Defendant appeals from a judgment of conviction on one count of unlawful sexual penetration in the first degree and two counts of sexual abuse in the first degree. On the first day of defendant’s trial, two medical expert witnesses testified that they had diagnosed the six-year-old child victim as having been sexually abused based on the credibility of her report; the next day, the Oregon Supreme Court issued its opinion in State v. Southard, 347 Or 127, 218 P3d 104 (2009), holding that such evidence is inadmissible. Defendant moved for a mistrial; the court denied the motion and instead instructed the jury to disregard the experts’ testimony regarding their diagnoses. On appeal, defendant assigns error to the trial court’s denial of his motion for mistrial, arguing that the jury instruction did not cure the prejudice caused by the erroneous admission of the experts’ testimony. We conclude that the medical diagnosis testimony was so prejudicial that defendant’s ability to obtain a fair trial was impaired. We therefore reverse and remand.
Because defendant was convicted, we summarize the facts in the light most favorable to the state. State v. Vidal, 245 Or App 511, 513, 263 P3d 364 (2011), rev den, 351 Or 761 (2012). Defendant visited the home of his wife’s sister (mother), where her two teenage sons and six-year-old daughter, B, were present. The boys were sitting on the floor playing video games while defendant and B sat on a couch behind them playing a game that involved tickling. Defendant left the house when mother returned. Later that day, B told mother that defendant had “put his fingers on me, underneath my shorts, and he was tickling me” and that defendant had touched her on “her back part.”
Mother took B to a pediatrician, Dr. McNamara, who performed an ano-genital examination. His only relevant finding was that B’s vagina had “no obvious hymenal rim and notching [was] present.” When he asked where defendant had touched her, B pointed to the area between her anus and vagina. McNamara referred B to CAKES, a child abuse assessment center, where a medical examiner and a social worker conducted an abuse assessment. McCready, a pediatric nurse practitioner, performed a physical evaluation of *498B, and after that evaluation, a social worker interviewed B while McCready observed from a separate room. During both the physical evaluation and the interview, B indicated that defendant had put his fingers in her anus, but she denied that he touched her vagina. The examination of B’s anus yielded no physical indication of abuse, though B’s hymen was “not of a vast quantity or a vast amount,” a nonspecific finding that, according to the testimony, was not by itself definitively indicative of past trauma.
Defendant was arrested and charged with one count of unlawful sexual penetration in the first degree, ORS 163.411, and two counts of sexual abuse in the first degree, ORS 163.427. The sexual penetration charge specified that it related to the penetration of B’s anus, and the charges for sexual abuse in the first degree related to the touching of B’s buttocks and genital area. Before defendant’s trial date, however, he was released on bail and fled the country. He was arrested on a warrant six years later, and this case was tried shortly thereafter.
At trial, McNamara and McCready testified for the state as medical experts regarding their findings. McNamara testified that he believed that B had been sexually abused, and McCready testified that she, too, had made a diagnosis of sexual abuse. Additionally, the jury watched B’s videotaped interview with the social worker at CARES. Also on the first day of trial, B testified that she remembered that defendant touched her vagina with two of his fingers, but she did not remember or recall many details.
On the second day of trial, defendant testified, and he denied that he had abused B. That same morning, the Oregon Supreme Court decided Southard.1 In that case, the Supreme Court considered whether “a diagnosis of ‘sexual abuse’ — i.e., a statement from an expert that, in the expert’s opinion, the child was sexually abused — is admissible in the *499absence of any physical evidence of abuse.” 347 Or at 142. The court held that, where “that diagnosis does not tell the jury anything that it could not have determined on its own, the diagnosis is not admissible under OEC 403.” Id. The court reasoned that, in such circumstances, the probative value of the diagnosis is “marginal” in that juries routinely make those types of credibility determinations, but that “[t]he risk of prejudice, however, [is] great.” Id. at 140. The diagnosis of the child in Southard was “particularly problematic,” the court explained,
“because the diagnosis, which was based primarily on an assessment of the [child’s] credibility, posed the risk that the jury will not make its own credibility determination, which it is fully capable of doing, but will instead defer to the expert’s implicit conclusion that the victim’s reports of abuse are credible.”
Id. at 141.
After the lunch break on the second day of trial, defendant’s counsel moved for a mistrial, arguing that, under OEC 403 and Southard, defendant had been unfairly prejudiced by the jurors having heard inadmissible evidence. He also requested that, if the motion for mistrial were denied, the court instruct the jury to disregard the portions of McNamara’s and McCready’s testimony diagnosing sexual abuse. The trial court denied the motion for a mistrial. After defendant finished testifying, the trial court instructed the jury:
“Ladies and gentlemen, yesterday the Oregon Supreme Court rendered a decision stating that a diagnosis of child * * * sexual abuse is not admissible in cases such as this. Therefore, you are to disregard only the witness’s statements that they made a diagnosis of child sexual abuse. All other portions of their testimony you are allowed to consider.”
Ultimately, the jury convicted defendant on all three counts.
Defendant now appeals, arguing that, in this case, the evidence of medical diagnoses of sexual abuse was so prejudicial that a curative instruction was insufficient and a mistrial should have been granted.
*500In support of that argument, defendant points out that this court has held in similar cases that the erroneous admission of a medical diagnosis is sufficiently prejudicial to be considered plain error under ORAP 5.45(1). Defendant also compares this case to Bruton v. United States, 391 US 123, 135-36, 88 S Ct 1620, 20 L Ed 2d 476 (1968), and similar Oregon cases in which a codefendant’s confession inculpating the defendant was erroneously admitted and an instruction to the jury to disregard the admission was insufficient to remove the prejudice.
The state responds that the curative instruction was sufficient to remove any prejudice. The state compares this case to State v. Brown, 310 Or 347, 364-65, 800 P2d 259 (1990), a case in which the Oregon Supreme Court held that prejudice from testimony by a police officer vouching for the credibility of another state witness could be cured with a limiting instruction. Further, the state argues that the trial court did not abuse its discretion in denying the motion for a mistrial because, in this case, unlike Southard, there was some physical evidence of sexual abuse and defendant’s decision to flee the country rather than face trial was circumstantial evidence of his consciousness of guilt.
We review a decision to deny a mistrial for abuse of discretion. State v. Bowen, 340 Or 487, 508, 135 P3d 272 (2006), cert den, 549 US 1214 (2007). ‘We will deem a denial of a mistrial as error only when a defendant’s ability to obtain a fair trial has been impaired.” State v. Vann, 158 Or App 65, 72, 973 P2d 354 (1999). Statements or testimony that the jury is instructed to disregard can be “so prejudicial that, as a practical matter, the bell once rung, cannot be unrung by such an admonishment.” State v. Jones, 279 Or 55, 62, 566 P2d 867 (1977) (internal quotation marks omitted). “Ultimately, we must decide whether, under the circumstances as a whole, defendant was denied the right to a fair trial, as a matter of law, by the events that transpired at trial.” State v. Davis, 345 Or 551, 583, 201 P3d 185 (2008), cert den,_US _, 130 S Ct 371 (2009).
We note first that defendant is correct that we have exercised our discretion in several cases to review as plain error the admission of evidence of expert medical diagnoses *501of sexual abuse. See State v. Volynets-Vasylchenko, 246 Or App 632, 638, 267 P3d 206 (2011); State v. Arriaza, 236 Or App 456, 237 P3d 222 (2010); State v. Clay, 235 Or App 26, 230 P3d 72 (2010); State v. Merrimon, 234 Or App 515, 228 P3d 666 (2010); State v. Lovern, 234 Or App 502, 228 P3d 688 (2010). However, we have not exercised our discretion to correct every error regarding admission of a diagnosis of sexual abuse. See, e.g., Vidal, 245 Or App 511 (declining to exercise discretion to review admission of medical diagnosis of sexual abuse of child as plain error because there was physical evidence in support of the expert’s opinion); State v. Martinez-Sanchez, 244 Or App 87, 260 P3d 599 (2011) (declining to exercise discretion to correct any error where the testimony regarding a sexual abuse diagnosis was elicited only after the defendant, on cross-examination, opened the door to that subject); State v. Childs, 243 Or App 129, 132, 259 P3d 77, rev den, 350 Or 573 (2011) (although allowing the testimony was error, it was harmless, because the case was tried to the court, the court expressly stated that it “did not particularly rely” on the diagnosis, and the testimony about abuse was corroborated by photographs); State v. Pickett, 246 Or App 62, 264 P3d 209 (2011), rev den, 351 Or 541 (2012) (declining to address Southard error that was harmless). However, Oregon courts do not recognize structural error, so the erroneous admission of a medical diagnosis does not necessarily require reversal in every case. See Ryan v. Palmateer, 338 Or 278, 295, 108 P3d 1127, cert den, 546 US 874 (2005) (“[Structural error is a doctrine that originated in federal criminal cases and * * * has not been adopted by this court as an aspect of Oregon law[.]”).
The question in this case is whether the admission of the witnesses’ medical diagnoses were so prejudicial that defendant was denied a fair trial, despite the court’s instruction to the jury. We conclude that they were. Here, there were two experts, and both drew conclusions based entirely on B’s statements, which were not corroborated by the physical findings. The only physical evidence in this case — nonspecific findings regarding B’s hymen — was, according to the experts, not indicative of vaginal penetration, and, in any event, defendant was not charged with vaginal penetration. Similarly, B’s own testimony, that defendant touched her vagina, *502contradicted her earlier statements to the pediatrician and at CARES that defendant put his fingers in her anus. The diagnoses were based primarily on an assessment of B’s credibility, which “posed the risk that the jury will not make its own credibility determination, which it is fully capable of doing, but will instead defer to the expert’s implicit conclusion that the victim’s reports of abuse are credible.” Southard, 347 Or at 141. Thus, the medical experts’ diagnoses in this case did not tell the jury anything that it was not equally capable of determining on its own. Further, they went directly to the dispositive issue in the case: whom to believe. Therefore, the risk of prejudice from McNamara’s and McCready’s testimony regarding their diagnoses was great.
We next consider whether the limiting instruction was sufficient to cure the prejudice. The state argues that the instruction in this case is similar to a curative instruction that was deemed sufficient in Brown, 310 Or 347. In that case, the defendant was tried for aggravated murder of a witness and felony murder, among other charges. Id. at 349. A detective testified about interviewing one of the defendant’s cellmates about statements that the defendant purportedly made about where he had put the victim’s missing body. Id. at 352-53, 362-64. The detective testified that he believed the cellmate “was being truthful,” to which the defendant objected. Id. at 364. The trial court immediately instructed the jury to disregard the detective’s testimony about his belief. Id. The Supreme Court held that the instruction was sufficient to dispel the jury’s possible contamination by the detective’s testimony about the credibility of the cellmate. Id. at 365.
In this case, however, the limiting instruction was not sufficient to cure the prejudicial diagnoses of sexual abuse. McNamara and McCready did not simply testify regarding whether they believed that B was truthful; rather, they stated their opinions as medical diagnoses from credentialed experts. They aligned their expertise on one side of a pure swearing contest, involving an incident in which there were no witnesses other than the parties, and in which each party’s testimony had vulnerabilities: B changed her account of the incident, and defendant fled the jurisdiction. Further, a significant amount of time lapsed between the admission of *503the prejudicial evidence and the trial court’s instruction. The jury had heard a full day of testimony by witnesses for the state, including McNamara and McCready, after which B testified, contradicting some of her original statements when she had been examined six years earlier. The jury went home for the night, and when they returned the next morning, defendant testified, denying that he had abused B. Another break occurred, and when the jury returned from lunch, now halfway through the second day of testimony, they were instructed to disregard the diagnoses from the day before. See Simpson v. Coursey, 224 Or App 145, 155, 197 P3d 68 (2008), rev den, 346 Or 184 (2009) (“To be effective, a curative instruction also must be timely.”).
We understand that jurors are “assumed to have followed their instructions,” except in cases that present an “overwhelming probability that they would be unable to do so.” State v. Terry, 333 Or 163, 177, 37 P3d 157 (2001), cert den, 536 US 910 (2002) (internal quotation marks omitted). For the reasons described above, this is such a case. Under the circumstances as a whole — a credibility determination, weaknesses in both accounts, and testimony from two medical experts directly addressing the core issue — and in light of Southard, defendant was denied the right to a fair trial, and the court abused its discretion in denying his motion for a mistrial.2
Reversed and remanded.
The bound Oregon Reports and Westlaw state that Southard was decided on October 1, 2009; however, the version of the case available on the Oregon Supreme Court website states the case was “filed September 30, 2009.” Additionally, in the trial transcript, the trial court states on October 1, 2009, that “I read the case yesterday. I knew it was there.” Nonetheless, it appears that the case was not available to the public until October 1, 2009.
Defendant also assigns as error the jury’s 11 to 1 verdict, arguing that non-unanimous verdicts are unconstitutional. We have rejected that argument in the past, State v. Cobb, 224 Or App 594, 596-97, 198 P3d 978 (2008), rev den, 346 Or 364 (2009), and we do so again.