Independent Expert Interpretation of Third-Party-Processed Data Does Not Violate the Confrontation Clause (Sixth Cir.)

I. Introduction

In United States v. Raef Hamaed (6th Cir. June 12, 2026) (recommended for publication), the Sixth Circuit affirmed the convictions and most sentencing determinations arising from a long-running healthcare-fraud scheme involving multiple pharmacists and five pharmacies in Michigan and Ohio. The scheme centered on billing insurers (including Medicare and Medicaid) for prescriptions that were not dispensed, using tactics such as failing to reverse claims for unpicked prescriptions, waiving copays to drive volume, substituting generics while billing for brand-name drugs, and falsifying patient signatures to survive audits.

The appeal raised a cluster of criminal-procedure issues, but the opinion’s most consequential analysis concerns the Confrontation Clause limits on expert testimony that relies on data collected, validated, or processed by non-testifying personnel—an increasingly common feature of modern fraud investigations that depend on large-scale claims data and analytics.

II. Summary of the Opinion

  • Confrontation Clause: No violation where Qlarant director Johanna Sullivan testified to her own analysis and interpretation of wholesaler and claims data, even though other team members had extracted and “verified” underlying data; defendants identified no specific out-of-court “statements” she conveyed for their truth.
  • Right to present a complete defense: No constitutional error in excluding (i) evidence that defendants also legitimately filled/billed some prescriptions (not a defense to a scheme not alleged to be “ceaseless” fraud), and (ii) generalized evidence of PBM industry bias/anticompetitive concerns (too speculative and marginally relevant).
  • “Expert” labeling before the jury: The district court and government erred by repeatedly calling Sullivan an expert in front of the jury, but defendants failed to show plain-error prejudice given limiting instructions and substantial independent evidence of guilt.
  • Variance (single vs. multiple conspiracies): Evidence supported one overarching conspiracy across pharmacies (common goal, common methods, overlapping participants), so no fatal variance.
  • Text-message interpretation: Co-conspirator Abdallah’s interpretations of texts were admissible as lay opinions under Rule 701 because they were rationally based on his perceptions and relationship with the sender.
  • Rule 17(c) subpoena: Denial of broad wholesaler subpoena affirmed as a speculative fishing expedition lacking specificity and necessity.
  • Juror letter: Post-verdict juror comment about “confusing” instructions could not impeach the verdict.
  • Sentencing: Loss calculations based on Qlarant’s invoice review upheld; inclusion of a pharmacy’s losses against a defendant affirmed under relevant conduct principles; sophisticated-means enhancement upheld for falsifying records/signatures to defeat audits.
  • Restitution: Even assuming procedural error in timing/pronouncement, any error was harmless where defendant had meaningful opportunities to contest the loss figure; joint-and-several restitution upheld.

III. Analysis

A. Precedents Cited

1. Confrontation Clause framework: “testimonial hearsay” and expert “surrogates”

The court situates the dispute within the modern Confrontation Clause line beginning with Crawford v. Washington, which limits the Clause to testimonial statements and rejects broad reliability-based exceptions. Building from Melendez-Diaz v. Massachusetts and Bullcoming v. New Mexico, the panel emphasizes that the government cannot evade confrontation by presenting an expert who merely relays another analyst’s testimonial assertions or results.

The opinion’s key interpretive lens is Smith v. Arizona, which the panel treats as the controlling articulation of the “two limits” of “testimonial hearsay”: (i) the statement must be testimonial, and (ii) it must be hearsay (offered for its truth). The Sixth Circuit adopts Smith v. Arizona’s insistence that courts must first “pinpoint” the out-of-court statements allegedly conveyed, and that the Confrontation Clause protects the right to cross-examine the declarant, not merely a witness who repeats the declarant’s assertions.

The opinion also relies on:

  • Davis v. Washington (testimonial limitation) and Ohio v. Clark / Michigan v. Bryant (primary-purpose test and contextual inquiry) for identifying what is “testimonial.”
  • Anderson v. United States and Fed. R. Evid. 801(c) for the hearsay “truth of the matter asserted” requirement.
  • Delaware v. Van Arsdall, Davis v. Alaska, and Pointer v. Texas to restate the core function of confrontation as meaningful cross-examination.
  • Delaware v. Fensterer to underline that the Sixth Amendment guarantees an opportunity for effective cross-examination, not the most effective cross-examination the defense might desire.
  • United States v. Contreras (4th Cir.) as persuasive authority that an expert’s “exposure to testimonial hearsay” does not itself bar the expert’s independent assessment.

2. Right to present a complete defense

The court frames the issue using Holmes v. South Carolina (meaningful opportunity to present a defense, tempered by evidence rules), United States v. Reichert (de novo review), and United States v. Odeh (right to competent, reliable exculpatory evidence). It then invokes Taylor v. Illinois and United States v. Reynolds to reaffirm that evidentiary exclusions with legitimate purposes—especially Fed. R. Evid. 403—do not automatically become constitutional violations.

For the constitutional test, the panel uses United States v. Scheffer: the defendant must show the exclusion was “arbitrary or disproportionate” and infringed a “weighty interest.”

3. “Expert” vouching/bolstering before the jury

The Sixth Circuit relies on its own internal admonitions against “tendering” a witness as an expert in front of the jury: United States v. Campbell, United States v. Johnson, and United States v. Maya. Yet the claim fails on plain-error prejudice analysis drawn from Greer v. United States and the “reasonable probability” standard of Molina-Martinez v. United States, with United States v. Sibley used to illustrate how curative instructions can mitigate the risk.

4. Conspiracy variance

The variance standard is developed through United States v. Robinson, United States v. Guerrero, and United States v. Siefert. The substantive multi-factor test—common goal, nature of the scheme, overlapping participants—is taken from United States v. Hughes, and the court cites United States v. Beals to reject the notion that subdividable ownership interests require a finding of multiple conspiracies.

5. Lay opinion on coded/ambiguous communications

On Abdallah’s interpretations of texts, the court applies Fed. R. Evid. 701 through United States v. Freeman and United States v. Reed, distinguishing the Second Circuit’s United States v. Kaplan (lack of personal experience with the defendant) because Abdallah had a lengthy relationship and participated in the conversations.

6. Subpoenas and post-verdict juror impeachment

The Rule 17(c) analysis follows United States v. Nixon (relevance, necessity, specificity, good faith; no fishing expeditions) as applied in Sixth Circuit cases like United States v. Theunick and United States v. Llanez-Garcia.

The juror-letter issue is resolved under the longstanding bar on juror impeachment of verdicts in McDonald v. Pless and United States v. Gonzales.

7. Sentencing and restitution

On sentencing, the court uses Gall v. United States for the overarching abuse-of-discretion framework and distinguishes procedural and substantive reasonableness with Sixth Circuit standards like United States v. Rayyan, United States v. Tristan-Madrigal, United States v. Boucher, and the within-Guidelines presumption described in Gall v. United States.

The loss-amount principles (reasonable estimate, methodology review) are developed via United States v. Betro, United States v. Wala, United States v. Matthews, and again United States v. Siefert, while relevant-conduct attribution is handled via United States v. Donadeo and United States v. Kennedy.

The sophisticated-means enhancement is anchored in United States v. Simmerman, United States v. Yousef, United States v. Crosgrove, and the “totality of the scheme” approach referenced through United States v. Betro.

On restitution, the panel draws the statutory boundary from United States v. Fike and the review standards from United States v. Clay. It relies heavily on United States v. Vandeberg for MVRA procedure, while using United States v. Hayden to acknowledge the due-process right to be present at sentencing. It then explains the Guidelines/MVRA distinction with United States v. Simpson, United States v. Kilpatrick, and affirms joint-and-several authority through United States v. Hunt.

B. Legal Reasoning

1. The central holding: Confrontation Clause requires identifying an actual out-of-court “statement” conveyed for its truth

The court’s Confrontation Clause analysis is method-driven: before debating “surrogacy,” it requires defendants to identify the out-of-court testimonial statements Sullivan allegedly transmitted. That step, drawn from Smith v. Arizona, becomes dispositive because defendants pointed to no concrete statements—only Sullivan’s generalized explanation of how Qlarant typically functions and a cross-examination exchange in which she said the data “had already been verified.”

The panel treats that “already been verified” remark not as the transmission of an analyst’s testimonial assertion, but as (at most) contextual information explaining why Sullivan did not personally redo validation work—without her becoming a “mouthpiece” for another’s substantive conclusion. Critically, the court distinguishes between:

  • testimonial hearsay conveyed for its truth (barred absent confrontation), and
  • an expert’s independent interpretation of admissible records, including records collected/processed by others (permissible).

2. Weight vs. admissibility: Data completeness challenges go to credibility, not confrontation

The panel emphasizes that even if completeness/chain-of-custody concerns matter, they typically affect evidentiary weight, not admissibility—invoking Melendez-Diaz v. Massachusetts’s footnote distinguishing confrontation from “everyone who laid hands on the evidence.” With defendants stipulating to admissibility of the underlying data and invoice review, the remaining dispute was whether Sullivan had to personally validate completeness for confrontation purposes. The Sixth Circuit answers no: defendants could attack reliability through cross-examining Sullivan about her process and involvement, consistent with Delaware v. Fensterer.

3. Distinguishing prohibited “surrogate analysts” (Smith/Bullcoming) from permissible “independent reviewers”

The court’s practical line-drawing is that Sullivan was not a mere surrogate because she “touched everything that went to the jury”: she reviewed inputs, ensured correct timeframes, confirmed data population into SAS, performed the final quality check, interpreted output shortages, and helped create and verify charts. In Smith v. Arizona and Bullcoming v. New Mexico, the testifying witnesses recounted another person’s testing and results; here, Sullivan testified to her own analysis of business records and software output, even if others performed earlier or clerical steps.

4. Constitutional defense evidence: legitimacy is not exculpation where the theory is “some fraud,” not “all fraud”

On the “legitimate prescriptions” issue, the court reasons from relevance and confusion concerns under Rule 403, but also from the logical structure of the charged conduct. Where the government does not allege that every claim was fraudulent, proof that some claims were legitimate does not negate an element of conspiracy to commit fraud. The panel relies on analogous reasoning in United States v. Ifediba, United States v. Betro, and United States v. Daulton: lawful conduct does not immunize unlawful conduct, and “good acts” are not a defense absent a charge of wholly criminal operation.

5. PBM “bias” evidence: the court rejects speculation piled on speculation

The attempt to introduce general distrust of PBMs and alleged market conflicts failed because defendants could not tie it to altered data or manipulation in this case. The court treated the proffer as inviting the jury to speculate that multiple PBMs conspired to frame defendants to gain competitive advantage. With no evidentiary bridge to the records at issue, exclusion was not “arbitrary or disproportionate” under United States v. Scheffer.

6. “Expert” labeling error: acknowledged, then defeated on plain-error prejudice

The Sixth Circuit candidly labels it “plain” error for the judge and prosecutor to call Sullivan an expert in front of the jury (per United States v. Johnson, United States v. Maya, and United States v. Campbell), because it risks judicial endorsement. But under Greer v. United States and Molina-Martinez v. United States, defendants must show a reasonable probability of a different outcome. The court finds none given limiting instructions and substantial non-Sullivan evidence (co-conspirator testimony and communications).

7. Sentencing and restitution: deference to supported estimates and MVRA procedure flexibility

For loss, the court reiterates that district courts need a “reasonable estimate” using “available information,” not precision. Defendants’ attacks on the invoice-review methodology failed because they offered speculation rather than counter-estimates or concrete missing records. The Ghussin-specific relevant conduct issue turned on foreseeability and profit-sharing benefit from Wayne Campus.

For restitution, the court treats the MVRA as a procedural framework that gives district courts flexibility (per United States v. Vandeberg). Even if the district court prematurely entered judgment including restitution, any error was harmless because the parties litigated actual loss at sentencing and again via motion practice, and the MVRA does not demand an oral restitution hearing in every case.

C. Impact

  • Data-driven prosecutions: The opinion strengthens the government’s ability to present analytics-based expert testimony where large datasets are gathered and processed by teams, so long as the testifying expert offers an independent interpretation and does not convey specific testimonial hearsay statements for their truth.
  • Defense strategy recalibration: Challenges to “completeness” or “integrity” of underlying records are framed as classic weight/credibility arguments (cross-examination, competing experts, counter-records), not automatic Confrontation Clause bars.
  • Trial practice warning: Despite affirmance, the court reiterates that “tendering” experts before the jury is improper and can become reversible if prejudice is shown—an important caution for district judges and prosecutors.
  • Scope of “complete defense” claims: The decision underscores that constitutional defense rights do not force admission of marginally relevant “industry critique” evidence untethered to case facts, and that “we did some legitimate business” evidence is often inadmissible where the charged scheme contemplates mixed lawful/unlawful activity.
  • Sentencing loss and restitution: The court continues a deferential posture to well-supported loss estimates and aligns restitution to the same actual-loss figure when that is what drove the Guidelines determination.

IV. Complex Concepts Simplified

  • Confrontation Clause: The Sixth Amendment generally requires that if the prosecution uses a witness’s testimonial statement against the defendant, the defendant must be able to cross-examine the person who made that statement.
  • Testimonial hearsay: A statement (1) made primarily for use in a criminal prosecution (testimonial), and (2) offered to prove what it says is true (hearsay).
  • “Surrogate expert” problem: The government cannot call an expert to simply repeat another analyst’s conclusions; the analyst who did the testimonial work generally must testify.
  • Weight vs. admissibility: Some problems (missing records, imperfect chain of custody, incomplete data) usually go to how persuasive evidence is (weight), not whether it can be presented at all (admissibility).
  • Variance (in conspiracy cases): A mismatch between what the indictment charged (e.g., one conspiracy) and what the trial evidence proved (e.g., multiple separate conspiracies). Not every mismatch is fatal; defendants must show the evidence can only support multiple conspiracies.
  • Rule 17(c) subpoena: A tool to obtain specific documents for trial, not a broad discovery device; courts can deny requests that are speculative or overly broad.
  • MVRA restitution: For fraud offenses, restitution is mandatory and aims to repay victims’ actual losses; procedures can be handled through briefing and post-sentencing processes if needed, so long as parties have a meaningful chance to be heard on disputed issues.

V. Conclusion

United States v. Raef Hamaed provides a publication-worthy clarification of confrontation principles in the era of team-based data analytics: the Constitution does not require the government to produce every person who helped extract or validate datasets when the testifying expert gives an independent analysis and does not transmit identifiable testimonial hearsay for its truth. The opinion simultaneously cautions trial courts against “expert”-label vouching in front of juries, reinforces limits on “complete defense” arguments grounded in marginal relevance or speculation, and affirms sentencing/restitution outcomes where defendants fail to rebut reasonable loss estimates with concrete proof.