Defendant appeals his convictions on two counts of criminally negligent homicide, ORS 163.145, two counts of fourth-degree assault, ORS 163.160, and one count of driving while intoxicated, ORS 813.010. Defendant argues that the trial court abused its discretion in excluding certain evidence under OEC 403, and that the trial court erred in allowing the state to call as an expert witness an accident reconstructionist who had originally been retained to investigate the accident for the defense but was not called to testify by the defense. Defendant asserts that the expert’s opinion concerning the causation of the accident that led to the charges against him were privileged under OEC 503. Because we agree with defendant’s second argument, we do not address his first assignment of error.

At about 9:00 p.m. on November 12, 1995, while defendant was driving his pickup truck across a bridge on Highway 38 near Elkton, he crossed the center line and struck an oncoming car. The driver, Keelie Garrison, and one of the passengers, Sarah Robbins, were killed instantly. Two other passengers were injured and another passenger had no injuries. A witness who arrived very shortly after the collision spoke with defendant and noticed an odor of alcohol on his breath. The witness reported his observation to a police officer who arrived shortly thereafter. Defendant denied having anything to drink that evening. The officer gave defendant several field sobriety tests and concluded that he was impaired. Defendant consented to an Intoxilyzer test, which was administered at 11:41 p.m. and which showed defendant’s blood alcohol level to be .088 percent. Blood and urine tests done several hours later also revealed the presence of alcohol, as well as marijuana metabolites, in defendant’s system.

At trial, a waitress from a nearby inn testified that she had served defendant beer earlier in the evening and tequila at about 8:00 p.m. Defendant drank some coffee, then left the inn between 8:45 and 9:00 p.m. At about 9:00 p.m., a motorist who was driving on Highway 38 at approximately 55 miles per hour was overtaken and passed by defendant’s pickup truck. The motorist and her passenger both testified *529that they believed that the truck was traveling at least 75 miles per hour. Less than a minute later, the motorist came on the accident scene on the bridge.

The state called as an expert witness an accident reconstructionist, Tom Fries, who testified that, in his opinion, the accident occurred because defendant had come around the corner before the bridge too fast and crossed the center line and then over corrected, causing him to strike the oncoming car. He estimated that defendant’s truck had been going between 53 and 80 miles per hour and that the car had been going between 42 and 49 miles per hour when the impact occurred.

Defendant called John Talbot, another accident reconstructionist, who testified that Fries’ reconstruction of the accident was physically impossible and that he believed that the collision had occurred because defendant’s steering had locked up. He estimated that defendant’s speed had been between 49 and 58 miles per hour at the time of the collision. Several of defendant’s friends testified that they had noticed problems with the steering on defendant’s truck prior to the collision.

The state sought to introduce evidence from another accident reconstructionist, John Myers, to rebut Talbot’s testimony. Defendant objected on the ground that Myers’ opinion was protected under OEC 503, because Myers originally had been retained by the defense to investigate the accident. The court then received testimony from Daniel Bouck, who had been defendant’s attorney at the time that Myers was retained to investigate the accident. Bouck testified that he had retained Myers to analyze data from the collision and to render an opinion as to the cause of the collision. Myers went with Bouck and a defense investigator to the scene of the collision to gather data. Bouck and Myers discussed a number of potential theories of the case, including the theory put forth by the state at the preliminary hearing. Bouck had not decided whether he would retain Myers to testify at trial, but he did intend to retain Myers to explain the mathematics and physics necessary for Bouck to be able to cross-examine the state’s accident reconstructionist. Bouck did not intend for Myers to disclose their theories to the state. Bouck related to *530Myers at least one statement that had been made by defendant and possibly more. The trial court ruled that Myers’ opinion as to the cause of the collision was not privileged under OEC 503, but the fact that Myers had been retained by the defense, had investigated the scene with defense counsel and had developed theories about the case with defense counsel would be inadmissible, as would defendant’s statements that had been related to Myers by defense counsel.

On rebuttal, Myers testified that he had examined the site of the collision, as well as police reports and photographs, and that, in his opinion, the collision occurred because defendant lost control of the vehicle, possibly due to overinflation of the tires and hydroplaning. He estimated the speed on defendant’s truck at the time of the collision to be 45 to 65 miles per hour. He also testified that he disagreed with Talbot’s reconstruction of the accident.

On appeal, defendant asserts that the trial court erred in admitting Myers’ testimony on rebuttal, because Myers’ opinion about the cause of the collision was developed for the defense in anticipation of litigation and, thus, fell within the privilege codified at OEC 503, or within the work product doctrine. “A client has a privilege to * * * prevent any other person from disclosing confidential communications made for the purpose of facilitating the rendition of professional legal services to the client * * * [bletween the client’s lawyer and the lawyer’s representative.” OEC 503(2)(b). “ ‘Representative of the lawyer’ means one employed to assist the lawyer in the rendition of professional legal services, but does not include a physician making a physical or mental examination under ORCP 44.” OEC 503(1)(e). The legislative commentary to the OEC states:

“The definition of ‘representative of the lawyer’ is consistent with present Oregon law. It recognizes that in rendering legal service, a lawyer may use advisors and assistants in addition to those employed in the process of communicating. The definition includes an expert who is hired to assist in rendering legal advice or to help in the planning and conduct of litigation, but not one employed to testify as a witness.” Legislative Commentary to OEC 503, reprinted in Laird C. Kirkpatrick, Oregon Evidence 209-10 (3d ed 1996) (emphasis supplied).

*531A “confidential communication” is a “communication not intended to be disclosed to third persons other than those to whom disclosure is in furtherance of the rendition of professional legal services to the client[.]” OEC 503(1)(b). Myers clearly was a “representative of the lawyer,” as he was retained to help defense counsel in the planning and conduct of litigation. The question before us is whether the expert opinion that Myers developed in the course of his employment by defense counsel, which he communicated to defense counsel, was a “confidential communication” within the meaning of OEC 503. The trial court concluded that the fact that the opinion was developed for defense counsel and with the assistance of defense counsel was privileged, but the contents of the communication were not privileged. We disagree.

A significant body of literature exists concerning this topic, and numerous cases have discussed it at length. Most jurisdictions treat expert opinions rendered in anticipation of litigation either as privileged or as falling within the work product doctrine of qualified privilege discussed below. The trial court apparently believed the question to be one of first impression in Oregon and, therefore, based its conclusion on a law review article that took the position that a privilege for expert opinion rendered in anticipation of litigation might not be necessary. See Edward J. Imwinkelried, The Applicability of the Attorney-Client Privilege to Non-Testifying Experts: Reestablishing the Boundaries Between the Attorney-Client Privilege and the Work Product Protection, 68 Wash U L Q 19, 21-22 (1990). That approach, however, is not consistent with Oregon law, as explained below.1

*532A discussion of whether material prepared in anticipation of litigation is subject to a privilege or a qualified privilege logically begins with Hickman v. Taylor, 329 US 495, 67 S Ct 385, 91 L Ed 451 (1947). Although Hickman was decided under the Federal Rules of Civil Procedure, its rationale has been almost universally adopted to one degree or another by most states, including Oregon, as discussed below. In Hickman, one party sought to discover through interrogatories various oral and written statements made by witnesses to an attorney for the opposing party. The Court concluded that the memoranda, statements and mental impressions sought did not, strictly speaking, fall within the attorney-client privilege. That conclusion, however, did not end the inquiry. The Court noted that the party seeking the information had full access to the witnesses whose statements had been taken, and concluded that it was “dealing with an attempt to secure the production of written statements and mental impressions contained in the files and the mind of the attorney Fortenbaugh without any showing of necessity or any indication or claim that denial of such production would unduly prejudice the preparation of petitioner’s case or cause him any hardship or injustice.” Id. at 508-09. The Court stated:

“Here is simply an attempt, without purported necessity or justification, to secure written statements, private memoranda and personal recollections prepared or formed by an adverse party’s counsel in the course of his legal duties. As such, it falls outside the arena of discovery and contravenes the public policy underlying the orderly prosecution and defense of legal claims. Not even the most liberal of discovery theories can justify unwarranted inquiries into the files and the mental impressions of an attorney.

“Historically, a lawyer is an officer of the court and is bound to work for the advancement of justice while faithfully protecting the rightful interest of his clients. In performing his various duties, however, it is essential that a *533lawyer work with a certain degree of privacy, free from unnecessary intrusion by opposing parties and their counsel. Proper preparation of a client’s case demands that he assemble information, sift what he considers to be the relevant from the irrelevant facts, prepare his legal theories and plan his strategy without undue and needless interference. That is the historical and the necessary way in which lawyers act within the framework of our system of jurisprudence to promote justice and to protect their clients’ interests. This work is reflected, of course, in interviews, statements, memoranda, correspondence, briefs, mental impressions, personal beliefs, and countless other tangible and intangible ways—aptly though roughly terms by the Circuit Court of Appeals in this case as the ‘Work product of the lawyer.’ ” Hickman, 329 US at 510-11.

The Court went on to qualify that not all work product of opposing counsel would necessarily be free from discovery in all cases. “Where relevant and non-privileged facts remain hidden in an attorney’s file and where production of those facts is essential to the preparation of one’s case, discovery may properly be had.” Id. at 511. The Court concluded that “the general policy against invading the privacy of an attorney’s course of preparation is so well recognized and so essential to an orderly working of our system of legal procedure that a burden rests on the one who would invade that privacy to establish adequate reasons to justify production through a subpoena or court order.” Id. at 512.

In a subsequent case, the Court stated that, although the work product doctrine most frequently is applied in civil litigation, “its role in assuring the proper functioning of the criminal justice system is even more vital.” United States v. Nobles, 422 US 225, 95 S Ct 2160, 45 L Ed 2d 141 (1975). Nobles, in fact, extended the work product doctrine in several regards. In Nobles, an investigator for defense counsel interviewed witnesses and produced a report concerning what the witnesses had said. Id. at 227. At trial, the witnesses testified in a manner inconsistent with what they had told the investigator, and defense counsel relied on the investigator’s report in attempting to impeach them on cross-examination. Id. at 227-28. The trial court ruled that defense counsel was required to disclose the report to the prosecutor when the investigator was called to testify about *534the witness’ statements. Id. The question before the Court was whether the defendant was required to produce the report under those circumstances. The Court considered the applicability of the work product doctrine, which it described as “a qualified privilege.” Id. at 237. The Court stated:

“One of those realities [of litigation in our adversary system] is that attorneys often must rely on the assistance of investigators and other agents in the compilation of materials in preparation for trial. It is therefore necessary that the doctrine protect material prepared by agents for the attorney as well as those prepared by the attorney himself. Moreover, the concerns reflected in the work product doctrine do not disappear once trial has begun. Disclosure of an attorney’s efforts at trial, as surely as disclosure during pretrial discovery, could disrupt the orderly development and presentation of his case.” Id. at 238-39 (footnote omitted).

In Nobles, the Court ultimately concluded that, by calling the investigator to testify, the defendant had “waived the privilege with respect to matters covered in his testimony.” Id. at 239. However, as shown by the quotations above, the Court extended the work product doctrine to cover reports prepared for an attorney by an agent in anticipation of litigation and made clear that the work product doctrine did not merely apply to pretrial discovery, but had application at trial as well.

Although Hickman and its progeny do not dictate state law concerning privileges and work product, they have provided the basis for some sort of work product doctrine of privilege or qualified privilege in most jurisdictions. In the majority of jurisdictions, “case law extends the attorney-client privilege to expert information.” Imwinkelried, Applicability of the Attorney-Client Privilege, 68 Wash U L Q at 21-22. The wisdom of extending the attorney-client privilege to expert opinions rendered to a party in anticipation of litigation is the subject of great debate. Such an application of the attorney-client privilege can lead to situations where one party essentially co-opts all of the reputable experts in a given field by seeking their opinions, preventing the opposing party from obtaining a reputable expert. On the other hand, allowing parties access to the opposition’s nontestifying *535experts can provide a strong disincentive for parties’ consultation of fair-minded experts who might render unfavorable opinions, as those opinions could then be used against them. Viewed from the perspective of economic theory, the application of the work product doctrine in this context maximizes the production of information used in litigation and increases efficiency, while overcoming the disincentive discussed above. See, e.g., Allen et al, A Positive Theory of the Attorney-Client Privilege and the Work Product Doctrine, 19 J Legal Stud 359 (1990) (discussing policy); Imwinkelried, Applicability of the Attorney-Client Privilege, 68 WashU L Q at 37-38 (same); Brink et ux v. Multnomah County, 244 Or 507, 519, 356 P2d 531 (1960) (concluding that a party could not obtain testimony of opposing party’s nontestifying expert, stating that “one litigant should not be permitted to make use of his opponent’s preparation of his case,” because “to do so would penalize the diligent and place a premium on laziness”).

As noted, the majority view is to extend some sort of work product protection to the opinions of nontestifying experts. The majority view has been articulated in United States ex rel Edney v. Smith, 425 F Supp 1038, 1047-49 (EDNY 1976), aff'd 556 F2d 556 (2d Cir), cert den 431 US 958 (1977). In Edney, the court concluded that, although the extension of the privilege to experts is not beyond criticism as the expert’s conclusions are based on more than just client communications that traditionally fall within the privilege, it is nonetheless desirable to do so because, in trial preparation, attorneys need to consult those who have expertise attorneys lack in order to evaluate the soundness of defenses and to prepare for adverse testimony on the subject.2 Another view, endorsed by Professor Imwinkelried in the journal article cited above, suggests that, although expert opinion prepared in anticipation of litigation should not be unconditionally privileged, it “is an ideal candidate for qualified protection *536under the work product privilege.” Imwinkelried, Applicability of the Attorney-Client Privilege, 68 Wash U L Q at 37. Under that approach, the opinions of opposing parties’ experts would not be available unless the party seeking the information made a showing of a special need for the information. That approach follows the model of Federal Rule of Civil Procedure 26(b), which allows discovery of certain materials if the party makes “a showing of exceptional circumstances under which it is impracticable for the party seeking discovery to obtain facts or opinions on the same subject by other means.” See also ORCP 36 B(3). That approach also is consistent with Hickman, which stated that where production of facts in an attorney’s files “is essential to the preparation of one’s case, discovery may properly be had.” Hickman, 329 US at 511.

Oregon law on this subject, although not plentiful, leads to the conclusion that Oregon extends the attorney-client privilege to the opinions of nontestifying experts who rendered those opinions in anticipation of litigation. The leading case on this topic is Brink et ux v. Multnomah County, 224 Or 507, 356 P2d 531 (1960), which is cited in the legislative commentary to the Oregon Evidence Code. Legislative Commentary to OEC 503, reprinted in Kirkpatrick, Oregon Evidence at 209. In Brink, the plaintiff in an inverse condemnation action called to the witness stand an appraiser who was

“previously employed for compensation by [defendant] Multnomah County here to observe this property in question and to act as a consultant and advisor to [the] Deputy District Attorney representing Multnomah County and in that respect ha[d] communication with [the Deputy District Attorney] in regard to certain data relative to this problem.” 224 Or at 509.

Although upholding the exclusion of this evidence on another basis, the court alternatively held that the evidence was inadmissible for the following reason:

“A communication made under the circumstances described by defendant’s counsel would fall within the privilege extended to a client for communications with his lawyer. A communication ‘by any form of agency employed or set in motion by the client is within the privilege.’ 8 Wigmore, *537Evidence § 2317, p 616 (3d ed 1940); City and County of San Francisco v. Superior Court, 37 Cal 2d 227, 231 P2d 26, 31 25 ALR 2d 1418 (1951).” Brink, 224 Or at 516-17 (emphasis in original).

The court concluded that the appraiser’s report was made in anticipation of litigation, and “this was sufficient to bring the report within the privilege.” Id. at 517.

The Brink decision follows the majority rule discussed above. The appraiser, a nontestifying expert who rendered an opinion to one party in anticipation of litigation, could not be called by the other party and examined concerning that opinion. Brink does not dispose of the question entirely, however, given that its alternative holding might be described as dictum and given that it was decided before the current codification of the rule of privilege at OEC 503. We, thus, turn to subsequent decisions concerning this topic.

In Nielsen v. Brown, 232 Or 426, 374 P2d 896 (1962), the court discussed at length the pros and cons of the Hickman work product doctrine and various courts’ application of it, concluding that none of the cases addressed “the precise question with which we are faced here: namely, whether in a personal injury action a physician employed and compensated by the defendant to examine the plaintiff as a part of the defendant’s preparation for trial can be prevented by the defendant from testifying on the trial as a witness for the plaintiff to an opinion already formed as the result of such examination.” Id. at 439. The court distinguished Brink and followed a line of cases indicating:

“In Rogers on Expert Testimony (3d ed), the author says at page 797:

“ “Where a person voluntarily submits to a physical examination by a physician at the instance of the adverse party he may call the latter as a witness and interrogate him relative to the examination. * * Id. at 440.

The court concluded that the plaintiff submitted to the examination with the implicit assumption that she would be able to call the physician as a witness. Id. at 444. Although the case is not directly on point, it is notable because it did not take an expansive view of the work product doctrine, stating *538that “what this court said in the Brink case on this subject amounts to no more than a holding that in the particular circumstances of that case, including the privileged character of Kolberg’s report and the use intended to be made of it by the plaintiffs, the claim of deprivation of a right to a fair trial was without foundation.” Nielsen, 232 Or at 444.

This court also has addressed the issue in the context of the prosecution calling defense expert witnesses. In State v. Moore, 45 Or App 837, 609 P2d 866 (1980), the trial court appointed a psychiatrist at the motion of the defendant, and the psychiatrist’s report was sent only to defense counsel. Id. at 840-41. This court held that the trial court erred in allowing the prosecution to call the psychiatrist as a witness, citing Brink for the proposition that the attorney-client privilege extends to agents of an attorney, and concluding that, because the expert’s opinion had been rendered “for the sole use and benefit of the defense,” the court erred in admitting the testimony. Id. at 841-42. Similarly, in State v. McGrew, 46 Or App 123, 610 P2d 1245, rev den 289 Or 587 (1980), this court addressed whether the state could call as witnesses several doctors who had examined the defendant. Although the question presented there was not whether the privilege existed, but whether the defendant had waived the privilege, this court noted: “It has been held that communications made to the attorney through other professionals engaged to aid the attorney also come within the privilege.” Id. at 126 n 2, citing Brink and Nielsen.

In 1981, the legislature enacted OEC 503, which, according to the legislative commentary, codified the current law concerning attorney-client privilege and extended it in certain ways. Legislative Commentary to OEC 503, reprinted in Kirkpatrick, Oregon Evidence at 207-12. As noted above, the commentary clearly states that the privilege extends to “an expert who is hired to assist in rendering legal advice or to help in the planning and conduct of litigation, but not one employed to testify as a witness.” Id. That view of the privilege also is expressed in State v. Brown, 297 Or 404, 687 P2d 751 (1984). Although Brown concerns the criteria for the admission of scientific evidence and, in particular, the admissibility of polygraph evidence, the court made the following observations:

*539“Defense counsel can shop for a friendly polygraph examiner, hire that examiner as a ‘consultant’ under the attorney-client privilege, see OEC 503 and Brink v. Multnomah County, 224 Or 507, 356 P2d 536 (1960), and if the defendant flunks that examination, the prosecution normally never knows the results of the examination. Even if the prosecution knew of such an examination, the state could not offer any evidence or comment about the examination.” 297 Or at 432 n 21 (emphasis supplied).

The most recent discussion of the application of the attorney-client privilege to a criminal defense expert was in State v. Bockorny, 125 Or App 479, 866 P2d 1230 (1993), on recons 126 Or App 504, 869 P2d 349, rev den 319 Or 150 (1994). In Bockorny, a murder case, an expert testified on behalf of the defendant concerning whether certain material found on a pair of scissors was blood. Id. at 484. In preparing for trial, the expert had contact with the prosecutor concerning the blood on the scissors. In the course of a discussion between the expert and the prosecutor, the prosecutor asked the expert about an entirely different subject—a biological staining method used to detect the presence of sperm heads in the victim’s mouth. Id. The expert later contacted the prosecutor and asked to examine the biological staining evidence, and the prosecutor agreed. Id. The defense called the expert witness to testify about the blood on the scissors, and the state called the same witness on rebuttal to testify about the biological staining method. This court stated:

“There was no evidence that biological staining was a subject for which the defense retained Grimsbo or that it had been a subject of discussion between Grimsbo and the defense. Therefore, Grimsbo’s testimony was not privileged under the attorney-client privilege, nor could it be called work product.5

*540Those cases lead to the conclusion that, whatever the relative merits and drawbacks of extending the attorney-client privilege to nontestifying experts, Oregon law clearly does recognize this privilege. The common law of privilege was set forth in Brink, and the common-law rule as expressed in Brink is reflected in the plain language, of OEC 503, the legislative commentary of OEC 503, and the more recent case law discussing this issue, such as Brown and Bockorny. The dissent accuses us of “enshrin[ing] in the law some of the most criticized aspects of the use of experts in litigation.” 155 Or App at 542. OEC 503 is not something that came into existence by being “enshrined” by this court; it came into existence by being “enacted” by the legislature. Whether the limitations on the privilege suggested by Professor Imwinkelried and embraced by the dissent are good ideas, or would be good public policy, is not for this court to say. Until the legislature chooses to amend OEC 503, we are bound to follow the law as it exists rather than to discard it in favor of interesting new ideas posited in journal articles.

Myers was hired in anticipation of litigation by defense counsel to render an opinion as to the cause of the collision. Myers was not retained to testify, but to analyze information about the accident and to give defense counsel an opinion about causation, in order for defense counsel to develop a theory of the case and to prepare to cross-examine expert witnesses concerning the cause of the collision. Defense counsel discussed potential theories of the case with Myers and related at least one of defendant’s statements to Myers. Myers was a representative of defendant’s lawyer, as that term is defined by OEC 503, because he was hired to assist in the planning of the litigation. His report to defense counsel concerning his opinions, theories or conclusions in connection with his investigation of the accident quite clearly would not be discoverable by the prosecution.3 Myers’ opinion *541concerning the accident reconstruction falls squarely within the attorney-client privilege as that privilege has been defined under Oregon law. The trial court erred in admitting Myers’ testimony.

Evidentiary error is not presumed to be prejudicial. OEC 103(1). The state argues that the error was harmless because the evidence against defendant was overwhelming, pointing to the evidence that defendant had been drinking and that defendant had lied to the police officer investigating the collision. The state further argues that Myers’ testimony was in some ways favorable to defendant, because Myers believed that hydroplaning or the condition of defendant’s tires could have contributed to the collision. Although the question is a close one, we agree with defendant that he was prejudiced by the error. Prejudicial error occurs if there is some likelihood of the error affecting the jury’s result. Jennings v. Baxter Healthcare Corp., 152 Or App 421, 430, 954 P2d 829, rev allowed 327 Or 317 (1998).

Defendant’s expert testified that he believed that the collision was a result of a defect in the steering mechanism of defendant’s truck. If the jury believed that expert’s evidence, it could have concluded that defendant’s consumption of alcohol was not causally related to the collision. Thus, the state’s argument that the evidence concerning alcohol was overwhelming does not hold up, because the key issue was whether a factor beyond defendant’s control, i.e., a failure of the steering mechanism, was the actual cause of the collision. We agree with the state that Myers’ testimony was not as unfavorable to defendant as was Fries’ testimony, because he believed that tire problems or hydroplaning could have caused the collision. However, it is undisputed that Myers’ testimony undermined the testimony of defendant’s own expert witness, Talbot. We conclude that Myers’ testimony did have a likelihood of affecting the jury’s verdict in this case. Thus, we are unable to say that the admission of that evidence was harmless error.

Reversed.

dissenting.

The majority holds that the trial court erred in allowing the jury to hear highly relevant testimony from *542Myers, a qualified expert. The only reason for its holding is that Myers developed his opinion while working for the side that did not call him as a witness. It does not matter to the majority that Myers based his testimony entirely on the police reports, his measurements at the accident scene, and other generally available information, or that he neither relied on nor revealed any confidential information that he received while he worked for defendant. An expert, in the majority’s view, is a hired gun who must remain forever loyal to the side that originally hired him, not an independent professional whose opinions may have some level of objectivity. The majority’s decision is both contrary to the authorities on which it relies and unnecessarily enshrines in the law some of the most criticized aspects of the use of experts in litigation. Because I believe that the trial court acted within its discretion, I dissent.

It is difficult to determine the precise basis for the majority’s holding. The majority appears to rely on both the attorney-client privilege and the work product doctrine. However, its opinion moves between the two concepts in a way that is difficult to follow. See, e.g., 155 Or App at 531 (states conclusion concerning scope of attorney-client privilege in one paragraph, in next paragraph supports conclusion by cases based on work product doctrine); id. at 535-36 (paragraph begins by referring to the work product doctrine, moves to a case that discusses the attorney-client privilege, and closes with an article and a case on the work product doctrine).1 It appears to rely on an amorphous “privilege” that seems to combine elements of both concepts, and it goes to great length to discuss matters, such as whether an expert’s communication to an attorney is privileged, that are not in dispute.2

*543In this dissent I will treat the attorney-client privilege and the work product doctrine separately, showing that neither justifies the majority’s result. Before doing so, I will emphasize two points that are central to my disagreement with the majority, neither of which it clearly discusses. The first is that the attorney-client privilege covers a communication between Myers and defendant’s former attorney and only the communication; it does not turn the contents of Myers’ mind into the attorney’s property, nor does it necessarily prevent Myers from communicating his conclusions to someone else or disqualify him as a witness for another party. It is the communication that is privileged, not the opinion. In contrast, the majority believes that, once it has decided that Myers’ communication of his opinion to defendant’s attorney was privileged, it has shown that Myers may not express the same opinion as a witness for the prosecution. As I will show, it has omitted a fundamental step from its analysis.

Second, the work product doctrine is primarily a limitation on compelled discovery of attorney work product; the focus is on protecting the attorney’s work. There is no issue of discovery in this case, nor is there any suggestion that Myers’ testimony compromised defendant’s attorney’s work product. In addition, the majority does not discuss the statutory embodiment of the doctrine.

I turn to the relevant facts. While driving his pickup truck, defendant was involved in an accident in which two persons in another vehicle were killed and two others were injured. He was convicted of two counts of criminally negligent homicide, one count of driving under the influence of intoxicants, and two counts of assault in the fourth degree. The crucial factual issue at trial was the cause of the accident, including the proper reconstruction of the physical events that occurred during it. The public defender who originally represented defendant obtained authority to hire Myers, an accident reconstruction engineer, to assist the defense on that subject. The public defender and his investigator went with Myers to the accident site, where they assisted Myers in taking measurements. The public defender passed on to Myers at least one thing that defendant had told *544him. Myers thereafter gave the public defender his conclusions, which apparently were not, as a whole, favorable to the defense. The defense did not call him to testify at the trial.

Before allowing Myers to testify for the prosecution on rebuttal, the court held an evidentiary hearing. At that hearing, the public defender described his contact with Myers and repeated Myers’ explanation of the information that he intended to use in reaching his conclusions:

“Q. * * * Did [Myers] indicate to you what part if at all your communication to him of what your client might have said had played a part in his opinion as to what occurred?

“A. Not in so many words. He said basically he’s going to look at this and he’s going to tell me scientifically what his opinion was as to what happened.

“Q. Based on the physical evidence, not on what someone might have said?

“A. Yeah. He was going to tell me what he—what his opinion based on measurements and things he saw, he was going to tell me, he said good or bad he was going to let me know what it is.” (Emphasis supplied.)

After the hearing, the court allowed Myers to testify on the following conditions: (1) he was not to mention his connection with the public defender or any communicationfrom the public defender; (2) he was not to refer to any statement by defendant or any testimony where defendant was the source of the information; and (3) he was not to discuss any sharing of theories of the case. In his testimony, Myers complied with those conditions, describing conclusions that he had reached as the result of his examination of the physical evidence and of his review of other public information. Neither defendant nor the majority asserts that the public defender’s communications with Myers had the slightest effect on any specific aspect of his testimony.

From these facts, the majority appears to conclude that allowing Myers to testify, under the restrictions that the court imposed and with no specific evidence of prejudice, violates both the attorney-client privilege and the work product doctrine. I would hold that it violates neither the privilege *545nor the doctrine. I begin with the attorney-client privilege, as it is embodied in OEC 503.

OEC 503(2)(b) creates a privilege for communications “[b]etween the client’s lawyer and the lawyer’s representative[.]” OEC 503(1)(e) defines “[Representative of the lawyer” as “one employed to assist the lawyer in the rendition of professional legal services,” excluding a physician making an independent medical examination under ORCP 44. According to the commentary to the evidence code, under that definition an expert who is hired to assist in the planning or conduct of the litigation is a representative of the lawyer, but one who is hired to testify is not. Laird C. Kirkpatrick, Oregon Evidence (3d ed 1996), 209-10, 222-23; see also Dyer v. R.E. Christiansen Trucking, Inc., 118 Or App 320, 329, 848 P2d 104 (1993), rev’d on other grounds 318 Or 391, 868 P2d 1325 (1994) (correspondence with expert before he was asked to testify is privileged under OEC 503(2)(b)).

I agree with the majority that any communications between Myers and the public defender were privileged. That, however, is only the beginning of the analysis. By its express terms, the privilege protects communications between the expert and the lawyer, not the entire contents of the expert’s mind. As the leading commentator on Oregon evidence law explains, “the fact that communications by the expert to the attorney may be privileged does not necessarily mean that the facts learned by the expert are within the privilege or the work product doctrine.” Kirkpatrick at 223. Myers did not testify concerning his communications with the public defender; in compliance with the court’s instructions, he did not mention that he had had any contact with either defendant or any of defendant’s attorneys.

Neither defendant nor the majority refers to any specific point in Myers’ testimony in which Myers relied on, or revealed the contents of, any confidential communication with the public defender or any information that could be traced, directly or indirectly, to defendant. It is not enough to justify excluding Myers’ testimony that he testified concerning the same subject matter for which the public defender had hired him and concerning the same information that he had discussed with the public defender. Information that *546Myers may have given the public defender in a privileged context is not necessarily itself privileged in all other contexts. The privilege applies to communications, not to information. Because Myers testified concerning the same information without relying on or revealing any communication with defendant or the public defender, there was no violation of the privilege.

The majority does not consider the distinction between a privileged communication and the information that is communicated, and it does not accept that information does not become privileged by being the subject of a privileged communication. The majority’s difficulty with those distinctions is obvious in its statement of what it believes the trial court held:

“The trial court concluded that the fact that the opinion was developed for defense counsel and with the assistance of defense counsel was privileged, but the contents of the communication were not privileged.” 155 Or App at 531.

That is not what I would conclude or, so far as I can determine, what the trial court concluded. There is no question that the contents of any communication between Myers and the public defender were privileged; what is not privileged are the contents of Myers’ mind. The fact that Myers said something to defendant’s attorney does not mean that he cannot say the same thing to someone else. Suppose, for example, that Myers had told the attorney, “My opinion is that the accident was the result of your client’s loss of control of his vehicle.” The fact that he made that statement to defendant’s attorney would be privileged in these circumstances. If that opinion were based on a confidential communication from the attorney or from defendant, the opinion itself would also be privileged, because it would derive from privileged information. That conclusion does not end the case, however. It does not prevent Myers from saying, when called by the opposing side, “My opinion, based entirely on the objective evidence, is that the accident was the result of defendant’s loss of control of his vehicle.”

What the majority does not recognize is that there are two steps to the analysis, not just the one that it sees. The first step is to determine whether any communications *547between Myers and the public defender were privileged. The majority and I agree that at least some were; otherwise there would be no issue. The next step is to determine whether Myers relied on or referred to those privileged communications in his testimony. As the public defender’s explanation and Myers’ testimony show, there is no credible argument that it did. The majority neither recognizes the need for, nor discusses, this second crucial step.

The majority’s failure to recognize those things is curious, in part because it discusses at some length a law review article whose thesis is that “one can draw a principled distinction between the client’s communications [to the expert or the attorney] and the rest of the expert’s information,” that “only communications warrant protection,” that “the rest of the expert information in its own right does not qualify under the attorney-client privilege,” and that “[privilege law does not allow parties to suppress all information in their possession.” Edward J. Imwinkelried, The Applicability of the Attorney-Client Privilege to Non-Testifying Experts: Reestablishing the Boundaries Between the Attorney-Client Privilege and the Work Product Protection, 68 Wash U L Q 19,24-25 (1990) (first and third quotations at 24; second and fourth quotations at 25).

Imwinkelried argues that it is indefensible to apply the attorney-client privilege to expert information that comes from sources other than communications from the client, id. at 28-29, and gives an example that is close to this case:

“[S]uppose that a civil or criminal defendant hires an accident reconstruction expert to analyze the design of the stretch of highway where a fatal collision occurred. The defense hopes to establish that the cause of the accident was the highway’s defective design rather than the defendant’s careless driving. After visiting the accident scene and reviewing public highway department engineering records * * *, the accident reconstruction expert forms her opinion. It is inappropriate to apply the attorney-client privilege to her knowledge because the information acquired does not constitute a communication from the client.Id. at 29-30 (citations omitted; emphasis supplied).

Imwinkelried then points out the dangerous public policy results of holding the part of the expert’s information that *548does not involve communications from the attorney or client to be privileged: “By expanding the attorney-client privilege to this extreme, the courts in effect would be allowing one party to place an expert in quarantine. ** * * The courts should not permit one party to corner the market on expert witnesses.” Id. at 32-35 (quotation at 35; citations omitted; emphasis supplied). One would think that the majority would at least consider and respond to these points.

The cases that the majority cites do not involve situations in which the second step is necessary and thus do not help its conclusion. In Brink et ux v. Multnomah County, 224 Or 507, 356 P2d 536 (1960),3 a condemnation case, the question was whether the plaintiff could compel an expert appraiser to testify about the contents of his report to the defendant. The Supreme Court first held that the trial court correctly sustained the objection to the evidence on unrelated grounds. It then discussed the privilege issue as an alternative ground for its decision, noting that a communication from the appraiser to the defendant’s attorney “would fall within the privilege extended to a client for communications with his lawyer[.]” 224 Or at 516. The plaintiffs’ only argument for the admission of the report was that the appraiser prepared it in the regular course of business without reference to an actual or potential lawsuit. The court rejected that view of the facts. It did not consider whether the appraiser could have testified, either voluntarily or under a court order, concerning his opinion, if he did so without reference to the report or to other communications with the defendant’s attorney.4 Brink, thus, involved only the first step of the analysis, *549because the issue was the admissibility of the report, which was itself a communication.

The remaining cases that the majority discusses primarily involve the work product doctrine. Those cases treat the issue as though it were a matter of general common law, with some reference to federal civil rules. However, in Oregon the work product doctrine in a criminal case is statutory, and that is where my analysis must begin. ORS 135.805 to ORS 135.873 provide for discovery in criminal cases; ORS 135.855(1)(a) makes certain material exempt from those discovery requirements:

“Work product, legal research, records, correspondence, reports or memoranda to the extent that they contain the opinions, theories or conclusions of the attorneys, peace officers or their agents in connection with the investigation, prosecution or defense of a criminal action[.]”

The statute does not define work product. The only statutory suggestion of what ORS 135.855(1)(a) might cover is ORCP 36 B(3), which is the civil work product provision. The rule defines its coverage as “trial preparation materials” and applies to “documents and tangible things * * * prepared in anticipation of litigation or for trial by or for another party or by or for that other party’s representative^]” (Emphasis supplied.) It does not apply to the in-court testimony of a qualified witness. That definition, and that scope of coverage, are generally consistent with the work product cases that the majority cites.

There are two obvious aspects of ORS 135.855(1)(a), neither of which the majority mentions.5 First, the statute has nothing to do the admissibility of evidence at trial or the qualifications of a trial witness. It is an exception to rules concerning pretrial discovery. That is also true of the common-law work product cases on which the majority relies, an aspect of those cases that it also fails to mention. Second, the focus of the statute is on protecting “the opinions, theories or conclusions of the attorneys, peace officers or their agents in *550connection with the investigation, prosecution or defense of a criminal action.” Myers’ testimony neither revealed nor threatened to reveal any of those things.

The statutory protection of work product does not directly apply to Myers’ testimony. The issue in this case is not the pretrial discovery of a document or other tangible thing, such as an expert’s report.6 Rather, it is whether an expert witness may testify at trial about his opinions on a matter in issue. The only way that the work product doctrine might apply is if Myers’ trial testimony intruded into those things that the work product doctrine protects, particularly the attorney’s trial preparation and strategy. There is no basis for concluding that it did so, in part because of the limitations that the court imposed, in part because Myers’ opinion was itself not based on work product, and impart because Myers had already disclosed his opinion to the prosecution. Although Myers may have developed his opinion while working for the public defender, he made it clear from the beginning that he would base his conclusions on his independent evaluation of the facts, not on communications from defendant or his attorney. Myers’ opinions were not the public defender’s work product, nor were they affected by it.

In discussing the work product doctrine, the majority again fails to distinguish between the content of a communication to the attorney and the independent contents of the expert’s mind. The leading federal cases that it discusses, Hickman v. Taylor, 329 US 495, 67 S Ct 385, 91 L Ed 451 (1947), and United States v. Nobles, 422 US 225, 95 S Ct 2160, 45 L Ed 2d 141 (1975), concern the compelled disclosure of written investigatory reports of interviews with witnesses, prepared by either the attorney or the attorney’s investigator. Those cases have nothing to do with whether a defense expert is a competent witness for the prosecution when the prosecution has properly learned of the expert’s existence and knowledge. As the Supreme Court noted in Nielsen v. Brown, 232 Or 426, 435-36 and n 3, 374 P2d 896 (1962), the *551purpose of the work product doctrine is to protect the attorney’s trial preparation materials, not the expert’s opinion. The United States Supreme Court held in Nobles that trial preparation materials include things prepared on behalf of the attorney, 422 US at 239 n 13; Nielsen makes it clear that things prepared on behalf of the attorney do not include the independent testimony of an expert’s opinion.

Nielsen is the first significant Oregon discussion of the work product doctrine. In that case, the issue was whether the plaintiff could require a physician who had examined her on behalf of the defendant to testify concerning the examination. The court first held that doing so did not violate the attorney-client privilege. 232 Or at 431-33. It then examined whether the physician’s opinions were work product of the defendant’s attorney. It noted that the issue before it did not involve requiring the production of an expert’s report; rather, it was whether the opposing party could call • the expert as a witness and compel him to express an opinion on a subject within his professional knowledge. It held that the “case does not involve the work product rule,” 232 Or at 435, and quoted a California court’s statement that in Hickman the material sought came entirely from the attorney’s files. It then cited the same court’s holding that an expert engineer had to give a pretrial deposition in which he would describe his conclusions and quoted the California court’s explanation:

“ ‘In our case it is the thought, research and effort of Cheek [the witness] which is sought by plaintiff. Although defense counsel may have exercised ingenuity in determining that “slipperiness” of the walk could be tested, this is not enough, as we read Hickman, to make the examination and tests of Cheek the work product of counsel.’ ”Id. at 436, quoting Grand Lake Drive In v. Superior Court, 179 Cal App 2d 122, 129, 3 Cal Rptr 621 (1960).

The court then held in Nielsen that “the mere fact that the doctor is compensated for his services by the defendant does not suffice to make him unavailable as a witness for the plaintiff.” Id. at 444. Although the court stated that it was not expressing an opinion as to the correct rule when the testimony is that of an appraiser, engineer, or similar expert, id., it clearly signaled the direction that it would have taken:

*552“Neither the Hickman case nor any other that we have seen is authority for the proposition that the information and knowledge in the mind of the expert must be kept there and away from the jury on the theory that they are the work product of the lawyer.” Id. at 437.

The Supreme Court thus understands the distinction between the expert’s communications to the attorney and the expert’s opinions that is completely missing from the majority’s opinion.7 Nielsen comes close to resolving this case against the majority’s result.

Defendant relies heavily on State v. Bockorny, 125 Or App 479, 866 P2d 1230 (1993), on recons 126 Or App 504, 869 P2d 349, rev den 319 Or 150 (1994), whose facts are in some respects similar to those in this case. In Bockorny, a murder case, the defendant called an expert to testify concerning whether material on a pair of scissors was blood. Before the expert’s testimony for the defense, the prosecutor had talked with him about that issue and had also asked him questions about the test that the state had used to determine the presence of sperm in the victim’s mouth. The expert thereafter asked to examine the relevant slides and concluded that they contained two sperm heads.

The prosecutor decided to call the expert in rebuttal to testify concerning the validity of the test for sperm. The expert had discussed the longevity of sperm and the presence of sperm heads on the slides with the defense, but he had not discussed the method of testing for sperm. The court allowed the expert to testify concerning stain methods, slide preparation, the specific test that the state had used, and related matters. He did not give an opinion on whether the slides contained sperm heads. In his closing argument, the prosecutor emphasized that the defendant’s own expert had said that the test was valid. 125 Or App at 483-85. We held that, because there was no evidence that the defense retained the expert on the issue of biological staining or discussed it with *553him, his testimony was neither within the attorney-client privilege nor work product. We recognized that, “if an expert is willing to give opinions to both sides, a litigant can be placed in a difficult, if not impossible, situation at trial. However, it is not a situation prohibited by law.” 125 Or App at 486.

Defendant argues that the foundation of our decision in Bockorny was that the defendant had not retained the expert on the subject about which he testified for the state. In this case, in contrast, Myers’ testimony was on the precise issue for which the public defender retained him. From that distinction defendant concludes that Myers’ testimony was inadmissible. Defendant misses the essential point of Bockorny, which is that the expert’s testimony was not based on attorney-client communications or work product. In Bockorny, the fact that the expert had not consulted with the defense on the subject of his testimony satisfied that requirement. That does not mean that there is no other way to make that essential showing. In this case both the public defender’s testimony before Myers testified and Myers’ testimony itself show that Myers did not rely on attorney-client or work product material.

I recognize that one side using the other’s expert may create a difficult situation for the side that originally hired the expert. If anything, that difficulty was greater in Bockorny than in this case, because in Bockorny the defendant had made the expert his own witness before the prosecution called him, a fact that the prosecutor used effectively in his closing argument. However, in both Bockorny and this case the trial courts established clear and appropriate restrictions before allowing the expert to testify, and the expert complied with those restrictions. We held that the court did not err in Bockorny, and the majority fails to explain why we should not reach the same conclusion here. Because its decision erroneously confuses the attorney-client privilege and the work product doctrine and erroneously expands them beyond their purposes and beyond their statutory limits, I dissent.