Cecil Tolbert was convicted of one count of armed kidnapping and two counts of sexual battery with the use or threatened use of a deadly weapon. The court sentenced Tolbert to life imprisonment for the kidnapping charge and thirty years each for the sexual battery charges, with all sentences to run concurrently. Tolbert appeals the trial court’s decision to allow an expert DNA analyst to testify that another DNA analyst found male DNA in a sample taken from the victim. We agree that this was inadmissible hearsay, but we find that the error was harmless.
On July 11, 1996, the victim was walking to a store at about 1:30 a.m. As the victim was walking, a man drove his car near the victim, threatened her with a gun, and told the victim to get into the car. The victim got into the car, and the man drove to an area behind a warehouse. The man forced the victim to perform oral sex on him and vaginally raped her. At some point, the back of the victim’s head was injured and she passed out or fell asleep. When the *293victim awoke, the man sexually assaulted her again. The man then released the victim.
The victim went to a hospital and was treated for the injury on the back of her head. The victim then went to a sexual assault treatment center, where a nurse practitioner took vaginal swabs. The swabs were sealed and sent to the Bro-ward Sheriffs Office (“BSO”).
Kevin Noppinger, a DNA analyst with the BSO, analyzed the swabs using restriction fragment length polymorphism (RFLP) testing. Noppinger found male DNA in the swabs taken from the victim, but the DNA profile did not match any known suspect.
By 2001, DNA technology had advanced to a method called short tandem repeat (STR).1 RFLP results and STR results cannot be compared, so older DNA samples that had been tested using the RFLP method had to be retested using the STR method. BSO had insufficient resources to retest the older samples, and in 2003 the federal government provided a grant to allow BSO to outsource the retesting of older DNA samples to reduce the backlog. BSO selected older DNA samples that had yielded RFLP results and outsourced the samples to Bowdy Technology (“Bowdy”) for STR testing. Bowdy tested the samples from the victim’s case. The Bowdy analysts found the victim’s DNA in the samples but, unlike BSO analyst Noppinger, the Bowdy analysts did not find male DNA in the portion of the victim’s samples they tested.
Bowdy reported its results to BSO. Lynn Baird, a BSO DNA analyst, hypothesized that there must have been male DNA in the samples taken from the victim because Noppinger had found it during the 1996 testing. Assuming that the STR testing performed by Bowdy may have simply failed to detect the male DNA, Baird herself retested the victim’s samples and, in fact, found male DNA. Using the STR method, Baird isolated the male DNA profile and provided it to the Combined DNA Index System (CODIS)2, a federal database of DNA profiles.
CODIS matched the male DNA profile from the victim’s sample to Tolbert’s DNA sample. BSO then located Tolbert and received a DNA sample from him, which Baird then compared to the male DNA sample she recovered from the victim’s sample. Baird determined Tolbert’s DNA was in the victim’s sample, leading to the state charging Tolbert in the instant case.
At the 2011 trial, the victim was unable to remember several portions of the 1996 incident. The victim was unable to identify Tolbert as the man who raped her and the state’s primary evidence of Tolbert’s involvement in the crime was the DNA evidence.
BSO DNA analyst Baird testified at trial. During her direct examination, the state asked her about Noppinger’s findings from the 1996 testing:
The state: Did you receive a sample of a DNA involving [the victim] and an unknown suspect?
[[Image here]]
Baird: I personally — it was submitted to the laboratory but I personally did not do the original analysis on this case.
The state: When — were you able to refer to the report that was done back then?
*294Baird: Yes.
The state: What was originally detected back in 1996?
The defense: Objection, hearsay.
The court: I’m gonna overrule that.
Baird: The analysis in 1996 was originally performed by Kevin Noppinger ... he was able to detect spermatozoa, that’s the male reproductive cell, on the vaginal swabs submitted as being in [the victim]. He did the RFLP DNA analysis on that item and he obtained a foreign DNA profile that was foreign to [the victim]. So at that point that was basically where the case stopped. We knew we had a foreign DNA profile there but we had no one to compare it to.
Baird then went on to describe how the samples were sent to Bowdy and that Bowdy did not find male DNA in the samples. Baird explained that the manner in which Bowdy tested the samples could account for the failure to find the male DNA in the sample. An analyst from Bowdy who tested the samples testified that occasionally a DNA profile was missed when only a portion of the samples were tested. The Bowdy analyst also testified that she herself had missed such samples in the past.
Baird explained to the jury that she retested the samples, found male DNA, and isolated the male DNA profile. Baird also testified that after Tolbert’s DNA profile was provided to her, she compared Tolbert’s DNA to the male DNA in the sample taken from the victim. Baird testified that Tolbert’s DNA was in the sample taken from the victim.
Martin Tracey, a biology professor specializing in population genetics, testified that the odds of “reaching into the human population and pulling out another individual with the same genetic sequence as Cecil Tolbert” were “approximately one in four hundred and eighty-seven quadrillion.”
On appeal, Tolbert argues the trial court reversibly erred by allowing Baird to testify regarding the 1996 DNA testing results. “The standard of review for admissibility of evidence is abuse of discretion, limited by the rules of evidence. Whether evidence falls within the statutory definition of hearsay is a matter of law, subject to de novo review.” Lucas v. State, 67 So.3d 332, 335 (Fla. 4th DCA 2011) (citations, alteration, and quotation marks omitted).
“ ‘Hearsay1 is a statement, other than one made by the declarant while testifying at the trial or hearing, offered in evidence to prove the truth of the matter asserted.” § 90.801(l)(c), Fla. Stat. (2011). Written reports are statements within the meaning of the hearsay rule. See Rivera v. State, 917 So.2d 210, 211 (Fla. 5th DCA 2005). A party may not evade the hearsay rules by having the witness summarize the statement rather than relay the statement verbatim. See Diaz v. State, 890 So.2d 556, 558 (Fla. 5th DCA 2005). Although an expert may rely on hearsay in reaching the expert’s opinion, “an expert’s testimony may not merely be used as a conduit for the introduction of the otherwise inadmissible evidence.” Linn v. Fossum, 946 So.2d 1032, 1037-38 (Fla.2006) (citations and quotation marks omitted).
We find that Baird’s testimony was hearsay. Baird clarified that she lacked personal knowledge of the 1996 testing and that she was relying on Noppinger’s written report. Baird summarized Noppinger’s findings from his written report, which was likewise hearsay.3
We turn now to the pivotal issue of whether the error was harmless. In assessing whether an error is harmless, “[t]he question is whether there is a rea*295sonable possibility that the error affected the verdict.” Bunche v. State, 5 So.Bd 38, 41 (Fla. 4th DCA 2009) (citation and quotation marks omitted).
Tolbert argues that Baird’s discussion of Noppinger’s findings was harmful because she improperly bolstered her finding that there was a male profile in the sample. “Bolstering” can be harmful error when the expert indicates that another expert reached the same conclusion as the testifying expert, or when the testifying expert testifies that the expert relied on a non-testifying expert’s opinion. See Miller v. State, — So.3d -, 2012 WL 6028048, 37 Fla. L. Weekly D2780, *5-6 (Fla. 4th DCA Dec. 5, 2012) (determining error was harmful when handwriting expert testified that non-testifying expert also determined handwriting belonged to defendant); Potts v. State, 57 So.3d 292, 294 (Fla. 4th DCA 2011) (determining error was harmful when fingerprint analyst testified that another fingerprint expert determined fingerprints belonged to defendant); Telfort v. State, 978 So.2d 225, 226-27 (Fla. 4th DCA 2008) (determining error was harmful when handwriting expert bolstered his testimony).
In contrast to Miller, Potts, and Telfort, in this case the discussion of the non-testifying expert’s findings did not directly implicate Tolbert. Baird’s discussion of Noppinger’s findings established why Baird chose to retest the victim’s sample after Bowdy did not find male DNA in the sample; Noppinger’s findings did not independently establish Tolbert’s guilt. Baird did not consult with Noppinger to reach her conclusion or rely on any of his conclusions; she independently tested the victim’s sample and determined that the male profile matched Tolbert. Baird’s discussion of the 1996 testing bolstered her expert opinion only to the extent that it supported her finding that there was male genetic material in the sample taken from the victim. The more important portion of Baird’s testimony was that the male profile in the victim’s sample matched Tolbert, and the 1996 test results were never matched to Tolbert.
Tolbert argues the hearsay evidence was harmful because the DNA evidence was essentially the only evidence of Tolbert’s guilt and because Bowdy did not find a male profile in the victim’s sample. Tol-bert argues that if the jury had not heard that Noppinger found male DNA in the victim’s sample in 1996, the jury would have disbelieved Baird’s assertion that there was male DNA in the sample or the jury would have believed the victim’s sample was contaminated.
Baird testified that she tested the samples taken from the victim, she found and isolated male DNA in the sample, and she determined that the male DNA from the victim’s sample matched Tolbert’s DNA. Baird and the analyst from Bowdy both testified that it was possible to miss a male DNA profile in a sample, they offered explanations for how the male profile was missed in this case, and the Bowdy analyst testified that she had missed such profiles in the past. Most importantly, Noppinger’s findings did not implicate Tolbert because Noppinger only found an unmatched DNA profile. We determine, under the facts of this case, that the error in admitting the hearsay evidence was harmless.
Affirmed.
MAY, C.J., and STEVENSON, J., concur.. Lynn Baird, a BSO DNA analyst, described the STR method as a technique that detects DNA fragments and makes millions of copies of the fragments, which allows the DNA analyst to compare very small amounts of DNA to a suspect's DNA.
. CODIS includes the National DNA Index System (NDIS), which is a database of DNA profiles contributed by federal and state forensic laboratories that was implemented in 1998. See 42 U.S.C. § 14132 (2006) (authorizing establishment of DNA index system).
. We note that the state did not lay a foundation for the business records exception or any *295other hearsay exception. See § 90.803(6), Fla. Stat. (2011).