United States v. Omnicare, Inc.

903 F.3d 78

• Court: Court of Appeals for the Third Circuit
• Decided: September 4, 2018
• Docket: 16-4418
• Status: Published

Opinion

CHAGARES, Circuit Judge.

*81 Plaintiff-relator Marc Silver appeals the District Court's grant of PharMerica Corporation's 1 motion for summary judgment and motion to dismiss his qui tam action filed under the False Claims Act ("FCA"), 31 U.S.C. §§ 3729 - 33, based on the FCA's public disclosure bar. That bar generally disallows qui tam actions that rely on allegations that are, at least in substantial form, already known to the public. Silver alleges that PharMerica - which owns and operates institutional pharmacies serving nursing homes - unlawfully discounted prices for nursing homes' Medicare Part A patients (reimbursed by the United States (hereinafter, "the Government") to the nursing home on a flat per-diem basis) in order to secure contracts to supply services to patients covered by Medicare Part D and Medicaid (reimbursed directly to the pharmacy by the Government on a cost basis) in the same nursing homes. This practice is known as swapping. Silver challenges the District Court's conclusion that the alleged fraud had already been publicly disclosed. Specifically, Silver asserts that the District Court erred by (1) treating public disclosures concerning the general risk of swapping in the nursing home industry as a bar to his specific allegations, supported by non-public information, that PharMerica was actually engaging in swapping, and (2) concluding that the fraud was publicly disclosed based upon Silver's deposition testimony that he depended upon publicly available documents, without undertaking an independent review to determine whether those documents sufficiently disclosed the fraud. As explained below, we agree with Silver and conclude that his allegations of fraud were not publicly disclosed. We therefore will reverse and remand.

I.

The incentive for a nursing home to swap arises because of the different payment structures noted above. 2 The Government *82 pays the nursing home a fixed per-diem rate for each Part A patient, and from this fixed amount, the nursing home must pay for all of the patient's care, including prescription drugs. Because the nursing home bears the financial risk for the amount of drugs dispensed to their Part A patients (who tend to be the sickest and so consume the most medication), nursing homes are motivated to negotiate with pharmacies for the lowest possible drug prices for those patients. In contrast, nursing homes are less concerned about the cost of drugs dispensed to Medicaid and Part D patients, because the pharmacies collect those payments directly from state Medicaid programs or from Part D prescription drug plan sponsors; the nursing homes bear no financial risk. This reimbursement structure may be viewed as incentivizing the nursing homes to "swap" with the pharmacies for lower drug prices for Part A patients in return for allowing the pharmacy to serve the more lucrative Part D patients. From the perspective of the pharmacies, it could be in their interest to provide drugs to Part A patients at even below-cost prices, because there are many fewer Part A patients than Part D patients, and the profit margins on the services provided to the Part D patients that the pharmacies would win the right to serve could compensate for the losses incurred serving the Part A patients.

Silver alleges that PharMerica did just that: agreed with various nursing homes to provide drugs to Part A patients at per-diem rates that were so low (as little as $8 per day) that they must have been below cost, in exchange for the right to service the nursing home's other residents at the market rate. Because these alleged below-cost payments would thereby serve as "remuneration ... to induce" the nursing homes "to refer an individual" - namely, Part D patients - "for the furnishing of any item or service for which payment may be made in whole or in part under a Federal health care program," 42 U.S.C. § 1320a-7b(b)(2)(A), Silver alleges that the swapping violated the Anti-Kickback statute. Silver accordingly brought these claims under the FCA 3 and its various state-law analogs, alleging that PharMerica fraudulently billed the federal government for services that it obtained through these alleged kickbacks by, among other things, falsely certifying in its reimbursement claims that it was complying with the Anti-Kickback rules.

After the District Court denied PharMerica's Federal Rule of Civil Procedure 12(b)(6) motion to dismiss - a ruling that is not before this Court on appeal - PharMerica filed dispositive motions relying upon the public disclosure bar in the FCA. Because the public disclosure bar was jurisdictional before it was amended on March 23, 2010, PharMerica moved to *83 dismiss Silver's pre-March 23, 2010 claims for lack of jurisdiction and moved for summary judgment on his later claims. The District Court granted both motions, determining - based on a number of publicly available documents that Silver admits he relied upon to deduce his allegation of fraud - that the transactions of fraud were publicly disclosed. Silver timely appealed.

II. 4

The public disclosure bar to the FCA, prior to March 23, 2010, provided that "[n]o court shall have jurisdiction over an action under this section based upon the public disclosure of allegations or transactions ... unless ... the person bringing the action is an original source of the information." 31 U.S.C. § 3730 (e)(4)(A) (2006). As amended effective March 23, 2010, 5 the disclosure bar is no longer jurisdictional and instead provides that a "court shall dismiss an action or claim under this section ... if substantially the same allegations or transactions as alleged in the action or claim were publicly disclosed ... unless ... the person bringing the action is an original source of the information." 31 U.S.C. § 3730 (e)(4)(A) (2010) ; see also United States ex rel. Moore & Co. v. Majestic Blue Fisheries, LLC , 812 F.3d 294 , 300 (3d Cir. 2016). Whereas an "allegation" of fraud is a specific allegation of wrongdoing, a "transaction" that raises an inference of fraud consists of both the allegedly misrepresented facts and the allegedly true state of affairs. See United States ex rel. Dunleavy v. Cty. of Del. , 123 F.3d 734 , 741 (3d Cir. 1997), abrogated on other grounds by Graham Cty. Soil & Water Conservation Dist. v. United States ex rel. Wilson , 559 U.S. 280 , 130 S.Ct. 1396 , 176 L.Ed.2d 225 (2010) ; Moore & Co. , 812 F.3d at 303 . As no one contends that, prior to Silver's suit, PharMerica had been publicly and explicitly accused of engaging in swapping, our task in this case is to ascertain whether the transactions raising an inference of that allegation of fraud were already publicly disclosed.

To determine whether a fraudulent transaction has been publicly disclosed by information contained in one of the enumerated public sources, 6 this Court employs a formula of sorts, where:

*84 "If X + Y = Z, Z represents the allegation of fraud and X and Y represent its essential elements. In order to disclose the fraudulent transaction publicly, the combination of X and Y must be revealed, from which readers or listeners may infer Z, i.e., the conclusion that fraud has been committed."

United States ex rel. Zizic v. Q2Administrators, LLC , 728 F.3d 228 , 236 (3d Cir. 2013) (quoting Dunleavy , 123 F.3d at 741 ). For a court to conclude that an inference of fraud [Z] has been publicly disclosed such that the public disclosure bar is triggered, then, "both a misrepresented [X] and a true [Y] state of facts must be publicly disclosed." United States ex rel. Atkinson v. PA. Shipbuilding Co. , 473 F.3d 506 , 519 (3d Cir. 2007). Where the fraud has been publicly disclosed - either because the public documents set out the allegation of fraud itself [Z] or its essential elements [X+Y] - a relator's claim will be barred so long as it is " 'supported by' or 'substantially similar to' [the] public disclosures." Zizic , 728 F.3d at 237 (quoting United States ex rel. Mistick PBT v. Hous. Auth. of City of Pittsburgh , 186 F.3d 376 , 385-88 (3d Cir. 1999) ); 31 U.S.C. § 3730 (e)(4)(A) (2010).

In this case, the parties agree that the allegedly "misrepresented" set of facts [X] is that PharMerica was complying with the Anti-Kickback statute, 7 and that the allegedly "true" state of facts [Y] is that PharMerica was in fact engaging in the fraudulent practice of swapping, which violates the statute. PharMerica argued - and the District Court found - that a number of publicly available reports and documents, upon which Silver testified that he relied to deduce the fraud, discussed swapping in the nursing home industry and accordingly that "the information cumulatively disclosed in the publicly available documents was sufficient to support an inference that PharMerica allegedly engaged in swapping transactions with nursing homes, and therefore the true state of facts (Y) was publicly disclosed." Appendix ("App.") 16. Finding that both X and Y were publicly disclosed, the District Court concluded that Silver's claim was barred. In reaching this conclusion, the District Court rejected Silver's argument that the public documents could not, on their own, disclose the fraud, and that to arrive at his allegations, Silver had relied on non-public contracts he had seen that indicated that PharMerica was offering below-price per-diem rates for Part A patients. Silver contends that the District Court erred in doing *85 so for two reasons. First, Silver argues that the District Court improperly determined that documents publicly describing the generalized risk of swapping in the nursing home industry served to bar his specific claim, which depended on non-public information that PharMerica was actually engaging in swapping in specific contracts. Second, Silver contends that the District Court ignored this Court's guidance when it concluded, on the basis of Silver's testimony, that he relied upon certain publicly available information to reach his conclusion and that the information itself disclosed the fraud, without independently determining that the relevant public document did, in fact, effectuate such a disclosure. We agree.

A.

As noted above, the District Court determined that various reports cumulatively disclosed the alleged fraudulent transactions. These reports consisted of:

1. A 1999 advisory opinion by the Health and Human Services - Office of the Inspector General ("HHS-OIG") concerning an ambulance company that wanted to provide steeply discounted services to a nursing home. The advisory opinion noted that, depending on the intent, such an offer might violate the Anti-Kickback Act because it provides a discount for services that the nursing home itself pays "in exchange for the opportunity to service and bill for higher paying Federal health care program business reimbursed directly by the program to the supplier." App. 700. The advisory opinion also noted that HHS-OIG had received "a considerable number of informal inquiries and anecdotal reports regarding discounts to [nursing homes] ... since the enactment of the [prospective payment system establishing per-diem reimbursement for Part A patients]" and that the inquiries "suggest that suppliers of a wide range of [nursing home] services" are offering Part A discounts that are "linked, directly or indirectly," to getting business that could be billed directly to the federal government via Medicare Part B. App. 700;

2. A 2000 HHS-OIG "Compliance Program Guidance for Nursing Facilities" published in the Federal Register that referenced the 1999 HHS-OIG advisory opinion and defined swapping as "when a supplier gives a nursing facility discounts on Medicare Part A items and services in return for the referrals of Medicare Part B business." App. 713;

3. A 2008 HHS-OIG "Supplemental Compliance Program Guidance for Nursing Facilities" reiterating that swapping violates the Anti-Kickback Act and cautioning that "a nursing facility should be careful that there is no link or connection, explicit or implicit, between discounts offered or solicited for business that the nursing facility pays for and the nursing facility's referral of business billable by the supplier or provider directly to Medicare or another Federal health care program." App. 734;

4. A 2004 report by the Lewin Group commissioned by the Centers for Medicare and Medicaid Services ("CMS") that discusses specifically the interactions between institutional pharmacies and nursing homes and notes that pharmacies provide many services to nursing homes at little or no cost. 8

*86 5. 2007 reports by the Harvard Medical School and the Medicare Payment Advisory Commission asserting that "[t]he [institutional pharmacy] market is highly concentrated, with the top three firms accounting for two-thirds of nursing home beds: Omnicare covers about 850,000 of the nation's 1.7 million beds (50 percent), PharMerica covers 220,000 (13 percent), and Kindred Pharmacy Services (KPS) covers 100,000 (6 percent)." App. 696; and

6. PharMerica's Form 10-k financial disclosures which delineated aggregate information such as PharMerica's costs, gross profits, and its bottom line.

The District Court's analysis relied most heavily on the Lewin Report - which the District Court viewed as linking the general statements about swapping in the nursing home industry with swapping between nursing homes and pharmacies in particular - and the 10-k disclosures that Silver supposedly relied upon as "the last piece of information he needed to conclude that PharMerica was, indeed, engaging in swapping." App. 17.

Neither of the documents, alone or considered together with the rest of the public documents, disclose the fraudulent transactions that Silver alleges, not least of which because the documents do not point to any specific fraudulent transactions directly attributable to PharMerica. See, e.g. , Atkinson , 473 F.3d at 528-29 (considering separately, for disclosure bar purposes, each specific "transaction" in which defendants were alleged to have misrepresented the true state of facts); United States ex rel. Feingold v. AdminaStar Fed., Inc. , 324 F.3d 492 , 495 (7th Cir. 2003) (explaining that a fraud is publicly disclosed "when the critical elements exposing the transaction as fraudulent are placed in the public domain"); see also Zizic , 728 F.3d at 237-38 (concluding that public disclosure bar applied because defendants were at least "directly identifiable" from the allegations that had already disclosed the specific fraudulent transaction). Rather, the documents merely indicate the possibility that such a fraud could be perpetrated in the nursing home industry, which is an allegation that would alone be insufficient to state a claim for fraud under the FCA and Federal Rule of Civil Procedure 9(b). Silver's more concrete claim, which set out specific facts suggesting that PharMerica in particular was actually engaged in swapping, relied upon these general disclosures but could not have been derived from them absent Silver's addition of the non-public per-diem information. As explained below, we hold that the FCA's public disclosure bar is not implicated in such a circumstance, where a relator's non-public information permits an inference of fraud that could not have been supported by the public disclosures alone.

*87 1.

Both the District Court and PharMerica accord too much weight to the Lewin Report. The District Court found that the Lewin Report "indicated that, as to long-term care pharmacies ... in particular, conditions were ripe for swapping transactions." App. 6. Likewise, PharMerica relies on the report as proof that the Government was "concern[ed] with swapping in the long-term care industry applied specifically to the provision of services by long-term care pharmacies like PharMerica" and commissioned the Lewin Report "to evaluate how long-term care pharmacies charge" for their services. PharMerica Br. 25. The Lewin Report explained that institutional pharmacies at the time provided many of their services to nursing homes at little or no cost and still achieved acceptable profit margins, because Medicaid was then the largest source of revenue for pharmacies and reimbursed for prescriptions at a sufficiently high rate so as to allow the pharmacies to offer these additional low-cost or free services. However, far from criticizing or noting concern about these free tie-in services, the reason that the Lewin Report was commissioned appears to have been to ascertain whether the pharmacies would be able to continue to provide these "customary services" that nursing homes had come to rely upon after Medicare Part D replaced Medicaid as the primary form of coverage for nursing home residents. See App. 741-42.

Contrary to PharMerica and the District Court's view, the Lewin Report appears to indicate that the Government was aware that pharmacies offered low-cost services bundled with their provision of drugs and services to Medicaid patients and that it hoped those low-cost services could continue after the transition to Part D. Rather than publicly disclosing the prevalence of or concern about swapping, the Lewin Report seems to indicate that the Government desired that pharmacies would continue to engage in conduct that plainly invited swapping and moreover that the Lewin Group was of the opinion that swapping was not a pervasive problem. See App. 759-60 (noting that pharmacies do not offer discounted drug pricing below the cost set by Medicaid because they are concerned about the risks of appearing to be engaging in swapping); App. at 763 (explaining that pharmacies "prefer fee-for-service reimbursement" rather than offering per-diem pricing and use true-up clauses to limit the risk they bear when they do provide per-diem pricing, by adjusting the per diem on a monthly basis to match the Medicaid rate). The Lewin Report simply gives no indication that, as PharMerica asserts, "CMS ... [was] concerned with and actively investigating swapping many years before [Silver] filed his lawsuit." 9 PharMerica Br. 26. Indeed, the Lewin Report does not appear at all to discuss discount pricing or swapping regarding prescription drugs. See supra note 8. The Lewin Report therefore does not *88 support or even hint at the inference that any institutional pharmacy - let alone PharMerica in particular - was swapping or would in the future be likely to swap, or that the Government was particularly concerned that the free tie-in services would lead to the scourge of swapping.

2.

The District Court also relied heavily upon PharMerica's 10-k disclosure form, which Silver testified that he consulted before filing his FCA claim. The majority of the District Court's analysis of whether the fraudulent transactions were publicly disclosed was dedicated to its determination that Silver had conceded that the aggregate financial information included in the 10-k was sufficient to support a "conclu[sion] that PharMerica had engaged in illegal swapping." App. 15. At no point did the District Court elucidate what information in the 10-k forms disclosed or suggested that PharMerica was engaged in swapping or how anyone could use the 10-k data in conjunction with information from the other public sources to reach such a conclusion. Rather, the District Court merely cited Silver's deposition testimony, in which he purportedly admitted that he relied on PharMerica's financial statements and that the information contained therein permitted him to make the "pretty easy" deduction that PharMerica was swapping. App. 639-40. The District Court rejected as "internally inconsistent" Silver's argument that this testimony was taken out of context and that for the disclosures themselves to support an inference of fraud, they would need to include more granular information about individual nursing homes, rather than average or aggregate data. App. 14-15. In reaching this conclusion, the District Court misapprehended Silver's testimony and the central importance of his non-public per-diem information to the plausibility of his allegation of fraud.

The crux of Silver's allegation is that the $8-10 per-diem rates that he discovered must have been below-cost (and so violate the Anti-Kickback Act) because if PharMerica had so low a cost to buy prescription drugs such that it was making money on services for Part A patients even with such a low reimbursement rate, then it would have been making an enormous profit on its significantly more numerous services to Part D patients, which Silver contends are reimbursed at a rate that is two-to-three times higher. But based on PharMerica's publicly stated profits in its 10-k, Silver deduced that PharMerica could not be making such enormous profits on their Part D patients because the company was simply not that profitable. Silver concluded that PharMerica must not in fact have such a low cost to purchase prescription drugs, meaning that it must be offering per-diem rates to Part A patients that are below its costs. Crucially, while this analysis depends on having a general sense of PharMerica's gross profitability (which is public information), the analysis would be impossible without first knowing what per-diem rate it was offering to Part A patients (which is not public information). This is because if the rate it was offering was, for example, $20 per day rather than $10 per day, its costs to purchase prescription drugs could be higher and it could still turn a profit on its Part A patients, while its profits from Part D patients - while still better than Part A profits - would not be so excessive such that it would not align with the reported gross profits. In order to allege plausibly that PharMerica was offering below-cost per-diem rates for Part A patients, then, Silver needed to know what the per-diem rate was. No one contends that this rate was publicly disclosed.

*89 With this understanding of how Silver deduced the alleged fraud, it becomes clear that the District Court erred in determining that the fraud was publicly disclosed via (1) documents indicating that swapping was a risk inherent in the nursing home business, (2) documents confirming that PharMerica was one of the major players servicing nursing homes, and (3) PharMerica's financial statements. In his deposition statements concerning his reliance on the financial statements, upon which the District Court based its conclusion that the fraud could be deduced by reliance on the information contained in those documents alone, Silver makes clear that his private knowledge of PharMerica's per-diem rates was the key to uncovering the fraud. Without this information, the public information that he consulted, which reported that swapping was a potential problem in the nursing home industry, would have been insufficient to disclose the actual fraud that Silver alleges. As the Court of Appeals for the Ninth Circuit recently observed, "[a]llowing a public document describing 'problems' - or even some generalized fraud ... across a swath of an industry - to bar all FCA suits identifying specific instances of fraud in that ... industry would deprive the Government of information that could lead to recovery of misspent Government funds and prevention of further fraud." United States ex rel. Mateski v. Raytheon Co. , 816 F.3d 565 , 577 (9th Cir. 2016). Although we have not explicitly said so, we clarify now that the FCA's public disclosure bar is not triggered when a relator relies upon non-public information to make sense of publicly available information, where the public information - standing alone - could not have reasonably or plausibly supported an inference that the fraud was in fact occurring. See, e.g. , United States ex rel. Shea v. Cellco P'ship , 863 F.3d 923 , 935 (D.C. Cir. 2017) (public disclosure bar not triggered where the relator "supplied the missing link between the public information and the alleged fraud" by "rel[ying] on nonpublic information to interpret each [publicly disclosed] contract," and where "[w]ithout [relator's] nonpublic sources ... there was insufficient [public] information to conclude" that the defendant actually engaged in the alleged fraud).

* * * * *

Having concluded that the publicly available information did not disclose the alleged true state of affairs that PharMerica was violating the Anti-Kickback law by engaging in swapping - what, in the terminology of our mathematical representation of the public disclosure analysis, we might title the "Y-factor" - the public disclosure bar is inapplicable to Silver's claims. The District Court erred in concluding otherwise.

B.

This conclusion is fully in keeping with our precedents applying the public disclosure bar to parasitic suits in which a relator uncovers a fraud based only on the application of background knowledge or experience to the publicly available facts, see United States ex rel. Stinson, Lyons, Gerlin & Bustamante v. Prudential Ins. Co. , 944 F.2d 1149 , 1160 (3d Cir. 1991), or to cases in which the relator relies "even partly" on publicly disclosed allegations of fraud, Zizic , 728 F.3d at 238 . In both such circumstances, a particular concrete allegation of fraud has already been disclosed in whole or in part, and the relator is merely extrapolating from or expanding on the allegation to include allegedly new fraudsters. When a free-standing allegation of fraud already exists in the public realm, the mere application of experience or deductive skills to such information or the addition of another allegation to the *90 already articulated accusation of fraud does not create a new, non-barred, claim of fraud. See, e.g. , Mateski , 816 F.3d at 579-80 ; United States ex rel. Winkelman v. CVS Caremark Corp. , 827 F.3d 201 , 210 (1st Cir. 2016) ("[A] complaint that targets a scheme previously revealed through public disclosures is barred even if it offers greater detail about the underlying conduct."). On the other hand, when, as here, the publicly disclosed information lacks relevant significance to the claim of fraud absent the addition of relator's non-public information, there are simply no publicly disclosed allegations of fraud upon which the relators claim could be based. Rather, the allegation exists solely by virtue of the relator's added information.

This distinction between concrete allegations of fraud and disclosures that might support such an allegation if supplemented by more particular information, likewise distinguishes this case from cases in which a fraudulent transaction was deemed disclosed even though the defendant itself was never mentioned in the public documents.

For instance, in United States ex rel. Gear v. Emergency Medical Associates of Illinois Inc. , 436 F.3d 726 (7th Cir. 2006), upon which PharMerica relies, the relator's claim regarding fraudulent billing practices at the teaching hospital that he worked at was barred based on the public disclosure that such fraudulent practices were taking place at teaching hospitals nationwide, even though relator's hospital - 1 of 125 such institutions operating at the time - was never mentioned in the disclosures. Id. at 728 . The Court of Appeals for the Seventh Circuit rejected the relator's contention "that for there to be public disclosure, the specific defendants named in the lawsuit must have been identified in the public records," and held that "[i]ndustry-wide public disclosures bar qui tam actions against any defendant who is directly identifiable from the public disclosures." Id. at 729 . PharMerica contends - and the District Court held - that the HHS-OIG and CMS documents suggesting that swapping may be occurring in the nursing home industry and acknowledging PharMerica's status as one of the three largest institutional pharmacies serving nursing homes, means that PharMerica was identified as a likely swapper, even without being directly named. In Gear , however, the allegations concerning the fraudulent practice were concrete and leveled directly at the industry at issue, and - most importantly - various hospitals had reached settlements with the Government concerning specific allegations that they engaged in the practice, and the Government was in the process of auditing dozens of additional hospitals. Id. at 728-29 . Given the public disclosures that the fraud was actually being perpetrated across the industry and the clear indication that the Government was already uncovering the culpable institutions, the relator's addition of information specifically identifying yet another hospital did not constitute relevantly new or undisclosed information. In the case at bar, however, the public disclosures lack any concrete indication that pharmacies were actually swapping, and the most on-point document seems to indicate that they were not doing so. Silver's allegation, supported by non-public contracts plausibly indicating below-cost per-diem pricing (which the Lewin Report specifically noted pharmacies would not offer) and clarifying the mechanism of the fraud (the true-up clauses that imply that the low per-diem rates are introductory prices subject to increase, but which PharMerica never "trues-up"), has relevance that was lacking in Gear , because it implicates participants in an industry that had, as yet, never been specifically accused of engaging in the fraud. See, e.g. , *91 United States ex rel. Dig. Healthcare, Inc. v. Affiliated Comput. Servs., Inc. , 778 F.Supp.2d 37 , 49-51 (D.D.C. 2011) (explaining that "while the government may be aware of fraud and improper payments being made by participants in the Medicaid program on a general level, it was not 'squarely on the trail' of the defendant," where the purported public disclosures "reveal some important background information," but do "not rise to the level of 'allegations or transactions' " (quoting United States ex rel. Fine v. Sandia Corp. , 70 F.3d 568 , 571 (10th Cir. 1995) ) ).

Similarly, in Zizic , this Court found that the fraud engaged in by two companies was publicly disclosed by a prior lawsuit even though the defendants were not named, because a prior suit alleged a specific fraud taking place in an industry - qualified independent contractors ("QICs") who review certain Medicare eligibility determinations - over a period of time, and only one QIC operated in the industry at any given period. Zizic , 728 F.3d at 238 . The fraud in Zizic was specifically alleged to have occurred in the industry, and the identity of each company was readily ascertainable because they "were the only QICs during their respective contractual terms" which took place during the time period alleged in the prior suit and anyone could look up what company served as a QIC at a given time. Id. The Court therefore concluded that, although the defendants "were not actually identified in the [prior] litigation, they were directly identifiable from that public disclosure." Id. Again, the same is not true here, where no specific allegations of fraud or disclosures of information which would raise an inference of fraud had been made against pharmacies servicing nursing homes.

Finally, our refusal to afford preclusive effect to information that discloses merely a potential or possibility of fraud, without any indication of who is perpetrating it or how they are doing so, accords with the heightened showing required by Federal Rule of Civil Procedure 9(b) when pleading a claim of fraud in FCA actions. See Foglia v. Renal Ventures Mgmt. , 754 F.3d 153 , 155 (3d Cir. 2014) ("[I]n alleging fraud or mistake, a party must state with particularity the circumstances constituting fraud or mistake.") (quoting Fed. R. Civ. P. 9(b) ). None of the publicly available documents in this case indicate that any institutional pharmacy is engaging in swapping or is likely to do so. The Lewin Report is the only document discussing pharmacies in particular, and that document at most explains the various payment structures that would make swapping possible or attractive. It does not imply that any pharmacy is suspected of engaging in swapping, and in fact asserts just the opposite - that pharmacies are wary of any prescription drug pricing that falls below the price set by the federal government.

A complaint based only on these publicly available documents would not be able to "support its allegations" with adequate factual detail needed to plead fraud with particularity. Moore & Co. , 812 F.3d at 307 . At a minimum, for an FCA relator to satisfy Rule 9(b), "he must provide 'particular details of a scheme to submit false claims paired with reliable indicia that lead to a strong inference that claims were actually submitted' "; "[d]escribing a mere opportunity for fraud will not suffice." Foglia , 754 F.3d at 157-58 (quoting United States ex rel. Grubbs v. Kanneganti , 565 F.3d 180 , 190 (5th Cir. 2009) ); see also, e.g. , United States ex rel. Baltazar v. Warden , 635 F.3d 866 , 868 (7th Cir. 2011) (holding that "reports documenting a significant rate of false claims by an industry as a whole - without attributing fraud to particular firms - do not prevent a qui *92 tam suit against any particular member of that industry ... because these reports do not so much as hint that any particular provider has submitted fraudulent bills" and so "do not disclose the allegations or transactions on which [Realtor's] suit ... is based" (citations omitted) ). In Foglia , this Court noted that an inference of illegality based on facts that could plausibly have either a legal or illegal explanation would be insufficient to meet Rule 9(b) 's burden, because a relator must "establish a 'strong inference' that false claims were submitted" and the possibility of a legitimate explanation undermines the strength of the inference of illegality. Foglia , 754 F.3d at 158 .

As we explained earlier, had Silver not been in possession of the non-public per-diem information, PharMerica's financial statements would not have raised a sufficiently strong inference of a false claim, because they would be just as consistent with PharMerica's use of higher per diems that were not below cost. Only with the addition of Silver's non-public per-diem information is the allegation of fraud raised with the necessary force. In other words, but for Silver's knowledge of the non-public contract information, the financial disclosures could not have provided specific enough detail to allege a fraud under Rule 9(b) 's pleading standard. No plaintiff could have come into court with only the publicly available information and survived a motion to dismiss, because even if the public documents identify a high likelihood of swapping in the nursing home industry and even if the institutional pharmacy sector is highly concentrated such that PharMerica is an obvious defendant, none of the documents indicate that PharMerica was actually engaging in swapping, as opposed simply to operating in an environment that makes swapping attractive. See, e.g. , Foglia , 754 F.3d at 158 ; Mateski , 816 F.3d at 577 ; Baltazar , 635 F.3d at 868 . We conclude that the public disclosures concerning the potential for swapping in the nursing home industry did not publicly disclose the actual fraud that Silver alleges, and his claim is accordingly not foreclosed by the FCA's eponymous bar.

III.

Silver also finds fault with the manner of the District Court's determination that Silver's admission that he relied upon certain public documents to deduce PharMerica's fraud meant that those documents had publicly disclosed the fraud. He argues that not only did the District Court err substantively as discussed above, but also that it erred procedurally by failing independently to determine whether the public documents at issue in fact contained sufficient information to disclose the fraudulent transactions. Instead, Silver contends, the District Court essentially took him at his word that his analysis of certain documents alerted him to the fraud, and accordingly determined that those documents must therefore have already publicly disclosed the fraud, thereby barring Silver's claim. Silver cites to our decision in Mistick where - discussing whether allegations that are "based upon" publicly available information must be actually derived from that information or instead just "supported by" or "substantially similar" to that information - we agreed with the majority of the Courts of Appeals that "the relator's independent knowledge of the information is irrelevant" if his allegations merely mirror allegations that were already publicly disclosed. 186 F.3d at 386 . Silver argues based on this that when determining whether an allegation or transaction is actually publicly disclosed, it is improper to rely upon what the relator says he relied on (because whether or not he relied on the public information is irrelevant), but instead that the court must *93 analyze the public documents to ascertain whether they disclose the fraud in sufficient detail. Again, we agree.

If the information that the relator relied upon is irrelevant to determining whether his allegations are based upon publicly available information, it would be anomalous to rely upon his characterization of the record to determine that the information was indeed public or that his allegations are in fact derived from those public documents. Although we have not specifically addressed the procedure to be followed when determining whether a given document relied upon by a relator publicly disclosed the fraud, in Atkinson , we mandated a two-step process to determine whether the public disclosure bar applies. "First, [the court must] determine whether the information was disclosed via one of the sources listed in § 3730(a)(4)(A). Second, [the court must] decide whether the relator's complaint is based upon those disclosures." Atkinson , 473 F.3d at 519 . For both steps, the court must reach its own conclusions based on the content of the record before it. As was made clear in Mistick , a relator's subjective belief that he relied upon certain information is immaterial to the court's decision, which must be based on an independent assessment of the scope of the information disclosed the by public documents. If the public documents disclose substantially the same fraud that the relator - even through non-public information - alleges, the allegation is deemed publicly disclosed, regardless of the relator's honest, but mistaken, belief to the contrary. 10 The court in such a situation owes no deference to the relator's understanding of how he arrived at his allegations, but instead must review the public documents and assess what relevant information can be gleaned from them. It follows that a relator's honest, but likewise mistaken, belief that certain public documents themselves disclose the alleged fraud - where in fact the documents only effect such a disclosure when read in light of proprietary or non-public information held by the relator - cannot be the sole basis for a court's determination that the documents disclosed the fraud. 11 Rather, as is the case when a relator claims not to rely on public disclosures, the court must first determine whether the publicly available documents in fact disclosed information sufficient to raise the inference of fraud, and second whether the relator's complaint objectively relied upon that disclosed information. See, e.g. , Shea , 863 F.3d at 934-35 (considering the content of public disclosures that relator suggested in his deposition provided him with the information needed to deduce the fraud, and independently determining that despite relator's apparent testimony, none of those public documents actually raised inferences of fraud).

Here, the District Court conflated these two steps, by basing its conclusion that the allegation of fraud was publicly disclosed *94 (step one) largely on Silver's apparent contention that he relied upon certain publicly disclosed documents (step two), rather than on an independent assessment of the scope of each disclosure. This is particularly clear in relation to PharMerica's form 10-k disclosures. Silver at various points in his deposition testimony admitted to relying on the aggregate financial information contained in the 10-k, which the District Court concluded was the "last piece of information" that Silver needed to make his allegation. App. 17. But the District Court did not explain how the information in the 10-k, even when combined with the other publicly available information, could lead to an inference of fraud. Neither could PharMerica, when pressed at oral argument, put forward any chain of reasoning based only on the 10-k and the publicly available information that would lead to Silver's allegation. Instead, in its brief and at oral argument, PharMerica returned continually to the fact that Silver said that he relied on the 10-k, and insisted that our analysis must stop there. But as we make clear now, in the context of the public disclosure bar, courts may not rest their conclusions based only on the relator's view of the state of the public disclosures. And as we held in Part II, infra , an independent analysis of the record leads to the conclusion that PharMerica's public financial disclosures could not, alone or in concert with the other disclosures, have uncovered PharMerica's alleged swapping. Such a conclusion instead depends necessarily upon Silver's non-public per-diem information. The District Court should have independently assessed the 10-k disclosures and explained what conclusions could reasonably be drawn therefrom - an exercise which likely would have alerted the District Court to the central flaw in PharMerica's argument. That it did not do so is a separate basis for our decision to reverse and remand.

IV.

Silver also argues that the District Court erred by refusing to assert supplemental jurisdiction over his state law claims. We review such a decision for abuse of discretion. See Elkadrawy v. Vanguard Grp. , 584 F.3d 169 , 172 (3d Cir. 2009). The District Court declined to exercise supplemental jurisdiction over Silver's state law claims based on its conclusion that it no longer retained any cause of action establishing federal jurisdiction. See 28 U.S.C. § 1367 (c)(3) ("The district courts may decline to exercise supplemental jurisdiction over a claim ... if ... the district court has dismissed all claims over which it has original jurisdiction."). Because we conclude that Silver's FCA claim is not foreclosed by the public disclosure bar and that his federal claim will remain pending before the District Court, we will also vacate and remand the District Court's order as to supplemental jurisdiction, to give the District Court an opportunity to consider exercising its jurisdiction over the claims brought under state law. 12

V.

For the foregoing reasons, we will reverse the District Court's Order and remand for proceedings consistent with this opinion.

1 PharMerica is the only active appellee in this matter. Omnicare, Inc., NNS Healthcare, Inc., and Neighborcare, Inc., were previously dismissed from the underlying suit, and Chem Rx Corporation is wholly owned by PharMerica.

2 Our description of how these distinct reimbursement policies may induce a nursing home to engage in swapping is derived from the parties' briefs. Neither party disputes this underlying incentive structure, which is amply corroborated by the documents in the record. See, e.g. , Appendix ("App.") 700 (Health and Human Services advisory opinion describing the "obvious motives for agreeing to trade discounts on [per diem reimbursement] business for referrals of non-[per diem reimbursement] business: the [nursing homes] minimize risk of losses under the [per diem reimbursement] system and [the service providers] secure business in a highly competitive market").

3 The FCA imposes civil liability on "any person who ... knowingly presents, or causes to be presented [to the federal government], a false or fraudulent claim for payment or approval," 31 U.S.C. § 3729 (a)(1)(A), and permits private persons to "bring a qui tam action on behalf of the government to recover losses incurred because of fraudulent claims," Hutchins v. Wilentz, Goldman & Spitzer , 253 F.3d 176 , 181-82 (3d Cir. 2001) (footnote omitted). "If the qui tam suit is ultimately successful, the private plaintiff, known as a relator, is entitled to up to 30% of the funds the government recovers." Id. at 182 . As noted earlier, Silver is the relator in this FCA case.

4 The District Court had jurisdiction under 28 U.S.C. § 1331 , and we have jurisdiction under 28 U.S.C. § 1291 . We review de novo both a district court's dismissal of an FCA claim for lack of subject matter jurisdiction, United States ex rel. Zizic v. Q2Administrators, LLC , 728 F.3d 228 , 234 (3d Cir. 2013), and its order granting a motion for summary judgment, United States ex rel. Spay v. CVS Caremark Corp. , 875 F.3d 746 , 752 (3d Cir. 2017).

5 Because the amendment, contained in the Patient Protection and Affordable Care Act of 2010 ("ACA"), Pub. L. No. 111-148 § 10104(j)(2), 124 Stat. 119 , 901, is not retroactive, see Graham Cty. Soil & Water Conservation Dist. v. United States ex rel. Wilson , 559 U.S. 280 , 283 n.1, 130 S.Ct. 1396 , 176 L.Ed.2d 225 (2010), claims based on conduct occurring before March 23, 2010 are still governed under the prior jurisdictional version of the statute.

6 The list of sources through which the disclosure of information would be deemed a public disclosure under the FCA was also amended and narrowed by the ACA. See, e.g. , Moore & Co. , 812 F.3d at 299 . The parties agree that the information alleged to have publicly disclosed the alleged fraudulent transactions in this case occurred through sources that would qualify as public disclosure sources under either version of the statute. See App. 12-13. The ACA's other relevant change - that the relator's alleged fraud need only be "substantially the same" as, rather than "based on," the publicly disclosed allegations or transactions in order to trigger the public disclosure bar - merely codified the law as it already existed in this Circuit. See United States ex rel. Atkinson v. Pa. Shipbuilding Co. , 473 F.3d 506 , 519 (3d Cir. 2007) ("To be 'based upon' the publicly revealed allegations or transactions the complaint need only be 'supported by' or "substantially similar to" the disclosed allegations and transactions." (quoting United States ex rel. Mistick PBT v. Hous. Auth. of City of Pittsburgh , 186 F.3d 376 , 385-88 (3d Cir. 1999) ) ). Accordingly, because the legal framework applicable to the determination of whether Silver's allegations were publicly disclosed is the same under either version of the statute, we need not consider separately the pre- and post-March 28, 2010 conduct.

7 We have recognized that "[f]alsely certifying compliance with the ... Anti-Kickback Act[ ] in connection with a claim submitted to a federally funded insurance program is actionable under the FCA." United States ex rel. Wilkins v. United Health Grp., Inc. , 659 F.3d 295 , 312 (3d Cir. 2011) (quoting United States ex rel. Kosenske v. Carlisle HMA, Inc. , 554 F.3d 88 , 94 (3d Cir. 2009) ). Moreover, as we noted in Wilkins , id. at 311 n.19,, Congress in the ACA, § 6402(f), 124 Stat. at 759, amended the Anti-Kickback Statute to state expressly that "a claim that includes items or services resulting from a violation of this section constitutes a false or fraudulent claim for purposes" of the FCA. See also 42 U.S.C. § 1320a-7b(g). Some of the fraudulent conduct alleged here, however, predates the ACA's effective date of March 23, 2010, and the amendment is not retroactive.

8 The District Court incorrectly described the Lewin Report as "indicat[ing] that long-term care pharmacies provide prescription drugs to nursing homes at little to no charge." App. 6 (emphasis added). The report is clear, however, that the pharmacies "provide many services to nursing facilities at little or no charge" and that they are able to do so specifically because they are reimbursed well for their provision of prescription drugs under Medicaid. App. 741; see also id. at 761 ("[Pharmacies] are able to offer many medication administration services at no additional charge because the Medicaid pharmacy reimbursement rates are high enough to cover the cost of these services. In essence, states are cross-subsidizing the cost of medication administration services through ingredient and dispensing rates rather than paying them directly through nursing facility rates"). There is no indication in the report that any prescription drugs were being provided at low cost. In fact, the Lewin Report indicates precisely the opposite, namely that pharmacies are hesitant to offer discounts on prescription drugs ( i.e. , offer drug prices lower than the rate set by Medicaid), lest they appear to be engaging in swapping. App. 759-60.

9 It is not the case, as PharMerica asserts, that "the government itself reported on widespread 'swapping' in the long-term care pharmacy industry." PharMerica Br. 33. Rather, the HHS-OIG documents discussed the risk of swapping in the nursing home industry, between nursing homes and their suppliers. Although pharmacies certainly fall into the category of suppliers of nursing homes, they were not specifically identified as suspected swappers. PharMerica itself recognized this distinction, as it initially (and correctly) describes the HHS-OIG documents as identifying the government's concern with "swapping arrangement in the long-term care industry" and the "pervasiveness of swapping transactions in the nursing home industry ," id. at 23-24, and then attempts in its argument section to recast these disclosures as concerning the pharmacies themselves.

10 Of course, if the relator actually relied on non-public information to reach his allegation of fraud, he may be eligible as an original source of the allegation and thereby "clear the [public disclosure] bar." Moore & Co. , 812 F.3d at 304 ; Atkinson , 473 F.3d at 520 .

11 Our conclusion does not render such an admission meaningless. If the district court makes the independent determination that information in certain documents publicly disclosed the fraud, then a relator's concession that he relied on that information could constitute a waiver of his ability to argue that he is an original source of the information, because in that case his information would not be "independent" of the public disclosure. See, e.g. , United States ex rel. Schumann v. Astrazeneca Pharm. L.P. , 769 F.3d 837 , 845 (3d Cir. 2014). But such an admission cannot itself establish that the information was publicly disclosed.

12 Given our determination that Silver's allegation was not publicly disclosed, we need not reach his alternative claim that, even assuming the public disclosure bar applied, the District Court erred when it determined that Silver failed to qualify for the FCA's "original source" exception.