Summary
upholding a conviction for healthcare fraud where the scheme involved a doctor "pre-filling" patient forms so that the eye condition was a default diagnosis before the doctor even met with a patient
Summary of this case from United States v. GladdenOpinion
No. 18-10991
07-31-2020
UNITED STATES of America, Plaintiff - Appellee, v. Salomon E. MELGEN, Defendant - Appellant.
John C. Shipley, Assistant U.S. Attorney, Emily M. Smachetti, U.S. Attorney Service - Southern District of Florida, U.S. Attorney Service - SFL, Miami, FL, Robert Benjamin Cornell, U.S. Attorney's Office, Fort Lauderdale, FL, for Plaintiff - Appellee John Kirk Ogrosky, Murad Hussain, Sina Mansouri, Arnold & Porter, LLP, Washington, DC, Matthew I. Menchel, Kobre & Kim, LLP, Miami, FL, Samuel A. Stern, Stern, LLC, Miami, FL, for Defendant - Appellant
John C. Shipley, Assistant U.S. Attorney, Emily M. Smachetti, U.S. Attorney Service - Southern District of Florida, U.S. Attorney Service - SFL, Miami, FL, Robert Benjamin Cornell, U.S. Attorney's Office, Fort Lauderdale, FL, for Plaintiff - Appellee
John Kirk Ogrosky, Murad Hussain, Sina Mansouri, Arnold & Porter, LLP, Washington, DC, Matthew I. Menchel, Kobre & Kim, LLP, Miami, FL, Samuel A. Stern, Stern, LLC, Miami, FL, for Defendant - Appellant
Before MARTIN, GRANT, and LAGOA, Circuit Judges.
GRANT, Circuit Judge: Salomon Melgen, an ophthalmologist practicing in Palm Beach County, was charged in a 76-count indictment broadly alleging that he operated a multi-year scheme to defraud Medicare. At trial, the government argued that Melgen had systematically diagnosed his patients incorrectly and prescribed medically unnecessary treatments. After nine counts were dismissed for multiplicity, the jury found Melgen guilty on each of the other 67 counts. The district court sentenced Melgen to 204 months of imprisonment (below the Guidelines range of 235–293 months) and also ordered restitution.
Melgen filed a notice of appeal. He also filed a motion for a new trial alleging newly discovered evidence, along with a motion for bond pending appeal. The court denied both of Melgen's motions, and this appeal followed. On appeal, Melgen brings us a laundry list of perceived bases for reversal—including challenges to a jury instruction about materiality, the introduction of summary charts at trial (alongside a host of other miscellaneous evidentiary issues), the sufficiency of the evidence, the court's denial of a new trial, and the reasonableness of his sentence. We affirm the district court's judgment in all respects.
I.
Salomon Melgen operated a high-volume practice as an eye doctor. A significant portion of his practice focused on age-related macular degeneration, or ARMD. There are two varieties of ARMD—"wet" and "dry." Typically, only 10 to 15% of those diagnosed with ARMD have the wet version of the disease. But Melgen's treatment records told a different story. Of the rather remarkable 97.8% of his patients that he diagnosed with ARMD, he also diagnosed 75.5% with wet ARMD in at least one eye. That included almost all of his African-American patients, even though expert testimony at trial indicated that wet ARMD is "nearly exclusively a disease of Caucasians," and thus almost never present in the African-American population.
Dry ARMD is basically untreatable, but wet ARMD may be slowed or stopped by so called anti-VEGF drugs. One recognized anti-VEGF treatment for wet ARMD is a drug called Lucentis. A single vial of Lucentis costs approximately $2,000. Between 2008 and 2013, Melgen's practice collected nearly $57 million from Medicare for administering Lucentis. By contrast, Melgen only rarely prescribed Avastin, another drug recognized as a treatment for wet ARMD that costs only $50.
VEGF stands for "Vascular endothelial growth factor."
Long before this criminal case began, Melgen was involved in litigation regarding his preferred method for prescribing and administering Lucentis—a method known as "multi-dosing." See Vitreo Retinal Consultants of the Palm Beaches, P.A. v. U.S. Dep't of Health & Human Servs. , 649 F. App'x. 684, 687 (11th Cir. 2016). Medicare's reimbursement rate to medical providers for Lucentis is based on the fact that each vial of Lucentis is intended to provide a single dose of solution for a single eye. See id. But Melgen argued that each vial held enough of the drug to safely administer multiple doses to separate patients from one vial—and indeed, this was how he administered the drug. This created a billing-and-compensation issue with Medicare. On the one hand, Melgen could argue that Medicare's total costs were the same regardless of whether he multi-dosed or single-dosed, because Medicare paid Melgen the ordinary per-patient rate for Lucentis either way. On the other hand, Medicare's rate of payment was based in part on the expected cost to the provider, so Melgen received a windfall relative to his expenditures by multi-dosing. And this windfall was substantial—by extracting up to three doses from a single vial, Melgen's practice "was ‘reimbursed’ for approximately $6,075 per single Lucentis vial, three times the average cost of the vial and three times the amount it would have received had it administered the drug according to the label." Id. at 688.
Melgen was instructed to repay Medicare millions of dollars for this practice in June of 2009 after Medicare notified Melgen that multi-dosing misrepresented his expenses and was medically unreasonable. Id. He then filed a suit seeking the return of those funds, arguing that Medicare's interpretation of its regulations was unreasonable. In 2016, we ruled against Melgen and affirmed Medicare's interpretations as not arbitrary or capricious. See id. at 687. As relevant to this appeal, the government suggested at trial that Melgen continued multi-dosing Lucentis until 2013.
Melgen also billed Medicare for numerous applications of focal laser photocoagulation, a procedure to treat wet ARMD where a high-intensity laser light is aimed at the eye. The government presented evidence at trial that the procedure is now almost never medically necessary given the effectiveness of anti-VEGF drugs.
Melgen was charged with Medicare fraud in a 76-count indictment. The charges generally outlined a scheme in which Melgen systematically over-diagnosed wet ARMD. The government alleged that Melgen billed Medicare for treatments to patients that did not need them—whether because they were completely healthy, because they had dry ARMD, or because the particular eye he claimed to treat in his records was either fully blind or a prosthetic. In one case, for example, Melgen billed Medicare 96 times for treatment on a single patient's prosthetic eye. The government also alleged that the scheme involved "pre-filling" some patient files so that ARMD was a default diagnosis even before Melgen met with a patient—including pre-drawing depictions of the patient's retina, even though those drawings purported to depict the condition of the eye as seen on a particular (and necessarily later) day.
Counts 1–46 of the indictment alleged that Melgen knowingly and willfully executed a scheme to defraud Medicare and to obtain, by means of materially false and fraudulent representations, money controlled by Medicare. Counts 47–65 alleged that Melgen knowingly made or caused to be made false and fraudulent Medicare reimbursement claims that were medically unreasonable and unnecessary. Counts 66–76 alleged that Melgen knowingly and willfully made and used materially false documents while knowing that they contained materially false and fraudulent statements and entries in connection with the delivery of and payment for healthcare benefits. These counts covered particular entries in Medicare charts wherein Melgen falsely diagnosed patients with wet ARMD. Melgen's case proceeded to trial. We will briefly recount those parts of the eight-week trial that are most relevant to the issues raised on appeal. As part of its case, the prosecution introduced summary charts of Medicare records under Rule 1006 to demonstrate that Melgen's practices were markedly different from similarly situated physicians. See Fed. R. Evid. 1006. Those records were compiled by drawing out particular doctors’ data from raw Medicare data. In order to make the summaries relevant, the government pulled the data for only those self-identified ophthalmologists who (1) billed Medicare for over 500 injections of Lucentis from 2008–2013, (2) had at least 2,000 Medicare patients during that time, and (3) billed at least one claim each of those years.
Melgen, who had sought to exclude the charts in limine, renewed his objection, arguing that the charts were prejudicial, that they were barred as testimonial hearsay, and that no evidence supported the comparison criteria. For its part, the government argued that it had explained its comparator criteria through the testimony of Dr. Stuart Fine, a retina specialist from Colorado. Dr. Fine endorsed the 500-injection cutoff—which would equal roughly 83 injections per year over a six-year time span—but he rephrased it as "a hundred a year, basically." The government also introduced testimony regarding that criterion from Dr. Julia Haller, an expert ophthalmologist based in Philadelphia. She testified that 500 injections of Lucentis over a six-year period would be a conservative estimate for identifying other retinal specialists. The district court admitted the charts. The witness who had prepared the charts then testified that the requirement that the comparators had treated 2,000 patients per year was based on Melgen's own patient population of slightly more than 2,000 patients during the relevant period, and that the requirement of treating one patient per year during the period ensured that the sample did not include doctors that had not practiced throughout the relevant period.
As part of the evidence indicating that Melgen had falsely treated patients, the government called two witnesses, Delores Griffith and her daughter Susanne Perry, who insisted that Griffith had never received a particular eye surgery from Melgen on a particular date. Melgen correctly countered that records from an anesthetist corroborated that the surgery occurred, and argued that the testimony was undisclosed extrinsic bad act evidence prohibited by Rule 404(b) in any event. In response, the district court issued a curative instruction:
On Thursday of last week, the prosecution introduced testimony from Delores Griffith and Susanne Perry in which both witnesses claimed that a surgical procedure performed by Dr. Melgen on May 21st of 2009, known as a vitrectomy, had not occurred. Evidence in Ms. Griffith's patient file indicates that the procedure was performed on that date, and billing records show that both Dr. Melgen and the separate surgical center had billed Medicare for the procedure. So you, as the jury, are to disregard the witnesses’ testimony about that procedure and you should strike it from your minds and give it no weight. So I ask you to follow that instruction.
Other evidence at trial included patient records taken from a sample of Melgen's bills. The parties contested whether that sample was statistically representative. The district court eventually instructed the jury that the sample was random, although not statistically guaranteed to be representative (later, at sentencing, the district court did conclude that the sample was representative to a 95% confidence interval).
The Eleventh Circuit's pattern jury instruction for Melgen's offenses says that a fact is material "if it has the capacity or natural tendency to influence a person's decision. It doesn't matter whether the decision-maker actually relied on the statement or knew, or should have known, that the statement was false." But before jury deliberations began, Melgen offered a lengthy new instruction based on a recent Supreme Court case addressing civil qui tam actions under the False Claims Act. See Universal Health Services, Inc. v. United States ex rel. Escobar , ––– U.S. ––––, 136 S. Ct. 1989, 2002, 195 L.Ed.2d 348 (2016). The district court denied the requested alteration, noting that no similar case had adopted the proposed instruction and that the proposed instruction lacked a factual basis.
Melgen was convicted on all counts after an eight-week jury trial. The district court denied Melgen's initial motions for judgment of acquittal and for a new trial.
Sentencing was next. Melgen had no prior convictions, and applying § 2B1.1 of the United States Sentencing Guidelines, the presentencing investigation report (PSR) set Melgen's base offense level at 6. It then found that Melgen was responsible for a loss of between $65 and $150 million, leading to a 24-level increase. U.S.S.G. § 2B1.1(b)(1)(M). The PSR also incorporated various adjustments and enhancements, including for the large number of victims and abuse of trust; those raised Melgen's total offense level to 42.
Melgen objected that the PSR applied the wrong loss calculation methodology and that it wrongfully included Medicare funds that he had already repaid to Medicare. The district court denied Melgen's objection to the loss calculation method and found that the starting point for calculating the loss amount was the amount sought in the fraudulent bills Melgen submitted to Medicare, not the allowed amount or amount that Medicare actually paid. The court next found that the government had presented credible evidence establishing a statistically reliable basis for concluding that the government's loss calculation was a reasonable estimate of Melgen's fraudulent billing. Because the record contained no sampling of patients from 2008, 2009, and 2013, the court decided to limit the loss calculation to the 2010 to 2012 time period. The court then found that the total amount Melgen fraudulently billed Medicare during those three years was $73,417,620. The court concluded that the mere fact that Melgen's treatment could possibly have benefitted other undiagnosed conditions his patients may (or may not) have had was insufficient to rebut the loss calculation; those other conditions were not specified in the patient's records and amounted to pure speculation in the court's view.
At the same time, some of the sentencing decisions were in his favor. The court sustained Melgen's objections to the PSR's two-level enhancement for abusing the vulnerable and to its two-level enhancement for conduct involving a conscious or reckless risk of death or serious bodily injury. It therefore concluded that Melgen had an offense level of 38 and was in criminal history category I, with a Guidelines range of 235–293 months.
The district court varied downward and imposed a 204-month sentence. It eventually ordered $52,997,442 in restitution. The court also later denied a motion for bond pending appeal and for a new trial due to alleged newly discovered Brady evidence. Melgen now appeals.
II.
Melgen's first argument on appeal is that the district court erred by giving the Eleventh Circuit's pattern jury instruction on materiality. We review de novo the legal correctness of a jury instruction, but review "questions of phrasing" and the denial of a requested instruction for an abuse of discretion. United States v. Dulcio , 441 F.3d 1269, 1275 (11th Cir. 2006) ; United States v. Prather , 205 F.3d 1265, 1270 (11th Cir. 2000) (citations omitted).
Our pattern jury instruction is based on the rule that a "false statement is material if it has a natural tendency to influence," or is "capable of influencing," the "decision of the decisionmaking body to which it was addressed." United States v. Henderson , 893 F.3d 1338, 1346 (11th Cir. 2018) (internal quotation marks omitted) (quoting Kungys v. United States , 485 U.S. 759, 770, 108 S.Ct. 1537, 99 L.Ed.2d 839 (1988) ). Melgen argues that the district court should have instead included his proposed language from Escobar —that materiality "looks to the effect on the likely or actual behavior of the recipient of the alleged misrepresentation." Escobar , 136 S. Ct. at 2002 (alteration adopted and citation omitted).
Escobar , however, does not lend him any aid. First, some background on that case. Under the False Claims Act the government (and often an employee or other knowledgeable person who reveals misdeeds) can recover for fraudulent claims against the government. One theory under which these suits sometimes proceed is known as "implied false certification." Escobar addressed the ins and outs of that theory, which is grounded in the idea that, "when a defendant submits a claim, it impliedly certifies compliance with all conditions of payment." Id. at 1995. "But if that claim fails to disclose the defendant's violation of a material statutory, regulatory, or contractual requirement, so the theory goes, the defendant has made a misrepresentation that renders the claim ‘false or fraudulent’ "—thereby triggering liability under the Act. Id. In that context, the Supreme Court noted that materiality looks to "the effect on the likely or actual behavior of the recipient of the alleged misrepresentation." Id. at 2002. That is the statement on which Melgen relies.
But it does not help him. To begin, we are not at all sure that Escobar didn't approve of the objective standard that our current materiality standard is based on. "Capable of influencing" is not so very different than looking to the "effect on the likely or actual behavior" of the actor. Moreover, the Escobar standard for materiality is not made out of whole cloth. Following the statement that Melgen relies on, the Supreme Court tied the concept to our understanding of materiality in tort and contract. See id. at 2002–03. As part of that discussion, it explicitly referenced the—objective—"reasonable man" standards in both tort and contract. Id. The Court explained, for instance, that in tort the materiality of a statement may be shown where "a reasonable man would attach importance to [it] in determining his choice of action" and that materiality "in contract law is substantially similar." Id.
Even if that were not sufficient—which we doubt— Escobar was addressing quite a different question than the one we face here. One of the key issues in that case was whether a misrepresentation or omission that technically qualified as a "condition of payment" would lead to liability even where the government would never actually refuse to pay on that basis. The answer was no, and the discussion of materiality was geared toward addressing that issue.
Which is, of course, a far cry from the criminal fraud statutes that Melgen faces. We are not the first to notice this distinction. In two cases since Escobar , the Fourth Circuit has examined whether the precise statement from Escobar that Melgen latches onto actually alters the long-standing objective materiality standard in criminal fraud cases. See United States v. Raza , 876 F.3d 604, 619–20 (4th Cir. 2017) ; United States v. Palin , 874 F.3d 418, 423 (4th Cir. 2017). According to that court? Doubtful. Like the Fourth Circuit, we think it unlikely that "the Court's examination of how materiality applies under ‘implied false certification’ FCA cases transfers to all cases charging fraud, or even all cases charging health care fraud." Palin , 874 F.3d at 423. And we have continued to rely on the standard articulated in our pattern jury instruction for materiality in criminal cases post- Escobar without requiring an alteration. See Henderson , 893 F.3d at 1346.
Moreover, Melgen cannot show a factual basis for his requested instruction even under his proposed interpretation of Escobar ’s standard. Melgen billed for certain services, and Medicare paid for certain services in reliance on those bills. Melgen did not put forward evidence that Medicare routinely pays for treatment based on an incorrect diagnosis, if only it is possible that the patient had some other condition that the treatment would have aided—far from it. And that is enough to conclude that Escobar does not help him deal with the materiality, or lack thereof, of his false statements. For these reasons, we conclude both that the district court did not err in refusing to give Melgen's proposed instruction and that any alleged error would have been harmless.
III.
Melgen next argues that the district court erred by allowing the introduction of summary charts comparing Melgen's billing to peer physicians. He sets out several potential rationales: the charts were not covered by Rule 1006, the evidence was testimonial hearsay in violation of the Confrontation Clause (or at least otherwise inadmissible hearsay), and the charts should have required expert evaluation under Rule 702. Evidentiary rulings are reviewed for both abuse of discretion and harmless error. See United States v. Maxwell , 579 F.3d 1282, 1295–96, 1298 (11th Cir. 2009). We review the admission of evidence for plain error where a defendant failed to make a particular objection at trial. United States v. Hoffman-Vaile , 568 F.3d 1335, 1341 (11th Cir. 2009). Confrontation Clause rulings are reviewed de novo and subject to constitutional harmless error analysis. United States v. Curbelo , 726 F.3d 1260, 1271–72 (11th Cir. 2013) ; United States v. Jones , 601 F.3d 1247, 1264 (11th Cir. 2010).
Where, as here, the underlying evidence is made up of voluminous Medicare claims, a district court has good reason to apply Rule 1006 to allow a summary chart. "Summary charts are permitted generally by Federal Rule of Evidence 1006 and the decision whether to use them lies within the district court's discretion." United States v. Richardson , 233 F.3d 1285, 1293 (11th Cir. 2000). Under that rule, "the essential requirement is not that the charts be free from reliance on any assumptions, but rather that these assumptions be supported by evidence in the record." Id. at 1294 (quoting United States v. Diez , 515 F.2d 892, 905 (5th Cir. 1975) ). Here, the 500-injections-over-six-years criterion was supported by the opinion of Dr. Haller (whom, we note, Melgen was able to cross-examine). The 2,000-patient cutoff reflected Melgen's own patient load. And the one-patient-each-year criterion matched Melgen's own consistent practice during the relevant period. We therefore find no abuse of discretion in the district court's decision to admit the charts under Rule 1006. Permitting the introduction of the underlying data under the business records exception to hearsay was also well within the district court's discretion.
The Confrontation Clause did not apply to Melgen's ability to cross-examine decisionmakers regarding the criteria that were used to make the summary charts. To begin with, the summaries were drawn from non-testimonial Medicare records that do not implicate the Confrontation Clause under Crawford v. Washington , 541 U.S. 36, 51, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004). Melgen, however, argues that the mere act of choosing selection criteria to decide which doctors’ data to include in the summary comparisons was a testimonial act. He argues that he has the right to cross-examine whoever selected the criteria. Here, that would be the members of the prosecution team that directed the creation of the exhibit.
For this reason, we also reject Melgen's general hearsay argument raised in his initial brief—one which, as the government points out, was never made at trial, and so is reviewed for plain error. Melgen is correct that "Rule 1006 is not a back-door vehicle for the introduction of evidence which is otherwise inadmissible." Peat, Inc. v. Vanguard Research, Inc. , 378 F.3d 1154, 1160 (11th Cir. 2004). But here the underlying Medicare data is admissible—so there is no concern that otherwise inadmissible testimony was snuck in as a summary.
That approach has no basis in our law. Attorneys routinely make decisions about which evidence they believe is relevant to establishing a particular point—decisions that may include, for example, which witnesses to call, or as here, which summaries to enter into evidence. The Confrontation Clause "guarantees a criminal defendant ‘the right ... to be confronted with the witnesses against him.’ " Al-Amin v. Warden , 932 F.3d 1291, 1302 (11th Cir. 2019) (quoting U.S. Const. amend. VI ). It does not reach back a step further to demand the opportunity to cross-examine an attorney over why they decided to call a particular witness—or, as in this case, about why they chose specific selection criteria in compiling the summary.
We note that Melgen did cross-examine the FBI analyst who presented the evidence, and questioned whether any errors might have been made in applying the chosen comparison criteria. The district court's application of Rule 1006 therefore introduced no Confrontation Clause issue.
Melgen also argues that a sufficiently qualified expert under Rule 702 should have been required as a basis for entering the comparison charts. His argument on this point relies mainly on an unpublished district court decision from the Northern District of Illinois that declined to recognize two statisticians as experts because they were not experts in so-called medical statistics. See United States v. Chhibber , 741 F.3d 852, 854 (7th Cir. 2014) (describing the procedural history in the trial court but affirming the defendant's conviction without addressing this issue). Whatever persuasive value that decision does or does not hold, we do not see the present case—where the district court had the testimony of medical experts on the rate of Lucentis injections that would indicate retinal practice specialty—as analogous. And as previously mentioned, the other comparison criteria were reasonably supported by the evidence. We therefore affirm the district court's admission of the summary charts.
IV.
We next address several of Melgen's miscellaneous arguments concerning his trial. First , Melgen claims that the district court should have excluded any evidence of Melgen's multi-dosing of Lucentis because the prosecution had ample other means for establishing Melgen's profit motive. But we have already explained that through multi-dosing, Melgen could get reimbursed by Medicare at a rate of incredible profit, receiving up to three times the cost of obtaining the drug. That fact was probative of his profit motive for falsely diagnosing patients with wet ARMD and then prescribing Lucentis—whether or not his creativity in administering multiple doses of Lucentis from one vial was lawful. The district court did not err in admitting the evidence of multi-dosing.
Second , Melgen asks that we reverse the district court's judgment because testimony from Delores Griffith (and her daughter) that Melgen never performed a particular surgery was later shown to be false. The district court denied Melgen's motion for a mistrial on that basis.
Denial of a defendant's request for a mistrial or new trial is reviewed for abuse of discretion. United States v. Vallejo , 297 F.3d 1154, 1163 (11th Cir. 2002). If the district court issued a curative instruction, this Court will reverse only if "the evidence is so highly prejudicial as to be incurable." United States v. Dodd , 111 F.3d 867, 870 (11th Cir. 1997). At the time that the court became aware that the witness had testified incorrectly, the district court immediately issued a thorough curative instruction. The objected-to witnesses’ testimony was not incurable—in fact, the district court ably explained to the jury that the witnesses had remembered falsely. Reversal is plainly not warranted.
Third , Melgen argues that the district court erred by declining to instruct the jury that the sample of patient files entered into evidence was not statistically random. As with the testimony from Delores Griffith, however, the court issued a curative instruction that avoided any prejudice on this point, telling the jury to disregard any statements concerning statistical confidence in the representative nature of the sample. That instruction was sufficient to avoid a mistrial.
Fourth , Melgen argues that the district court erred by providing the jury with twelve unredacted copies of the indictment—in particular, he argues that the court violated Federal Rule of Criminal Procedure 30(b). But that Rule only requires the court to notify the parties before closing arguments of "how it intends to rule on the requested instructions." Fed. R. Crim. P. 30(b). Allowing the jury to receive twelve copies (rather than one) does not concern any ruling on the requested jury instructions—and thus does not implicate Rule 30(b).
Melgen does cite cases in which courts have warned that providing a copy of the indictment outlining the prosecution's theory of the case may be unfairly prejudicial. But those cases are a far cry from this one. In one example, the judge had not given the jury any "neutral written jury instructions to guide them in their deliberations"—but did give them the indictment. United States v. Van Dyke , 14 F.3d 415, 423 (8th Cir. 1994). The potential for prejudice was obvious there; nothing comparable existed here. Moreover, because Melgen did not object when he was informed that twelve copies of the indictment were sent to the jury—instead saying "that's fine"—any challenge to the number of copies given to the jury would be reviewed for plain error, if at all. And Melgen identifies no authority for concluding that the district court plainly erred by providing multiple copies of the indictment. Nor can he establish any effect on his substantial rights, as required to meet the standard for plain error, because the court repeatedly told the jury that the indictment was not evidence of his guilt. Fifth , Melgen argues that the district court erred by not asking more questions or granting relief after learning of contact between a government witness and two defense witnesses. Melgen asserts that the government witness "intimidated" the defense witnesses. A district court's decisions about whether to grant a new trial and whether to conduct an evidentiary hearing regarding alleged witness intimidation are reviewed for an abuse of discretion. See United States v. Perez-Oliveros , 479 F.3d 779, 782 (11th Cir. 2007) ("We review the denial of a motion for a new trial for abuse of discretion."); United States v. Arbolaez , 450 F.3d 1283, 1293 (11th Cir. 2006) ("Generally, a court's decision about whether to hold an evidentiary hearing lies within that court's sound discretion and will be reviewed only for an abuse of discretion."). After the district court learned of the contact, the court conducted a brief hearing before deciding that the contacts had not been prejudicial in the end because no testimony had been altered. Melgen offers no explanation for how either witness's testimony would have been different, beyond the possibility that one witness's demeanor slightly changed after the contact. Cf. United States v. Thompson , 130 F.3d 676, 687 (5th Cir. 1997) ("The defendant bears the burden of showing that testimony would have been different."). The court did not abuse its discretion by concluding that no mistrial was required.
V.
Melgen also briefly challenges the sufficiency of the evidence supporting his convictions. "A jury's verdict cannot be overturned if any reasonable construction of the evidence would have allowed the jury to find the defendant guilty." United States v. Capers , 708 F.3d 1286, 1297 (11th Cir. 2013) (quoting United States v. Herrera , 931 F.2d 761, 762 (11th Cir. 1991) ). On appeal, Melgen echoes the same argument he made to the jury—that the evidence supported a finding that any "mistakes" in diagnosing patients were not willfully false and that he reasonably believed the treatments were required. But the "evidence need not be inconsistent with every reasonable hypothesis except guilt, and the jury is free to choose between or among the reasonable conclusions to be drawn from the evidence presented at trial." Id. (quoting United States v. Poole , 878 F.2d 1389, 1391 (11th Cir. 1989) ).
That rule guides our conclusion that Melgen's sufficiency of the evidence challenge fails. Melgen's arguments go to the weight of the evidence, and not its sufficiency. The jury heard hours of evidence about Melgen's motive and means for fraudulently billing Medicare for an expensive drug. It's true that Melgen offered alternative explanations for some of the evidence against him; we are thinking, for example, of the claim that pre-prepping diagnoses before Melgen saw patients might have led to honest mistakes. But those explanations were for the jury to weigh, not us. After all, the jury also could have seen pre-prepping as a sign of Melgen's willful choice to treat the vast majority of his patients for ARMD, whether or not it was medically necessary. The scope of the scheme was easily enough for the jury to conclude that Melgen had engaged in systematic fraud, rather than committing isolated mistakes. We find the evidence sufficient to uphold the jury's verdict.
And because the district court did not commit errors in its evaluation of the evidence, those non-errors cannot lead to the application of the cumulative error doctrine. We affirm Melgen's conviction on all counts. VI.
Before addressing Melgen's sentence, we turn to his requests for a new trial under Brady and Federal Rule of Criminal Procedure 33. First, the Brady claim. Alleged Brady violations are reviewed de novo; it is the defendant's burden to show all the elements of a violation. United States v. Jones , 601 F.3d 1247, 1266 (11th Cir. 2010). To succeed on his Brady argument, Melgen must show that (1) the government possessed evidence favorable to him; (2) he did not possess the evidence and could not obtain the evidence with any reasonable diligence; (3) the prosecution suppressed the favorable evidence; and (4) had the evidence been disclosed to him, there is a reasonable probability that the outcome would have been different. Vallejo , 297 F.3d at 1164.
In denying Melgen's motion for a new trial on this basis, the district court explained that the "information upon which Defendant relies was not newly discovered, and assuming it was, it probably would not have changed the outcome of the trial." We agree. The main piece of purported Brady evidence was a statement that a witness for the government, Dr. Berger, made during sentencing. The content of the statement was the medical fact that the use of an anti-VEGF drug could control leakage of wet ARMD, hiding signs of the disease. That is not the kind of evidence that could not have been obtained before trial—it is medical testimony that Melgen could have introduced himself with a variety of medical witnesses. Perhaps recognizing that fact, Melgen claims that the new evidence is not the contention itself but that one of the government's witnesses agreed with the contention. But this inappropriately focuses on the who , rather than the what , of the testimony—which does not satisfy Brady ’s requirement that the evidence be unobtainable without reasonable diligence. Melgen could have asked any of the government's witnesses if they agreed with that point during trial, or called his own witnesses. Really, the mere fact that a particular government witness agreed with a fact that Melgen finds useful to his defense is not "new evidence." Nor is it likely, in light of all the evidence, that Berger's opinion would have changed the outcome at trial.
Melgen possessed another piece of purported Brady evidence, an internal e-mail from within CMS, before trial began—which automatically defeats his Brady claim. See Felker v. Thomas , 52 F.3d 907, 910 (11th Cir. 1995) (no Brady violation where the defendant knows or should have known of the allegedly exculpatory information before trial).
Melgen also asked for a new trial under Rule 33. But relief under that Rule cannot be predicated on supposed new evidence that merely is impeaching. United States v. DiBernardo , 880 F.2d 1216, 1224 (11th Cir. 1989). The district court did not err in denying Melgen's motion for a new trial under Rule 33 because the "new" evidence was merely evidence that might, at best, have been used as weak impeachment evidence against the government's witnesses.
VII.
Melgen next appeals his sentence as procedurally and substantively unreasonable. We review the reasonableness of a sentence under an abuse-of-discretion standard. United States v. Irey , 612 F.3d 1160, 1188–89 (11th Cir. 2010) (en banc). In reviewing a sentence for reasonableness, we first consider whether the district court committed any significant procedural error, and next consider whether the sentence was substantively reasonable. Gall v. United States , 552 U.S. 38, 51, 128 S.Ct. 586, 169 L.Ed.2d 445 (2007). The party challenging the sentence bears the burden to show that the sentence is unreasonable considering the record and the § 3553(a) factors. United States v. Tome , 611 F.3d 1371, 1378 (11th Cir. 2010). Those factors include the need to reflect the seriousness of the offense, promote respect for the law, provide just punishment for the offense, deter criminal conduct, and protect the public from the defendant's future criminal conduct. See 18 U.S.C. § 3553(a)(2). After evaluating for reasonableness, we will only vacate a defendant's sentence if left with the firm conviction that the district court committed clear error in weighing the § 3553(a) factors and imposing a sentence outside the range of reasonable sentences based on the facts. Irey , 612 F.3d at 1190.
Here, the district court did not commit procedural error in determining Melgen's sentence. Melgen's procedural challenge hinges on several determinations that went into the district court's loss calculation. We review a district court's determination of the loss amount under the Sentencing Guidelines for clear error. United States v. Baldwin , 774 F.3d 711, 727 (11th Cir. 2014). We will conclude that a finding of fact is clearly erroneous only we are "left with a definite and firm conviction that a mistake has been committed." United States v. Pierre , 825 F.3d 1183, 1191 (11th Cir. 2016) (quoting United States v. Rothenberg , 610 F.3d 621, 624 (11th Cir. 2010) ). Because the district court is in a unique position to assess the evidence and estimate the loss based upon that evidence, its loss determination is entitled to appropriate deference. U.S.S.G. § 2B1.1 cmt. n.3(C). The Sentencing Guidelines define "loss" as the greater of "actual" loss or "intended" loss. Id. § 2B1.1 cmt. n.3(A). "Actual loss" is defined as the reasonably foreseeable pecuniary harm that resulted from the offense, and "intended loss" is defined as the pecuniary harm that was intended to result from the offense, including pecuniary harm that would have been impossible or unlikely to occur. Id. Losses that insurance companies and patients sustain that result from Medicare fraud are "relevant conduct that may be considered by the district court when calculating the amount of loss." Hoffman-Vaile , 568 F.3d at 1344.
Here, the district court did not clearly err in calculating the loss amount. The government presented enough evidence that the sample patient group was sufficiently representative of Melgen's patient population between 2010 and 2012 for the district court to make a reasonable estimate based on that sample. And those years are only a portion of the timeframe covering Melgen's scheme—suggesting that he surely obtained even more Medicare funds than the district court accounted for. Melgen's expert witness conceded that, while the government was missing the "seed" number that was used by a computer program to generate that sample, there was no reason to question the functionality of that program. The district court was also entitled to consider both the 80% of treatment cost that Medicare ordinarily pays as well as the 20% that has to be covered by other payment sources. Id. at 1343–44.
Finally, the district court did not clearly err in reaching its final loss determination, despite Melgen's argument that some patients in fact had diseases that required treatments for which Medicare would have offered some degree of reimbursement. Again, the district court need only reach a reasonable estimate of loss. Under the commentary to the Guidelines, when a defendant is convicted of a federal healthcare offense involving a government healthcare program, the aggregate dollar amount of fraudulent bills constitutes prima facie evidence of the amount of the intended loss, if not rebutted. U.S.S.G. § 2B1.1 cmt. n.3(F)(viii). The district court did not err in concluding that Melgen had failed to rebut the prima facie evidence of loss. As the court stated, the "mere fact that Defendant's treatment could have possibly benefitted another condition his patients may have had, but which he did not indicate he was treating, is insufficient to rebut the government's proof. It is pure speculation that any of Defendant's patients who were improperly diagnosed and treated for conditions that they did not have actually benefitted from the treatments they did receive." Cf. United States v. Dehaan , 896 F.3d 798, 808 (7th Cir. 2018) (treating amounts paid by Medicare to a fraudulent provider as a loss even though some of the physician's patients may have qualified for treatment).
For similar reasons, we reject Melgen's brief suggestion that the district court clearly erred by declining to consider amounts that he may have repaid during the civil multi-dosing litigation.
Nor was Melgen's (below-Guidelines) sentence substantively unreasonable. In considering the substantive reasonableness of a sentence, we consider the totality of the circumstances and whether the sentence achieves the sentencing purposes stated in § 3553(a) and described above. United States v. Sarras , 575 F.3d 1191, 1219 (11th Cir. 2009). "A district court's sentence need not be the most appropriate one, it need only be a reasonable one." Irey , 612 F.3d at 1191. Here, the court expressly considered Melgen's "age," his "lack of criminal history," and his "medical conditions" before concluding that Melgen deserved a downward variance to 204 months but no further. In light of the extent of Melgen's fraudulent scheme, the sentence the district court imposed was more than reasonable.