dissenting.
{¶ 45} The majority opinion misconstrues the applicable case law in reaching its conclusion. I conclude that a forensic interview cannot be both testimonial and nontestimonial without violating a defendant’s Sixth Amendment right to confront the witnesses against him.
{¶ 46} The majority opinion ably explains the law of the Sixth Amendment’s Confrontation Clause as elucidated by various federal and Ohio cases. I will not redescribe these cases. The majority opinion also fairly characterizes the case *304law from other states that it summarizes in the section titled “Other State Supreme Court Decisions.” I dissent, not based on the majority opinion’s understanding of the law, but because of the way the majority opinion applies the law to this case.
{¶ 47} The majority opinion acknowledges that many of the questions asked by the forensic interviewer, Kerri Marshall, were asked “to gather forensic information” and are, therefore, testimonial. In Davis v. Washington (2006), 547 U.S. 813, 828-829, 126 S.Ct. 2266, 165 L.Ed.2d 224, the court held that statements of a witness who is unavailable for cross-examination should be redacted or excluded to avoid violating the defendant’s right to confront witnesses against him. The testimonial statements in this case were not redacted or excluded. Furthermore, the testimonial statements in this case are different from those discussed in Davis. In Davis, the testimonial statements were made after a series of nontestimonial statements had concluded. Id. The interrogator in that case had elicited statements to assist the police in meeting an ongoing emergency: those statements were nontestimonial, and their admission as evidence was permissible. Id. After eliciting the initial statements, the interrogator asked a series of questions attempting to elicit information about the alleged crime. Id. at 828. The court found “no great problem” with this approach because the nontestimonial statements were separate and distinct from the testimonial statements. Id. at 829. The court stated that questioning could evolve from addressing an emergency to eliciting forensic information. Id. at 828. According to the Supreme Court, testimonial statements may not be introduced as evidence and if they are part of a transcript or other document, they must redacted. Id. at 829.
{¶ 48} The majority opinion makes creative use of the Davis opinion. First, it concludes that the concept of nontestimonial statements evolving into testimonial statements applies when an interrogator has a dual purpose. Davis does not support this conclusion. The interrogator in Davis did not have dual purposes: she had two separate purposes. She completed the questions regarding the ongoing emergency and then moved on to elicit information that could be used as evidence. Marshall’s questions, which elicited, in the opinion of the majority, both testimonial and nontestimonial statements, were interspersed, rendering it difficult to distinguish those that should be redacted from those that need not be redacted. Second, in this case there was no ongoing emergency. The emergency occurred the night before, so there was no occasion for the questioning to evolve from eliciting nontestimonial statements to eliciting testimonial ones. Third, the testimonial statements in this case, which the majority opinion concedes exist, were not redacted. The majority opinion relies on Davis, but only to the extent that Davis suits its purposes.
*305{¶ 49} Although remanding the cause to enable the court of appeals to determine whether the error is harmless is better than finding it harmless, we should do neither. It is clear from the record that the error in this case was not harmless. In State v. Conway, 108 Ohio St.3d 214, 2006-Ohio-791, 842 N.E.2d 996, ¶ 78, we stated that the determination of whether a constitutional error is harmless “is not simply an inquiry into the sufficiency of the remaining evidence. Instead, the question is whether there is a reasonable possibility that the evidence complained of might have contributed to the conviction.” It is patently obvious that the testimonial statements in this case “might have contributed to the conviction.”
{¶ 50} Another troubling aspect of the majority opinion is its implicit conclusion that Marshall’s questions were medically necessary. Although I concede that they may have been helpful, they were not necessary. First, a doctor had examined M.A. the previous evening. That doctor would have done and asked everything necessary to treat M.A. at that time. See State v. Hooper (2007), 145 Idaho 139, 141, 176 P.3d 911, about which the majority opinion states, “Because the interview occurred after the child met with and was examined by the physician, the subsequent interview served a forensic, not a medical or treatment-oriented, purpose.” Majority opinion at ¶ 23. Second, as discussed below, a nurse practitioner examined M.A. after Marshall’s questioning. She would have asked all medically relevant questions during her examination.
{¶ 51} The testimonial statements in this case were neither redacted nor harmless. Nevertheless, the majority opinion concludes that the testimonial statements do not violate the defendant’s right to confront witnesses. I will now explain why I believe that all of the statements elicited by Marshall were testimonial and, therefore, improperly admitted into evidence.
Marshall’s Interview with M.A.
{¶ 52} Kerri Marshall is a licensed social worker employed by CCFA as a medical forensic interviewer. Marshall described her job duties as interviewing children who are alleged to be victims of sexual or physical abuse. She testified that law-enforcement personnel customarily observe the interviews that she conducts and that the children are not aware that they are being observed. Marshall’s interview with M.A. was contemporaneously broadcast to another room over closed-circuit television, where it was viewed by several people, including a police detective. It was also recorded on a DVD.
{¶ 53} During her interview with M.A., Marshall asked many questions about the events of the previous evening. Some of the questions were not relevant to an ongoing medical emergency or to medical treatment. For example, Marshall asked M.A., “How did your underwear get off?” “Did daddy’s pee-pee touch your pee-pee?” and “Were you laying down or sitting up when daddy played pee-*306pees with you?” Marshall subsequently prepared a report of the interview, entitled “Medical Forensic Interview Summary.” In this report, Marshall noted that Arnold “fled the home by stealing [Otto’s] purse and her car” after Otto confronted him. Marshall recommended that M.A. be “protected from any contact with alleged perpetrator as this investigation continues.”
Statement made in the course of a police interrogation
{¶ 54} The issue in this case, as it was in Stahl and Siler, is to determine whether the hearsay statements that were offered by the prosecution and that the defendant argued violated his right under the Sixth Amendment to confront a witness are testimonial. A threshold question, however, is whether the statements were made in the course of a police interrogation. See Davis, 547 U.S. at 822, 126 S.Ct. 2266, 165 L.Ed.2d 224; Siler, 116 Ohio St.3d 39, 2007-Ohio-5637, 876 N.E.2d 534, at ¶ 30-31. What constitutes “police interrogation” for purposes of Confrontation Clause analysis has not been addressed by the United States Supreme Court. See Davis at 823, fn. 2.
{¶ 55} It is, of course, plainly obvious that a police officer did not conduct the interrogation in this case; social worker Kerri Marshall conducted the interrogation. The question becomes: was Marshall an agent of law enforcement when she conducted the interrogation? Id. For the reasons that follow, I conclude that she was.
{¶ 56} Although the state argues that Crawford and Davis apply only when the interviewer is a law-enforcement officer, the cases do not support such a narrow interpretation. I am persuaded that Crawford and Davis define a broader constitutional protection from out-of-court statements that are obtained primarily to assist in a criminal prosecution, regardless of whether the interrogator is a police officer or an agent of the police. Davis, 547 U.S. at 822-823, 126 S.Ct. 2266, 165 L.Ed.2d 224; Crawford, 541 U.S. at 51-53, 124 S.Ct. 1354, 158 L.Ed.2d 177. See Crawford at 50 (the principal evil at which the Confrontation Clause was directed was the civil-law mode of criminal procedure, and particularly its use of ex parte examinations as evidence against the accused).
{¶ 57} In Siler, we stated that courts have consistently applied the primary-purpose test to statements that a child declarant made to police or those determined to be police agents. Siler, 116 Ohio St.3d 39, 2007-Ohio-5637, 876 N.E.2d 534, at ¶ 29. In one of those cases, the Supreme Court of North Dakota stated. “In cases since Crawford, other states with the functional equivalent of the Children’s Advocacy Center (‘CAC’) involved in this case have held that similar statements made by a child with police involvement inevitably are testimonial.” State v. Blue (2006), 199 N.D. 50, 717 N.W.2d 558, ¶ 15. A Florida court of appeals has considered four factors to determine whether the interrogation at issue was “the functional equivalent of a police interrogation. These four *307factors are (1) the effect of the Florida statutes pertinent to the establishment and functioning of the CPT [the Florida equivalent of a CAC], (2) the nature and extent of law enforcement involvement in the examination of the child by [the nurse practitioner] at [the hospital], (3) the purpose of the examination performed by [the nurse practitioner] in her capacity as a member of the CPT, and (4) the absence of any ongoing emergency at the time [the nurse practitioner] conducted her examination of the child.” Hernandez v. State (Fla.App.2007), 946 So.2d 1270,1280. Although I would not adopt this four-part test, the factors are helpful in determining whether Marshall was acting as an agent of the police when she interrogated M.A.
{¶ 58} First, the statutory scheme that authorized the creation of CACs contains provisions that establish a link between the CACs and law enforcement. R.C. 2151.426 and 2151.427. See Ohio Adm.Code 5101:2-33-26. Second, a police detective watched the interrogation as it was happening and the interrogation was recorded and saved to a DVD. Third, focusing primarily on issues that were not medical, the interview was memorialized as a “Medical Forensic Interview Summary,” suggesting that the purpose was forensic, not medical. Fourth, there was no ongoing emergency while the interview was conducted.
{¶ 59} Furthermore, Marshall is not a medical professional; her job title is “medical forensic interviewer.” “Forensic” means “[u]sed in or suitable to courts of law or public debate.” Black’s Law Dictionary (9th Ed.2009) 721. The Michigan Department of Human Services has stated that “[t]he goal of a forensic interview is to obtain a statement from a child * * * that will support accurate and fair decision-making in the criminal justice and child welfare systems,” and that “the interview is not part of a treatment process.” State of Michigan, Forensic Interviewing Protocol, at http://www.michigan.gov/documents/dhs/DHS-PUB-0779_211637_7.pdf (accessed May 25, 2010).
{¶ 60} I conclude that Marshall was an agent of the police when she conducted her forensic interview of M.A. See Blue, 2006 ND 134, 717 N.W.2d 558, ¶ 14-16, and the cases cited therein; State v. Mack (2004), 337 Or. 586, 593, 101 P.3d 349 (Department of Human Services caseworker was a proxy for the police).
Application of the primary-purpose test
{¶ 61} The next step is to determine whether the primary purpose of the interrogation was to enable police assistance to meet an ongoing emergency. Siler, 116 Ohio St.3d 39, 2007-Ohio-5637, 876 N.E.2d 534, at paragraph one of the syllabus, quoting Davis, 547 U.S. at 822, 126 S.Ct. 2266, 165 L.Ed.2d 224. First, the interview involved a description of past events. The alleged abuse had occurred the previous evening, and the questioning specifically attempted to obtain a description of the abuse. Second, a reasonable observer would not perceive an ongoing emergency at the time of questioning. The patient had been *308discharged from the hospital the previous evening. At oral argument, counsel conceded that no medical emergency existed at the time of Marshall’s interview. Third, the questioning was not necessary to resolve an emergency because there was no ongoing emergency. Finally, the interview was rather formal, more akin to the videotaped, planned interview of Crawford than to the frantic 9-1-1 call or the sequestered but spur-of-the-moment interview recounted in Davis. Each factor independently suggests that there was no ongoing emergency; collectively, the conclusion is manifest.
{¶ 62} Because the primary purpose of the Marshall interview was not to meet an ongoing emergency, the next step is to evaluate the entirety of the factual circumstances surrounding the interview to establish whether its primary purpose was “to establish or prove past events potentially relevant to later criminal prosecution.” Davis, 547 U.S. at 822, 126 S.Ct. 2266, 165 L.Ed.2d 224.
{¶ 63} A CAC can be established only by a children’s services agency, law enforcement, or a prosecutor, and the CAC is responsible for assembling a multidisciplinary team. R.C. 2151.426 and 2151.427(A). The multidisciplinary team must include law enforcement and prosecuting attorneys as members. Id. The statutory connection between CACs and law enforcement suggests that CACs are not solely medical-treatment providers and that a CAC interviewer can be an agent of the police.
{¶ 64} The circumstances of the interview indicate that its primary purpose was “to establish or prove past events potentially relevant to later criminal prosecution.” Davis, 547 U.S. at 822, 126 S.Ct. 2266, 165 L.Ed.2d 224. See Siler, 116 Ohio St.3d 39, 2007-Ohio-5637, 876 N.E.2d 534, paragraph one of the syllabus. Police observed the interview, which the state concedes is a customary practice. A DVD recording of the interview was preserved, a strong indication that the purpose of the interview was to obtain evidence for use by the prosecution. I am unaware of doctors videotaping patient interviews to assist them in medical treatments or of doctors allowing police officers to routinely observe them when they examine their patients.
{¶ 65} Furthermore, many of the questions asked were investigatory in nature and similar to the questions asked in a direct examination in a judicial proceeding. See Davis, 547 U.S. at 830, 126 S.Ct. 2266, 165 L.Ed.2d 224. For example, questions about how the underwear was removed, who did so, and the specific positions — standing up or lying down — in which the alleged abuse occurred represent an attempt to gain specific details of past events. If the questions have a medical purpose, it is secondary to their investigatory purpose. I might view Marshall’s questions differently if she were a nurse, as in Stahl, but she is not. Furthermore, the nurse practitioner would have made all inquiries relevant to medical treatment during the physical examination after the social worker *309questioned M.A. That the nurse practitioner stated that Marshall’s interview “guides my exam” is no doubt true to some degree. But nurse practitioners are highly educated professionals; they do not need an intermediary. In my opinion, the intermediary was interjected in order to elicit forensic evidence, not to assist in the medical examination.
{¶ 66} It is objectively apparent from the record that Marshall asked questions to assist in the police investigation. The circumstances of this case are quite different from State v. Muttart, 116 Ohio St.3d 5, 2007-Ohio-5267, 875 N.E.2d 944, ¶ 62, in which hearsay statements obtained by medical personnel in the course of treatment survived a Sixth Amendment challenge. In Muttart, police did not observe the interview and the interview was not videotaped. The state argues that the questions that Marshall asked M.A. helped to assess the need for future counseling but fails to show whether any counseling occurred. Even if it had occurred, a recommendation for counseling alone would be insufficient to establish that the interview was primarily for medical purposes.
{¶ 67} The critical evidence in this case is Marshall’s report, something the majority opinion does not address. It uses the words “perpetrator” and “allegations” and includes a witness list, an item typically not found in a medical report. The report states that the patient should have no contact with the “alleged perpetrator as this investigation continues.” The report indicates that Marshall believed that she was assisting an ongoing investigation targeting a particular criminal suspect. When interviewers believe themselves to be participants in an investigation that has targeted a particular criminal suspect, they conduct precisely the type of ex parte examinations that the Confrontation Clause protects against. Crawford, 541 U.S. at 50-53, 124 S.Ct. 1354, 158 L.Ed.2d 177.
Conclusions reached by sister states
{¶ 68} Since Crawford was decided, many state supreme courts have addressed the issue before us. In at least eight cases, state supreme courts have concluded that out-of-court statements by child sexual-assault victims to various non-law-enforcement personnel were nontestimonial. Seely v. State (2008), 373 Ark. 141, 282 S.W.3d 778; People v. Vigil (Colo.2006), 127 P.3d 916; State v. Arroyo (2007), 284 Conn. 597, 935 A.2d 975; Commonwealth v. DeOliveira (2006), 447 Mass. 56, 849 N.E.2d 218; State v. Krasky (Minn.2007), 736 N.W.2d 636; Hobgood v. State (Miss.2006), 926 So.2d 847; State v. Spencer (2007), 339 Mont. 227, 2007 MT 245, 169 P.3d 384; State v. Vaught (2004), 268 Neb. 316, 682 N.W.2d 284. Each of these cases turned on factual determinations that are not present in this case. See, e.g., Seely, 373 Ark. at 156, 282 S.W.3d 778 (the primary purpose of an interview conducted by a social worker “was medical treatment”); Vigil, 127 P.3d at 927 (statements to doctor were for purposes of medical diagnosis); Hobgood, 926 So.2d at 852 (statements were made to people who “were not working in *310connection with the police” or were made for the purpose of seeking medical treatment).
{¶ 69} At least nine state supreme courts have concluded that out-of-court statements by child sexual-assault victims to non-law-enforcement personnel are testimonial. State v. Contreras (Fla.2008), 979 So.2d 896; Hooper, 145 Idaho 139, 176 P.3d 911; In re Rolandis G., 232 Ill.2d 13, 327 Ill.Dec. 479, 902 N.E.2d 600; State v. Bentley (Iowa 2007), 739 N.W.2d 296; State v. Henderson (2007), 284 Kan. 267, 160 P.3d 776; State v. Snowden (2005), 385 Md. 64, 867 A.2d 314; State v. Justus (Mo.2006), 205 S.W.3d 872; Blue, 2006 ND 134, 717 N.W.2d 558; Mack, 337 Or. 586, 101 P.3d 349. Each of these cases involves an interviewer who performed in circumstances substantially similar to the facts before us. See, e.g., Blue, 2006 ND 134, 717 N.W.2d 558, ¶ 2-3 (a forensic interviewer conducted the interview while a police officer watched; the officer was given a videotaped recording of the interview); Contreras, 979 So.2d at 905 (interview by child-protection-team coordinator was watched by police officer and recorded); Bentley, 739 N.W.2d at 297, 300 (interview by counselor at child-protection center was watched by police officers, who took videotaped copy of interview with them). My conclusion in this case is bolstered by the fact that the majority of our sister courts that have considered substantially the same issue have reached the same conclusion that I reach.
Conclusion
{¶ 70} I conclude that the primary purpose of Marshall’s forensic interview was to establish or prove past events potentially relevant to later criminal prosecution. I conclude, therefore, that the statements were testimonial and that their admission violated the Confrontation Clause of the Sixth Amendment to the United States Constitution. I would reverse the decision of the court of appeals. I dissent.