OPINION

KLAPHAKE, Judge.

On appeal from his conviction for second-degree unintentional felony murder, Minn.Stat. § 609.19, subd. 2(1) (1998), Gordon Dennis Weaver challenges several of the district court's evidentiary rulings, including its decision to allow the assistant medical examiner to testify regarding laboratory test results on the carbon-monoxide level in the victim's blood. Appellant also challenges the district court's imposition of a 300-month sentence, which represents a double upward departure, based on the jury's finding that appellant acted with particular cruelty.

Because the laboratory test results were testimonial in nature under Crawford, and because appellant had no opportunity to cross-examine the laboratory technician, whose identity was unknown, the district court erred in allowing the assistant medical examiner to testify regarding the test results. And because this error was not harmless beyond a reasonable doubt, appellant is entitled to a new trial. We therefore reverse appellant's conviction and remand for a new trial.

FACTS

On the morning of October 16, 1999, a fire was reported at appellant's home in White Bear Lake. Police and firefighters arrived within minutes. The firefighters found the family dog in an upstairs closet and appellant's wife, Jean Weaver, face down on the basement floor with her head close to a cement laundry tub. Her body was partially burned and her head was matted with blood. She was carried outside, where paramedics unsuccessfully attempted to revive her.

The state fire marshal soon determined that the fire had been intentionally set with an accelerant, and Jean Weaver's death was deemed a homicide. Appellant was arrested and released on bail, but he disappeared in March 2000, shortly after he was indicted for first-degree intentional felony murder and second-degree unintentional felony murder under Minn.Stat. §§ 609.185(a)(3) (intentionally causing death "while committing or attempting to commit" arson), 609.19, subd. 2(1) (1998) (unintentionally causing death while committing felony). Four years later, authorities located appellant living in Oregon under a false name, and he was returned to Minnesota to face trial.

Several days into trial, the defense moved to exclude the testimony of Dr. Susan Roe, the Assistant Ramsey County Medical Examiner who performed the autopsy, and who was to testify on the results of carbon-monoxide testing on blood samples taken from the victim and the family dog. The defense argued that the test results lacked foundation because the identity of the laboratory technician who conducted the tests was unknown and the blood samples had been destroyed. The defense also argued that the reports were inadmissible hearsay and that admission of the test results would violate appellant's right to confrontation under Crawford v. Washington, 541 U.S. 36, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004).

The state explained that the medical examiner's office sent the blood samples to Regions Hospital Toxicology Department, where the tests were performed, and that in accordance with its policy, Regions destroyed the data after two years. The state insisted that "any unavailability . . . has been caused by the fact that this defendant absented himself" from the state for four years. The state further argued that the test results were admissible as the basis for their expert's opinion under Minn. R. Evid. 703.

The state called Dr. Roe in limine to make an offer of proof. Dr. Roe testified that she has performed more than 4,000 autopsies as a medical examiner. She testified that all tissue and fluid specimens she obtained during autopsies were sent to toxicology laboratories outside the medical examiner's office, like the one at Regions Hospital. Dr. Roe acknowledged that she lacked the expertise to perform such tests. But from her general knowledge and from her experience with the laboratory, she knew that Regions Hospital and its laboratory were accredited and met applicable national standards, that the instruments were calibrated on a routine basis, and that other medical examiner's offices in the metro area used hospital laboratories for such tests. Dr. Roe testified that she was confident in relying on these test results because the laboratory had been reliable in the past; the results from the blood samples were consistent, and the measurements were not unexpected under the circumstances. She further assumed that the laboratory would have known that the medical examiner's office was a medical-legal operation and that, with the chain-of-possession forms, there were legal implications. Dr. Roe testified that she received the results from these tests in two ways: either the laboratory called her office the day after the autopsy or it sent paperwork confirming the results.

The district court ruled that even though the laboratory results were hearsay, they were admissible under Minn. R. Evid. 703. The court further ruled that because the laboratory technician was unavailable, an oral report of the test results could be communicated to the jury by Dr. Roe under Minn. R. Evid. 804(5).

Dr. Roe then testified before the jury. While she appears to have referred to her autopsy report, it was not admitted into evidence. Dr. Roe testified that the victim received the injury to the back of her head when she hit her head on the concrete laundry tub. Dr. Roe explained that the force of impact caused a "contrecoup" injury in which the victim's brain began to swell and herniate in the area of the brainstem.

Dr. Roe also observed soot in the openings of the victim's nose, lips, tongue, and ears. During the internal exam, she found soot extending past the trachea and into the bronchi. Based on the presence of soot in those body recesses, Dr. Roe concluded that the victim was alive and breathing when the fire started.

Dr. Roe was then asked about the blood samples sent to Regions Hospital for carbon-monoxide testing. She testified that the dog's blood measured 85.3% and that the sample taken from the victim measured 61.4%, a lethal concentration.

Dr. Roe concluded that, based on the autopsy and test results, she believed that the victim died of asphyxia, or lack of oxygen due to carbon-monoxide poisoning caused by the fire. Dr. Roe acknowledged that while the head injury could have caused death, she did not believe it did so in this case, given evidence that the victim was alive when the fire started, had a lethal carbon-monoxide level, and showed incomplete brain herniation.

On cross-examination, Dr. Roe acknowledged that the head injury would have caused unconsciousness and profuse bleeding and that the victim was already in the process of dying from the head injury. Dr. Roe further acknowledged that she did not know who performed the carbon-monoxide testing or the type of machine on which the testing was done, and that people have survived carbon-monoxide poisoning at levels greater than the levels measured here.

The defense presented expert testimony from neurologist Dr. David Ketroser. He testified that the victim suffered a traumatic head injury that was consistent with falling backwards and hitting a cement laundry tub. In addition, Dr. Ketroser testified that all herniations of the brain are partial and that there is no such medical term as a "complete" herniation. He claimed that it was not possible to render a medical opinion that a head injury did not cause death merely because the herniation was not complete. Dr. Ketroser further testified that the victim's death was likely caused by both the head injury and carbon-monoxide poisoning, and that it was not possible to limit the cause of death to only one source with any degree of medical certainty. Finally, Dr. Ketroser was asked about methods for testing for carbon-monoxide poisoning and he opined that the method used here failed to provide a reliable measurement.

Appellant testified that on the morning of the fire, he and his wife were in the basement having a discussion when he became annoyed with her and pushed her with his forearm. He testified that she took a step back, bumped into a plastic rack used for drying clothes, and collapsed the rack, causing her to fall and hit the back of her head against the laundry tub. Appellant testified that his wife took a few deep breaths, went limp, and did not appear to be breathing or have any pulse. Appellant believed that she was dead.

Appellant testified that he panicked and decided to burn the house to destroy its contents and his wife's body. He poured accelerant on everything in the laundry room, lit a fire on a shelf, and left. He never considered calling for help or going to the police and tried to act as if nothing was wrong. He conceded that he had committed arson and had caused his wife's death.

During closing arguments, the defense argued that in order to convict appellant of first or second-degree felony murder, the jury had to find that it was the "act of arson that caused the death." The defense argued that if "you're not convinced beyond a reasonable doubt that carbon-monoxide poisoning [was] the actual cause of death . . . then you pass on to manslaughter."

The jury found appellant guilty of second-degree unintentional felony murder. The jury then heard arguments and was instructed on the aggravating sentencing factors of particularly cruelty to and particular vulnerability of the victim. The jury returned a special verdict finding that appellant had acted with particular cruelty. The district court thereafter sentenced appellant to a double upward durational departure of 300 months.

ISSUES

1. Did the district court commit reversible error by allowing Dr. Roe to testify regarding the laboratory test results and to render an opinion based in part on those results because admission of that evidence violated appellant's right to confront witnesses?

2. Did the district court err or abuse its discretion with respect to the other issues raised by appellant?

ANALYSIS I.

"[E]videntiary rulings — including the admission of chemical or scientific test reports — are within the discretion of the district court and will not be reversed absent a clear abuse of discretion." State v. Caulfield, 722 N.W.2d 304, 308 (Minn. 2006) (holding that Bureau of Criminal Apprehension (BCA) laboratory report identifying a substance as cocaine was testimonial and implicated defendant's right to confrontation). "But whether the admission of evidence violates a criminal defendant's rights under the Confrontation Clause is a question of law [which] this court reviews de novo." Id.

Testimonial Nature of Laboratory Test Results

The threshold question in any Confrontation Clause analysis is whether the disputed evidence was "testimonial" under Crawford v. Washington, 541 U.S. 36, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004). Here, the parties and the district court initially discussed whether the test results were properly admissible under Minn. R. Evid. 703(a) (allowing expert to rely on facts or data if they are of types "reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject"). The state also characterized the laboratory test results as admissible hearsay under the business records exception, Minn. R. Evid. 803(6).

The rules of evidence, however, cannot justify admission of evidence that otherwise violates the Confrontation Clause. See Caulfield, 722 N.W.2d at 310 (rejecting argument that laboratory report is not testimonial because it falls within hearsay exception for public or business records). Thus, the constitutional issue of whether admission of the laboratory test results through the testimony of Dr. Roe violated appellant's confrontation rights under Crawford takes precedence over the issue of whether the laboratory test results are admissible under the rules of evidence.

The state has the burden to prove that the laboratory test results are not testimonial. Caulfield, 722 N.W.2d at 308. The critical determinative factor in assessing whether a statement is testimonial is whether it was "prepared for litigation." Id. at 309.

Here, the laboratory test results were obtained at the request of the medical examiner, Dr. Roe, during an autopsy that occurred in the course of a homicide investigation. The results were specifically relied on by Dr. Roe in reaching her conclusion on the cause of death. And the information was relayed to the jury in lieu of testimony at trial. See id. (concluding that BCA laboratory report "bears characteristics of each of the three generic descriptions" of testimonial evidence offered by Crawford).

The Crawford Court outlined three general formulations of what should be considered "testimonial" statements. Crawford, 541 U.S. at 51-52, 124 S.Ct. 1354. But neither the Supreme Court nor our state supreme court has adopted a "comprehensive definition of `testimonial.'" Caulfield, 722 N.W.2d at 308. In the absence of further guidance, we are left with the holding in Caulfield that a BCA laboratory report identifying a substance as cocaine was testimonial. Id. at 310. We note that the medical examiner has duties independent of the police and of criminal investigations generally, unlike the BCA. See Minn. Stat. § 390.11, subd. 2 (2006). But here a homicide investigation had been started in which the carbon-monoxide testing would likely provide relevant evidence, although not evidence as clearly decisive as the drug testing in Caulfield.

Moreover, despite the state's argument that the laboratory technician was nonadversarial and removed from the prosecutorial process, Caulfield specifically rejects an approach that focuses on the intent of the declarant. Caulfield, 722 N.W.2d at 309. And despite the state's argument that the laboratory tests were not conducted in anticipation of litigation, the blood samples were sent to the laboratory after police and the medical examiner had preliminarily determined that arson had occurred and after appellant had been arrested as a suspect. Dr. Roe acknowledged that the toxicology request form that she submitted along with the samples included chain-of-possession notes, that the laboratory technician performing the tests would have known that the medical examiner's office was a medical-legal operation, and that the laboratory technician would have known that the test results might have legal implications. Dr. Roe's testimony regarding the laboratory test results was introduced as a basis for her medical opinion that the cause of death was carbon-monoxide poisoning. We therefore conclude that the laboratory test results admitted through the testimony of Dr. Roe were testimonial in nature.

Waiver or Forfeiture of Right to Confrontation

The state insists that appellant "waived" or "forfeited" the right to cross-examine the laboratory technician by fleeing and evading prosecution for four years. The state notes that it could not produce the laboratory technician to testify because the hospital, following its normal policy, destroyed the test records after two years. Indeed, the district court agreed that the records would have been available at trial had appellant not fled the state.

Under certain circumstances, usually involving trial strategy, a defendant may waive the right to confront adverse witnesses by voluntarily leaving the court-room once trial has commenced. State v. Worthy, 583 N.W.2d 270, 277-78 (Minn. 1998). But waiver requires the voluntary relinquishment of a known right; we cannot conclude that appellant knew he would lose his right to confront the laboratory technician because the laboratory reports and underlying raw data would be destroyed after two years. See Caulfield, 722 N.W.2d at 313. And while the doctrine of forfeiture by wrongdoing continues to apply to Confrontation Clause violations under Crawford, that doctrine is usually applied in cases involving witness tampering or some other type of threats to witnesses to procure their unavailability. See Caulfield, 722 N.W.2d at 311-12; see also State v. Alvarez-Lopez, 136 N.M. 309, 98 P.3d 699, 703-05 (2004), cert. denied, 543 U.S. 1177, 125 S.Ct. 1334, 161 L.Ed.2d 162 (2005) (defendant did not forfeit his right to confrontation under Crawford by absconding from state and remaining fugitive for seven years).

The loss of evidence from a delay in trial is often predictable. See generally Black v. State, 725 N.W.2d 772, 776 (Minn.App. 2007) (holding that motion to withdraw guilty plea was untimely when made more than two years later, given that drug evidence had been destroyed and witnesses' memories had likely faded). Thus, a defendant may be deemed to forfeit his right to confrontation by absconding, such as when there is a terminally ill witness. Here, however, while appellant left the state after being indicted, there is no indication that he knew Regions Hospital would destroy its records after two years. We therefore reject the state's argument that appellant has forfeited or waived his right to confront the laboratory technician by fleeing and evading trial for four years.

Harmless Error Analysis

The harmless error doctrine continues to apply to violations of the Confrontation Clause. Caulfield, 722 N.W.2d at 314. The "constitutional harmless error analysis is not a matter of analyzing whether a jury would have convicted the defendant without the error, but rather whether the error reasonably could have impacted upon the jury's decision." Id. (quoting State v. Juarez, 572 N.W.2d 286, 292 (Minn. 1997)). Thus, the analysis requires more than just consideration of the sufficiency or strength of the other remaining evidence of guilt.

In determining "whether a jury verdict was surely unattributable to an erroneous admission of evidence," we must consider the "manner in which the evidence was presented, whether it was highly persuasive, whether it was used in closing argument, and whether it was effectively countered by the defendant." State v. Al-Naseer, 690 N.W.2d 744, 748 (Minn. 2005). A final factor to consider involves whether other evidence of guilt was overwhelming. Id.

With regard to the first factor, the manner in which the evidence was presented, the trial in this case was lengthy, with a "plethora of other evidence." Caulfield, 722 N.W.2d at 314. Nevertheless, the test results were a key part of Dr. Roe's testimony and formed the basis, in part, for her opinion that the victim died from carbon-monoxide poisoning, not from the head injury she received when she was pushed by appellant. As with the BCA report involved in Caulfield, the laboratory test results were "presented in a way that was designed to secure the state's verdict." Id.

With respect to the second factor, the evidence was "highly persuasive" and not merely cumulative of other evidence presented at trial. As with the BCA test results in Caulfield, the laboratory test results at issue here have "the appearance of being conclusive proof" of the cause of the victim's death. Id. at 315.

As to the third factor, the test results were discussed by the state in its closing argument to the jury. The state cited the test results, along with the other physical observations made by Dr. Roe, as support for the conclusion that the victim was breathing and alive for at least 10 minutes after the fire was started. As in Caulfield, measured by these first three factors, "it cannot be said that the verdict was surely unattributable to the erroneous admission" of the laboratory test results. Id.

Finally, we are left with the fourth factor, whether the evidence was effectively countered by the defense, and the fifth factor, whether the other evidence of guilt was overwhelming. Although unrebutted evidence has greater impact on the verdict, the defense here vigorously attacked Dr. Roe's reliance on the laboratory test results by cross-examining her, comparing the questionable nature of the laboratory tests conducted by the hospital with the safeguards and protocol followed by the BCA analyst and by the fire marshals, and presenting the testimony of Dr. Ketroser. During closing argument, the defense spent substantial time discussing and attacking the reliability of the laboratory test results. The test results became a central and critical focus of the defense, which argued that if the state did not prove that the victim died from carbon-monoxide poisoning, then the jury must find appellant not guilty of felony murder and guilty of some lesser offense.

The state argues on appeal that the defense focus on the cause of death was irrelevant because the jury was properly instructed that it need only find that appellant caused the victim's death while "committing or attempting to commit" arson. The state appears to be correct on the law. See State v. Nielsen, 467 N.W.2d 615, 618 (Minn. 1991) (holding that felony-murder rule applies whenever felony and homicide are parts of one continuous transaction, and rejecting defendant's argument that he should have been acquitted of felony murder because rape was afterthought to homicide and the autopsy did not disclose whether rape took place before or after death). But the prosecutor did not present this theory or make this specific argument to the jury, which was left with the jury instructions and with defense counsel's insistence that a finding of death from carbon-monoxide poisoning and not from the head injury was necessary for a felony murder conviction.

With respect to the final factor, even if other evidence of guilt was overwhelming, that does not necessarily mean that the error was harmless. Caulfield, 722 N.W.2d at 316. Here, the laboratory test results became so critical to the defense and were so thoroughly discussed that they surely attributed to the verdict. While the jury could have reached the same verdict without the laboratory test results, that evidence surely influenced the verdict in some way. We therefore conclude that the error in admitting this evidence was not harmless and that appellant is entitled to a new trial.

II.

Given our decision to grant appellant a new trial, we need not discuss in detail the remaining issues raised by appellant. We nevertheless mention the sentencing issue to give some guidance to the district court on retrial.

At sentencing, appellant concedes that the district court followed the procedure endorsed by the supreme court in State v. Chauvin, 723 N.W.2d 20, 27 (Minn. 2006) (holding that district courts have inherent judicial authority to impanel sentencing jury for purposes of departing from presumptive sentence). Appellant merely challenges the instructions given to the jury on the aggravating factor of "particular cruelty." The district court did not provide the jury with a definition for "particular cruelty" and merely read the special verdict form, which tracked the language of the guidelines and the standard jury instructions, as follows: "The first [aggravating factor] that's being submitted to you is whether the victim . . . was treated with particular cruelty for which the defendant should be held responsible." See Minnesota Sent. Guidelines II. D.2.b.(2); 10 Minnesota Practice, CRIMJIG 8.01 Spec. Verdict Form (2006). This instruction fails to provide sufficient guidance to the jury.

When trial judges relied on their collective experience or collegial knowledge of typical cases, a definition of "particular cruelty" was unnecessary. See State v. Smith, 541 N.W.2d 584, 589-90 (Minn. 1996). With sentencing juries, however, "particular cruelty" is a relative term that requires a uniform meaning irrespective of the jurors' lay understanding of the term. The failure to define "particular cruelty" raises a multitude of problems, as identified in a recent law review article. See Kevin S. Burke, State v. Dettman: The End of the Sentencing Revolution or Just the Beginning?, 33 Wm. Mitchell L.Rev. 1331, 1341-44 (2007).

While we agree with appellant that the instruction given to the jury needs to precisely define "particular cruelty," we do not agree with appellant's proposed instruction, which requires conduct that is done with intent to harm. See State v. Bicek, 429 N.W.2d 289, 292 (Minn.App. 1988), review denied (Minn. Nov. 23, 1988). In general, a defendant's conduct must be significantly more cruel than that usually associated with the offense of which he was convicted. See, e.g., Holmes v. State, 437 N.W.2d 58, 59 (Minn. 1989) (departure unjustified when conduct not significantly different from that typically involved in crime); State v. Hanson, 405 N.W.2d 467, 469 (Minn.App. 1987) (departure not warranted when defendant did not commit manslaughter in a manner significantly more serious than a typical manslaughter). In the past, judges have imposed upward sentencing departures based on "particular cruelty" when the defendant's conduct included threats, degradation of the victim, or the gratuitous infliction of pain. See, e.g., Smith, 541 N.W.2d at 590. Courts have also found "particular cruelty" to exist when a defendant leaves the victim to die alone without notifying emergency personnel, sets fire to a victim who is still alive, or attempts to conceal or destroy a victim's body. See, e.g., State v. Folkers, 581 N.W.2d 321, 327 (Minn. 1998); State v. Jones, 328 N.W.2d 736, 738 (Minn. 1983); State v. Gurske, 424 N.W.2d 300, 305 (Minn.App. 1988); State v. Dircks, 412 N.W.2d 765, 767-68 (Minn.App. 1987), review denied (Minn. Nov. 24, 1987).

We acknowledge the possibility that a new criterion may be necessary, one that is easily defined and readily understood by jurors. But we decline to provide an exact definition of the phrase "particular cruelty," and leave the matter to the district court and attorneys in this matter should the issue arise again.

DECISION

Because the laboratory test results were testimonial under Crawford and because their admission through the testimony of Dr. Roe violated appellant's right to confront the laboratory technician who performed the tests, the district court erred in allowing admission of this evidence. Moreover, because the erroneous admission of this evidence was not harmless beyond a reasonable doubt, appellant is entitled to a new trial. We therefore reverse the conviction and remand for a new trial.

Reversed and remanded.