Sixth Circuit: Confrontation Clause Permits Expert’s Independent Interpretation of Team-Processed Data (No Right to Cross-Examine Data Validators)

Introduction

United States v. Raef Hamaed (6th Cir. June 12, 2026) arises from a years-long pharmacy scheme spanning five pharmacies in Michigan and Ohio—Harper Drugs, Wayne Campus Pharmacy, LLC (“Wayne Campus”), Heartland Pharmacy, LLC (“Heartland 1”), Heartland Pharmacy 2 (“Heartland 2”), and Eastside Pharmacy (“Eastside”). The government alleged that multiple pharmacist-owners and operators billed Medicare/Medicaid and other insurers for prescriptions they did not dispense, concealed shortages through forged signatures, waived copays to increase volume, and at one location substituted generics while billing brand-name drugs.

After trial, a jury convicted Raef Hamaed, Kindy Ghussin, Ali Abdelrazzaq, and Tarek Fakhuri of conspiracy to commit healthcare fraud and wire fraud under 18 U.S.C. § 1349, and convicted Fakhuri of one substantive healthcare fraud count under 18 U.S.C. §§ 1347 and 2 (Abdelrazzaq’s substantive convictions were later set aside). On appeal, defendants raised issues centered on (1) the Confrontation Clause and expert testimony based on an audit-style “invoice review” performed with team support; (2) limits on defense evidence (legitimate scripts; PBM bias); (3) judicial “expert” bolstering; (4) single vs. multiple conspiracies; and (5) sentencing/restitution methodologies (loss, sophisticated means, MVRA procedure).

Summary of the Opinion

The Sixth Circuit affirmed all convictions and challenged sentencing rulings. The court held that:

  • Qlarant director Johanna Sullivan’s testimony did not violate the Confrontation Clause because she gave an independent interpretation of admissible records and did not act as a surrogate “mouthpiece” for non-testifying analysts.
  • Excluding evidence that defendants also filled legitimate prescriptions, and excluding broad “PBM industry bias” evidence, did not violate the right to present a complete defense.
  • Although it was error to repeatedly refer to Sullivan as an “expert” in front of the jury, defendants failed to show the plain-error prejudice required for reversal.
  • Trial proof supported a single overarching conspiracy, so there was no variance from the indictment.
  • The district court did not abuse its discretion in admitting cooperator Abdallah’s lay interpretations of text messages, denying a sweeping Rule 17(c) subpoena request, rejecting juror-impeachment efforts, calculating loss, applying sophisticated means (Ghussin), and imposing restitution (including joint-and-several liability).

Analysis

Precedents Cited

1) Confrontation Clause framework and “testimonial hearsay” limits

The court grounded its analysis in the modern Confrontation Clause line beginning with Crawford v. Washington, emphasizing that only “testimonial hearsay” triggers the Clause’s bar. It drew the two-part formulation from Smith v. Arizona—the statement must be (i) testimonial and (ii) hearsay. It then used:

  • Melendez-Diaz v. Massachusetts (government cannot replace live testimony with a document memorializing an analyst’s opinion; chain-of-custody gaps typically go to weight, not admissibility).
  • Bullcoming v. New Mexico (no “surrogate” analyst to present another analyst’s testimonial assertions).
  • Ohio v. Clark and Michigan v. Bryant (primary-purpose test; consider surrounding circumstances).
  • Davis v. Washington (testimonial/non-testimonial boundary) and Anderson v. United States (hearsay defined as offered for truth).
  • Delaware v. Van Arsdall, Davis v. Alaska, and Pointer v. Texas (centrality and functions of cross-examination).
  • Delaware v. Fensterer (Confrontation Clause guarantees an opportunity for effective cross-examination, not cross-examination “effective” to the defense’s preferred extent).

The court also cited United States v. Contreras (4th Cir. 2025) to reinforce that an expert’s “exposure to testimonial hearsay” does not automatically bar testimony if the expert provides an “independent assessment.”

2) Right to present a complete defense and evidentiary limits

Using Holmes v. South Carolina and United States v. Scheffer, the court applied the “arbitrary or disproportionate” plus “weighty interest” test, and emphasized that evidentiary rules may exclude confusing/misleading evidence under Rule 403. It relied on:

  • United States v. Reichert (standard of review) and United States v. Reynolds (Rule 403 as legitimate limit).
  • United States v. Odeh (scope of the defense right).
  • Taylor v. Illinois (right is not unfettered).
  • United States v. Dimora, United States v. Ifediba, United States v. Betro, and United States v. Daulton (no defense that some conduct was lawful when the charged fraud theory does not require “ceaseless criminal conduct”).
  • McKee Foods Corp. v. BFP Inc. (PBM anticompetitive practices context), used to frame but not validate the speculative “PBM bias” theory.

3) Improper “expert” bolstering—error but no plain-error reversal

The panel reiterated Sixth Circuit caution that parties should not tender a witness “as an expert” in front of a jury and courts should not declare an expert “qualified” in a way that signals judicial endorsement, relying on United States v. Campbell, United States v. Johnson, and United States v. Maya. Nonetheless, under Greer v. United States and the prejudice standard in Molina-Martinez v. United States, defendants failed to show a reasonable probability of a different outcome given limiting instructions and other evidence (with a harmlessness comparison to United States v. Majors and mitigation reasoning akin to United States v. Sibley).

4) Variance—single vs. multiple conspiracies

The court applied longstanding Fifth Amendment principles from Stirone v. United States and United States v. Miller, and Sixth Circuit variance standards from United States v. Siefert, United States v. Guerrero, and United States v. Robinson. For the number-of-conspiracies inquiry, it used the three-factor approach from United States v. Hughes (common goal, nature of scheme, overlapping participants) and rejected a “subdividing” argument under United States v. Beals.

5) Evidence law: lay opinion; Rule 17(c); juror impeachment

  • Lay interpretation of messages: the court relied on Rule 701 principles and cases including United States v. Freeman and United States v. Reed to hold cooperator Abdallah could interpret texts based on personal experience; it distinguished United States v. Kaplan (2d Cir. 2007) due to Abdallah’s extensive relationship with Abdelrazzaq. (Standard of review from United States v. Fox.)
  • Rule 17(c) subpoenas: applying United States v. Nixon, with Sixth Circuit support from United States v. Theunick and United States v. Llanez-Garcia, the court rejected a broad, speculative wholesaler subpoena as a “fishing expedition.”
  • Juror impeachment: the court invoked McDonald v. Pless and United States v. Gonzales to bar using a juror letter complaining of confusing instructions to impeach the verdict.

6) Sentencing loss, relevant conduct, sophisticated means, and restitution

  • Reasonableness framework: Gall v. United States plus Sixth Circuit gloss from United States v. Rayyan, United States v. Tristan-Madrigal, and United States v. Boucher.
  • Loss methodology: the court emphasized “reasonable estimate” standards (including United States v. Wala and clear-error framing from United States v. Matthews), and approved reliance on audit-style comparisons where defendants offered no concrete rebuttal (cf. United States v. Washington).
  • Relevant conduct for conspiracy loss: it applied U.S.S.G. § 1B1.3 and Sixth Circuit authority like United States v. Donadeo and foreseeability reasoning akin to United States v. Kennedy.
  • Sophisticated means: using United States v. Simmerman, United States v. Crosgrove, and the then-recent United States v. Yousef, the court held falsified patient-signature practices supporting audit evasion can qualify as “sophisticated means.”
  • Restitution: applying United States v. Clay and United States v. Fike (authority), the court addressed presence/procedure concerns via United States v. Hayden and especially United States v. Vandeberg (MVRA procedure; hearing not mandated; harmlessness where opportunity to object exists). It also cited United States v. Simpson (loss and restitution often align when based on actual loss), United States v. Kilpatrick (no “exact precision” requirement), and United States v. Hunt (joint-and-several discretion).

Legal Reasoning

A. The operative “new” rule: post-Smith Confrontation Clause boundary for team-based expert work

The opinion’s central doctrinal contribution is its practical application of Smith v. Arizona to large-scale, team-processed audit data: the Confrontation Clause is not violated when a testifying expert personally analyzes, quality-checks, and interprets admitted underlying data—even if other staff collected, extracted, validated, or input that data—so long as the expert does not transmit a specific absent analyst’s testimonial assertions for their truth.

The court’s reasoning proceeds in three steps:

  1. Identify the supposed out-of-court “statements.” Following Smith v. Arizona, the panel first required defendants to pinpoint the testimonial hearsay allegedly conveyed. They could not identify any concrete statements by Qlarant analysts that Sullivan repeated.
  2. Separate “process testimony” from “surrogate testimony.” The panel treated Sullivan’s discussion of Qlarant’s general procedures (how Qlarant “typically functioned,” its “standards, practices, and procedures”) as permissible background based on personal knowledge. It distinguished impermissible surrogacy (as in Smith v. Arizona and Bullcoming v. New Mexico) from testimony about one’s own review and interpretation.
  3. Reframe “validation” disputes as weight, not confrontation. Even if other team members validated data completeness, defendants’ inability to cross-examine those validators did not create a constitutional bar. Relying on Melendez-Diaz v. Massachusetts (chain-of-custody and similar foundational gaps usually go to weight), and Delaware v. Fensterer (opportunity, not ideal cross-examination), the court held defendants could probe the issue through Sullivan, who performed the “heart” of the analysis: comparing purchases to billed claims and interpreting shortages.

    Notably, defendants had stipulated to admissibility of the key records and the invoice review results, narrowing the dispute to whether Sullivan’s testimony “incorporated” non-testifying analysts’ opinions. The panel treated that claim as inferential and unsupported by actual testimonial hearsay in the record.

    B. Excluding “we also did lawful work” and “PBMs are biased” defenses

    The opinion reinforces a recurring fraud-trial distinction: where the government does not allege that every transaction is fraudulent, defendants generally cannot rebut the charge by proving some subset was legitimate. Citing United States v. Ifediba and United States v. Betro, the panel reasoned that evidence of legitimate prescriptions would likely confuse jurors (Rule 403) while offering little probative value against a scheme-based theory.

    On “PBM bias,” the panel accepted the general proposition that bias evidence can be relevant (Delaware v. Van Arsdall), but found the proffer too speculative and attenuated: defendants offered no evidence PBMs manipulated the particular claims data at issue, and the theory implied an implausible multi-PBM framing conspiracy.

    C. “Expert” bolstering: acknowledged error constrained by plain-error prejudice

    The panel candidly held that the district court should not have repeatedly called Sullivan an “expert” before the jury, consistent with United States v. Johnson, United States v. Maya, and United States v. Campbell. But under Greer v. United States, the defendants could not show outcome-changing prejudice, especially given limiting instructions that jurors were free to reject Sullivan’s opinions and the existence of substantial independent evidence (cooperator testimony and communications among conspirators).

    D. Single conspiracy finding and sentencing/restitution deference

    On variance, the panel applied the usual factors (common profit goal, shared modus operandi, and overlapping personnel/ownership) and found the evidence supported one conspiracy despite different ownership slices. This is a pragmatic, enterprise-style view of conspiracy that emphasizes functional interdependence (profit sharing; shared methods; shared relationships) over formal corporate boundaries.

    On sentencing, the panel reaffirmed that loss determinations can be “reasonable estimates” and need not be precise (United States v. Wala), especially when the defense offers speculation rather than concrete counterproof. For sophisticated means, it treated audit-evasion through falsified patient signatures and records as sufficient concealment sophistication under United States v. Crosgrove and United States v. Yousef.

    On restitution, the panel’s key holding is procedural: even if the district court entered judgment including restitution before fully resolving objections in the manner contemplated, any error was harmless under United States v. Vandeberg where the defendant had a meaningful opportunity to contest the pivotal issue (actual loss) at sentencing and again through post-judgment motion practice.

    Impact

    • Healthcare fraud prosecutions (audit analytics). The opinion is likely to be cited to defend the admissibility of government experts whose conclusions rely on large datasets assembled by teams (PBMs, CMS contractors, forensic units), so long as the testifying witness performs the key interpretive analysis and does not introduce specific testimonial hearsay statements from non-testifying analysts.
    • Defense strategy shift. The decision channels challenges away from confrontation and toward evidentiary weight: defendants should marshal concrete examples of missing invoices, incorrect inputs, or alternative calculations, rather than arguing that every upstream handler must testify.
    • Trial management and “expert” labels. While the panel labels the “expert” designation in front of the jury as error, its prejudice analysis suggests reversals will be uncommon absent a close case or absent limiting instructions—potentially reducing incentives to litigate this issue unless the record can show a decisive impact.
    • Conspiracy charging across multi-entity businesses. The court’s single-conspiracy analysis supports broader indictments where participants share profit distribution and methods, even if day-to-day operations or ownership vary by location.
    • MVRA practice. The restitution discussion reinforces that MVRA procedures are flexible and that harmless-error principles can sustain restitution even when the district court’s sequencing is imperfect—especially where loss was litigated in substance.

    Complex Concepts Simplified

    Confrontation Clause
    The Sixth Amendment generally requires the prosecution to present witnesses in court for cross-examination. But it applies mainly to testimonial hearsay—out-of-court statements created for use at trial and offered for their truth.
    “Surrogate” expert vs. independent expert analysis
    A surrogate expert repeats another analyst’s conclusions (barred by Bullcoming v. New Mexico and Smith v. Arizona). An independent expert uses admissible data to form and testify to their own conclusions (permitted here).
    Rule 403
    A judge can exclude even relevant evidence if it is likely to confuse the issues or mislead the jury. Here, proof of legitimate prescriptions risked distracting from whether defendants also ran a fraudulent billing scheme.
    Variance (single vs. multiple conspiracies)
    A variance occurs if the indictment charges one conspiracy but the proof shows several unrelated conspiracies. The court looks for a shared goal, shared scheme features, and overlapping participants.
    Rule 17(c) subpoenas
    They are not broad discovery tools. Under United States v. Nixon, the requesting party must show relevance, necessity, and specificity, and cannot merely speculate that helpful material might exist.
    Loss amount (Guidelines) vs. restitution (MVRA)
    Guidelines “loss” can involve intended loss; MVRA restitution is tied to actual loss. When the court’s loss finding is based on actual loss, the two numbers often converge (as noted with United States v. Simpson).

    Conclusion

    United States v. Raef Hamaed clarifies a post-Smith v. Arizona Confrontation Clause boundary important to modern data-driven fraud prosecutions: the Sixth Circuit will allow a testifying expert to present conclusions drawn from large datasets assembled by a team, provided the expert offers an independent analysis and does not transmit specific testimonial assertions from absent analysts. The decision also reaffirms limits on “complete defense” claims (no meaningful defense in proving some lawful conduct where the scheme includes unlawful conduct), sustains broad single-conspiracy proof across multi-location businesses, and underscores the judiciary’s latitude in loss and MVRA restitution procedures, constrained primarily by reasonableness and meaningful opportunity to contest the key factual predicates.