EASTERBROOK, Chief Judge.
Medicare pays teaching hospitals for work by residents (that is, recent graduates still in training) on a fee-for-service basis only when a teaching physician supervises the residents. (Technically the payments are “for” the services rendered by the teacher, in the role of the patient’s attending physician, but the recipient is the hospital rather than either the teaching physician or the resident.) The costs of providing an extended education to the residents are reimbursed through grants rather than by payments for specific services they perform for patients. During the 1990s, the Department of Health and Human Services concluded that many if not all of the 125 teaching hospitals affiliated with medical schools were billing for unsupervised services that residents performed, thus receiving double compensation. HHS began to audit teaching hospitals’ invoices and demand reimbursement. The General Accounting Office (now the Government Accountability Office) conducted its own study and concluded that HHS was right. See GAO, Medicare: Concerns With Physicians at Teaching Hospitals (PATH) Audits (July 1998).
Private litigation has addressed the same topic. Relators may pursue qui tam suits under the False Claims Act, 31 U.S.C. §§ 3729-33, on behalf of the United States and collect a bounty. The risk that unnecessary “me too” private litigation would divert funds from the Treasury led to a proviso in § 3730(e)(4)(A): suits cannot be “based upon the public disclosure of allegations or transactions” in public agencies’ official reports unless the relator is an “original source of the information.” (This language was altered in 2010, but that change is not retroactive. See Graham County Soil & Water Conservation District v. United States ex rel. Wilson, — U.S. -, 130 S.Ct. 1396, 1400 n. 1, 176 L.Ed.2d 225 (2010). We use the language in force when the events underlying this suit took place.)
United States ex rel. Gear v. Emergency Medical Associates of Illinois, Inc., 436 F.3d 726 (7th Cir.2006), concludes that the 1998 GAO report and similar public documents disclose that billing for unsupervised work by residents was an industry-wide practice. This led us to hold that an allegation that a particular teaching hospital had billed for residents’ unsupervised work was “based upon” that disclosure, and that only an “original source” of the information could pursue qui tam litigation. Gear was not an original source and lost. We added in Glaser v. Wound Care Consultants, Inc., 570 F.3d 907, 920 (7th Cir.2009), that a private suit is “based upon” a public disclosure when the allegations are “substantially similar,” even if the private relator adds details. That understanding increases the importance of the “original source” exception. But to qualify as an original source, the relator not only must discover the fraud independently but also must disclose it to the government before filing suit. Gear failed to do that — and Robert S. Goldberg and June Beecham, the relators in this litigation, likewise failed.
Goldberg and Beecham filed this suit against a teaching hospital in 2004, two years before Gear and five years before Glaser. They have revised their complaint several times, trying to plead around those decisions. The district court concluded that they failed, and it dismissed the suit. 748 F.Supp.2d 917 (N.D.Ill.2010). Relators believe that they succeeded, and they also rely on United States ex rel. Baltazar v. Warden, 635 F.3d 866 (7th Cir.2011), which they believe narrows the scope of Gear.
As finally revised, relators’ complaint alleges that Rush University Medical Center submitted fee-for-service bills to the Medicare program on account of unsupervised work that residents had performed in the hospital’s operating theaters. The district court concluded that this kind of allegation tracks the 1998 GAO report and is blocked by § 3730(e)(4)(A). Relators say, however, that they have been more specific about how this hospital billed for unsupervised services. As relators see things, the 1998 GAO report, and the PATH audits more generally, dealt with bills submitted for services that residents had performed all by their lonesome. This suit, according to relators, arises from residents’ services that were supervised, but inadequately— and, perhaps more importantly, that the hospital certified had been supervised. Relators contend that an audit carried out under the PATH protocol would not have detected the fraud being practiced at Rush University Medical Center, and that the GAO report also did not describe this kind of deception.
According to relators, Rush University permitted teaching physicians to supervise multiple operations simultaneously. If Medicare is to pay for the procedure as a teaching physician’s work, “the teaching physician must be present during all critical portions of the procedure and immediately available to furnish services during the entire service or procedure.” 42 C.F.R. § 415.172(a)(1). The complaint alleges that Rush scheduled teaching physicians for multiple surgeries simultaneously, so that even if the teaching physician were present for the “critical” portion of one (indeed, for all “critical” portions of all surgeries in a time slot), the surgeon could not have been “immediately available” for the rest of each procedure. If a teaching physician supervised the “critical” portions of a procedure in Surgery A and then left to do the same in Surgery B, that physician was not “immediately available” to the resident in Surgery A; “critical” procedures required his presence in Surgery B. Rush University says that there is nothing fraudulent about representing compliance with the regulation if some other teaching physician was “immediately available” to the resident in Surgery A, but that is an issue on the merits. The question we must decide now is whether the allegations of this complaint are “substantially similar” to, and thus “based on,” the disclosures in the PATH audits and the GAO report.
The district court answered affirmatively, because the audits and report were about bills for unsupervised work by residents, and the allegations of this complaint concern one means for work to be deemed “unsupervised.” In other words, the court understood “public disclosure of allegations or transactions” (the statutory language) at a high level of generality. This is where Baltazar becomes relevant. We held in Baltazar that a very high level of generality is inappropriate, because then disclosure of some frauds could end up blocking private challenges to many different kinds of fraud. Public reports disclosed that more than half of all chiropractors in an audited sample had submitted improper bills to the Medicare and Medicaid programs. We held that this did not disclose a particular fraud by a particular chiropractor, because no one could use the published reports as the basis of litigation; the government could not seek reimbursement without showing that a particular chiropractor had committed a particular fraud in a particular way, and we held in Baltazar that someone who supplied those vital details could not be thrown out of court under § 3730(e)(4)(A).
Similarly, no one who read the GAO report, or followed the progress of the PATH audits, would know or even suspect that Rush University was misrepresenting the “immediate availability” of teaching physicians during concurrently scheduled procedures. The allegations in Gear parroted the GAO report; Gear added nothing to the public disclosure except the name of a teaching hospital, and as the GAO report suggested that all (or almost all) teaching hospitals billed for unsupervised work by residents, Gear had not added anything of value. Goldberg and Beecham, by contrast, allege a kind of deceit that the GAO report does not attribute to any teaching hospital. Unless we understand the “unsupervised services” conclusion of the GAO report and the HHS audits at the highest level of generality— as covering all ways that supervision could be missing or inadequate — the allegations of these relators are not “substantially similar.” Given Baltazar, boosting the level of generality in order to wipe out qui tarn suits that rest on genuinely new and material information is not sound.
Relators’ allegations may be incorrect— and, to repeat, Rush University has done nothing wrong if a teaching physician was “immediately available” during all parts of the surgeries, even if the principal teacher was making a circuit of operating theaters. But these are questions on the merits, not potential defects in the complaint.
The judgment of the district court is vacated, and the case is remanded for proceedings consistent with this opinion.