatives if summary judgment is affirmed and they have no cause of action. Then in that situation, I think that I agree with Your Honor. In that situation under Gill, the court could say that they are not adequate class representatives and it could deny the class certification appeals as moot on that basis. But I just want to be clear that that's a different question than whether the district court abused its discretion in deciding to resolve class cert, in deciding to issue the summary judgment decision right before the class certification decision, and I don't think there was any error there. Thank you, Your Honor. Thank you, Your Honors. I just have four short points I want to make. First, the issue has been raised about whether you have to speculate, and there is some speculation in any time you're dealing with a but-for world. I mean, that's the nature of a but-for world. That's not a reason for granting summary judgment. The issue is whether the speculation is based on evidence. Well, Butrin was a case where there was no evidence for the speculation that was requested. There are two theories for why the generic would have been able to enter. One was that it would be able to enter, that Andrix, the generic, would be able to enter by litigation, but the only evidence in the record was that the patent holder, Andrix, was going to win 80% of the time. So that's why that patent barred the entry by Antion, the generic. And as Your Honor recognized in the Fresenius case, when there was evidence to be done for a patent analysis, the patent wasn't an automatic bar. So the patent theory, the patent win theory, did not work to get the generic on the market. The second was whether there was an agreement, and the court said, well, this is a case where the plaintiff has said that the five agreements that were actually entered into would not exist in the but-for world. That's not what we're complaining about here. The plaintiffs say here that there would have been a settlement, but the actual settlement that was entered into would have been different. It would have had a different date. So it's not completely speculative. I think the more appropriate case for the court to look at from its precedents rather than Will Butrin is Danny Kresge. Danny Kresge was a case where there was a conspiracy. There was a jury trial. The jury found that there was a conspiracy, but the district court granted a directed verdict because it said that it wasn't certain that the plaintiff would have gotten two concerts that it was complaining about. It was a conspiracy between a concert promoter and music acts, and the court said, well, it doesn't have to be certain. It's enough evidence here for the jury to reach a determination as to whether those concerts were a loss because the plaintiff testified that his terms that he would have offered for those acts were better than the terms that the defendant offered. I mean, that's much less than the kind of evidence we have here, and we have evidence here of exactly what the FDA was looking at, and not only that the FDA targeted November 30th, but that the FDA succeeded in reaching that goal. I mean, if it was targeting another goal, the jury would have been entitled to find that it would have succeeded by reaching it by that date, just like it did in the actual world. There were seven months between the time that the FDA wrote a memo saying we're bringing an exception to the AIP and the time when it actually granted approval. Certainly during that time, it could have been a day earlier, which was the only issue that the district court allowed us to discover. My second point is that there's evidence here that the AIP exception was granted because of the settlement here, and this is in the July 27th labeling approval. It's a JA 605 to 609. This is what the FDA said. This is the reason why the FDA granted the AIP. It says, the agency determined that OGD, that's the Office of Generic Drugs, the part of FDA that does the approvals, should review this ANDA because of the major amendments, second, Rambaxi's settlement with Pfizer, and Rambaxi's 180 days of exclusivity. So that settlement was a main reason why the FDA granted this AIP, and that's something that a jury would be entitled to consider and to consider in drawing inferences. The other issue that's been raised, third, is that there are other drugs where the FDA didn't succeed in granting approval by the designated date. Well, those drugs are different. There's no reason to believe that those drugs are good analogs for Lipitor. Lipitor was the largest blockbuster ever at the time. There's evidence here that the FDA specifically targeted it, treated it special, and we have a record here that far exceeds the record in any of the other cases that we've talked about. We talked about Zettia and Colchris. Those are district court cases where there were theories that the FDA would have granted approval earlier. Those cases involved much less evidence than we have here. In those cases, the arguments were being made that the plaintiffs would have done different things before the FDA. They would have acted differently if they had been looking at a different date, because those cases didn't have the kind of developed FDA record that we have here. Ordinarily, you don't have a detailed FDA record where the FDA writes memos describing its exact thinking and why it was doing things, why it was granting the AIP, why it was expediting review, why it actually granted the approvals. We don't usually have that kind of detail. I think that doesn't help you. Well, I guess. Because they're showing their work, and they're explaining why they took steps when they took them, including the reasons you've described for granting the exception to the AIP. So I don't know if that necessarily helps you. I guess I disagree with you, Your Honor. I mean, I think the reasons for their timing was because of the date that had been agreed to, the date that Pfizer paid Rambaxi to accept. The timing of this deal, the reason why they were doing this in 2011, was because they were concerned that Rambaxi wasn't going to get on the market and that Rambaxi was going to block the entry of other generics. That date had a principal role. The FDA says over and over again, everybody realizes the significance of November 30. They said the other generics recognize the significance of November 30. It was the concern about no generic being on the market on November 30 that drove the timing of the approval of that FDA. The FDA recognized that it wasn't a guarantee. It had to tick off all the reasons, had to make sure that this was a safe drug, that it was bioequivalent, that the labeling was good. It recognized that it had to do all of that. But it recognized that it was a big benefit for the public to get that done by November 30. If the date was different, there was enough time in the existing schedule with the existing AIP date to get it done at least one day earlier. But there's more time in there, and the AIP also could have been earlier if everybody had been targeting that earlier date. Thank you, Your Honor. Very quickly, Judge Schwartz, you asked my friend Ms. Lalonde about the sequencing question. I want to turn there first. We think this is true regardless of whether the court affirms or reverses, but certainly if the court affirms the summary judgment, I think everybody agrees that the district court was correct when it said there is no cause of action, and accordingly there is no class. Our quarrel is we think the district court should have stopped there, that the 56-page opinion that followed that was all advisory. But I think we agree, and I think I heard Ms. Lalonde say, that it would be perfectly appropriate for this court to say the district court was correct, that once it had decided that there were no class claims, there was no class to certify. And then the question is, what do you do about that? We think this court's decision in Una Lashtigo Band versus Corzine controls in that scenario. That was a case about an intervention motion that was entertained once the underlying case had dropped out, and the holding is once the underlying case drops out, there is no need to address that issue. That's an advisory opinion, and so the appropriate thing to do is to vacate that decision, and then you don't have all of these erroneous statements about ascertainability floating out there. They can't be used to thwart the certification of a class in the future. So we would urge the court to vacate under the Una Lashtigo case, and we would urge also that decisions like Gale and Lusardi be limited to the circumstances in which that exception arose. They recognize a general rule under Article III that the case is moot once the named plaintiff's claims have dropped out. The purpose of that exception is it's a pro-plaintiff exception to ensure that when you have absentees that might have an interest in the case, that their interests are protected. But, of course, in this scenario where there's a global summary judgment in favor of the defendants and there's no class claim, that exception does not apply. Thank you. Your Honor, my opposing counsel did not have the opportunity to address direct purchase of class, so I'll just make a couple of points. Just to be clear, what was before the district court was both the $3.7 million in out-of-pocket costs of a comparable case, which is actually larger than the number you asked about, divided by 63 is $59,000 apiece. What was also in front of the district court was that here six class members have maximum claims below $10,000, eight below $70,000. I want to emphasize that's a maximum theoretical number if we won a trial, got treble damages. So obviously, given the realistic odds of trial, the values of these claims are lower. That was all in front of the district court. And that's comparable to one of the prior cases I mentioned, the value drug case, also in front of the district court. Twenty-one of the 27 class members there who did not join had among the lowest claims, 21 below $300,000 each. And finally, we also put in front of the district court that information, and a number of those class members overlap with this class here. So thank you. Counsel, we thank you for the briefing, the arguments. They've all been very helpful. The court will take the matter under advisement.