Nos. 25-1056/1099/1104/1157/1353

Appeal from the United States District Court for the Eastern District of Michigan at Detroit. No. 2:20-cr-20162-2 — Linda V. Parker, District Judge.

Argued: March 19, 2026

Decided and Filed: June 12, 2026

Before: MOORE, THAPAR, and MATHIS, Circuit Judges.

ARGUED: Andrew Goetz, MILLER JOHNSON, Detroit, Michigan, for Appellant Raef Hamaed. Mark K. Kriger, LARENE & KRIGER, P.L.C., Detroit Michigan, for Appellant Tarek Fakhuri. James R. Gerometta, LAW OFFICE OF JAMES R. GEROMETTA, Lake Orion, Michigan, for Appellant Ali Abelrazzaq. Kassem M. Dakhlallah, HAMMOUD, DAKHLALLAH & ASSOCIATES, PLLC, Dearborn, Michigan, for Appellant Kindy Ghussin. Tory D. Roberts, UNITED STATES DEPARTMENT OF JUSTICE, Washington, D.C., for Appellee. ON BRIEF: Andrew Goetz, MILLER JOHNSON, Detroit, Michigan, for Appellant Raef Hamaed. Mark K. Kriger, N. C. Deday LaRene, LARENE & KRIGER, P.L.C., Detroit Michigan, for Appellant Tarek Fakhuri. James R. Gerometta, LAW OFFICE OF JAMES R. GEROMETTA, Lake Orion, Michigan, for Appellant Ali Abelrazzaq. Kassem M. Dakhlallah, HAMMOUD, DAKHLALLAH & ASSOCIATES, PLLC, Dearborn, Michigan, for Appellant Kindy Ghussin. Tory D. Roberts, Jeremy R. Sanders, UNITED STATES DEPARTMENT OF JUSTICE, Washington, D.C., for Appellee.

Page 2MATHIS, Circuit Judge. Six pharmacists operating five pharmacies engaged in a yearslong scheme to bill their patients’ insurers for prescriptions they did not dispense. After a grand jury charged them with numerous offenses, including conspiracy to commit healthcare fraud and wire fraud, Raef Hamaed, Kindy Ghussin, Ali Abdelrazzaq, and Tarek Fakhuri exercised their right to a jury trial. A jury convicted each of them on one count of conspiracy to commit healthcare fraud and wire fraud and convicted Fakhuri on one count of healthcare fraud. Defendants appeal their convictions. Hamaed, Ghussin, and Abdelrazzaq also challenge aspects of their respective sentences. Because Defendants have not shown a reversible error, we affirm.

Hassan Abdallah and Raef Hamaed opened a series of independent pharmacies in Michigan and Ohio. They recruited friends and former employees — Tarek Fakhuri, Kindy Ghussin, Balhar Singh, Ali Abdelrazzaq, and Hassan Khreizat — to help at each one. At the height of their business relationship, they co-owned five pharmacies across the two states: Harper Drugs; Wayne Campus Pharmacy, LLC (“Wayne Campus”); Heartland Pharmacy, LLC ('Heartland 1'); Heartland Pharmacy 2 ('Heartland 2'); and Eastside Pharmacy ('Eastside'). Fraudulent billing practices took place at all five pharmacies.

Prescription billing is relatively straightforward. Before a pharmacist fills a prescription, he must first communicate with the patient's insurer. Pharmacists usually do not contact the insurance company directly; instead, they communicate through intermediaries called pharmacy benefit managers ('PBMs'). PBMs help manage prescription drug benefits for Medicare, Medicaid, and private insurance companies like Blue Cross Blue Shield of Michigan.

Once a PBM receives a patient's prescription information, it verifies the patient's eligibility, approves the claim, and authorizes payment to the pharmacy. This transaction is Page 3instantaneous, but the payment takes a few weeks to process. If the PBM approves the claim, the pharmacy then prints the prescription's label, puts the medication in a container, affixes the label, and notifies the patient that it is ready for pickup. Prescription refills work the same way.

When a patient picks up her medicine, the pharmacy is usually required to collect her signature. The insurance company may also require the pharmacy to collect a copayment from the patient. Sometimes the pharmacy can waive a copay. So long as the insurer permits these waivers — Medicare does not — the pharmacy may implement them.

But what happens when a patient does not pick up her duly filled prescription? For most pharmacists, this creates more work — they must go back into the electronic system, reverse the PBM claim, and cancel payment. Where others see extra labor, these pharmacists saw the potential for a payout.

Rather than reversing claims for prescriptions they did not dispense, these coconspirators pocketed the profit. They targeted patients who were noncompliant with their medications. Each pharmacy had a place to keep the labels for prescriptions that would never be dispensed. The pharmacists told their staff not to process reversals for those prescriptions.

The pharmacists also manufactured ways to increase the volume of their business and thereby increase their chances of capitalizing on forgetful or neglectful patients. They waived copays on medications that were more expensive like insulin and brand-name drugs. And at Heartland 2, the pharmacists billed insurance for brand-name drugs that they secretly replaced with the generic form.

The coconspirators made millions in profits. They each received thousands of dollars in monthly profit-sharing checks. Dividends this large were possible only because of their fraudulent scheme. Abdallah and Hamaed joked that no “legal” business made as much money as they did. R. 263, PageID 3123.

These high rewards came at great risk. The PBMs audited in-network pharmacies. Sometimes the PBMs collected patient signatures to ensure that the patients picked up their paid prescriptions. To conceal their scheme from these audits, the coconspirators often forged patient Page 4signatures. Other times the PBMs collected the pharmacy’s wholesaler invoices to ensure that the pharmacy purchased the medication that it billed to patients' insurers. Large shortages revealed that the pharmacy billed for more prescriptions than they filled.

Although the coconspirators were able to manipulate their records for the PBM audits, they could not evade detection forever. Qlarant eventually caught on.

Qlarant contracts with the federal government to ensure the fiscal and clinical integrity of healthcare systems. The Centers for Medicare and Medicaid Services (“CMS”) awarded Qlarant the Investigations Medicare Drug Integrity Contract. Under this contract, Qlarant is tasked with detecting and investigating allegations of Medicare fraud.

Like the PBMs, Qlarant reviews a pharmacy’s wholesaler invoices and compares them to the pharmacy's Medicare billing records. After reviewing the invoices for each of the five pharmacies, Qlarant determined that all five had fraudulently billed Medicare and Medicaid for more medication than they ordered and dispensed. Qlarant calculated an approximate $13,196,752 loss to the government.

A grand jury indicted Abdallah, Hamaed, Fakhuri, Ghussin, Singh, and Abdelrazzaq for one count of conspiracy to commit healthcare and wire fraud, in violation of 18 U.S.C. § 1349. It also charged Abdallah, Fakhuri, and Abdelrazzaq with substantive counts of healthcare fraud, in violation of 18 U.S.C. §§ 1347 and 2, based on specific prescriptions they billed to Medicare or Medicaid but never dispensed.

Abdallah and Singh pleaded guilty to the conspiracy charge, but Hamaed, Fakhuri, Ghussin, and Abdelrazzaq went to trial. At trial, the government called an expert witness to testify about large discrepancies between the pharmacies' wholesaler invoices and their prescription-claims data. That witness — Johanna Sullivan — worked for Qlarant. Relevant here, Sullivan reviewed the pharmacies' data and determined that all five pharmacies fraudulently billed Medicare and Medicaid for more medication than they ordered and dispensed.

Page 5After a fifteen-day trial, the jury convicted Defendants of the conspiracy count. It also convicted Fakhuri of one substantive count of healthcare fraud and Abdelrazzaq of two. Abdelrazzaq moved for judgment of acquittal, and the district court granted it as to both substantive offenses.

The district court sentenced Defendants and ordered them to pay restitution. The district court sentenced Hamaed to 120 months' imprisonment, Fakhuri to 84 months' imprisonment, Ghussin to 65 months’ imprisonment, and Abdelrazzaq to 24 months’ imprisonment. It ordered joint-and-several restitution, with Hamaed liable for $12,803,048.97, Fakhuri liable for $6,540,225.51, Ghussin liable for $3,356,299.65, and Abdelrazzaq liable for $607,648.34.

These appeals followed.

Defendants challenge their convictions on various constitutional, evidentiary, and procedural grounds. Hamaed, Abdelrazzaq, and Ghussin also contend that their sentences are unreasonable. We start with the convictions and then turn to the sentences.

Defendants first argue that Johanna Sullivan's testimony about Qlarant's investigation violated the Confrontation Clause. They assert that her expert opinion incorporated the opinions of nontestifying Qlarant analysts who verified that the data Sullivan relied on was complete. According to Defendants, Sullivan’s testimony violated their confrontation rights because they could not cross-examine those out-ofcourt analysts about the data’s integrity. We disagree.

The parties dispute the standard of review applicable to Defendants' confrontation claims. We generally review alleged Confrontation Clause violations de novo. United States v. Taylor , 127 F.4th 1008, 1013 (6th Cir. 2025). But the government argues that Defendants forfeited their confrontation challenge because they failed to timely object to Sullivan’s testimony on constitutional grounds. We review an unpreserved confrontation Page 6challenge for plain error. United States v. Burrell , 114 F.4th 537, 554 (6th Cir. 2024). We need not resolve this dispute because Defendants’ arguments fail under either standard.

The Sixth Amendment’s Confrontation Clause guarantees criminal defendants the right 'to be confronted with the witnesses against him.' U.S. Const. amend. VI. It requires the prosecution to present the testimony of its witnesses at trial. Melendez-Diaz v. Massachusetts , 557 U.S. 305, 324 (2009). The only exceptions to the right of confrontation are those “established at the time of the founding.” Crawford v. Washington , 541 U.S . 36, 54 (2004). To that end, the Clause bars the admission of certain statements from an absent witness “unless the witness is unavailable and the defendant had a prior chance to subject her to crossexamination.” Smith v. Arizona , 602 U.S. 779, 784 (2024).

The Supreme Court has described the opportunity for a defendant to cross-examine the prosecution’s witnesses as “the main and essential purpose of confrontation.” Delaware v. Van Arsdall , 475 U.S. 673, 678 (1986) (citation modified). The Confrontation Clause thus enshrines the Framers' 'judgment' that nothing 'beats and bolts out the truth much better' than cross -examination. Crawford , 541 U.S . at 61 – 62 (citation modified). It allows a defendant to reveal a witness’s “possible biases, prejudices, or ulterior motives.” Davis v. Alaska , 415 U.S. 308, 316 (1974). And crossexamination “is an essential and fundamental requirement for the kind of fair trial which is this country’s constitutional goal.” Pointer v. Texas , 380 U.S. 400, 405 (1965).

But not all out-of-court statements trigger our confrontation alarm bells. Crawford , 541 U.S. at 51. The Clause prohibits only 'testimonial hearsay.' Davis v. Washington , 547 U.S. 813, 823 –24 (2006). That “two -word phrase” contains “two limits.” Smith , 602 U.S. at 784.

First, the statement must be testimonial. Crawford , 541 U.S. at 51. A statement is testimonial if the “primary purpose of the conversation was to create an out -of-court substitute for trial testimony.” Ohio v. Clark , 576 U.S. 237, 245 (2015) (citation modified). So we look to 'the primary purpose of the statement' and 'how it relates to a future criminal proceeding.' Smith , 602 U.S. at 800 (citation modified). In doing so, we consider 'all of the relevant Page 7circumstances” surrounding the statement. Michigan v. Bryant , 562 U.S. 344, 369 (2011). The Court has described the “core class of testimonial statements” as follows:

Ex parte in-court testimony or its functional equivalent — that is, material such as affidavits, custodial examinations, prior testimony that the defendant was unable to cross-examine, or similar pretrial statements that declarants would reasonably expect to be used prosecutorially; extrajudicial statements contained in formalized testimonial materials, such as affidavits, depositions, prior testimony, or confessions; statements that were made under circumstances which would lead an objective witness reasonably to believe that the statement would be available for use at a later trial.

Melendez-Diaz , 557 U.S. at 310 (citation modified).

Second, the statement must be hearsay. Smith , 602 U.S. at 785. An out-of-court statement is hearsay when it is offered “to prove the truth of the matter asserted.” Id. (quoting Anderson v. United States , 417 U.S. 211, 219 (1974)); Fed. R. Evid. 801(c).

The Confrontation Clause trumps the rules of evidence. Crawford , 541 U.S. at 61. This means that hearsay statements admissible under the rules of evidence will survive a confrontation challenge only if they are not testimonial. Melendez-Diaz , 557 U.S. at 324. In that sense, a witness’s status as an expert witness does not permit her to convey testimonial hearsay under the guise of providing opinion testimony. Bullcoming v. New Mexico , 564 U.S. 647, 661 – 63 (2011); see Fed. R. Evid. 703. The Confrontation Clause grants the defendant the right to cross-examine the person making the statement, not just the expert witness who repeats it. Smith , 602 U.S. at 803.

The Supreme Court has remained steadfast in its application of this constitutional mandate. It has held that the government cannot substitute an expert's live testimony with a document memorializing her opinion. Melendez-Diaz , 557 U.S. at 310 – 11. And the government cannot call “any old analyst” to testify to that opinion; it must call the one who did the work necessary to form the opinion, “not a surrogate merely reading from her records.” Smith , 602 U.S. at 798, 800; accord Bullcoming , 564 U.S. at 663.

Page 8Defendants argue that the government violated their confrontation rights by using Sullivan as one of these impermissible surrogates. According to them, Sullivan conveyed other analysts’ opinions to the jury.

Sullivan served as Qlarant's director of clinical operations and the lead pharmacy specialist on its CMS contract. In that role, she managed a team of pharmacists and subjectmatter experts who worked to detect fraud in the Medicare Part D program.

Sullivan investigated allegations of fraud at Defendants' pharmacies by conducting an invoice review. The invoice review involved comparing Defendants’ prescription -claims data against their wholesaler records. The prescription-claims data shows the prescriptions billed to the insurers, while the wholesaler records show the medications ordered to fill those prescriptions. The data itself is processed by a Statistical Analysis System (SAS) software program. The subject-matter experts input the data into the SAS program, which calculated any shortages between the medications ordered and the prescriptions billed.

Sullivan reviewed the data for each of Defendants’ pharmacies and determined that none of them purchased enough medications to justify their claims. In other words, they billed for more prescriptions than they filled and pocketed the insurers' payments for prescriptions they never purchased.

But Sullivan was not the first Qlarant witness the government disclosed — she was the fifth. And she “did not extract the data” used for the invoice review; the “people on her team” did that. R. 249, PageID 1969. Sullivan testified that she 'verified the time frames' were correct, but she “didn’t go back through the data because it wasn’t needed at [the time she got involved]; it had already been verified.” Id. at 2025. According to Defendants, this created a confrontation issue because nontestifying witnesses performed some of the work necessary for Sullivan to form a conclusion.

Indeed, law enforcement and Qlarant's subject -matter experts usually do most of the clerical work required for Qlarant’s invoice reviews. They identify and collect the relevant data.

Page 9The subject-matter experts also validate the integrity of that data to make sure it is accurate and complete. Sullivan is familiar with Qlarant’s data -collection process — she is usually involved in it —and she testified about her team’s general operating procedures. But she was not involved in the data-collection process for this case because the records were requested and obtained years before she got involved.

Importantly, Defendants do not challenge the admission of the data underlying Sullivan’s conclusion or even the invoice review itself — at trial, they stipulated to the admissibility of all the relevant records and the results of the review. Instead, they challenge Sullivan’s ability to testify about the invoice review because she did not personally ensure the completeness of the data on which she relied, and they could not confront the people who did.

Our first task: We must pinpoint the out-ofcourt statements from Qlarant's subject -matter experts that Sullivan allegedly conveyed to the jury. See Smith , 602 U.S. at 801. Defendants do not point us to any specific statements. And we cannot find any.

Defendants lament large swaths of Sullivan’s testimony where she spoke in broad terms about Qlarant’s general data -collection process. But Sullivan’s testimony makes clear that she spoke about her understanding of Qlarant’s general process based on her involvement in other cases. Sullivan, based on her personal knowledge, could testify about how Qlarant 'typically functioned,' including its 'standards, practices, and procedures,' as well as the company's “guidelines and techniques,” without running afoul of the Confrontation Clause. See id. at 799. She did not testify to what other people did in this case. Indeed, the district court did not allow her to do so.

The closest Sullivan came to conveying any testimonial hearsay was on crossexamination. Fakhuri's attorney questioned Sullivan about her role in the data -collection process. In answering his question about whether she verified that the records were complete, Sullivan testified that there was no need for her to do that because the records “had already been verified” by the time she got involved. R. 249, PageID 2025.

Still, it is hard to see how Sullivan conveyed another person's out -of-court statements during this exchange. Sullivan said that others had reviewed the integrity of the data, but she did Page 10not act as a 'mouthpiece' for their opinions about its completeness. Smith , 602 U.S. at 800. Defendants make an inferential leap that we will not. They suggest that Sullivan necessarily conveyed to the jury that those records were complete because others told her they were. But Sullivan never said that the records were complete.

Nor was Sullivan's ultimate conclusionthat Defendants had not purchased enough medication to cover their insurance claims — impermissibly informed by any testimonial hearsay. The Confrontation Clause does not require the presence of everyone “whose testimony may be relevant' to establish the 'authenticity' or 'accuracy' of the data underlying an expert's conclusion. Melendez-Diaz , 557 U.S. at 311 n.1. Whether the records were complete or not goes to the weight the jury should give Sullivan's conclusion, not its admissibility. Cf. id. (recognizing that “gaps in the chain of custody normally go to the weight of the evidence rather than its admissibility” (citation modified)). And Defendants had the opportunity to minimize the weight of Sullivan's testimony by emphasizing her lack of involvement in the data -collection process. Indeed, they thoroughly took advantage of that opportunity — their cross-examinations lasted twice as long as Sullivan’s direct examination.

As Defendants point out, any missteps in the data-collection process would have undermined Sullivan's conclusion. But their inability to cross -examine everyone involved in that process does not create a Confrontation Clause violation. Defendants stipulated to the admissibility of the data that Sullivan relied on, and they did not need to cross-examine the subjectmatter experts to “weed out” any errors in Sullivan’s analysis. See id. at 319. If any wholesaler records were missing from Sullivan's invoice review or if any data entries were incorrect, Defendants had 'a full and fair opportunity to probe and expose these infirmities' through Sullivan. See Delaware v. Fensterer , 474 U.S. 15, 22 (1985) (per curiam). After all, she reviewed the records and the inputs, and Defendants had access to them.

Even so, Defendants argue that cross-examination was ineffective because Sullivan was “a last -minute, thirdorder substitute” who “conducted little of the analysis that formed the basis for her conclusions.' Hamaed Br. at p.39. Their attempts to minimize Sullivan's hands -on involvement in the invoice review presented at trial are unpersuasive.

Page 11To be sure, Sullivan did not validate the completeness of the data. But she analyzed and interpreted it. She conducted the final “quality check” on all the information. R. 249, PageID 1991. She verified that Qlarant received records for the right time and ensured that the invoice review focused on that period. She also confirmed that the SAS system incorporated all the data from all the wholesalers and that the relevant data correctly populated into that system. True, the subject-matter experts corrected any typographical errors that the system flagged. But Sullivan personally checked their work. She went 'over all that stuff . . . [w]ith a fine -tooth comb.' R. 254, PageID 2101.

Sullivan also touched everything that went to the jury. To that end, she was personally involved in the invoice review presented at trial. She analyzed the wholesaler records, the prescription-claims data, and the spreadsheets containing that information, and she interpreted the results of the SAS comparison. She helped create charts submitted to the jury summarizing the losses Medicare and Medicaid suffered because of the scheme. And she compared those charts to the invoice-review analysis to ensure the data was consistent.

That the government disclosed other experts before Sullivan does not change things. Sullivan conducted her own review and reached her own results. In that regard, she is not the kind of “substitute” expert that the Court prohibited in Smith and Bullcoming . In those cases, the testifying experts did not personally participate in any of the testing they recounted to the jury. Smith , 602 U.S. at 790; Bullcoming , 564 U.S. at 657. The expert in Smith relayed the results another person said she obtained based on tests she said she performed. 602 U.S. at 790. The expert in Bullcoming did the same. 564 U.S. at 651. But Sullivan testified to the work she personally performed. It does not matter that other people performed the same work before her. See id. at 666 (opinion of Ginsburg, J.).

Defendants wish to cross-examine everyone who validated the data. But the Confrontation Clause promises 'an opportunity for effective cross-examination, not crossexamination that is effective in whatever way, and to whatever extent, the defense might wish.” Fensterer , 474 U.S. at 20. And it has never required the presence of “everyone who laid hands on the evidence.” Melendez-Diaz , 557 U.S. at 311 n.1.

Page 12It should go without saying that it took a team of people to perform the clerical work necessary for Qlarant’s investigation, which included the collection and input of “hundreds of thousands of claims lines.” R. 256, PageID 2327. But the data -collection process was not “the most important part[] of [Sullivan’s] analysis,” as Defendants contend. See Hamaed Br. at p.42. The most important part of her analysis was the actual analysis — the comparison of the medications purchased against the prescriptions billed. That was the heart of Sullivan's testimony, and her opinion was based on her personal assessment of the data, her review of its input into the SAS program, and her interpretation of the program’s calculations.

* * *

In sum, Defendants have not shown that Sullivan’s testimony violated the Confrontation Clause. The right of confrontation did not prevent Sullivan from testifying to her independent interpretation of data provided by others. See United States v. Contreras , 149 F.4th 349, 363 (4th Cir. 2025) (reasoning that an expert's 'exposure to testimonial hearsay' does not automatically prevent the introduction of their 'independent assessment' (quotation omitted)). The district court did not err in rejecting Defendants’ Confrontation Clause claims.

Defendants also argue that the district court violated their right to present a complete defense by excluding evidence of (1) legitimately billed prescriptions and (2) the PBMs' financial bias. We review this constitutional challenge de novo. United States v. Reichert , 747 F.3d 445, 453 (6th Cir. 2014).

“Whether rooted directly in the Due Process Clause . . . or in the Compulsory Process or Confrontation Clauses of the Sixth Amendment, the Constitution guarantees criminal defendants a meaningful opportunity to present a complete defense.” Holmes v. South Carolina , 547 U.S. 319, 324 (2006) (citation modified). This 'generally includes the right to the admission of competent, reliable, exculpatory evidence to negate an element of the offense.” United States v. Odeh , 815 F.3d 968, 977 (6th Cir. 2016) (citation modified). Like most constitutional rights, this one is not “unfettered.” Taylor v. Illinois , 484 U.S. 400, 410 (1988).

Page 13The district court's ability 'to exclude evidence from a criminal trial' under well -established rules of evidence tempers the right to present a complete defense. United States v. Reynolds , 86 F.4th 332, 350 (6th Cir. 2023). The district court may exclude evidence from a criminal trial if it has a legitimate purpose for doing so. Holmes , 547 U.S. at 324 – 25. To that end, a district court may properly exclude unfairly prejudicial, confusing, or potentially misleading evidence offered by a defendant. Id. at 326; Fed. R. Evid. 403.

A defendant must make two showings to establish a constitutional violation. First, he must show that the exclusion of his evidence was “arbitrary or disproportionate to the purposes [the evidence rule was] designed to serve.” United States v. Scheffer , 523 U.S. 303, 308 (1998) (citation modified). Second, the exclusion must have “infringed upon a weighty interest” of the defendant. Id. Defendants have not met this “demanding” burden. Reynolds , 86 F.4th at 351.

Start with the evidence of legitimate billing. Before trial, the government sought to exclude specific instances of Defendants' legitimate billing practices and their provision of legitimate pharmaceutical services. The district court granted the motion, concluding (in part) that the evidence would be confusing to the jury under Rule 403. Consistent with its order, the court prevented Defendants' witnesses from discussing prescriptions they received from Defendants’ pharmacies.

The district court did not err in excluding evidence that these pharmacists sometimes operated a legitimate business. We assess the evidence considering “the government’s theory” of culpability. United States v. Dimora , 750 F.3d 619, 630 (6th Cir. 2014). Relevant here, the government did not allege that every prescription the defendants billed was fraudulent. See United States v. Ifediba , 46 F.4th 1225, 1238 (11th Cir. 2022). So “it was no defense” that they legitimately billed the insurers for some prescriptions.¹ Id. ; accord United States v. Betro , 115 F.4th 429, 453 (6th Cir. 2024). “Evidence of noncriminal activities would only be relevant if the indictment charged the defendants with ceaseless criminal conduct.” United States v. Daulton , 266 F. App’x 381, 386 (6th Cir. 2008) (citation modified).

Page 14Now consider evidence of bias in the PBM industry. At trial, Defendants sought to admit evidence of an industrywide “lack of trust” in PBMs, which has spawned “all sorts of reforms.” R. 254, PageID 2173. This alleged lack of trust stems from the PBMs' financial ties to pharmacies that compete with independent pharmacies and the general perception that they engage in industry-wide anticompetitive practices. See, e.g. , McKee Foods Corp. v. BFP Inc. , 173 F.4th 242, 250 (6th Cir. 2026). The district court properly excluded this line of questioning.

Defendants cannot connect their broad policy-based concerns to the data at issue here. None of them presented any evidence that the PBMs manipulated their claims data. A jury would have to make a series of speculative leaps to conclude that a group of PBMs conspired together to frame Defendants for healthcare fraud by overstating their claims data, all to get a competitive advantage in the market. Evidence of bias is relevant. But the district court did not err in excluding this evidence, which is “only marginally relevant” at best. Cf. Van Arsdall , 475 U.S. at 679.

Defendants also assert that both the district court and the government impermissibly bolstered Sullivan’s credibility by referring to her as an expert in front of the jury.

Some background is in order. During Sullivan's testimony, Hamaed objected to Sullivan’s testimony based on lack of personal knowledge. Afterward, the parties discussed how to proceed with her testimony. The government suggested qualifying her as an expert. Hamaed agreed, so the government moved to qualify her as an expert in front of the jury. Defendants did not object, and the district court allowed Sullivan to proceed. The court later realized that it had not explicitly ruled on the government's motion. At the conclusion of Sullivan's direct examination, the court made the 'record crystal clear' by clarifying that it had granted the motion to “qualify her as an expert.” R. 249, PageID 2016.

During crossexamination, Fakhuri’s counsel questioned Sullivan about her knowledge of the datacollection process used in this case. Abdelrazzaq’s counsel repeatedly objected to her lack of personal knowledge in front of the jury. He twice asked for her testimony to be stricken —“every testimony, every figure, everything”—because she had “no idea” what she was Page 15talking about. Id. at 2025. The government asked to move the conversation outside the hearing of the jury.

The district court denied the motion to strike and agreed to give a curative instruction on Abdelrazzaq's speaking objection. The court advised the parties of the contents of that instruction before reading it to the jury. No one objected. The court told the jury:

[Fakhuri’s counsel] was questioning Ms. Sullivan last week, and then there w[ere] some objections that took place pertaining to documents that have been reviewed, and personal knowledge and those kinds of things. And so, I want you all to know that Ms. Sullivan’s testimony that you heard last week is permissible, and it is consistent with the Rules of Evidence which allow experts, which Ms. Sullivan is an expert with specialized knowledge, allows experts with specialized knowledge to rely upon data that has been pulled by others, provided that the conclusions reached by Ms. Sullivan were based on her own personal analysis. All right? And so, I have, in fact, denied the request to strike her testimony. Her testimony, ladies and gentlemen, is, in fact, permissible.

R. 254, PageID 2098 – 99.

Defendants argue that the court erred by referring to Sullivan three times as an expert. They concede that they forfeited these arguments, so we review them for plain error. See Greer v. United States , 593 U.S. 503, 507 (2021). To establish plain error, Defendants must show that (1) there was an error, (2) the error was “plain,” (3) the error affected their “substantial rights,” and (4) the error “had a serious effect on the fairness, integrity or public reputation of judicial proceedings.” Id. at 507 – 08 (citation modified).

This court has described the correct process for a party to qualify a witness as an expert at trial. The party should lay the appropriate foundation and then elicit the witness's opinion testimony. United States v. Campbell , 135 F.4th 376, 395 (6th Cir. 2025). The party should not tender the witness as an expert in front of the jury or ask the court to do so. Id. Nor should the district court declare 'that a witness is qualified as an expert or to render an expert opinion.' United States v. Johnson , 488 F.3d 690, 697 (6th Cir. 2007) (quotation omitted). Why? Because referring to a witness as an expert in front of the jury “lends a note of approval to the witness that inordinately enhances the witness's stature and detracts from the court's neutrality and detachment.” Id.

Page 16The court should not have referred to Sullivan as an expert in front of the jury, and the government should not have asked it to do so. See United States v. Maya , 966 F.3d 493, 505 (6th Cir. 2020); Johnson , 488 F.3d at 697 – 98. That much is plain. But Defendants must do more than point to a plain error; they must also show that the error affected their substantial rights. See Greer , 593 U.S. at 507 –08. This means that they “must show a reasonable probability that, but for the error, the outcome of the proceeding would have been different.” Molina-Martinez v. United States , 578 U.S. 189, 194 (2016) (citation modified).

For two reasons, Defendants have not met their burden. First, the district court mitigated its error. The district court instructed jurors that they 'd[id] not have to accept' Sullivan's opinions —rather, they should decide independently “how much of [her] testimony to believe, and how much weight it deserves.” R. 289, PageID 5137. And second, substantial evidence, in particular the testimony of Defendants’ coconspirators and written communications among the conspirators, supported Defendants' convictions outside of Sullivan's testimony. Cf. United States v. Majors , No. 23-3444, 2024 WL 3429411, at *3 (6th Cir. July 16, 2024).

Defendants argue that the district court's repeated references to Sullivan as an expert prejudiced them because the court sanitized their cross-examinations, which focused on her lack of personal knowledge about the datacollection process. The district court's instructions, however, “minimized any risk that the court’s reference to [Sullivan] as an expert might have improperly influenced the jury.” United States v. Sibley , 681 F. App'x 457, 46061 (6th Cir. 2017). Also, Defendants never grapple with the significant evidence against them outside of Sullivan’s testimony. So this challenge fails.

Abdelrazzaq and Fakhuri argue that the government's proof at trial varied from the conspiracy alleged in the indictment. At best, they say, the government’s proof showed that the fraud at each pharmacy constituted separate conspiracies, not an overarching conspiracy involving all the pharmacies. We review de novo whether a variance occurred. United States v. Robinson , 99 F.4th 344, 364 (6th Cir. 2024). We consider the trial evidence about the number ofPage 17 conspiracies 'in the light most favorable to the government.' United States v. Guerrero , 76 F.4th 519, 524 (6th Cir. 2023) (quotation omitted).

The Fifth Amendment’s Grand Jury Clause provides that “[n]o person shall be held to answer for a capital, or otherwise infamous crime, unless on a presentment or indictment of a Grand Jury.” U.S. Const. amend. V. This provision guarantees criminal defendants the right to be 'tried only on charges that had in fact been passed on by a grand jury.' United States v. Miller , 471 U.S. 130, 135 (1985). 'The very purpose of the requirement that a man be indicted by grand jury is to limit his jeopardy to offenses charged by a group of his fellow citizens acting independently of either prosecuting attorney or judge.” Stirone v. United States , 361 U.S. 212, 218 (1960). Thus, 'after an indictment has been returned its charges may not be broadened through amendment except by the grand jury itself.” Id. at 215 – 16.

A variance occurs when 'the evidence proves facts materially different from those alleged in the indictment.” United States v. Siefert , 161 F.4th 379, 393 (6th Cir. 2025) (citation modified). This can happen in a conspiracy case when 'the indictment charges a single conspiracy and the evidence at trial instead demonstrates multiple, separate conspiracies.' Guerrero , 76 F.4th at 524.

We will conclude that the trial evidence has varied from the indictment “only where . . . the evidence can reasonably be construed only as supporting a finding of multiple conspiracies.” Siefert , 161 F.4th at 395 (quotation omitted). To decide how many conspiracies the proof supported, we look to the 'existence of a common goal, the nature of the scheme, and the overlapping of the participants in various dealings.” United States v. Hughes , 505 F.3d 578, 587 (6th Cir. 2007) (quotation omitted). Abdelrazzaq and Fakhuri have not shown a variance.

Start with the common goal. As the government points out, the common goal of the charged conspiracy was to increase the pharmacies' profits. To achieve this goal, the coconspirators billed insurers for prescriptions they did not fill. Each defendant — even Singh, who had no ownership interest in any of the pharmacies — received a profit-sharing check inflated by their fraudulent billing practices.

Page 18Next, take the nature of the scheme. Each defendant furthered the conspiracy’s common goal by billing for prescriptions that they did not dispense. They all used identical strategies, including falsifying patients’ signatures and cultivating a designated area for fraudulently billed prescription labels. And they all targeted specific kinds of medications, such as inhalers and insulin.

Finally, consider the overlap among the coconspirators for the five pharmacies at issue. The following chart outlines each defendant's ownership interest in each pharmacy and the designated pharmacist in charge.² Harper Drugs (Detroit, MI) Wayne Campus (Detroit, MI) Heartland 1 (Kettering, OH) Heartland 2 (Moraine, OH) Eastside 3 (Detroit, MI) Hassan Abdallah 33.3% 18% 25% 33.3% 50% Pharmacist in charge Raef Hamaed 33.3% 18% 25% 33.3% 25% Tarek Fakhuri 4 33.3% Pharmacist in charge 18% 25% Kindy Ghussin 16% 25% Pharmacist in charge 33.3% Balhar Singh Pharmacist in charge Ali Abdelrazzaq 30% Pharmacist in charge

As shown above, Hamaed was involved in all five pharmacies, and Fakhuri and Ghussin were involved in three. True, Abdelrazzaq had an ownership interest in Wayne Campus only, but he Page 19co-owned Wayne Campus with every defendant other than Singh. And they all knew each other -each defendant joined the conspiracy because of their prior relationships with one another. Moreover, the pharmacies often shared the same personnel.

Contrary to Abdelrazzaq's and Fakhuri's assertions, the fact that the coconspirators' ownership interests can be “subdivided” into individual pharmacies does not mandate the finding of multiple conspiracies. United States v. Beals , 698 F.3d 248, 259 (6th Cir. 2012) (quotation omitted). The evidence showed that all of the defendants acted knowingly to achieve a common goal -that proof was sufficient for the jury to find a single conspiracy. Abdelrazzaq's and Fakhuri’s variance arguments thus fail.

Abdelrazzaq contends that the district court erred in permitting Abdallah to testify to his interpretations of text messages that Defendants sent to Abdallah. We review a district court’s evidentiary rulings under the abuse-of-discretion standard. United States v. Fox , 134 F.4th 348, 381 (6th Cir. 2025) (per curiam).

During Abdallah's testimony, the government asked him about his understanding of several text messages from Abdelrazzaq. During one text exchange, Abdallah expressed his frustration about Abdelrazzaq's lack of contribution to the profit -sharing checks: 'Of all pharmacies, our income from [Abdelrazzaq’s] does not pay the mortgage for most of us.” R. 266, PageID 3179. Abdelrazzaq responded, “I am doing my best and I will keep doing that.” Id. at 3180. According to Abdallah, Abdelrazzaq’s message meant that he was “trying to commit fraud.” Id. Abdelrazzaq objected, calling Abdallah’s statement speculation. The district court overruled the objection. Abdallah continued to testify to his interpretation of text messages like these, with one overarching theme: the text-message participants were talking about committing fraud.

Federal Rule of Evidence 701(a) permits lay witnesses to testify to opinions that are “rationally based on the witness’s perception.” This testimony is permissible because it draws upon the witness’s “experiential observations.” United States v. Freeman , 730 F.3d 590, 595 (6th Cir. 2013) (quotation omitted). For example, we have recognized that law-enforcement Page 20officers may testify as lay witnesses about their interpretations of conversations when they either participated in the conversation, observed it, or have 'personal knowledge of the facts being related.” United States v. Reed , 163 F.4th 338, 356 (6th Cir. 2025) (quotation omitted).

The district court did not abuse its discretion in permitting Abdallah to testify to his interpretation of the text messages. Abdallah participated in those conversations, knew about the facts relayed, and based his perceptions on his lengthy business relationship with Abdelrazzaq. See id. at 358. That is enough foundation to support a lay opinion under Rule 701.

Abdelrazzaq asks us to reach a different result based on the Second Circuit’s decision in United States v. Kaplan . 490 F.3d 110 (2d Cir. 2007). But Kaplan does not help Abdelrazzaq. There, the court held that the witness’s interpretation was not rational because he had no personal experience with the defendant. Id. at 119. By contrast, Abdallah had plenty of experience with Abdelrazzaq. His personal relationship with Abdelrazzaq provided the foundation necessary for him to testify about his interpretation of their conversations.

Abdelrazzaq also argues that the district court erred in denying Hamaed's motion to subpoena wholesaler records under Federal Rule of Criminal Procedure 17(c). We usually review a district court’s order denying the issuance of a subpoena duces tecum for an abuse of discretion. United States v. Theunick , 651 F.3d 578, 591 (6th Cir. 2011). Under that standard, we will not disturb the district court’s decision “[w]ithout a determination of arbitrariness or that the trial court finding was without record support.” United States v. Nixon , 418 U.S. 683, 702 (1974). But the government argues that Abdelrazzaq did not preserve this claim because he did not file his own Rule 17(c) motion or join Hamaed’s, so we should apply plain -error review. We need not resolve this standard-ofreview dispute because Abdelrazzaq's challenge fails under either standard.

Under Rule 17(c), a party may seek a subpoena to order a “witness to produce any books, papers, documents, data, or other objects the subpoena designates.” Fed. R. Crim. P. 17(c)(1). Rule 17(c) paves the way for a defendant to subpoena records from third parties. But the rule does not operate as a generalized means of securing pretrial discovery. Nixon , 418 U.S. at 698.

Page 21Generally, before a defendant can obtain a subpoena for third-party documents, he must show that the records: (1) are relevant; (2) cannot reasonably be procured otherwise; (3) are necessary to prepare for trial; and (4) are sought in “good faith,” not “as a general fishing expedition.” Id. at 699 – 700 (citation modified).

Before trial, Hamaed sought to subpoena 'all orders, invoices, and purchase history records' related to the five pharmacies from 44 wholesalers for a 10 -year period. R. 144, PageID 659 – 60. The government had already obtained and provided those records to Defendants, but Hamaed asserted that he needed to independently verify that the government’s record collection and production were complete.

The district court did not err in denying the motion for a Rule 17(c) subpoena for two reasons. First, neither Hamaed nor Abdelrazzaq showed that they could not otherwise obtain the records sought by the subpoena. The government said it had produced the records in discovery. Also, Hamaed or Abdelrazzaq could have checked their own pharmacies' records. Second, Hamaed’s subpoena request lacked “specificity.” Nixon , 418 U.S. at 700. It was motivated by pure speculation that the government either did not obtain all the relevant documents or did not disclose them all. And the subpoena request was not tailored to procure the allegedly missing records; instead, Hamaed cast a wide net to dredge up anything and everything the wholesalers had to offer in hopes of catching something that might impeach Sullivan. This kind of fishing expedition is plainly impermissible. See United States v. Llanez-Garcia , 735 F.3d 483, 494 (6th Cir. 2013).

Ghussin asks us to remand for a jury poll because he received a letter from a juror who called the jury instructions 'confusing.' Ghussin Br. at p.32. Because Ghussin did not preserve this claim below, we review it for plain error.

We reject Ghussin’s challenge to the jury’s verdict. Generally, “the losing party cannot, in order to secure a new trial, use the testimony of jurors to impeach their verdict.” McDonald v. Pless , 238 U.S. 264, 269 (1915). '[A] juror is incompetent to impeach his or her verdict.' United States v. Gonzales , 227 F.3d 520, 523 (6th Cir. 2000). Thus, a party cannot attack the Page 22integrity of a unanimous jury verdict by relying on a juror’s testimony to show a “misinterpretation of the court’s instructions.” Id. Because Ghussin attempts to rely on a juror’s post -verdict testimony to do just that, his claim fails.⁵

We turn now to the sentencing issues. We review sentences imposed by the district court for an abuse of discretion. Gall v. United States , 552 U.S. 38, 51 (2007). The sentence imposed must be both procedurally and substantively reasonable. Id.

A proceduralreasonableness challenge attacks the district court’s method of calculating a sentence. Id. A district court imposes a procedurally unreasonable sentence by 'failing to calculate (or improperly calculating) the [Sentencing] Guidelines range, treating the Guidelines as mandatory, failing to consider the [18 U.S.C.] § 3553(a) factors, selecting a sentence based on clearly erroneous facts, or failing to adequately explain the chosen sentence.” Id.

A defendant’s substantive -reasonableness challenge asserts that the sentence imposed is “too long.” United States v. Rayyan , 885 F.3d 436, 442 (6th Cir. 2018). We analyze “whether the length of the sentence is greater than necessary to achieve the sentencing goals set forth in 18 U.S.C. § 3553(a).” United States v. Tristan-Madrigal , 601 F.3d 629, 632 – 33 (6th Cir. 2010) (citation modified). A sentence can be greater than necessary if the court 'placed too much weight on some of the § 3553(a) factors and too little on others.” Rayyan , 885 F.3d at 442. “T he district court’s decision to assign more or less weight to a given factor is a matter of reasoned discretion, not math.' United States v. Boucher , 937 F.3d 702, 707 (6th Cir. 2019) (citation modified). We presume that sentences within the Guidelines range are substantively reasonable. Gall , 552 U.S. at 51.

Hamaed, Abdelrazzaq, and Ghussin argue that their sentences are procedurally unreasonable because the district court erred in calculating their respective loss amounts.

Page 23Ghussin separately asserts that the district court erred in applying a sophisticated-means enhancement.

We start with the challenges to the loss amounts. When a defendant is convicted of fraud, his advisory Guidelines range reflects the financial harm he caused. U.S.S.G. § 2B1.1(b). This is because the “Guidelines enhance a defendant’s sentence to correlate to the amount of loss caused by his fraud.' Siefert , 161 F.4th at 400 (citation modified). The greater the loss, the greater the offense level. See U.S.S.G. § 2B1.1(b).

The Guidelines define loss as “the greater of actual or intended loss.” Betro , 115 F.4th at 454 (quotation omitted). Relevant here, actual loss is 'the reasonably foreseeable pecuniary harm that resulted from the offense.” U.S.S.G. § 2B1.1(b)(1)(C)(i). In other words, it is the true financial loss the victim suffered from the fraud. Id.

The district court's calculation of actual loss 'need not be determined with precision.' Siefert , 161 F.4th at 400 (quotation omitted). That said, the calculation should reflect a “reasonable estimate of the loss.” United States v. Wala , 166 F.4th 583, 596 – 97 (6th Cir. 2026) (quotation omitted). The district court should base its reasonable estimate on 'available information.” Siefert , 161 F.4th at 400 (quotation omitted). To that end, the court must “actually find facts, and it must do so by a preponderance of the evidence.” Id. (quotation omitted).

We review the district court’s factual findings for clear error and the methodology behind its calculation de novo. Betro , 115 F.4th at 454. Factual findings are clearly erroneous when “we are left with a definite and firm conviction that the district court made a mistake.” United States v. Matthews , 155 F.4th 845, 850 (6th Cir. 2025) (citation modified). The party challenging the district court's loss determination bears a 'heavy burden' to show that the court’s “evaluation of the loss was not only inaccurate, but was outside the realm of permissible computations.” Wala , 166 F.4th at 597 (quotation omitted).

Page 24The district court properly calculated the loss amounts attributable to each defendant at their sentencing hearings, and its calculations were supported by a preponderance of the evidence. We consider each challenge below.

Hamaed . Based on the proof at trial, the district court calculated Hamaed’s loss amount as $13 million, which warranted a 20-level enhancement under U.S.S.G. § 2B1.1(b)(1)(K). This calculation covered the losses caused by all five pharmacies. The court based its loss calculation on Sullivan's invoice review. Recall that Sullivan's review compared the wholesaler records against the prescription-claims data.

The district court did not err in calculating the loss amount based on Sullivan's testimony. The evidence substantially supported her calculations, and Hamaed provided no evidence to materially contradict her estimates. See United States v. Washington , 715 F.3d 975, 985 (6th Cir. 2013) (upholding the district court's loss determination where the defendant “provided no estimates” to contradict the “available information” that the court relied on).

On appeal, Hamaed argues that the district court relied on a flawed methodology — that is, the invoice review. True, the original review included loss amounts for prescriptions that other government witnesses testified were not fraudulent. But the court accounted for this error by noting that even without those prescriptions, the Guidelines calculation would not have changed.

According to Hamaed, the district court erred in failing to deduct even more based on his speculation that the overall invoice review was systemically flawed. He asserts that Sullivan’s calculations overestimated the losses attributable to him, but he provided no examples that would have a material impact on his Guidelines calculation. He merely suggests that there are missing wholesaler records that could show additional medication purchases, but he produced none of this to the district court. And he further argues that the invoice review was unreliable, but he provided no evidence to support this allegation.

Hamaed has not shown that Qlarant’s invoice review was wrong. So he failed to meet his heavy burden to show that the district court was wrong to rely on it.

Page 25Abdelrazzaq . Abdelrazzaq adopted Hamaed's arguments. But unlike Hamaed, Abdelrazzaq did not object to the court's loss calculation at his sentencing. Regardless, his challenge fails for the same reasons that Hamaed’s does.

Ghussin . Like Hamaed and Abdelrazzaq, Ghussin argues that the district court erred in relying on Sullivan’s calculations. And like those two, Ghussin failed to show that Sullivan’s calculations were inaccurate.

But Ghussin makes another argument specific to him. He asserts that the district court erred in holding him responsible for the losses caused by Wayne Campus, a pharmacy he did not operate.

In calculating a defendant's loss amount, the court can consider 'relevant conduct.' U.S.S.G. § 1B1.3(a). Relevant conduct includes 'loss that resulted from certain conduct of others,” like a coconspirator. United States v. Donadeo , 910 F.3d 886, 894 (6th Cir. 2018). The actions of a defendant's coconspirators are attributable to the defendant when those acts were '(i) within the scope of the jointly undertaken criminal activity, (ii) in furtherance of that criminal activity, and (iii) reasonably foreseeable in connection with that criminal activity.' U.S.S.G. § 1B1.3(a)(1).

Because the losses caused by the fraud committed at Wayne Campus were reasonably foreseeable to Ghussin, the district court properly included them in Ghussin's loss amount. Ghussin had an ownership interest in Wayne Campus and profited from the fraud committed there through the profit-sharing checks he received. He even talked to the other coconspirators about Wayne Campus's profitability and expressed frustration at Abdelrazzaq's failure to generate more money.

Ghussin’s attempts to now minimize his involvement in the pharmacy are unpersuasive. The evidence supports the district court’s finding that the fraud committed at Wayne Campus Page 26was reasonably foreseeable to him. See United States v. Kennedy , 714 F.3d 951, 961 (6th Cir. 2013).

Ghussin also argues that his sentence was procedurally unreasonable because the district court erred in applying an enhancement for his use of sophisticated means to conceal the conspiracy. We review the district court's factual findings for clear error and its legal conclusions de novo. United States v. Yousef , 170 F.4th 498, 507 (6th Cir. 2026). We have not yet “resolved whether we review the application of the enhancement to those factual findings for clear error or de novo.” Id. at 507 – 08. But the exact standard of review is not dispositive here because Ghussin’s challenge fails under either. See id.

The Guidelines call for a twopoint enhancement to a defendant’s offense level if he uses “sophisticated means” to commit his offense. U.S.S.G. § 2B1.1(b)(10)(C). The enhancement applies when the proof shows 'especially complex or especially intricate offense conduct pertaining to the execution or concealment of an offense.' United States v. Simmerman , 850 F.3d 829, 833 (6th Cir. 2017) (quoting U.S.S.G. § 2B1.1 cmt. n.9). To decide whether the evidence supports the enhancement, we look to “the totality of a defendant’s conduct— the entire scheme.' Betro , 115 F.4th at 456 (quotation omitted). In that regard, we do not isolate individual portions of the defendant’s conduct. See id. And “[a]n otherwise simple theft may nevertheless trigger the enhancement if the means to conceal the theft were sophisticated.' Yousef , 170 F.4th at 508.

The district court properly applied the sophisticated-means enhancement based on evidence that Ghussin falsified patient records to obscure his fraudulent conduct. Ghussin tricked patients into signing for prescriptions they did not receive, and he used those forged signatures to satisfy PBM audits. Deceptive practices such as falsifying records exemplify the use of sophisticated means. See United States v. Crosgrove , 637 F.3d 646, 667 (6th Cir. 2011) (collecting cases).

Ghussin argues otherwise. He asserts that this enhancement should apply only to 'cunning and scheming defendants.' Ghussin Br. at p.22. According to him, he is neither.

Page 27But the evidence shows that Ghussin participated in a healthcare-fraud scheme that survived for almost a decade, spanned five pharmacies, and involved six defendants. Part of the reason the scheme lasted so long was because of Ghussin’s sophisticated method to evade detection. See Yousef , 170 F.4th at 508 – 09. So the court did not err in applying the enhancement.

Ghussin also challenges the substantive reasonableness of his sentence. He asserts that his below-Guidelines sentence was longer than it needed to be. His argument lacks merit.

The record shows that the district court weighed the relevant § 3553(a) factors and imposed a presumptively reasonable below-Guidelines sentence. It contemplated mitigating factors like Ghussin's age, health, and family network. And it varied downward based on Ghussin’s low risk of recidivism. But the district court noted that healthcare fraud is a “serious offense,” and it imposed a sentence that would “send a message” to other pharmacists. R. 417, PageID 7451 – 52.

Ghussin concedes that the district court considered appropriate mitigating factors, but he argues that it should have given them more weight than it did. Absent evidence of arbitrariness, however, a mere 'assertion that the district court should have balanced the § 3553(a) factors differently . . . is simply beyond the scope of this court's appellate review.' United States v. Frei , 995 F.3d 561, 567 – 68 (6th Cir. 2021) (citation modified). And Ghussin has not shown that the district court acted arbitrarily. His challenge thus fails.

Lastly, Hamaed argues that the district court erred in imposing restitution. We disagree.

We review the propriety of the district court’s restitution order de novo. United States v. Clay , 162 F.4th 757, 773 (6th Cir. 2025) (per curiam). “Because federal courts have no inherent power to award restitution, restitution orders are proper only when and to the extent authorized by statute.” United States v. Fike , 140 F.4th 351, 354 – 55 (6th Cir. 2025) (quotation omitted). When restitution is statutorily authorized, “we review the amount of restitution ordered under the abuse of discretion standard.” Clay , 162 F.4th at 773.

Page 28The Mandatory Victims Restitution Act (“MVRA”) requires the district court to order restitution when the defendant has been convicted of 'any offense committed by fraud' that caused an identifiable victim to suffer a loss. 18 U.S.C. § 3663A(c)(1). Because Hamaed was convicted of conspiracy to commit healthcare and wire fraud, the MVRA required the district court to order restitution.

Hamaed does not contest the fact of restitution so much as the district court’s procedure in ordering it. He asserts that the court violated his right to be present at his sentencing by ordering restitution in its judgment without first discussing it or pronouncing it at his sentencing hearing.

Relevant here, the district court raised the restitution issue at the sentencing hearing but allowed the parties to brief the issue afterward. It later changed course and entered judgment on all aspects of Hamaed’s sentence, including restitution, before he filed a brief on the restitution issue. The court imposed joint and several liability on Hamaed in the amount of $12,803,048.97, based on its actual loss calculation. Hamaed asked the court to reconsider its order, and it refused.

To be sure, a defendant has a dueprocess right “to be present at his sentencing.” United States v. Hayden , 102 F.4th 368, 371 (6th Cir. 2024). But the imposition of restitution under the MVRA differs from the imposition of a term of imprisonment. The MVRA “does not mandate” that the district court 'conduct a hearing to obtain relevant evidence and afford the parties an opportunity to present oral argument.” United States v. Vandeberg , 201 F.3d 805, 813 (6th Cir. 2000).

Instead, the MVRA 'delineates a panoply of procedures pertinent to the issuance and enforcement of restitution.” Id. ; see 18 U.S.C. § 3664. It “grants a district court discretion to choose the procedures that will best aid the court in assessing the amount of loss.” Vandeberg , 201 F.3d at 813. It even allows the court to defer ruling on restitution until after the sentencing hearing if the victim’s losses are not ascertainable by ten days before sentencing, so long as it resolves the restitution issue within 90 days. Id. If the parties dispute any issues, the court must Page 29then give them an opportunity to be heard and rule on the disputed issue. Id. at 814; see also 18 U.S.C. § 3664(c) (incorporating Federal Rule of Criminal Procedure 32(c) (1996)).

Here, the district court did not rule on Hamaed’s objections to the restitution amount and nonetheless imposed restitution. But even assuming that was error, any error was harmless. After all, the court gave the parties an opportunity to be heard on the crux of the disputed issue — they discussed the actual loss amount at length during the sentencing hearing. And the court then gave Hamaed another opportunity to be heard in his motion to correct the restitution amount. After considering Hamaed’s motion and the government’s response, the court rejected Hamaed's argument. It did not find his proposed loss amount 'reliable or accurate' and determined that he had provided “no proof.” R. 428, PageID 7514. Despite having over a year to review Qlarant’s loss calculations, Hamaed recycled his speculative argument that Qlarant’s review was somehow wrong. Because the court provided Hamaed an “opportunity to object to the amount,” any error was harmless. Vandeberg , 201 F.3d at 814.

None of Hamaed's other arguments convinces us to overturn the district court's restitution award. To be sure, the methods for calculating the loss amount under the Guidelines and restitution under the MVRA can differ. The Guidelines permit the district court to base the loss amount on the defendant’s intended losses if they exceed the victims’ actual losses. Betro , 115 F.4th at 454. By contrast, the MVRA is based exclusively on actual loss. Clay , 162 F.4th at 774; see also 18 U.S.C. § 3664(f)(l)(A) (stating that 'the court shall order restitution to each victim in the full amount of each victim's losses'). But Hamaed's concern about these differences is misplaced. Where, as here, the loss amount is based on the victim’s actual losses, the two figures are generally the same. See United States v. Simpson , 538 F.3d 459, 465 – 66 (6th Cir. 2008).

Hamaed posits that the district court needed to come up with a more precise amount. But “the MVRA does not require courts to calculate restitution with exact precision.” United States v. Kilpatrick , 798 F.3d 365, 388 (6th Cir. 2015). To be sure, ' some precision is required.” Id. To that end, Qlarant’s calculations were substantially precise— Sullivan based her invoice review on concrete data, and Hamaed failed to show that Qlarant’s calculations were inaccurate. Thus, it cannot be said that the court meted out “rough justice.” Id. (quotation omitted).

Page 30Nor did the district court err in imposing restitution on a joint-and-several basis. The MVRA entrusts the district court to choose how to apportion restitution. 18 U.S.C. § 3664(h). The coconspirators all worked together to cause the losses, and the district court decided they are all responsible for paying them back. Nothing shows that the court abused its discretion in applying joint and several liability. See United States v. Hunt , 521 F.3d 636, 649 (6th Cir. 2008).

For these reasons, we AFFIRM the district court’s judgments.