WALLACE, Circuit Judge

Leah Spero (argued), Spero Law Office, San Francisco, California, for Defendant-Appellant.

Joshua Halpern (argued), Attorney, United States Department of Justice, Washington, D.C.; Merry Jean Chan, Chief, Appellate Section; David L. Anderson, United States Attorney; United States Attorney's Office, San Francisco, California; for Plaintiff-Appellee.

Before: J. Clifford Wallace and Ryan D. Nelson, Circuit Judges, and James S. Gwin, District Judge.

WALLACE, Circuit Judge: David Lague, a former physician's assistant, was convicted of thirty-nine counts of distributing controlled substances outside the usual course of professional practice and without a legitimate medical purpose to five of his former patients, in violation of 21 U.S.C. § 841(a)(1), (b)(1)(c), and (b)(2). Lague appeals from his judgment of conviction, arguing that the district court erred in allowing the government to present evidence of his uncharged practice-wide prescriptions. We have jurisdiction under 28 U.S.C. §§ 1291 and 1294, and we affirm.

I.

In 2007, Lague began working as a physician's assistant at a chronic pain-management medical practice in San Leandro, California. He was licensed to prescribe controlled substances including opioids.

In 2016, after Lague's patient SL was arrested for possession with the intent to distribute opioids that Lague had prescribed to him, SL agreed to cooperate with the Drug Enforcement Administration's (DEA) investigation into the clinic.

At the direction of the DEA, SL recorded his future visits to the clinic. During one visit in 2016, SL offered cash to Lague in exchange for doubling his prescription for oxycodone. Lague wrote the double prescription, falsely recording in his patient notes that SL had asked for it simply to save money on his copay for the following month. Lague and SL discussed how Lague would write the prescription to avoid scrutiny from the pharmacy. SL would fill the prescription at one pharmacy but would refill his prescription the next month at a different pharmacy.

In his patient notes, Lague claimed to monitor SL's compliance through urine testing. But the urine tests revealed that SL had not been taking any of his prescriptions. Lague never confronted SL about the negative urine test results, and falsely wrote in his notes that SL was following his opioid agreement.

In March 2017, the DEA executed a search warrant at the clinic, seizing over one hundred patient files. Based on those patient files and on SL's recordings, the government charged Lague with thirty-nine counts of unlawfully distributing Schedule II and Schedule IV controlled substances to five former patients: SL, DL, KO, JF, and MCM. The government also charged Lague with seven counts of healthcare fraud and conspiracy to commit healthcare fraud for unlawfully prescribing fentanyl to MCM.

At trial, both parties presented a medical expert. The government called Dr. Charles Szabo. Lague called Dr. Gary Martinovsky. The experts opined on whether the charged prescriptions were within the usual course of professional practice.

The experts focused on Lague's charged prescriptions, testifying about various pain-management standards from the California Medical Board Guidelines, the American Pain Society Guidelines, and the Center for Disease Control and Prevention Guidelines. These guidelines provide recommended prescribing amounts based on generally accepted medical standards.

Medical standards also warn of the risks of consuming controlled substances in certain combinations. For example, drug addicts combine opioids like oxycodone and hydrocodone with a benzodiazepine for an enhanced but dangerous "high." Drug addicts may take this combination with a muscle relaxant, forming the "holy trinity," for an even more dangerous "high." The "holy trinity" of drugs rarely serves a legitimate medical purpose.

To monitor patients’ pill-seeking behaviors, medical professionals perform urinalysis testing. The testing is designed to detect the consumption of unprescribed substances (a sign of drug addiction), and the nonconsumption of prescribed medications (a sign of illegal sales). Professionals also rely on other warning signs such as a patient seeking an early prescription refill.

At trial, the government presented evidence that Lague had prescribed enormous quantities of controlled substances in dangerous combinations to the five patients covered by the Second Superseding Indictment. The government presented the recordings of SL's visits. The government also presented the patient files of Lague's five patients. Two of Lague's former patients testified at trial, corroborating SL's testimony that Lague had falsified patient files and had not examined patients before prescribing controlled substances.

The government also introduced Lague's statements in his interview with the DEA and his testimony before the grand jury. In his interview with the DEA, Lague said that he did not want to be a "policeman" with his patients. He also said that it was "possible" that he had falsified his patients’ files. Before the grand jury, Lague acknowledged that the level of opiates prescribed at the clinic, especially starting in 2015, was higher than appropriate.

In addition to this patient-specific evidence, the government introduced Lague's practice-wide prescription data from 2015 and 2016 to show how Lague's prescription levels compared to that of other opioid prescribers, including Dr. Martinovsky. The data concerned Lague's prescriptions for 458 patients unrelated to the Second Superseding Indictment. The prescription data showed that Lague had prescribed opioids at among the highest rates compared to other pain management prescribers in California.

Robert Gibbons testified about the prescription data. For his testimony, Gibbons, a statistician at the U.S. Department of Health and Human Services, relied on Medicare's Integrated Data Repository. Gibbons presented a series of charts comparing Lague's practices to three groups of practitioners: providers who prescribe opioids to over 50 Medicare patients, providers specializing in pain management and anesthesiology, and providers specializing in cancer treatment. The data showed that in 2016, Lague issued more opioids than any other Medicare prescriber in California. Gibbons testified that Lague's prescription data "made him an outlier" and that Lague's prescribed opioids were "quite a bit higher" than 99 percent of prescribers Gibbons compared.

Paul Short also testified about Lague's practice-wide prescription data. Short relied on California's Controlled Substances Utilization Review and Evaluation System (CURES), an aggregator of controlled substances filled by California pharmacies. Short presented charts that showed that Lague had prescribed 1.4 million Schedule II pills in 2016, that Lague's methadone and oxycodone prescriptions exceeded the maximum recommended dosages, that Lague often prescribed a combination of opioids and benzodiazepines, and that Lague prescribed the "holy trinity" to some of his patients.

The government recalled Short as a rebuttal witness to compare Lague's prescription practices with those of Lague's expert, Dr. Martinovsky using the CURES data. The rebuttal testimony showed that Lague's prescription amounts dwarfed Dr. Martinovsky's.

After the conclusion of the trial in July 2018, the jury found Lague guilty of the unlawful distribution charges. Lague was convicted for doubling SL's opioid prescriptions so that he could sell the excess, and for prescribing controlled substances in enormous quantities and dangerous combinations to DL, KO, MCM, and JF. Lague was acquitted of the healthcare fraud charges. The district court sentenced Lague to 120 months imprisonment for his unlawful Schedule II prescriptions and to 60 months imprisonment for his unlawful Schedule IV prescriptions, to be served concurrently. This appeal followed.

II.

We review the question whether specific evidence falls within the scope of Federal Rule of Evidence 404(b) de novo. See United States v. Carpenter , 923 F.3d 1172, 1180–81 (9th Cir. 2019) (citation omitted). We review the district court's admission of "other act" evidence for an abuse of discretion. Id. If the district court abuses its discretion under Rule 403, we ask "whether the government successfully bore its burden of proof that the error in admitting the evidence was harmless." United States v. McElmurry , 776 F.3d 1061, 1070 (9th Cir. 2015).

III.

A.

It is generally "unlawful for any person knowingly or intentionally ... to manufacture, distribute, or dispense, or possess with intent to manufacture, distribute, or dispense, a controlled substance." 21 U.S.C. § 841(a)(1). A medical professional's prescription of a controlled substance is lawful only if "issued for a legitimate medical purpose by an individual practitioner acting in the usual course of his professional practice." 21 C.F.R. § 1306.04 ; see also United States v. Moore , 423 U.S. 122, 124, 96 S.Ct. 335, 46 L.Ed.2d 333 (1975).

Lague argues that the district court erred, under Federal Rule of Evidence 404(b), by granting the government's motion in limine to present data of his practice-wide prescriptions. He contends that these uncharged prescriptions do not support an inference that he intended to write the charged prescriptions outside the usual course of professional practice and without a legitimate medical purpose.

We begin with the text of Federal Rule 404(b). See United States v. Boulware , 384 F.3d 794, 807 (9th Cir. 2004). Under Federal Rule 404(b), "[e]vidence of a crime, wrong, or other act is not admissible to prove a person's character in order to show that on a particular occasion the person acted in accordance with the character." Fed. R. Evid. 404(b)(1). But other act evidence may be admissible to prove "motive, opportunity, intent, preparation, plan, knowledge, identity, absence of mistake, or lack of accident." Fed. R. Evid. 404(b)(2).

We apply a four-part test to determine whether "other act" evidence is admissible. A district court may admit other act evidence if: (1) the evidence tends to prove a material point; (2) the other act is not too remote in time; (3) the evidence is sufficient to support a finding that defendant committed the other act; and (4) (in certain cases) the act is similar to the offense charged. See United States v. Bailey , 696 F.3d 794, 799 (9th Cir. 2012) (citation omitted). The government "has the burden of proving that the evidence meets all of the above requirements." United States v. Arambula-Ruiz , 987 F.2d 599, 602 (9th Cir. 1993) (citation omitted).

Lague argues that the government failed to meet its burden under the first part of our Rule 404(b) analysis, i.e. , that his practice-wide evidence did not tend to prove a point material to the unlawful distribution charges because there was no evidence that those underlying prescriptions were issued unlawfully. We agree with Lague that, under Rule 404(b), "the government ... bears the burden of proving a logical connection between appellant's purported involvement in the previous [act] and a material fact at issue in the crime with which he was charged." United States v. Mayans, 17 F.3d 1174, 1183 (9th Cir. 1994).

As is relevant here, the "material fact at issue" is whether Lague intended to prescribe controlled substances to the five patients covered by the Second Superseding Indictment without a legitimate medical purpose. United States v. Rendon-Duarte , 490 F.3d 1142, 1144–45 (9th Cir. 2007). If Lague's aberrational prescription data is probative of his intent to prescribe the underlying, uncharged prescriptions without a legitimate medical purpose, there is a logical connection between the "other" prescriptions and the charged prescriptions.

But we have not yet decided whether a medical professional's practice-wide prescription data is probative of unlawful intent in a section 841 charge. We therefore now turn to our sister circuits for guidance.

The government relies on the Eleventh Circuit's decision, United States v. Merrill, 513 F.3d 1293 (11th Cir. 2008). The defendant in Merrill was a physician charged with distributing controlled substances in violation of section 841(a). Id. at 1297. At trial, the physician insisted that his charged prescriptions were issued for a legitimate medical purpose. Id. at 1299. The district court allowed the government to introduce evidence of more than 33,000 prescriptions the physician had written during the relevant three-year period. Id. After his conviction, the defendant argued on appeal that the district court had abused its discretion by admitting the uncharged prescriptions under Rule 404(b). Id.

But the Eleventh Circuit upheld the physician's conviction and concluded that the district court "did not abuse its discretion in admitting either the summary or the individual prescriptions underlying" the practice-wide data. Id. at 1303. The Eleventh Circuit explained that the "evidence of the quantity and combination of prescriptions" the physician had written was "directly related to" whether he was "relieved of liability under the Controlled Substances Act because he acted in the ‘usual course of a professional practice.’ " Id. This was because a "jury may consider prescription data sets outside those specifically charged in the indictment to determine whether a physician has exceeded the legitimate bounds of medical practice.’ " Id. , citing United States v. Harrison , 651 F.2d 353, 355 (5th Cir. 1981).

Lague, for his part, relies on the Eighth Circuit's decision, United States v. Jones , 570 F.2d 765 (8th Cir. 1978). The physician in that case was also charged with intentionally distributing a Schedule II controlled substance under section 841. Id. at 766. At trial, the district court allowed the government to introduce evidence of 478 other prescriptions for Schedule II drugs the physician had written for his former patients as evidence of his unlawful intent to write the charged prescriptions. Id. Upon being convicted, the physician argued on appeal that the district court had erred in admitting the evidence of the uncharged prescriptions. Id.

The Eighth Circuit agreed with the physician and reversed the section 841 conviction. Id. The Eighth Circuit observed that the "other" prescriptions could be logically connected to the crime charged only if the physician wrote those "other" prescriptions "outside the bounds of professional medical practice." Id. But unlike the Eleventh Circuit in Merrill , the Eighth Circuit held that, without specific evidence of the treatment of the patients underlying those "other" prescriptions, the quantity of the prescriptions was not probative of whether the physician had "acted unprofessionally." Id.

Lague and the government ask us to distinguish the case before us from Merrill and Jones respectively. We now turn to that issue.

We disagree with Lague that Merrill is different from this case. Lague contends that Merrill is inapposite because the government there had to prove a scheme to defraud involving excessive quantities of drugs. We acknowledge that the Eleventh Circuit's opinion in Merrill had referenced its earlier discussion that "evidence of the quantity and combination of prescriptions ... during the relevant period is directly related to the issue of whether [the physician] committed health care fraud." Merrill, 513 F.3d at 1303. But in Merrill , the Eleventh Circuit independently concluded that the physician's practice-wide prescription data was admissible under Rule 404(b) because it tended to prove the intent element of the section 841(a) charges, i.e. , whether the physician intended to act "in the usual course of professional practice." Id. We read Merrill to affirm the admission of practice-wide uncharged prescriptions under Rule 404(b) irrespective of any nexus to a healthcare fraud charge. We also disagree with the government that this case is different from Jones . The government asserts that Jones was decided against the backdrop of the clear-and-convincing standard the government was required to overcome when seeking to admit "other act" evidence before the Federal Rules of Evidence was codified. We disagree. In Jones , the Eighth Circuit acknowledged that the government's burden of proof to have "other acts" admitted into evidence had been relaxed by the Federal Rules of Evidence. Jones , 570 F.2d at 768 (explaining the evolution of a proponent's burden of proof under Rule 404(b) ).

Simply put, Merrill and Jones are irreconcilable. Faced with this split of authority, and after carefully examining the law of our circuit, we hold that the Eleventh Circuit's opinion in Merrill better comports with the text and purpose of Rule 404(b).

" Rule 404(b) is a rule of inclusion—not exclusion—which references at least three categories of other ‘acts’ encompassing the inner workings of the mind: motive, intent, and knowledge." United States v. Curtin , 489 F.3d 935, 944 (9th Cir. 2007) (en banc). Under our "low threshold test of sufficien[cy]," United States v. Dhingra , 371 F.3d 557, 566 (9th Cir. 2004), the government "need not prove Rule 404(b) evidence by a preponderance of the evidence," Bailey , 696 F.3d at 799. Instead, the government need only lay a factual foundation from which a "jury could reasonably conclude that [the defendant] committed the allegedly-similar bad acts," and that he possessed the requisite intent in committing those bad acts. Id. , citing Huddleston v. United States , 485 U.S. 681, 685, 108 S.Ct. 1496, 99 L.Ed.2d 771 (1988) ; see also Fed. R. Evid. 104(b). In deciding where "other act" evidence is relevant to prove intent, we defer to the "district judge's own experience, general knowledge, and understanding of human conduct and motivation." Curtin , 489 F.3d at 948, quoting McCormick on Evidence § 185 (6th ed. 2006) (emphasis and alteration omitted).

Applying this relaxed standard, we hold that uncharged prescriptions of controlled substances in enormous quantities, and in dangerous combinations, support a reasonable inference that the underlying prescriptions were issued outside the usual course of professional practice and without a legitimate medical purpose. Lague's practice-wide evidence was therefore probative of his unlawful intent, undermining his defense at trial that the charged prescriptions amounted to "a few bad judgments." Because the prescription data made the intent element of the section 841 charges more probable, the district court properly admitted Lague's uncharged prescriptions under Rule 404(b).

B.

Next, Lague contends that the district court abused its discretion, under Federal Rule of Evidence 403, by failing to preview the underlying prescription data before admitting it into evidence. See Curtin , 489 F.3d at 958 (holding that a district court "does not properly exercise its balancing discretion under Rule 403 when it fails to place on the scales and personally examine and evaluate all that it must weigh"). We assume, without deciding, that the district court abused its discretion by failing to preview all of the prescription data before granting the government's motion in limine. ^, We hold that any error was harmless based on the overwhelming evidence of guilt against Lague.

The burden to show that the evidentiary trial error was harmless falls on the government, and our review begins with a "presumption of prejudice." Bailey , 696 F.3d at 803. Reversal is not required if "there is a ‘fair assurance’ of harmlessness or, stated otherwise, unless it is more probable than not that the error did not materially affect the verdict." Id. , quoting United States v. Morales , 108 F.3d 1031, 1040 (9th Cir. 1997) (en banc); see also United States v. Rendon-Duarte , 490 F.3d 1142, 1144–45 (9th Cir. 2007). We have "found harmless error despite the erroneous admission of evidence" where "the properly admitted evidence was highly persuasive and overwhelmingly pointed to guilt." Bailey , 696 F.3d at 804 (citations omitted).

Although Lague's prescription data was presented through two witnesses and highlighted in the government's opening statement and closing argument, the focus of the nearly two-week trial was on the charged prescriptions. The government admitted the patient files, presented the testimony of one patient's father, a patient's former surgeon, and investigators. Thus, even without the uncharged prescription data, the case was not as close as Lague suggests.

The jury also had access to the patients’ medical charts underlying the unlawful distribution charges, showing continued "red flags" such as use of illegal drugs and, most importantly, the prescriptions for the charged patients that showed copious prescribed controlled substances. For example, the evidence revealed that Lague doubled SL's opioid prescriptions without asking SL about his pain, and that he covered it up by falsely telling the pharmacy that it was for a two-month prescription. The evidence also showed that Lague had prescribed opioids to DL, a drug addict, multiple times the CDC's limit for exercise-induced shoulder pain, despite his urine test showing that he was using cocaine and unprescribed morphine and Xanax. Lague also prescribed DL a benzodiazepine and an amphetamine on top of the opioids. Based on the patient-specific evidence, the government certainly cleared the "benchmark for criminal liability" by proving that Lague "intentionally ... distributed controlled substances for no legitimate medical purpose and outside the usual course of professional practice." United States v. Feingold, 454 F.3d 1001, 1010 (9th Cir. 2006). We disagree with Lague that the prescription data "was impossible to defend against." Lague's trial counsel successfully cabined the weight of the prescription data, inducing the government's witnesses to concede that the prescription data was not highly probative of Lague's guilt. If Lague had rebuttal evidence that the uncharged prescriptions were legitimate, he could have presented it.

We also disagree with Lague that the district court was required to give a specific limiting instruction after the government introduced the prescription data. The district court read a general limiting instruction to the jury before their deliberations. Lague did not request a more specific instruction during trial. This general instruction mitigated the prejudice of admitting the "other act" evidence. See United States v. Hardrick , 766 F.3d 1051, 1056 (9th Cir. 2014). That the jury acquitted Lague of healthcare fraud and was able to compartmentalize the evidence on the various charges also militates against Lague's claim of prejudice. See Park v. California , 202 F.3d 1146, 1150 (9th Cir. 2000).

Thus, the admissible evidence at trial shows that Lague "gave inadequate physical examinations or none at all," that he "ignored the results of the tests he did make," that he took minimal "precautions against [the] misuse and diversion" of controlled substances, and that he prescribed "as much and as frequently as the patient demanded." Moore , 423 U.S. at 142–43, 96 S.Ct. 335. We reject Lague's characterization of the trial as one based on the credibility of two competing expert witnesses. We therefore hold that it was more probable than not that any Rule 403 error in admitting the prescription data did not materially affect the jury's verdict.

IV.

Lague's practice-wide prescription data was admissible under Rule 404(b)(2) to prove his unlawful intent to distribute controlled substances outside the usual course of professional practice. Even if we assume that the district court abused its discretion, under Rule 403, by failing to preview all of the underlying prescription data admitted at trial, the result would be the same. The patient-specific evidence overwhelmingly pointed to Lague's guilt, and thus, any Rule 403 error would be harmless.

AFFIRMED .