*696 After practicing medicine for over thirty years, Doctor Richard Evans decided to branch into pain management. Seven years, eleven thousand oxycodone prescriptions, and a couple of million dollars later, Evans faced federal criminal charges. According to the Government, Evans's pain-management clinic was really a "pill mill"-an operation that prescribes drugs with no legitimate medical purpose. After a lengthy jury trial, Evans was convicted of distributing controlled substances, money laundering, mail fraud, and conspiracy.

Evans now appeals. He claims the evidence at trial was insufficient for the jury to convict on some counts, legal flaws in the indictment tainted others, the district court bungled two evidentiary rulings, and his Confrontation Clause rights were violated. All Evans's claims lack merit. With respect to Evans's Confrontation Clause challenge, we assume without deciding that his rights were violated but nevertheless conclude that any error was harmless. Thus, we AFFIRM all Evans's convictions and his sentence.

I.

A.

In 2010, Doctor Richard Evans started seeing pain-management patients at his office located in Houston. Many of these patients transferred in from two other Houston pain-management clinics where Evans had practiced since 2008. Before that, Evans had primarily worked with car accident, workers' compensation, and cancer patients.

Around the time Evans's pain-management patients transferred in, two women who worked at a dental office in Evans's building, Kristi Smith and Donna Epley, started noticing changes. Cars with Louisiana license plates and loaded with up to six people would show up at Evans's office. These cars would fill the lot, outnumbering the cars with Texas plates. Epley, who would usually arrive early in the morning and unlock the building, would see people waiting outside when she arrived. Evans's office would be packed all day long, and people would loiter around the building. The pair noticed that the patient load would be heavy Monday through Thursday. The flow would ease up on Fridays, which-according to a sign in Evans's waiting room-was the day that no prescriptions would be given out.

Smith and Epley noticed things about the newcomers. Some were unkempt or unhygienic. Others would use the wrong restroom or try to take baths in the sinks. One time, Smith watched as a woman washed and dried her hands over and over and over. Epley thought, based on her past work with addicts, that many of the newcomers seemed impaired, on something, or addicted. Once, a patient left Evans's office, shaking a prescription with glee because he "got it."

The duo brought their concerns about Evans's new clientele to the building's management and Evans's receptionist. Evans responded with a set of building rules. The rules regulated parking ("Patients who are in the parking lot prior to 8:15am may not be seen"), prohibited loitering ("Do NOT ' hang out ' in any of our parking lots" or "in the hallways"), imposed a dress code ("Though it should go without saying, personal grooming/hygiene/clothing are important"), and limited conversation topics in the waiting areas ("[ D ] iscussions about medication/treatment may result in discharge ").

After reading a newspaper article about pill mills popping up in Houston when Louisiana tightened its regulations on *697 pain-management clinics, Smith and Epley grew suspicious. The pair started snooping through Evans's trash. In it, they would find torn and re-taped prescriptions and handwritten letters requesting prescriptions (dosage and pill count often included). They also found a flyer, which explained the procedure patients could follow to get prescriptions via mail and for multiple months. According to the flyer, patients could either mail in a $240 money order for a month of prescriptions or bring in the money for additional months of prescriptions when they came in for their appointments (saving them $40). Patients were directed to "talk to David at Briargrove Pharmacy about mailing" their medications to them.

Smith and Epley relayed their suspicions to law enforcement and the Texas Medical Board. In September 2012, federal agents raided Evans's office (which, Smith and Epley reported, stopped the flow of patients from Louisiana). Patient files, computers, and prescription pads were seized from Evans's office. After this, Evans still saw patients, but just the "normal traffic load."

B.

In 2015, a federal grand jury charged Richard Evans and David Devido, the owner of Briargrove Pharmacy (that filled many of Evans's prescriptions), with several crimes: conspiracy under 18 U.S.C. § 371 , six counts of distributing controlled substances without a legitimate medical purpose and outside the usual course of professional practice under 21 U.S.C. § 841 , five counts of money laundering under 18 U.S.C. § 1957 , and eight counts of mail fraud under 18 U.S.C. § 1341 . Devido was also charged with four counts of health-care fraud for false claims under 18 U.S.C. § 1347 .

The six distribution counts pertained to six prescriptions for oxycodone Evans wrote for five of his patients from Louisiana. The five patients were Marvin Wampole (whose count was voluntarily dismissed by the Government at trial), Kimberly Richardson (whose prescriptions the Government charged in two separate counts), Stacy Cash, Audie Decoteau, and Shane Roper. The five money-laundering counts were based on five withdrawals Evans made from his account with Amegy Bank. The eight mail-fraud counts charged Evans with making false representations by writing prescriptions without adequate medical examinations, without proper recordkeeping, and without providing the proper standard of care. The mail was used in Evans's scheme, per the indictment, when Evans received money orders through the mail from four patients and caused drugs to be sent via mail to the same four. The conspiracy count charged Evans and Devido with conspiring to do two crimes: distributing controlled substances and mail fraud.

C.

The day Richard Evans and David Devido's trial was to start, Devido struck a deal with the Government and pleaded guilty to obtaining possession of a controlled substance through misrepresentation in violation of 21 U.S.C. § 843 (a)(3). No deal was struck with Evans, and his case went forward to trial.

The Government began its case against Evans by calling Smith and Epley, who reported their observations and suspicion that Evans ran a pill mill (a conclusion Evans unsuccessfully objected to on the ground that it was outside their personal knowledge). Next, the Government called Brenda Clayton-one of Evans's longtime medical assistants-to relay how Evans's *698 practice worked. Clayton, who has no advanced degree and only on-the-job training as a medical assistant, said that Evans's other assistant, Rhoda Mann, would handle the new patients. Follow-up visits were Clayton's business.

Clayton discussed her role during these follow-up visits. She would ask the returning patient if he had any complaints and if the current dosage was adequate. If it was adequate, Clayton would fill out a prescription based on the patient's chart. After the prescription was written up, Evans would enter the examination room, see the patient, and sign the already filled-out prescription. If, on the other hand, a patient asked for an increase in his dosage, Clayton would note it, tell Evans, and then Evans would decide what to do.

Clayton also explained how a patient could get medication without an office visit. When a patient first started seeing Evans, if the patient wanted to stay on medication, he would have to come in every 30 days to get a 30-day prescription. But eventually, the patient would be told she could start coming in every 90 days.

Two systems were devised to make sure the patient got 90 days of pills from only one visit. Some patients would get three 30-day prescriptions per visit. One prescription could be filled immediately and the other two would have earliest-fill-by dates projected one and two months in advance. Other patients would get just one 30-day prescription from the visit. When their month-long supply of pills ran low, the patients would mail in $240 money orders along with notes requesting a refill.

Clayton and Mann handled these letters. They would check the incoming mail to see if a money order was attached and then match the letter and money order to the patient's files. They would then fill out the prescription based on the patient's past dosage. These prescriptions were either signed later by Evans or had already been signed.

These pre-signed prescriptions, according to Clayton, came about in two ways. Evans would keep some pre-signed prescriptions in a safe that the assistants could access. Alternatively, when the assistants were busy, they would leave Evans a prescription for him to sign but not fill in along with the file of the patient who sent in the money order. Evans, Clayton said, "didn't change medications at mail ins." Clayton also confirmed that the money order would be for a prescription, not an office visit. Once a prescription was ready to go, someone from the office would either mail it directly to the patient or take it to Briargrove Pharmacy to get it filled. From there, Briargrove would mail the pills to the patient.

To explain what happened at Briargrove, the Government called Devido, who was the former owner and chief pharmacist of Briargrove. In late 2009, Devido noticed an increase in the number of prescriptions for opioids and other controlled substances coming from Evans. As Devido was unfamiliar with Evans, Devido called Evans to inquire into Evans's practice. Evans told Devido that he was practicing pain medicine full time and discussed his procedure for admitting patients. Based on this conversation, Devido was satisfied that everything was above board.

But over time, Devido started to notice that Evans was frequently prescribing a suspicious mix-the combination of an opioid, a tranquilizer, and a muscle relaxer. This assortment, according to Devido, was a well-known and highly abused drug cocktail. One of Devido's employees, Chris Helmke, noticed as well. Helmke wrote Devido a letter expressing concern about the number of prescriptions coming from Evans's office. Helmke's letter, which *699 called Evans a "pill-mill" doctor whose patients' "unsavory" appearance was attracting attention, was admitted over Evans's hearsay and relevance objections. Devido admitted that he basically "turned a blind eye" to the volume of prescriptions because they brought in so much money.

A DEA database confirmed what Devido said about Briargrove's volume. Briargrove was the number one pharmacy in Texas for purchasing oxycodone and in the top fifteen for the nation in 2011. Evans's prescriptions made up over 50 percent of Briargrove's total. Briargrove's output was not as high for 2012, but it was on pace until Evans's office was raided in September 2012. In total, from 2010 through 2012, Evans wrote over 11,000 prescriptions for oxycodone, coming out at over 1.6 million pills. Over that same period, Evans brought in over $2.4 million in cash, money orders, and checks from patients receiving those oxycodone prescriptions.

In addition to volume and revenue, Devido talked about his role in Evans's mail-order scheme. In January 2011, Rhoda Mann, one of Evans's assistants, called Devido. Mann said she was calling on behalf of Evans and told Devido to expect a lot more patients from Louisiana. She also asked if Devido could start shipping the drugs to Louisiana so that Evans could establish a one-visit-every-three-months system. According to Devido, Mann told him that during a visit, Evans would fill out three prescriptions, give one to the patient and hold on to the other two. At the end of each month, Evans would call the patient to see if the patient needed the next month's prescription. If the patient did, someone from Evans's office would bring the prescription to Briargrove where it would get filled and sent to the patient in Louisiana. Devido agreed to the plan and started shipping the drugs to Louisiana.

Another key witness for the Government was Doctor Graves Owen, who reported on what and how much Evans was prescribing. For this part, it helps to know a few things about drugs and federal drug laws. Under federal law, controlled substances are classified into five Schedules, I through V. See 21 U.S.C. § 812 (a). As the schedule number decreases, the potential for abuse and the addictive properties increase. See id. § 812(b). Schedule I drugs-cocaine, heroin, methamphetamine, and the like-are deemed to have no medical use and thus cannot be legally prescribed under federal law. See id. § 812(b)(1) ; 21 C.F.R. § 1308.11 . Schedule II drugs, while legal, are addictive and have a high potential for abuse. See 21 U.S.C. § 812 (b)(2). Such drugs may generally only be obtained via a written prescription by a doctor. See id. § 829(a). Oxycodone, a synthetic opioid sold under the name Roxicodone, is in Schedule II. See 21 C.F.R. § 1308.12 (b)(1)(xiii). Hydrocodone is a semisynthetic opioid, which usually gets combined with a pain reliever like acetaminophen and sold under names like Vicodin, Lorcet, Lortab, and Norco. While hydrocodone is currently in Schedule II, see 21 C.F.R. § 1308.12 (b)(1)(vi), at the time Evans was prescribing, it was in Schedule III. Although oxycodone and hydrocodone are different at the molecular level, their effects on the human body are very similar.

According to Owen, Evans was prescribing high doses of oxycodone and hydrocodone, which, when combined with the other drugs he consistently prescribed, formed a popular drug cocktail. The cocktail included an opioid, a tranquilizer-usually Valium or Xanax-and a muscle relaxant-like Soma or Mobic. Owen said that a doctor should know that this combination packed a punch and formed an abused drug cocktail.

*700 Owen explained the medical community's standard for prescribing opioids and the deficiencies in Evans's practice. Owen said that by at least 2011, opioids were seen by the medical community as a high-risk treatment of last resort. Accordingly, more conservative options-like exercises, psychotherapy, and pain-relieving shots-should be exhausted before turning to opioids. And just because a patient comes in on opioids does not mean he should automatically get more. The doctor would need to examine past medical records to confirm that: conservative options were actually exhausted; the opioid was being taken; the patient was not acting aberrantly (doing things like using other drugs, reporting drugs lost or stolen, asking for early refills, or getting drugs from other doctors); and the drugs were improving the patient's pain, function, and life.

With this summary of the medical standard, Owen turned to the deficiencies he found when he studied 17 or 18 of Evans's patient files. The files had "very superficial" medical history and inadequate medical records. While they contained MRIs and pharmacy pages with the patients' most recent rounds of prescriptions, they did not hold records from past doctors who prescribed the drugs.

Evans's files also rarely documented any benefits from the drugs, a sign-according to Owen-that Evans was prescribing without heed. In fact, many patients reported increased or similar pain after being on medication for a sustained period, indicating that the drugs were not helping and other treatment was called for. Owen also saw that patients were exaggerating their symptoms-say, by regularly reporting that their pain was 10 (or even "10+") out of 10. Such self-reporting is suspect, per Owen, and likely reflects the patient's emotional or psychological suffering rather than physical pain. For such patients, Evans failed to dig deeper to explain what was being reported or to refer the patient to someone who could.

Owen further reported that the files contained few documented physical examinations and the examinations that were performed were inadequate. This was troubling, Owen said, because physical examinations are needed to confirm a patient's symptoms and to track the physical effects of the drugs and other therapies. Owen acknowledged that some of the files had range-of-motion tests. But for those that did, it was unclear who performed the tests, and the results were often too precise to be credible. And, Owen pointed out, training in physical therapy-a qualification no one in Evans's office possessed-would be required to perform an accurate range-of-motion test.

Owen also found Evans's risk-assessment tools to be subpar. Two such tests were used-the Screener and Opioid Assessment for Patients with Pain (SOAPP) and the Current Opioid Misuse Measure (COMM). But the results from these tests were not used for anything. Nor were any red flags-like a report of past drug problems-followed up on. And, Owen added, other risk factors not revealed by SOAPPs or COMMs should be considered, like the patient's age, environmental stressors, and family or personal history of disorders or addictions. Out of the files Owen reviewed, only one had a drug-test result (which was positive for marijuana) and the test needed confirmation.

Concluding, Owen did not see any evidence that the practice of medicine was going on. "It looked more like these were customers." He found that Evans "failed to perform the basic responsibilities of a physician practicing medicine."

D.

The defense case was that Evans ran a legitimate pain-management clinic.

*701 Through cross-examination of Government witnesses and calling former employees, Evans drew out the parts of his practice that did not involve prescribing drugs, the precautions his office took, and that many of his patients had suffered debilitating injuries and experienced real pain.

Evans established that his practice involved more than pain-management patients. He still saw some car accident, workers' compensation, and cancer patients, as well as some patients requiring minor surgeries and seeking help with their weight. He was also certified to administer Suboxone, a medication designed to treat opioid addiction, and saw some patients suffering from addiction.

To explain why his pain-management patients were driving the several hours from Louisiana to Houston, Evans established that the cost of appointments in Houston was lower. In 2008, Louisiana tightened its laws on pain medications, making it difficult for pain patients to see doctors and increasing the cost of doctors' visits. Texas, on the other hand, did not impose such hurdles, and Evans could charge low-income and uninsured patients substantially less.

Evans also drew out evidence about his clinic's practices and precautions. When a new pain-management patient came in, the patient would have to fill out paperwork, submit a recent MRI, and present a pharmacy printout of current prescriptions. To verify that the MRI was correct, a member of Evans's staff would call the MRI facility to make sure it was legitimate. If an MRI was confirmed as a fake, the patient's appointment would be cancelled. If an MRI was too old, Evans would order a new one.

Among the forms new patients had to sign were a patient contract and a pain-management agreement. Both agreements warned of the dangers of drug abuse and set forth general rules which, if violated, would result in the end of treatment. Under the agreements, patients were prohibited from "doctor shopping" (that is, visiting multiple doctors to obtain multiple prescriptions), "pharmacy shopping" (same idea as doctor shopping), selling their drugs, or using drugs at greater than the prescribed rate. Patients also promised to honestly communicate, submit to drug tests, and bring excess drugs on visits. Evans also used a prescription-drug-monitoring database to confirm that patients were not doctor shopping. Several patients' files reflected that they were discharged for violating these agreements. On top of these agreements, Evans had his patients perform two types of risk-assessment tests-the SOAPP and the COMM.

Evans also established that his pain-management practice involved more than just churning out prescriptions. New pain-management patients were given brochures on the benefits of stretching and another on nonsurgical treatment for soft-tissue injuries. On visits, before a patient would see Evans, they would be weighed, have their blood pressure taken, and be shown exercises to help with their pain. Patients might then receive a massage, ultrasound, electric stimulation, or peg board (a device used to loosen tight muscles) treatment to help with the pain. Patients were also encouraged to do stretching at home and were given a journal to track their stretching. Patients would be asked to rate their pain, and the staff would keep track of it.

After all these preparations, Evans would see the patient in an examination room. Evans called Rhoda Mann, his other assistant, to testify about what happened in the room. According to Mann, Evans would perform a physical examination, checking painful areas. Evans would then, *702 according to Mann, decide whether to prescribe the patient a controlled substance. Mann admitted that most of the time, she would fill out the prescription and then have Evans sign it. On occasion, the assistant would retrieve a pre-signed prescription from Evans's safe. According to Mann, Evans would pre-sign prescriptions so that when he was out of town his patients would not go into withdrawal.

Following up on that last point, Mann admitted during cross-examination that she thought the majority of Evans's patients receiving opioids were chemically dependent. Mann also confirmed that after a patient's initial visit, she could get a prescription or drugs mailed to them via money order. Patients would "order[ ]" a prescription, and a prescription would be written up and delivered to Briargrove or sent directly to the patient.

The jury was unconvinced by Evans's defense and found him guilty on all counts. The district court imposed a below-Guidelines sentence of 60 months' incarceration and three years' supervised release. It also imposed a fine of $250,000, and ordered forfeiture of $268,336. Evans now appeals.

E.

On appeal, Evans claims that his convictions for distributing controlled substances, money laundering, and mail fraud should be reversed for lack of sufficient evidence. He also claims that the indictment was legally inadequate, necessitating reversal of his mail-fraud and conspiracy convictions. Next, Evans argues the district court erred in two evidentiary rulings-one allowing Smith and Epley to state that they thought Evans ran a "pill mill" and another admitting Helmke's letter stating the same thing. Finally, he argues that his Confrontation Clause rights were violated when the district court limited the scope of his cross-examination of Clayton.

We consider and reject each claim in turn.

II.

Evans's three sufficiency-of-the-evidence challenges are evaluated "with substantial deference to the jury verdict." United States v. Delgado , 672 F.3d 320 , 330 (5th Cir. 2012) (en banc). We affirm if a reasonable juror could conclude that the elements of the crime were established beyond a reasonable doubt. See United States v. McDowell , 498 F.3d 308 , 312 (5th Cir. 2007). We are obliged to view "the evidence in the light most favorable to the verdict" and draw "all reasonable inferences from the evidence to support the verdict." Id. (quoting United States v. Ragsdale , 426 F.3d 765 , 770-71 (5th Cir. 2005) ). "The evidence need not exclude every reasonable hypothesis of innocence or be wholly inconsistent with every conclusion except that of guilty, and the jury is free to choose among reasonable constructions of the evidence." United States v. Lugo-Lopez , 833 F.3d 453 , 457 (5th Cir. 2016) (per curiam) (quoting United States v. Salazar , 66 F.3d 723 , 728 (5th Cir. 1995), abrogated in part by United States v. Sorrells , 145 F.3d 744 (5th Cir. 1998) ).

As Evans's sufficiency challenges were not properly preserved by an appropriately timed motion for acquittal, we review for plain error. See United States v. Jimenez , 509 F.3d 682 , 690 (5th Cir. 2007). To demonstrate plain error, Evans must show a clear or obvious legal error that affects his substantial rights and "seriously affect[s] the fairness, integrity, or public reputation of judicial proceedings." See Puckett v. United States , 556 U.S. 129 , 135, 129 S.Ct. 1423 , 173 L.Ed.2d 266 (2009) (alteration in original). For a sufficiency challenge, an error is "clear or obvious"

*703 only when the record is "devoid of evidence pointing to guilt" or "the evidence of a key element of the offense [i]s so tenuous that a conviction would be shocking." See United States v. Suarez , 879 F.3d 626 , 630 (5th Cir. 2018) (alteration in original) (quoting McDowell , 498 F.3d at 312 ).

With this, we consider Evans's three sufficiency challenges in turn. All three fail.

A.

Evans challenges the sufficiency of the evidence supporting three of his five convictions for distributing a controlled substance under 21 U.S.C. § 841 . To convict on these counts, the Government was required to prove (1) that Evans "distributed or dispensed a controlled substance, (2) that he acted knowingly and intentionally, and (3) that he did so other than for a legitimate medical purpose and in the usual course of his professional practice." See United States v. Brown , 553 F.3d 768 , 780-81 (5th Cir. 2008) (quoting United States v. Norris , 780 F.2d 1207 , 1209 (5th Cir. 1986) ). Evans does not quarrel with the evidence on elements one and two, so all the action turns on the last element-whether the three charged prescriptions had no legitimate medical purpose or were written outside the course of Evans's professional practice.

The three convictions under scrutiny relate to three oxycodone prescriptions Evans wrote for three different patients-Stacy Cash, Audie Decoteau, and Shane Roper. Evans's attack is two-pronged. First, he asserts that the evidence supporting these convictions was not sufficiently particularized. Relying primarily on a Fourth Circuit case, United States v. Tran Trong Cuong , 18 F.3d 1132 (4th Cir. 1994), Evans claims that the criminality of a particular prescription can only be shown through the patient's own testimony or expert testimony specifically addressing the patient's case. To permit otherwise would allow the jury to find a particular prescription criminal based solely on a pattern or association. In this case, neither Cash, Decoteau, nor Roper testified, and Doctor Graves Owen, the Government's medical expert, did not specifically address their cases. So, Evans says, the jury had no particularized basis to conclude that the three prescriptions were medically illegitimate. Evans bolsters this argument with his second point-that his clinic had all the trappings of a legitimate pain-management practice.

We are unconvinced. Put briefly, our review of the Government's main evidence on the three convictions-the three patient files themselves-leads us to conclude that the jury could rationally find that Cash's, Decoteau's, and Roper's prescriptions were written without a legitimate medical purpose and outside the usual course of professional practice.

All three patient files have troubling features Owen isolated as symptomatic of illegitimate prescribing practices. All three patients are Louisianans who came to Evans complaining of a longstanding injury and severe pain. Despite their complaints, no record of any physical examination exists-besides a range-of-motion test on Decoteau's last visit. When they started seeing Evans, all three reported that they were already on an opioid for the pain. Cash's prescription sheet was particularly loaded: she came to Evans on fentanyl, oxycodone, Ambien, Cymbalta, acetaminophen, Lyrica, naproxen, and tramadol, among other things. The prior medical records and histories of all three were barebones. Each provided only an MRI, gave no indication of how long he or she had been on opioids, and did not state what non-opioid alternatives he or she had tried.

*704 In all three cases, this was enough for Evans to prescribe more than a year and a half of Roxicodone along with a mix of other drugs which formed a popular drug cocktail. Their dosages were not reduced (Decoteau's was actually bumped up twice-once without even a doctor's visit). While seeing Evans, no patient reported a meaningful abatement of pain. In fact, Decoteau's reported pain got substantially worse on his last two visits, with no documented response from Evans. In general, Evans's notes in each file are illegible, cursory ("Any signs of misuse: 0 "; "Do benefits outweigh the risks of treatment: Y "), or completely missing. The notes from Evans's employees are also extraordinarily repetitive, stating over and over that the patient is doing "really good" with stretching and that his or her "pain level is about the same."

Each patient also reported concerning feelings and behaviors. Cash reported that she often took more pills than prescribed and sometimes had mood swings, felt she needed more drugs, and got overwhelmed to the point she thought she could not handle things. Decoteau reported that he often felt a need for more pills and "seldom" ran out of pills early, took more than he was supposed to, or borrowed pills from others. He also reported mood swings, anger issues, impatience with doctors, and feeling overwhelmed. Roper reported that others sometimes expressed concern about his use of medication and that he seldom ran out of pills early, took more than prescribed, or felt consumed by the need to get more. No documented response exists for any of these red flags. None of the three received a drug test, and, by the end, all three were seeing Evans only once every three months. From these files, the jury could reasonably conclude that the particular prescriptions written for Cash, Decoteau, and Roper had no legitimate medical purpose or were outside the course of Evans's professional practice.

Evans's heavy reliance on Tran Trong Cuong is misplaced. As an out-of-circuit case we have not expressly endorsed, it cannot be used to show clear or obvious error, the second prong of plain-error review. See United States v. Fuchs , 467 F.3d 889 , 901 (5th Cir. 2006) (finding no plain-error when there was no "clearly established law in the Fifth Circuit" on the point). Of greater significance, the evidence in this case meets the level of particularity the Tran Trong Cuong court demanded.

In Tran Trong Cuong , the Fourth Circuit struck down 80 distribution convictions based on prescriptions written for 20 non-testifying patients. 18 F.3d at 1141-43 . At trial, the Government's only evidence on the 80 counts was a medical expert's testimony, a summary report prepared by the expert on patient files taken from the defendant's office (not the files themselves), and copies of the 80 prescriptions. Id. at 1141 . The summary report only stated the number of visits, what the visits were for, and what was prescribed. Id. The expert never examined or interviewed any of the 20 relevant patients, and, on direct, he made no mention of the prescriptions they received. Id. On cross-examination, the expert admitted that for some of these patients he had "no way of judging whether [the prescriptions] were valid or not because there was not enough ongoing relationship." Id. With respect to one patient, he admitted that the medication given "could be justified" in the abstract, but taken together with the pattern of the other cases, it could not be justified. Id.

The Fourth Circuit held that this evidence was insufficient to support the 80 distribution convictions. Id. at 1143 . The court refused to conclude the prescriptions were improper "simply because they followed *705 a pattern." Id. at 1141 . Instead, the court expressed concern that the "defendant may have been found guilty of some counts by association." Id. This was impermissible, per the Fourth Circuit, because the defendant was "entitled to individual consideration of every count in [the] indictment by the jury and evidence sufficient to convict on each count beyond a reasonable doubt." Id. at 1142 .

Fourth Circuit cases interpreting Tran Trong Cuong demonstrate that it is the evidence's connection to the particular prescription-not the type of evidence-that is key to confirm a distribution conviction against a prescribing doctor. In United States v. Singh , the Fourth Circuit explained that "[t]here are no specific guidelines concerning what is required to support a conclusion that an accused acted outside the usual course of professional practice." 54 F.3d 1182 , 1187 (4th Cir. 1995) (alteration in original) (quoting United States v. August , 984 F.2d 705 , 713 (6th Cir. 1992) ). "Rather, the courts must engage in a case-by-case analysis of evidence to determine whether a reasonable inference of guilt may be drawn from specific facts." Id. (quoting August , 984 F.2d at 713 ). According to the Singh court, the reason Tran Trong Cuong's convictions were reversed was "not because the victims did not testify, but rather because their lack of testimony was not replaced by any substantive evidence." Id. at 1188. This contrasted with the situation in Singh , where even though two of the relevant patients did not testify, the defendant's sufficiency challenge was denied because an expert witness filled in the gaps for those two patients. Id.

We have similarly interpreted Tran Trong Cuong as focusing on the evidence's connection to the particular patient, not the precise type of evidence. In United States v. Armstrong , we described the holding in Tran Trong Cuong (without endorsing it) as such: "[E]vidence regarding the particular patient visit or treatment giving rise of the § 841 charge should be presented at trial in order for a conviction on the charge to withstand a motion for acquittal." 550 F.3d 382 , 391 (5th Cir. 2008). We found the evidence particular enough in Armstrong because the Government had the relevant patients (who turned out to be undercover agents) testify, played recordings of the relevant visits, and put on the relevant patients' medical records. Id. We did not, however, hold that this level of evidence was necessary for a distribution conviction to withstand attack on appeal. See id.

Turning back to Evans's case, the Government's evidence satisfies Tran Trong Cuong 's admonition that the particular prescription must be medically invalid, not just fit a larger pattern. In this case, the Government offered the three relevant patient files, not merely a cursory summary report. See Tran Trong Cuong , 18 F.3d at 1141 . And unlike the Government expert in this case, the expert in Tran Trong Cuong specifically testified that he was relying on trends rather than specific patient files to find the prescriptions medically invalid. See id. Here, Owen testified that every file he reviewed-which included Cash's, Decoteau's, and Roper's-contained inadequate documentation to justify the level of opioids prescribed.

Evans's next move is to argue that he generally ran a legitimate pain-management clinic. Given his practice's legitimate trappings, the safeguards he had in place, and the various appropriate actions he took, Evans argues that any overprescribing was merely negligent and not criminal. To support this theory, Evans runs through the various indicia of his practice's validity. According to Evans, he: (1) required MRIs and prescription records; (2)

*706 did physical examinations; (3) adjusted some patients' dosage downward and rejected others; (4) treated patients for addiction with Suboxone ; (5) created treatment plans that went beyond prescribing opioids, including massages, stretching, and peg-board therapy; (6) monitored drug abuse via SOAPPs and COMMs; and (7) treated real patients with real injuries and real pain. He contrasts all this with the typical distribution case, where the doctor might never see the patients or hands prescriptions out of a car in a back-alley.

We address each point below. But the bottom line is this: in light of the substantial evidence that Evans ran an illegitimate practice, every fact Evans brings to his defense does little to move the needle in his favor. The jury could reasonably dismiss each precaution or prudent act Evans cites as either wholly inadequate, a vestige of his prior legitimate practice, or in fact a sham designed to hide his operation's nefarious purpose.

Just a sampling of the circumstantial evidence corroborating this view will suffice. Evans routinely prescribed a high dosage of opioids along with other drugs which formed a well-known and highly abused drug cocktail. Evans wrote the majority of opioid prescriptions for Briargrove which, in turn, was Texas's top opioid distributor. Most of his pain-management patients came from Louisiana. These patients started coming when Louisiana tightened its pain-medication regulations and stopped coming when Evans's office was raided. Under Evans's mail-order scheme, his patients would be seen only once every three months and would get prescriptions merely for sending in enough money. A medical expert's review of 17 or 18 of Evans's patient files left the impression that all of the patients were really customers. There was also a lot of evidence that Evans abused his prescription pad. Both of Evans's assistants admitted that Evans would sign blank prescriptions for them to fill out later-a fact corroborated by a stack of signed but otherwise blank prescriptions seized from Evans's office. Rhoda Mann, Evans's medical assistant, thought that the majority of Evans's patients who received oxycodone were chemically dependent and needed opioids to avoid withdrawal.

The case of Kimberly Richardson-the relevant patient for the two distribution convictions Evans does not challenge-is particularly damaging to Evans's case that his safeguards were adequate or enforced. At trial, Richardson testified that after back and neck surgery in the early 2000s, she became addicted to pain medications. Her addiction lasted until the mid-2000s, when she got clean. Richardson continued to suffer serious pain and started seeing Evans in 2010 after hearing about him via word of mouth in Louisiana.

Richardson's first visit with Evans lasted five minutes. Despite disclosing that she had a prior drug problem, she received a prescription for Roxicodone, Lortab, and Valium. Richardson quickly became addicted again.

When she first started seeing Evans, Richardson would drive to Evans's office every month, where he would see her for two or three minutes. One of Richardson's appointments was caught on video. During the visit (which Richardson called typical), Rhoda Mann filled out the prescription and Evans signed it. Evans did not physically examine Richardson during the visit, even though she complained that the pain from her Lupus was getting worse. (Owen would later testify that the symptoms of Richardson's Lupus could have been exacerbated by opioids, causing her greater pain.) In total, Evans saw Richardson for less than two minutes.

*707 Richardson's monthly dosage of Roxicodone fluctuated. After seeing Richardson for a few months, Evans reduced her pill count. Richardson stayed on the lower dosage for a few months, but eventually got into a fight with Evans, demanding he up her drugs. She told him that if he did not do so, she wanted her money back. Following the argument, he bumped her Roxicodone prescription up past where it had been before. Richardson thought her new dosage was "a lot of medication for anybody" and if she "used all the medication," instead of selling some of it, she "would be in a coma or dead." At one point, her drugs were confiscated by the postal service, and she returned to Evans's office asking for more. She got another prescription. In sum, the jury could reasonably view Richardson's case as strong evidence that Evans's safeguards were either not tailored to stop drug abuse, dramatically under-enforced, or simply a sham.

Going one by one through the facts Evans marshals in his defense, the jury could and did reasonably reject each. Evans's patients did have to submit MRIs and prescription sheets. But without more, Owen explained, such medical records are inadequate. They do not reveal the full scope of the past treatments. Nor do they document the patients' full medical history. Thus, neither demonstrates that more conservative options were exhausted, a prerequisite-according to Owen-before turning to a risky tool of last resort like opioids. Nor does Evans's claim that he performed physical examinations inspire much faith. Only one documented examination occurred for the three relevant patients over a total of 27 appointments. In any event, the jury could credit Owen's testimony that the only recorded examinations-range-of-motion tests-were inadequate, incredible, and performed by staff without adequate qualifications.

Next, Evans did dismiss some patients for things like submitting fake MRIs, doctor shopping, and violating patient agreements. He also reduced some of his patients' prescriptions. These dismissals and reductions, however, are irrelevant to the legitimacy of Cash's, Decoteau's, and Roper's treatment. And the jury could reasonably infer that patients were rejected only when they became troublesome and with little thought given to their wellbeing. Take the case of one patient that was discharged for injecting her pain medication. She was sent on her way with a reduced prescription for Roxicodone and not referred to drug treatment. In fact, Clayton could not recall an instance where a patient was referred to a drug-rehabilitation program, and only two of Evans's 956 patients who received oxycodone were ever put on Suboxone to treat addiction.

The jury could reasonably conclude that Evans's non-opioid based treatments-like massages, peg-board therapy, and stretching-were cookie-cutter treatment plans that did not validate his simultaneous overprescription of opioids. Even if things like massages and stretching help, the jury could reasonably credit Owen's testimony that opioids should be a tool of last, not first, resort. And by the end, patients would only receive these allegedly helpful services once every three months.

With respect to the patient agreements, the SOAPPs, and the COMMs, the jury could reasonably conclude that these were mere formalities. The risk-assessment tools meant little as they were never followed up upon. And self-reporting alone may not be enough. As Owen explained, drug tests should be imposed for long-term opioid recipients.

Finally, the fact that Evans's patients were in real pain and needed real help does not show that the strong doses of opioids they all received for a sustained *708 period were legitimate. In fact, it shows the opposite. The three relevant patients who entrusted their health and care to Evans did not report a meaningful abatement in their pain or increase in function. Instead, they reported concerning thoughts and feelings about their medication with no follow up.

In sum, all three patient files contain numerous signs of medically illegitimate prescribing practices. This all rests atop the strong circumstantial evidence that Evans acted outside the usual course of his professional practice. Altogether, the record is not "devoid of evidence pointing to guilt," nor is the evidence "so tenuous that a conviction is shocking." See Suarez , 879 F.3d at 632-33 (quoting United States v. Phillips , 477 F.3d 215 , 219 (5th Cir. 2007) ).

B.

Evans next attacks the sufficiency of the evidence on all five of his convictions for money laundering under 18 U.S.C. § 1957 . We reject his challenge as we cannot say that letting his convictions stand would be plain error.

Section 1957(a) makes it a crime knowingly to "engage in a monetary transaction in criminally derived property of a value greater than $10,000 and is derived from" certain specified crimes. 18 U.S.C. § 1957 (a). Broken down, the crime has three elements: "(1) property valued at more than $10,000 that was derived from a specified unlawful activity, (2) the defendant's engagement in a financial transaction with the property, and (3) the defendant's knowledge that the property was derived from unlawful activity." Fuchs , 467 F.3d at 907 . "[C]riminally derived property" is " 'property constituting, or derived from, proceeds obtained from a criminal offense,' such as drug trafficking." United States v. Alaniz , 726 F.3d 586 , 602 (5th Cir. 2013) (quoting 18 U.S.C. § 1957 (f)(2) ).

In this case, the five withdrawals the Government charged all emanated from a single bank account-Evans's account with Amegy Bank. This account was commingled-that is, it held Evans's clean and tainted money. "Obviously, when tainted money is mingled with untainted money in a bank account, there is no longer any way to distinguish the tainted from the untainted because money is fungible." United States v. Davis , 226 F.3d 346 , 357 (5th Cir. 2000). To deal with this "commingling problem," we have adopted a clean-funds-out-first rule. See id. Under the clean-funds-out-first rule, the defendant is deemed to first withdraw her clean funds in a commingled account before reaching into tainted funds. See id. ; cf. United States v. Poole , 557 F.2d 531 , 535-36 (5th Cir. 1977) (reversing a conviction for interstate transfer of funds obtained by fraud based on use of commingled funds because the check in question was for less than the amount of clean funds in the account). Thus, "where an account contains clean funds sufficient to cover a withdrawal, the Government can not prove beyond a reasonable doubt that the withdrawal contained dirty money." United States v. Loe , 248 F.3d 449 , 467 (5th Cir. 2001).

But this poses a problem when a defendant makes several withdrawals, each individually for less than the clean-fund total in his account. See Davis , 226 F.3d at 357 . Viewed individually, a particular withdrawal would only use clean money, even though in aggregate the defendant would have had to dip into tainted funds. See id. To cope with this problem, we aggregate the transactions-"when the aggregate amount withdrawn from [the] account ... exceeds the clean funds, individual withdrawals may be said to be of tainted money, even if a particular withdrawal was less *709 than the amount of clean money in the account." See Fuchs , 467 F.3d at 907 (alterations in original) (quoting Davis , 226 F.3d at 357 ). So, under this aggregation rule, the Government need only show aggregate withdrawals in excess of $10,000 above the amount of clean funds in the account to validate a money-laundering conviction. See Davis , 226 F.3d at 357 ; see also Fuchs , 467 F.3d at 907 ("Because the total amount of the financial transactions ($4 million) exceeded the amount of clean funds ($3 million), the government sufficiently demonstrated financial transactions involving the proceeds of unlawful activity in violation of § 1957.").

Turning to the case at hand, the Government's theory of money laundering was that Evans used the ill-gotten proceeds of his practice to invest in a retirement account and to purchase real estate. These charged transactions were based on withdrawals from Evans's bank account with Amegy. To trace the flow of Evans's money and prove this theory, the Government called an expert who had reviewed the patient files seized from Evans's office. This expert reported that of the roughly 2,000 patient files seized, 956 patients received Schedule II prescriptions. The expert called these 956 patients Evans's "Schedule II patients." The expert then matched copies of prescriptions in the patient files with money orders patients delivered to Evans as well as daily monetary receipts and deposit slips. With this raw data, the expert created a database tracking the number of Schedule II prescriptions going out and the flow of money in. From 2010 to 2012, Evans brought in over $2.4 million in cash, money orders, and checks from his Schedule II patients. While Evans received income from other sources over that time period, the expert was able to segregate the money Evans received from Schedule II patients.

The expert next looked into where this $2.4 million went, creating charts showing the flow of money in and out of Evans's various accounts. Evans had two main bank accounts (one with Chase and another with Amegy), and one main investment account with Raymond James. The expert tracked deposits, withdrawals, and monthly balances from the Chase account beginning in January 2010 until Evans closed it in October 2011. At the start of 2010, most of the money coming in was from workers' compensation payments. But over time, payments from Schedule II patients became the predominant source of deposits. By September 2011, most deposits were from Schedule II patients. The expert then did a similar analysis of the Amegy account, which Evans opened in August 2011. In aggregate, a little more than $1.13 million was deposited into the Amegy account. Of that, "the majority, if not almost all" came from Schedule II patients. In turn, an "overwhelming majority" of Schedule II payments were made from patients from Louisiana, who made up 805 of the 956 Schedule II patients. The expert also tracked aggregate withdrawals from the Amegy account, concluding aggregate withdrawals were $221.13 shy of aggregate deposits.

The expert then testified to the five money-laundering transactions charged in the indictment. Evans was charged with making two deposits into his Raymond James retirement account (one for $14,439 and another for $15,000) and three withdrawals from his Amegy account ($64,000, $59,000, and $48,663) to buy some real estate. The expert confirmed that these five charged transactions occurred, that the two charged deposits into the Raymond James account were based on checks drawn from Evans's Amegy account, and that the three charged withdrawals from Evans's Amegy account were used to purchase real estate.

*710 On appeal, Evans quarrels with the Government's method. He points out that the Government's expert assumed that all Schedule II prescriptions were medically illegitimate and thus any proceeds from them were tainted. But, according to Evans, this conclusion cannot be rationally drawn from the evidence. The Government's medical expert only reviewed 17 or 18 patient files, far shy of the 956 Schedule II patients. Other Government witnesses admitted that some of Evans's patients might have been legitimate. And Evans even called one of his former patients, who swore that Evans's prescriptions helped him. Evans also argues that the Government failed to calculate his irrefutably clean money, which came from patients not receiving pain-management therapy-like his car-accident, workers' compensation, and cancer patients. Evans synthesizes his argument by claiming that the Government asked the jury to do a math formula without evidence on a key variable-total clean money. The jurors could not know if the aggregate money withdrawn was more than the total clean funds because they had no idea how much of Evans's money was tainted and how much was clean.

We are unconvinced. It is true that the record does not reveal exactly how much of Evans's money was tainted. But despite this, the jury could rationally find that a sufficient amount of his practice's proceeds were tainted to justify all five money-laundering convictions.

The reason for our confidence is that the amount of tainted funds the Government needed to prove up was tiny relative to the amount of money Evans's practice brought in and how much went out. Recall that Evans's aggregate deposits and withdrawals from Amegy were both a little over $1.13 million. The difference between aggregate deposits and withdrawals was $221.13. Under the clean-funds-out-first rule, this gives Evans a small $221.13 buffer of un-touched funds. Under the aggregation rule, to prove all five money-laundering counts, the Government needed to show that more than $50,221.13 of the $1.13 million in deposits was criminally derived. That would be enough to convict on all five counts as it would satisfy § 1957(a) 's $10,000 statutory threshold five-times over while accounting for the $221.13 buffer of untransferred funds.

Given this low requisite threshold-just $50,221.13 of $1.13 million, or a little less than five percent-Evans faces an uphill battle to show his money-laundering convictions were plainly erroneous. True, Evans's Amegy account contained indisputably clean proceeds from non-Schedule II patients. But the Government's expert, who traced the flow of prescriptions and money orders, said that most, if not all, of Evans's money in his Amegy account came from Schedule II patients. Also true, the record does not reveal precisely how many Schedule II prescriptions were medically invalid. But from the evidence we have discussed in detail on Evans's challenge to his distribution convictions, a rational juror could conclude that the percentage of Evans's money coming in from bogus prescriptions was over five percent. It may be the case that trends and tendencies alone cannot support a conviction on a particular distribution count. See Tran Trong Cuong , 18 F.3d at 1141-43 . But here, the Government was not required to show that any particular prescription was criminal, just five percent of the total. We cannot say that the record is "devoid of evidence pointing to guilt," or that the evidence is "so tenuous that a conviction is shocking." See Suarez , 879 F.3d at 632-33 (quoting Phillips , 477 F.3d at 219 ).

C.

Evans's third and final sufficiency-of-the-evidence challenge goes to *711 his eight mail-fraud convictions under 18 U.S.C. § 1341 . Relatedly, he attacks the legal sufficiency of indictment on those same mail-fraud counts. Again, we review the attacks for plain error, 1 and, once again, conclude there was none.

Mail fraud under 18 U.S.C. § 1341 has three elements: "(1) a scheme to defraud; (2) use of the mails to execute that scheme; and (3) the specific intent to defraud." United States v. Lucas , 516 F.3d 316 , 339 (5th Cir. 2008) (quoting United States v. Dotson , 407 F.3d 387 , 391-92 (5th Cir. 2005) ). The action here pertains to the first and last elements-"scheme to defraud" and "intent to defraud"-as Evans does not contest his use of the mails.

We begin by defining our terms. "Scheme to defraud" is tricky to define, "but it includes any false or fraudulent *712 pretenses or representations intended to deceive others in order to obtain something of value, such as money, from the [entity] to be deceived." United States v. Saks , 964 F.2d 1514 , 1518 (5th Cir. 1992) (internal citation omitted). Such falsity must be material, see United States v. Caldwell , 302 F.3d 399 , 409 (5th Cir. 2002), that is, it must have "a natural tendency to influence, or [be] capable of influencing, the decision of the decisionmaking body to which it was addressed," Neder v. United States , 527 U.S. 1 , 16, 119 S.Ct. 1827 , 144 L.Ed.2d 35 (1999) (quoting United States v. Gaudin , 515 U.S. 506 , 509, 115 S.Ct. 2310 , 132 L.Ed.2d 444 (1995) ). We have held in the context of a prosecution for making false statements that this "natural tendency" test "is an objective one focused on whether the statement is 'of a type capable of influencing a reasonable decision maker.' " United States v. Abrahem , 678 F.3d 370 , 375 (5th Cir. 2012) (quoting United States v. McBane , 433 F.3d 344 , 351 (3d Cir. 2005) ). The test focuses on the "intrinsic qualities" of the statement itself and "transcend[s] the immediate circumstances in which it is offered." Id. at 376 (quoting McBane , 433 F.3d at 351 ). Thus, a crazed man demanding to see a patient being kept under guard by federal agents makes a materially false statement when he tells the agents he is the patient's lawyer. Id. at 371-73, 376 . Although the circumstances dispel any chance that the agents might buy the lie, the representation that one is a lawyer "is of a type that would naturally tend to influence or is capable of influencing the decision maker." Id. at 376 .

Moving along, "intent to defraud" requires "an intent to (1) deceive, and (2) cause some harm to result from the deceit." United States v. Moser , 123 F.3d 813 , 820 (5th Cir. 1997) (quoting United States v. Jimenez , 77 F.3d 95 , 97 (5th Cir. 1996) ). "A defendant 'acts with the intent to defraud when he acts knowingly with the specific intent to deceive for the purpose of causing pecuniary loss to another or bringing about some financial gain to himself.' " United States v. Umawa Oke Imo , 739 F.3d 226 , 236 (5th Cir. 2014) (quoting United States v. Akpan , 407 F.3d 360 , 370 (5th Cir. 2005) ); accord United States v. St. Gelais , 952 F.2d 90 , 96 (5th Cir. 1992).

Before we can turn to Evans's particular attacks, we must elaborate on the standard we use to judge indictments. The point of an indictment "is to inform a defendant of the charge against him." United States v. Hoover , 467 F.3d 496 , 499 (5th Cir. 2006). Thus, "[p]ractical, not technical, considerations govern" our review, and the test "is 'not whether the indictment could have been framed in a more satisfactory manner, but whether it conforms to minimal constitutional standards.' " United States v. Chaney , 964 F.2d 437 , 446 (5th Cir. 1992) (footnote omitted) (quoting United States v. Webb , 747 F.2d 278 , 284 (5th Cir. 1984) ). Those minimum constitutional standards are threefold: the indictment must (1) have the essential elements of the offense charged for each count, (2) describe the elements of the offense with particularity, and (3) be specific enough to protect the defendant from subsequent prosecution for the same offense. See United States v. Simpson , 741 F.3d 539 , 547 (5th Cir. 2014).

As previewed, Evans's attack goes to the particularity with which the indictment charged the "intent to defraud" and "scheme to defraud" elements of mail fraud. His argument with respect to the first-intent-fails straight out of the gate as we have held that intent "need not be specifically charged in the indictment." See United States v. Ratcliff , 488 F.3d 639 , 644 n.4 (5th Cir. 2007). With respect to the *713 latter-"scheme to defraud" and its requirement of materiality-our caselaw holds that the indictment need not "establish materiality," see Hoover , 467 F.3d at 500, but must simply raise an "inference," see Caldwell , 302 F.3d at 409 . "[T]he proper inquiry is whether the allegation [in the indictment] is 'potentially capable of being proved material by the government at trial,' and whether the allegation is sufficient to support an inference of materiality." United States v. Bieganowski , 313 F.3d 264 , 286 (5th Cir. 2002) (quoting United States v. McGough , 510 F.2d 598 , 602 (5th Cir. 1975) ).

Evans's twin-attacks on the indictment and the sufficiency of evidence at trial share a common core. His basic argument underlying both challenges is that the indictment alleged and evidence showed that his patients got exactly what they wanted-opioids and other drugs. Thus, Evans's false representations that the drugs were prescribed for a medically legitimate purpose could not be a material falsehood-i.e., one that might influence his patients. Nor could he have intended to defraud his patients. They got the benefit of the illicit deal they struck.

We reject both arguments. With respect to Evans's attack on materiality, the indictment raised an inference and the evidence allowed the jury reasonably to conclude that Evans's false representation-that he was following the proper standard of care when prescribing-was material. Such a falsehood would have a "natural tendency" to influence his patients' decision making. See Neder , 527 U.S. at 16 , 119 S.Ct. 1827 . A reasonable person faced with the reality that his doctor is prescribing powerful opioids without a medicinal purpose might find a new doctor. See Abrahem , 678 F.3d at 375 . 2 While the "immediate circumstances" do not indicate that the four particular patients named in the mail-fraud counts would have been concerned about the quality of care they received, the "intrinsic qualities" of the representation are surely material. See id. at 376 . The representation that one is a doctor prescribing needed medicine is "of a type that would naturally tend to influence" patients. See id.

Turning to the intent element, the evidence at trial was enough to allow the jury to conclude that Evans's misrepresentations about his prescription writing practices were intentional and that he contemplated that "some harm" would arise from these misrepresentations. See Moser , 123 F.3d at 820 . The jury heard evidence that Evans's overprescribing practices caused some patients to get addicted, fed others' preexisting addictions, and at the very least denied them the proper medical care they paid for. 3 And even if Evans thought he was ultimately helping his patients, the purpose of his deception was certainly to bring "about some financial gain to himself." See Umawa Oke Imo , 739 F.3d at 236 (quoting Akpan , 407 F.3d at 370 ).

Evans argues that the particular harm he may have contemplated-getting his patients hooked, feeding their addictions, or denying them proper care-was not the *714 type of harm required to show intent to defraud. Rather, he argues that "the harm contemplated must affect the very nature of the bargain itself." See United States v. Starr , 816 F.2d 94 , 98 (2d Cir. 1987) ; United States v. Regent Office Supply Co. , 421 F.2d 1174 , 1180 (2d Cir. 1970). Evans argues that if he really was a drug dealer, his patients got the benefit of the bargain they struck with him. But, we have not adopted the rule from Starr and Regent Office , nor have many other circuits. See United States v. Bereano , 161 F.3d 3 , 1998 WL 553445 , at *3 & n.5 (4th Cir. 1998) (unpublished table decision) (remarking that the circuits are split over whether any contemplation of harm is even necessary); United States v. Butler , 704 F.Supp. 1351 , 1354 (E.D. Va. 1989) (noting that the rule from Starr "is in opposition to the rule in the majority of the Circuits"). Evans thus cannot rely upon Starr to show a "clear or obvious" error. See Fuchs , 467 F.3d at 901. 4

To recap, we discern no plain error, whether by lack of evidence or defects in the indictment, with respect to any of Evans's convictions. We now turn to a pair of evidentiary rulings Evans contests on appeal.

III.

Evans asks us to review and reject two evidentiary rulings made by the district court. First, he contends that the district court improperly allowed Kristi Smith and Donna Epley-the two dental assistants who worked in Evans's building-to say that they suspected Evans ran a pill mill. Per Evans, this testimony was inadmissible under Federal Rule of Evidence 602 as pure speculation, not based on Smith or Epley's personal knowledge. Next, Evans argues that Chris Helmke's letter to David Devido-Helmke's boss at Briargrove Pharmacy at the time-was hearsay and thus inadmissible under Rule 802. In that letter, Helmke accused Evans of running a pill mill and described Evans's patients as "unsavory." For both attacks, Evans claims the admission of the challenged evidence influenced the result of what he thinks is a close case.

We review the two rulings for abuse of discretion. See United States v. Lockhart , 844 F.3d 501 , 512 (5th Cir. 2016). If error occurred, it can be excused if harmless. See United States v. El-Mezain , 664 F.3d 467 , 525-26 (5th Cir. 2011) ; see also Fed. R. Crim. P. 52(a). Here, because the alleged errors are nonconstitutional, such errors are harmless unless they had a "substantial and injurious effect or influence in determining the jury's verdict." See El-Mezain , 664 F.3d at 526 (quoting United States v. Lowery , 135 F.3d 957 , 959 (5th Cir. 1998) (per curiam) ). "Under this standard, we ask 'whether the error itself had substantial influence' on the jury in light of all that happened at trial; if we are 'left in grave doubt, the conviction cannot stand.' " Id. (quoting Kotteakos v. United States , 328 U.S. 750 , 765, 66 S.Ct. 1239 , 90 L.Ed. 1557 (1946) ).

We consider each of Evans's evidentiary challenges in turn, starting with Smith and Epley's opinion testimony and then moving to Helmke's letter. We ultimately reject both, finding no error or harm from the first and no harm from the second.

A.

"A witness's testimony," under Federal Rule of Evidence 602, "must be *715 based on personal knowledge." Id. at 495. "Personal knowledge can include inferences and opinions, so long as they are grounded in personal observation and experience." United States v. Cantu , 167 F.3d 198 , 204 (5th Cir. 1999) (quoting United States v. Neal , 36 F.3d 1190 , 1206 (1st Cir. 1994) ). Similarly, under Rule 701, a lay witness may state his ultimate opinion, provided that opinion is "based on personal perception," "one that a normal person would form from those perceptions," and "helpful to the jury." United States v. Ebron , 683 F.3d 105 , 136-37 (5th Cir. 2012). Such opinions "must be the product of reasoning processes familiar to the average person in everyday life." Id. (quoting United States v. Yanez Sosa , 513 F.3d 194 , 200 (5th Cir. 2008) ).

Here, Smith and Epley's shared opinion that Evans ran a pill mill was "based on [their] personal knowledge," and therefore admissible. See El-Mezain , 664 F.3d at 495 . Their conclusion was based on over two years of observations, which included: a major influx of people from Louisiana soon after Louisiana tightened its regulations, the fact that some of the newcomers were disheveled and unhygienic, and that some of the newcomers would arrive early, loiter, and act strangely. Epley said that some of the new people looked impaired and addicted, and she saw a person waving a prescription, saying "I got it." The pair's search through Evans's trash revealed prescriptions, letters requesting prescriptions, and a flyer explaining how to get prescriptions and medication via mail. They saw that Evans's office was not heavily trafficked on the one day when prescriptions were not given out. Smith testified that after a raid on Evans's office, the flow of people from Louisiana slowed down. In total, these observations would naturally lead a person to suspect that the doctor's office next door might have transitioned into a drug front.

True, Smith and Epley's opinions were not drawn from the patient charts or what happened in the examination room. But they were not required to be. Instead, their opinions were properly drawn from commonsensical inferences a "normal person would form" based on years of observations. See Ebron , 683 F.3d at 137 . This is not a case where a witness forms an opinion based on hearsay rather than personal observation. Cf. United States v. $92,203.00 in U.S. Currency , 537 F.3d 504 , 508 (5th Cir. 2008). Nor is it one where a witness fancifully speculates. Cf. Visser v. Packer Eng'g Assocs., Inc. , 924 F.2d 655 , 659 (7th Cir. 1991) (en banc). Thus, admission of Smith and Epley's ultimate opinion-that Evans ran a pill mill-was proper.

But even if it was not, such error was harmless. Smith and Epley's properly admitted testimony of their observations alone would naturally lead the jury to infer that Evans ran a pill mill. Cf. El-Mezain , 664 F.3d at 526 (finding error harmless when the improperly admitted evidence was cumulative). And viewed in relation to "the entire trial proceedings"-namely, the evidence of what happened in the examination room-Smith and Epley's comments were not prejudicial. See id. To prove Evans's guilt, the Government: called two patients to testify that Evans's appointments were cursory, played a video of one of those cursory appointments, and put on patient charts and expert testimony detailing those charts' deficiencies. In the end, even if Smith and Epley's testimony had been improperly admitted, there is no "reasonable possibility" that it "contributed to the conviction." See El-Mezain , 664 F.3d at 526 (quoting United States v. Williams , 957 F.2d 1238 , 1242 (5th Cir. 1992) ).

*716 We next consider Helmke's letter and conclude that its admission, even if erroneous, was harmless.

B.

Our analysis of Evans's challenge to Helmke's letter will be brief. Assuming arguendo that Helmke's letter was inadmissible hearsay, its admission was nevertheless harmless. As explained above, the record contains substantially stronger evidence of Evans's illegitimate practice than Helmke's letter. Accordingly, we cannot conclude the letter's admission had a "substantial and injurious effect or influence" on the jury's verdict. See El-Mezain , 664 F.3d at 526 (quoting Lowery , 135 F.3d at 959 ).

Having decided that neither of Evans's evidentiary challenges is valid, we move to his last argument-that his Confrontation Clause rights were compromised. We determine this claim fails as well.

IV.

Evans claims that the district court violated his Confrontation Clause rights by refusing to permit cross-examination into a target letter Brenda Clayton-one of Evans's medical assistants and a Government witness-received from the Government. The Government counters by arguing that the district court's limit on cross-examination was constitutionally permissible, and, either way, any violation was harmless. We assume without deciding that the limit violated the Confrontation Clause, but hold that any potential error was harmless.

From the record, it appears that Clayton was a very nervous witness. On direct, after a line of questioning where Clayton swore that she did not see patients come in early, loiter, or dress inappropriately, she teared up and the district court ordered a break. During the break, the court questioned Clayton's lawyer about her emotional status (outside the presence of the jury), remarking that she seemed "a little tender." Her lawyer explained that Clayton cried every time they met. The prosecution then requested to have Clayton declared a hostile witness, a request the court denied.

Back on direct, the prosecutor restarted by apologizing for upsetting Clayton. Clayton confirmed that she had spoken with the prosecutor three times before the trial. The prosecutor continued his direct examination by telling her (a total of three times during his direct) that she had done nothing wrong and was not in trouble.

The next day of trial, near the end of Clayton's cross-examination, Clayton confirmed she had received a letter from the Government stating that she was "a target of a federal grand jury investigation." (A "target," per the United States Attorneys' Manual, is "a person as to whom the prosecutor or the grand jury has substantial evidence linking him or her to the commission of a crime and who, in the judgment of the prosecutor, is a putative defendant." U.S. Dep't of Justice, United States Attorneys' Manual § 9-11.151.) Before Clayton could answer, the Government objected based on relevance. After hearing from the defense, the court initially upheld the relevance objection and directed the jury to disregard the question. Later, during a sidebar, the court gave the defense another crack at making its case.

The defense noted that the day before, Clayton was a fragile and tentative witness. From this, the jury "could infer that her demeanor was caused in part by the fact that she was testifying against her former boss." To bolster this inference, the defense pointed out that the prosecution reassured her that the Government did not think she had done anything wrong. Per the defense, this was an effort to make the jury believe the Government had no criminal *717 interest in her, when in fact the Government had previously sent her a target letter. Based on this, the defense argued "there was a misimpression left with the jury" which the "defense was entitled" to correct by bringing "the full picture of the fact that she was testifying under a target letter." Bringing this fact to light might explain why Clayton was not a model witness. This did not sway the district court, and it reconfirmed its prior ruling.

On appeal, Evans claims that the district court's ruling violated his Sixth Amendment right under the Confrontation Clause to expose Clayton's potential motive to "shad[e her] testimony in an effort to please the prosecution" so as to avoid future prosecution. See Greene v. Wainwright , 634 F.2d 272 , 276 (5th Cir. Jan. 1981) ; cf. Davis v. Alaska , 415 U.S. 308 , 317-18, 94 S.Ct. 1105 , 39 L.Ed.2d 347 (1974) (protecting questioning about "undue pressure" based on the witness's "possible concern that he might be a suspect in the investigation"); United States v. Croucher , 532 F.2d 1042 , 1046 (5th Cir. 1976) (protecting questioning of "bias and prejudice stemming from criminal charges which could still be pursued against" the witness).

Even assuming a Confrontation Clause violation, any such violation was harmless. Our harmless error analysis for a Confrontation Clause violation is informed by five nonexclusive factors originally set forth by the Supreme Court in Delaware v. Van Arsdall :

(1) "the importance of the witness' testimony in the prosecution's case"; (2) "whether the testimony was cumulative"; (3) "the presence or absence of evidence corroborating or contradicting the testimony of the witness on material points"; (4) "the extent of cross-examination otherwise permitted"; and (5) "the overall strength of the prosecution's case."

United States v. Jimenez , 464 F.3d 555 , 564 (5th Cir. 2006) (quoting Delaware v. Van Arsdall , 475 U.S. 673 , 684, 106 S.Ct. 1431 , 89 L.Ed.2d 674 (1986) ). This "harmless-error analysis first requires us to 'assum[e] that the damaging potential of the cross-examination [was] fully realized.' " Id. at 563 (alteration in original) (quoting Van Arsdall , 475 U.S. at 684 , 106 S.Ct. 1431 ). We look at each Van Arsdall factor in turn, considering each de novo. See id. at 558.

The first three Van Arsdall factors circle around a related set of questions. Was Clayton's testimony important? Unique? Non-cumulative? Otherwise uncorroborated? See Van Arsdall , 475 U.S. at 684 , 106 S.Ct. 1431 . We determine these factors on balance favor the Government.

At first blush, Clayton's testimony could seem crucial. She had been inside the examination room and witnessed how Evans handled his patients there. She was an "insider," unlike any of the Government's other witnesses. But her testimony largely did not touch on the key issue on the distribution counts-whether the particular charged prescriptions were medically legitimate. The Government showed the deficiency of Evans's treatment of those particular patients mainly by calling one of the patients, putting on a medical expert, and presenting the patients' files.

What is more, numerous independent sources corroborated the basic scheme-that Evans would accept money orders by mail and in exchange disburse prescriptions to Briargrove Pharmacy. On top of Clayton's testimony, the prosecution had at least four witnesses describe the scheme-including David Devido, Rhoda Mann, and two patients-and put on a flyer found in Evans's trash which laid out the basics of the scheme. Other parts of *718 Clayton's story were also well corroborated. The Government introduced several prescriptions Clayton confirmed she wrote out, several patient letters requesting prescriptions with accompanying money orders, patient files with money orders and matching prescriptions inside, several otherwise blank prescriptions signed by Evans found in the patient files, and a pad of 29 pre-signed prescriptions.

The Government did refer to Clayton's testimony in opening and closing. But the Government also emphasized the testimony of several other witnesses; the physical evidence, including the patients' charts, the flyer, the pre-signed prescriptions, and the recording of Richardson's appointment; the money-order scheme; and the large number of prescriptions going out and vast sum of money coming in until the raid.

Plus, the testimony of the other insider-Rhoda Mann-was more devastating to Evans's case. Mann admitted that Evans kept blank prescriptions for her to fill out so that patients would not go into withdrawal when he left town. She added that she thought the majority of Evans's patients on opioids were chemically dependent. 5 Evans is able to pinpoint a single fact that only Clayton testified to-that she would fill out prescriptions for returning patients before Evans had even seen them. Though this was strong evidence against Evans, in light of the numerous sources that confirmed that Evans pre-signed prescriptions and made them available for his assistants, it does not make a measurable difference.

On the fourth Van Arsdall factor-the extent of the cross-examination allowed-neither party gets much traction. Evans was permitted to cross-examine Clayton at length, but not about her potential motive to shade her testimony.

On the fifth and last Van Arsdall factor-the overall strength of the Government's case-we find that it favors the Government. Our prior review of the evidence from the sufficiency challenges confirms that the Government's case was robust and multidimensional. It included strong circumstantial evidence-the flood of patients after the Louisiana clinics closed, the explosion of prescriptions which abated after the raid, and the money-orders-for-drugs-by-mail scheme. It included strong direct evidence that Evans was churning out prescriptions without medical need-the testimony of two patients and the files of many more who received extended periods of opioids with little else being done to alleviate their pain or underlying condition. And, as we previously discussed in the sufficiency section, the Government's evidence strongly supported the conclusion that the trappings of Evans's practice were wholly ineffective or simply a sham.

In sum, any alleged Confrontation Clause violation was harmless.

* * *

*719 For the foregoing reasons, we AFFIRM all of Evans's convictions and his sentence.