This case presents a question recently resolved by Greene v. Wainwright, 634 F.2d 272 (5th Cir.1981). Greene involved an identical issue of law and an almost identical array of facts. Also as in Greene, we must overturn convictions, obtained after a lengthy trial, because the trial court's rulings deprived criminal defendants of their fundamental rights of confrontation and cross-examination.

Dennis Slater and Joanne Lindstrom appeal convictions and sentences for mail fraud, 18 U.S.C. Sec. 1341 and Sec. 1342, and conspiracy to commit mail fraud, 18 U.S.C. Sec. 371. Neither appellant challenges the sufficiency of the evidence when viewed in the light most favorable to the government. Glasser v. United States, 315 U.S. 60, 80, 62 S. Ct. 457, 469, 86 L. Ed. 680 (1942). Slater and Lindstrom raise six issues on appeal,1 but we conclude that it is necessary that we treat only two: whether preindictment delay deprived appellants of their right to due process and whether the district court erred in imposing restrictions on appellants' ability to review psychiatric information bearing on the credibility of the government's key witness and in limiting cross-examination of that witness. Because we find that the trial court's restrictions on access to documents and on cross-examination denied appellants the right to confront their accusers, we reverse.

The convictions centered around the activities of Bay Therapy, Inc., a Florida corporation purporting to provide physical therapy treatment to injured persons pursuant to doctors' prescriptions. Joanne Lindstrom was a legal secretary employed by Dennis Slater, who was a senior trial attorney with a Tampa law firm. Slater, Lindstrom and David Webster, also an attorney, formed Bay Therapy, Inc. in the summer of 1976, each owning a one-third interest. The three agreed that Lindstrom would oversee the clinic's operation. She leased a building, acquired the necessary equipment and employed Rosamond Sloan, a licensed practical nurse, to operate the clinic. In October 1976, Sloan was replaced by the person who became the government's star witness at trial.

After she had been operating Bay Therapy for about nine months, Sloan's replacement contacted her brother, a former FBI agent who was then employed by the Fraud Division of the Florida Insurance Commissioner's office. At the suggestion of the investigators, she began attempting to learn incriminating information from Lindstrom, Slater and Webster. She also initiated meetings with federal investigators.

Eventually, she became the key witness at the Bay Therapy trial. During 1978, 1979 and 1980, two successive federal grand juries heard extensive evidence on the operations of Bay Therapy before handing down an indictment. The indictment charged that the appellants, as part of a scheme to inflate medical costs and defraud insurance companies, caused patients to be sent to Bay Therapy for treatment they did not need and often did not even receive. The trial lasted for three weeks and involved the testimony of eighty-six witnesses.

The government's key witness testified that during the period when she was overseeing operations at Bay Therapy, she, Lindstrom and Slater had discussed alteration of records, that she and Lindstrom had in fact changed records, that Slater and Lindstrom had ordered her to duplicate billing cards and that patients signed up for treatments they did not receive. Other witnesses testified about Slater's attempts to secure business for the clinic, and a number of former patients related their divergent experiences with Bay Therapy. Insurance claims managers debated whether increased therapy bills would in fact result in higher settlements, and an attorney outlined the factors he customarily considered in settling personal injury cases. Both appellants testified at the trial, denying all charges.

The jury found Slater and Lindstrom guilty of conspiracy to commit mail fraud and seventeen substantive counts of mail fraud. The district court sentenced Slater to concurrent sentences of five years imprisonment on all counts, but the court suspended all but six months of the sentence and placed Slater on four years probation. Lindstrom was placed on three years probation.

(1)

Lindstrom and Slater assert that the three-year period between the initiation of the investigation and the rendering of the indictment was excessive. Appellants contend that the delay, during which two witnesses died, deprived them of their fifth amendment right to due process and their sixth amendment right to speedy trial. We cannot agree.

The speedy trial guarantee of the sixth amendment does not apply to preindictment delay. United States v. Lovasco, 431 U.S. 783, 788, 97 S. Ct. 2044, 2047, 52 L. Ed. 2d 752 (1977); United States v. Marion, 404 U.S. 307, 320, 92 S. Ct. 455, 463, 30 L. Ed. 2d 468 (1971). The due process clause of the fifth amendment, however, requires dismissal of an indictment if a defendant makes a twofold showing: (1) that the delay caused actual prejudice to the conduct of his defense, and (2) that the delay was the product of deliberate action by the government designed to gain a tactical advantage. United States v. Marion, 404 U.S. at 324, 92 S.Ct. at 465; United States v. Townley, 665 F.2d 579, 581-82 (5th Cir.), cert. denied, --- U.S. ----, 102 S. Ct. 2305, 73 L. Ed. 2d 1307 (1982); United States v. Hendricks, 661 F.2d 38, 39-40 (5th Cir.1981); United States v. Nixon, 634 F.2d 306, 310 (5th Cir.), cert. denied, 454 U.S. 828, 102 S. Ct. 120, 70 L. Ed. 2d 103 (1981); United States v. Willis, 583 F.2d 203, 207-08 (5th Cir.1978).

Appellants urge that they have demonstrated prejudice and that they have been denied the ability to demonstrate prosecutorial bad faith. They argue that the deaths of two witnesses, Dr. L.J. Cordrey and William Hapner, caused them "actual prejudice and not merely 'the real possibility of prejudice inherent in any extended delay.' " United States v. McGough, 510 F.2d 598, 604 (5th Cir.1975), quoting United States v. Marion, 404 U.S. at 326, 92 S.Ct. at 466. We agree that appellants have made a prima facie showing of prejudice. Of the seven patient-clients who were the subject of the mail fraud counts, five were treated or examined by Dr. Cordrey. Slater had talked to Dr. Cordrey concerning his treatment of numerous patients whom the government alleged had been sent to Bay Therapy to unnecessarily boost their medical expenses. In many instances, Dr. Cordrey was the only physician to have examined these patients during their association with Bay Therapy. Further, appellants contend that William Hapner would have testified about his discussions with Slater on certain ethical and legal implications of Slater's involvement with Bay Therapy. Slater argues that Hapner would have testified as to why Slater referred numerous patients to certain defendants in the case and that Slater received no fees for the referrals.

Prejudice, however, is not enough: "proof of prejudice is generally a necessary but not sufficient element of a due process claim ... [T]he due process inquiry must consider the reasons for the delay as well as the prejudice to the accused." United States v. Lovasco, 431 U.S. at 790, 97 S.Ct. at 2048. An appellant's showing of prejudice triggers a "sensitive balancing of the government's need for an investigative delay ... against the prejudice asserted by the defendant." United States v. Brand, 556 F.2d 1312, 1317 n. 7 (5th Cir.1977), cert. denied, 434 U.S. 1063, 98 S. Ct. 1237, 55 L. Ed. 2d 763 (1978). Delay caused by a good faith ongoing investigation will not offend "fundamental conceptions of justice," United States v. Lovasco, 431 U.S. at 791, 97 S.Ct. at 2049. Delay that prejudices a defendant will, however, require dismissal of an indictment if the delay was an intentional device to gain tactical advantage over the accused. United States v. Marion, 404 U.S. at 324, 92 S.Ct. at 465.

Appellants have not suggested what tactical advantage the government could have hoped to gain by delay. As in United States v. Hendricks, 661 F.2d 38, 40 (5th Cir.1981) appellants have produced "no evidence even tending to show that the delay was a deliberate tactical maneuver by the government." Lindstrom and Slater contend that they could have demonstrated the cause of the delay had they been given access to the files of the United States Attorney. Appellants had requested the court to subpoena the files of the government, pursuant to United States v. Surface, 624 F.2d 23, 25 n. 2 (5th Cir.1980), to determine whether the delays were for permissible reasons.

We do not know the extent to which Lindstrom and Slater were in fact granted access to the government's files. The United States tells us that "in response to defense requests, the Court allowed the defense to rummage through prosecution files to determine if preindictment delays were properly motivated." Appellants reply that "at no time did the Court permit the defense to 'rummage through prosecution files.' "

The government contended in its Answer to Motion to Dismiss for Pre-Indictment Delay that the period of time between investigation and indictment was reasonable. This was a complex case involving numerous documents and records. Accord United States v. Singer, 687 F.2d 1135, 1143 (8th Cir.1982). Two successive grand juries participated in the preindictment investigation. Targets of investigation were afforded opportunities to appear before the grand jury and time to consider the offers. The government doubtless spent considerable time locating the 199 Bay Therapy patients themselves.

Although this is a close question, we are unable to conclude that appellants have made a showing sufficient to require an in camera inquiry. Under the dictates of Surface, the court need not examine government files until the defense at least has suggested what tactical advantage the government could gain from the delay. Appellants have presented no evidence whatever that the delay was intentionally designed to obtain some tactical advantage. Because we can discern no advantage to the government in delay, we decline to dismiss the indictment or to remand for further hearings on the issue.

We are most reluctant to fashion rules of decision that have the effect of encouraging the government to secure premature indictments. We do not wish to "pressure prosecutors into resolving doubtful cases in favor of early--and possibly unwarranted--prosecutions. The determination of when the evidence available to the prosecution is sufficient to obtain a conviction is seldom clear-cut, and reasonable persons often will reach conflicting conclusions." United States v. Lovasco, 431 U.S. at 793, 97 S.Ct. at 2050. Further,

In our view, investigative delay is fundamentally unlike delay undertaken by the Government solely "to gain tactical advantage over the accused," precisely because investigative delay is not so one-sided. Rather than deviating from elementary standards of "fair play and decency," a prosecutor abides by them if he refuses to seek indictments until he is completely satisfied that he should prosecute and will be able promptly to establish guilt beyond a reasonable doubt. Penalizing prosecutors who defer action for these reasons would subordinate the goal of "orderly expedition" to that of "mere speed." This the Due Process Clause does not require.

Id. at 795-96, 97 S.Ct. at 2051 (footnotes and citations omitted).

(2)

Appellants contend that the district court improperly (1) placed limitations on defense questioning of the government's chief witness relating to her prior psychiatric treatment and confinement, and (2) denied the defense access to medical records suggesting that the government's witness suffered from psychiatric illnesses, including delusions. In the alternative, appellants assert error in the district court's refusal to order an independent psychiatric examination of the witness.2 Because we agree that the district court erroneously restricted cross-examination and defense access to documents pertaining to the psychiatric history of the government's witness, we need not reach appellants' alternative argument.

The sixth amendment to the United States Constitution mandates that a criminal defendant has the right "to be confronted with the witnesses against him." The Supreme Court has repeatedly emphasized that its "cases construing the [confrontation] clause hold that a primary interest secured by it is the right of cross-examination." Davis v. Alaska, 415 U.S. 308, 94 S. Ct. 1105, 39 L. Ed. 2d 347 (1974) quoting Douglas v. Alabama, 380 U.S. 415, 418, 85 S. Ct. 1074, 1076, 13 L. Ed. 2d 934 (1965); Ohio v. Roberts, 448 U.S. 56, 63, 100 S. Ct. 2531, 2537, 65 L. Ed. 2d 597 (1980). The right of cross-examination is an essential safeguard of factfinding accuracy in an adversary system of justice and "the principal means by which the believability of a witness and the truth of his testimony are tested," Davis v. Alaska, 415 U.S. at 316, 94 S.Ct. at 1110.

In its brief and during oral argument, the government urged that the scope of cross-examination is a matter within the discretion of the district court. We have repeatedly held, however, that "this discretionary authority ... comes into play only after there has been permitted as a matter of right sufficient cross-examination to satisfy the Sixth Amendment." Greene v. Wainwright, 634 F.2d 272, 275 (5th Cir.1981), quoting United States v. Bass, 490 F.2d 846, 858 n. 12 (5th Cir.1974). The Supreme Court stated in Alford v. United States, 282 U.S. 687, 691, 51 S. Ct. 218, 219, 75 L. Ed. 624 (1931) that "it is the essence of a fair trial that reasonable latitude be given the cross examiner."

One goal of effective cross-examination is to impeach the credibility of opposing witnesses. In Davis v. Alaska, 415 U.S. at 316, 94 S.Ct. at 1110, the Court observed that "the cross-examiner is not only permitted to delve into the witness' story to test the witness' perceptions and memory, but the cross-examiner has traditionally been allowed to impeach, i.e., discredit, the witness." Similarly, in United States v. Williams, 592 F.2d 1277, 1281 (5th Cir.1979), we noted that cross-examination in "matters relevant to credibility ought to be given wide scope."

Certain forms of mental disorder have high probative value on the issue of credibility. Although the debate over the proper legal role of mental health professionals continues to rage,3 even those who would limit the availability of psychiatric evidence acknowledge that many types of "emotional or mental defect may materially affect the accuracy of testimony; a conservative list of such defects would have to include the psychoses, most or all of the neuroses, defects in the structure of the nervous system, mental deficiency, alcoholism, drug addiction and psychopathic personality." Juviler, Psychiatric Opinions as to Credibility of Witnesses: A Suggested Approach, 48 Cal.L.Rev. 648, 648 (1960). Mental illness may tend to produce bias in a witness' testimony. A psychotic's veracity may be impaired by lack of capacity to observe, correlate or recollect actual events. A paranoid person may interpret a reality skewed by suspicions, antipathies or fantasies. A schizophrenic may have difficulty distinguishing fact from fantasy and may have his memory distorted by delusions, hallucinations and paranoid thinking. A paranoid schizophrenic, though he may appear normal and his judgment on matters outside his delusional system may remain intact, may harbor delusions of grandeur or persecution that grossly distort his reactions to events. As one commentator succinctly summarized the interplay between mental disorders and legal issues of credibility:

[T]he delusions of the litigious paranoiac make him believe he has grievances, which he feels can be corrected only through the courts. His career as a litigant is frequently touched off by a lawsuit or legal controversy whose outcome left him dissatisfied. Often he will insist on conducting his own case, quoting voluminously from cases and statutes. Because he is likely to be of better-than-average intelligence, he may mislead a jury that is uninformed about his paranoiac career and actually convince them that his cause is just.

Trivial incidents and casual remarks may be interpreted in a markedly biased way, as eloquent proof of conspiracy or injustice. In his telling them, these trivial incidents may by retrospective falsification be given a grossly distorted and sinister significance. Even incidents of a decade or more ago may now suddenly be remembered as supporting his suspicions, and narrated in minute detail.

On the other hand, so far as the power of observation is concerned, the paranoid witness may be quite as competent as anyone, and perhaps more than most; his suspiciousness may make him more alert and keen-eyed in watching what goes on.

Delusions of persecution may evoke intense hatred. This may lead to counter-accusations resting on false memory, which may be very real to the accuser and be narrated by him with strong and convincing feeling. And indeed they may have a kernel of truth; because of his personality and his behavior, many people probably do dislike him. As Freud said, a paranoid does not project into a vacuum. Such a person not infrequently feels the need for vengeance.

Weihofen, Testimonial Competence and Credibility, 34 Geo.Wash.L.Rev. 53, 82 (1965) (footnotes omitted).

The government in this case contends that psychiatric evidence merely raises a collateral issue. But such labels cannot substitute for analysis. Whether called "collateral" or not, the issue of a witness' credibility is committed to the providence of the jury. Although the use of psychiatric evidence "does not fall within the traditional pattern of impeachment, the law should be flexible enough to make use of new resources." Weihofen, supra at 68. By this late date, use of this kind of evidence can hardly be termed "new."

At trial the defense sought to show that the key witness for the government was not credible, arguing that her motive for initiating and pursuing the investigation of Bay Therapy was based on hatred of the appellants.4 Lindstrom and Slater argued to the district court that this witness was carrying out a vendetta against them because she had not received a promised percentage of Bay Therapy when the business was sold. Appellants further sought to impeach the witness' credibility by demonstrating that her alleged vendetta resulted from a continuing mental illness, for which she had been periodically treated and confined. From public sources and from psychiatric records which the district court permitted defense counsel to review, the defense gathered material suggesting that in 1971 the witness was hospitalized following a serious suicide attempt; that in 1977 the witness, while she was running Bay Therapy, offered a patient of Bay Therapy $3,000 to murder the wife of the witness' alleged lover; that in 1978 she was involuntarily committed under Florida's Baker Act5 after taking an overdose of drugs; that in 1980 she was arrested and charged with aggravated assault for having allegedly fired a shotgun through the window of her purported lover's house; that following this incident she was briefly placed in a stockade until, at the urging of her psychiatrist, she was transferred to Hillsborough County Hospital where she was involuntarily committed under the Baker Act; that during this confinement she was diagnosed "schizophrenic reaction, chronic undifferentiated type" and described by the Chief of Psychology at Hillsborough as being "immature, egocentric, [and] manipulative," having superficial relationships causing "marital problems and sexual conflicts in general" and seeing authority as something to be manipulated for self gratification and as an obstacle; that an unsigned chart entry noted that the patient had a "history of hallucinations" and was "suicidal--homicidal and delusional." Through effective questioning in these areas, appellants contend that they could have shown the witness' past pattern of aggressive and manipulative conduct toward persons close to her and that they could have demonstrated the witness' motivation and determination in pursuing a vendetta.

The trial court, fearing that the defense would attempt to put the witness herself on trial, stated:

I think we should discuss, too, the extent of the cross-examination of this witness in regard to these activities. I am trying to think this through. I am not fully convinced it's a 608-B question, although I think it's akin to that. But any questions along this line probably would go further than what seems to be envisioned by 608-B, because you're testing the witness's credibility is what you're attempting to do, I would suppose.6 So, within reason, there are two or three properly framed questions I'm going to allow you to at least let this jury know about the fact that this witness has had some mental and emotional problems in the past.

I'm not going to allow the defense to try this witness, so to speak.

Three examples suffice to illustrate the extremely narrow limits within which the district judge permitted cross-examination of the witness.7 First, the court refused to allow the defense to question the witness about the murder contract that she allegedly offered to a Bay Therapy patient. The defense proffered testimony by the patient to the effect that the witness had approached him and offered him $3,000 to shoot the wife of the witness' purported lover. The court sustained the government's objection to the questions, and also denied a defense request to ask the witness the questions out of the hearing of the jury. Secondly, the district court denied appellants the opportunity to cross-examine the witness about her own alleged attempt to shoot a shotgun through the window of her purported lover's house, after which she was committed under the Baker Act. Thirdly, the judge sustained the government's objection to defense questions focusing on whether, during her commitment, the witness had told hospital personnel that she had attempted suicide for the purpose of manipulating and punishing her boyfriend. Defense counsel proffered hospital records showing that the witness had in fact made such statements. During the ensuing side bar conference, defense counsel stated that the witness' history of manipulation was "the whole point of our defense. It's those people that cross her, Dennis Slater, Joanne Lindstrom, [or her alleged lover], she goes out to get them with a vendetta. She manipulated them. She did it to him on numerous occasions. She's doing the same thing to him. She's told two people that I have a witness for that she's out to get Dennis Slater."

These rulings by the district court constituted an abuse of discretion contradicting Supreme Court and former fifth circuit authority on the right of confrontation in general and the right to examine the psychiatric history of adverse witnesses in particular. In Greene v. Wainwright, 634 F.2d 272 (5th Cir.1981), the district court dismissed a petition for habeas corpus claiming a violation of the right to confrontation. The state trial judge had prohibited the defendant from questioning the state's chief witness about his "mental condition and about certain bizarre criminal actions." Id. at 274. "His credibility was crucial to the state's case." Id. at 275. The fifth circuit reversed, holding that the denial of an opportunity to present evidence regarding the "alleged recent history of mental instability ... exceed[ed] any possible trial court discretion." Id. at 276.

We find Greene controls this issue. The restrictions on cross-examination in this case are even more egregious than those recently condemned in Greene, because the disputed medical records suggested a history of psychiatric disorders, manifesting themselves in violent threats and manipulative and destructive conduct having specific relevance to the facts at issue. As in Greene, the witness in question was the chief witness for the prosecution. She initiated and pursued the investigation of Bay Therapy. She was an insider to the fraud scheme, who testified in detail about the operation and about the activities of Slater and Lindstrom. Accord United States v. Summers, 598 F.2d 450, 460 (5th Cir.1979) ("where the witness the accused seeks to cross-examine is the 'star' government witness, providing an essential link in the prosecutor's case, the importance of full cross-examination to disclose possible bias is necessarily increased"). We find the district court committed reversible error in unconstitutionally depriving appellants of their sixth amendment guarantee of the right to confrontation and cross-examination.

In addition to claiming error in limiting their ability to cross-examine the government's witness, appellants contend that the trial judge abused his discretion in denying them access to the psychiatric records of that same witness. Lindstrom and Slater argue that this material was necessary to test the perception, credibility and motivation of the witness effectively. We agree.

Prior to trial, appellants sought subpoenas duces tecum to obtain records of the witness' psychiatric treatment. The district court, after in camera examination, permitted some of the records to be used for cross-examination, but declined to allow any records admitted into evidence. The trial judge denied appellants any opportunity to review four sets of records, Court's Exhibits 1, 2, 3, and 4.

Court's Exhibit 1 consisted of the 1972 records of Dr. Charles DeMinico. Dr. DeMinico wrote in a consultation report that his initial impression of the government's witness was:

Conversion Reaction with dissociative traits. [The patient] has all the classical symptoms of a Hysterical Personality in all aspects of her past life history, her interpersonal relationships, her marital histories, and her personal psychological maladjustments. [She] admitted, to a certain degree, the manipulation of one fasting blood sugar test and passed that off casually with some sort of vague rationalization. However, she denied any conscious knowledge of having taken insulin to modify the results of the glucose tolerance test. To explain this she wove an intricate fabrication in a bland, detached way that is typical of "la belle indifference" of a classical Freudian hysteric.

The Exhibit also included a summary of a Minnesota Multiphasic Personality Inventory (MMPI) Report:

The unwillingness of this patient to admit to the relatively minor faults which most people have suggests that she is a person with strong needs to see herself, and to be seen by others, as an unusually virtuous person. Such people tend to be rigid, defensive, and uncompromising individuals who stress moral issues and emphasize their own integrity. They tend to be frustrated, insecure people who have little insight, and who are unaware of their own stimulus value. It is doubtful that these tendencies have invalidated the patient's test results, but they may have caused her to receive somewhat reduced scores on the clinical scales.

* * *

She is inclined to be overly sensitive to the responses and intentions of those around her. She chronically misinterprets the words and actions of others, which leads to difficulties in her interpersonal relationships.

(emphasis added).

Court's Exhibit 2 consisted of the records of Dr. Maruicio Rubio, who treated the witness following a 1971 suicide attempt. This Exhibit, as well as parts of Court's Exhibit 1, is substantially cumulative of the psychiatric materials disclosed to appellants at trial.

Court's Exhibit 3 consisted of reports made by Dr. Daniel Sprehe, who was the witness' treating psychiatrist from 1979 until the time of trial. The Exhibit contained a discharge summary dated November 7, 1979, noting that the witness had been admitted voluntarily because she was "mulling over thoughts of suicide." On the first evening after her admission, the witness "became very hysterical and demanded to sign out, so temporarily she was held on a Baker Act involuntary commitment." However, within twenty-four hours she calmed down and was able to sign as a voluntary patient. The Baker Act commitment thus was not carried out.

In a letter dated May 22, 1979, Dr. Sprehe described the patient as "very cynical about both doctors and lawyers and regales one with many very involved intrigues and treacheries of which she feels she has evidence in the medical and legal profession. She predicts that eventually she will kill herself. She is obviously very emotionally unstable with marked hysterical features."

In a History and Physical Examination dated February 11, 1980, approximately a month after the witness' aggravated assault on her alleged lover, Dr. Sprehe noted:

She has really deteriorated in the past couple of months and the weeks prior to coming into this hospital. There was an attempt to get her to agree to go into the hospital because she was calling nightly with suicide threats, homicide threats toward her boy-friend and on at least one occasion I sent a Fire Rescue Truck out to her house because she stated that she had just taken an overdose when she talked to me on the phone. She also was getting erratic about keeping her office appointments and in general was making paranoid demands. She fired her lawyer, then rehired him, fired myself on several occasions as her psychiatrist. She felt that everyone was against her, etc.

* * *

... She has also worked as a legal researcher and because of this she fancies herself as being a legal wizard and smarter than most lawyers, and also smarter than most doctors and this makes her in a position of constantly second-guessing her doctors and lawyers, and trying to manage her case herself ....

* * *

Pseudo-neurotic schizophrenia with marked paranoid trends and with questionable competency at the present time. She has marked impulsiveness and possible suicidal tendencies. It is felt that she should be watched closely.

A Discharge Summary dated February 29, 1980 observed that her "hospital course was quite stormy with a great deal of paranoid distrust of everyone. She felt that her doctor was in league with her ex-boyfriend." She was stated to be "near maximum medical improvement ... I would rate her to be 25% partial disability from psychiatric standpoint ...."

Court's Exhibit 4 consisted of records made by St. Joseph's Hospital. A Physician's Progress log dated November 11, 1979 noted that Dr. Sprehe convinced the witness to voluntarily enter the hospital on that date following repeated serious suicide threats. The entry for that day noted that she "insists on signing out of the hospital for a variety of reason[s], all rather infantile and some of them delusional. Says she will go home and kill self to spite everyone."

The district court's rationale for denying appellants access to these records was that they were cumulative of records already disclosed to the defense; that some were remote in time from the events at issue; that the evidence might confuse the jury and that access to the confidential professional materials would violate the witness' privacy rights. We find none of these reasons sufficient to justify withholding this material evidence from the appellants.

This issue is controlled by United States v. Partin, 493 F.2d 750, 763-64 (5th Cir.1974), cert. denied, 434 U.S. 903, 98 S. Ct. 298, 54 L. Ed. 2d 189 (1977). The former fifth circuit held in Partin that the exclusion of "portions of a hospital record showing statements of the patient-witness, notation of symptoms, and treatment," id. at 763-64, constituted reversible error.8 The witness, who had voluntarily committed himself for psychiatric care approximately six months before the event about which he sought to testify, had been diagnosed as being an undifferentiated schizophrenic. The court reasoned, "In simple language the defendant has the right to explore every facet of relevant evidence pertaining to the credibility of those who testify against him," and evidence on mental capacity may be especially probative of the ability to "comprehend, know and correctly relate the truth," id. The Partin court concluded that:

The readily apparent principle is that the jury should, within reason, be informed of all matters affecting a witness's credibility to aid in their determination of the truth.

Partin had the right to attempt to challenge Rogers' credibility with competent or relevant evidence of any mental defect or treatment at a time probatively related to the time period about which he was attempting to testify.

Id. at 762, 763 (citations omitted).

We need not reach the issue of admissibility of these records in this case. The district court's denial of defendants' access to the psychiatric materials is alone sufficient to constitute reversible error. Following Partin, the fourth circuit recently reversed a conviction because defendants had been denied access to psychiatric records revealing that the principal witness was "delusional and hallucinatory with poor judgment and insight" and had been "secluded for his own welfare." United States v. Society of Independent Gasoline Marketers of America, 624 F.2d 461 (4th Cir.1979), cert. denied, 449 U.S. 1078, 101 S. Ct. 859, 66 L. Ed. 2d 801 (1980). After an in camera inspection, the trial court informed counsel that the records reflected two periods of hospitalization (one during the charged conspiracy and one prior to it) and that the hospitalization involved a "mental disorder or illness at the time." Id. at 467. The trial judge permitted the defense to question the witness about the two periods of hospitalization but did not make the records themselves available. On appeal, the fourth circuit reversed:

The ability of defense counsel to impeach [the witness] regarding his ability to properly perceive events about which he testified was severely limited by counsel's inability to examine the hospital records. We can think of no more relevant or significant material than a hospital record indicating that a witness who is testifying against his former employer had been under treatment for mental illness which rendered him at that time delusional and hallucinatory with poor judgment and insight.

Id. at 469.

The reasoning of Partin and Society of Independent Gasoline Marketers applies even more strongly in this case. The cumulative evidence of the psychiatric records suggests that the key witness was suffering from an ongoing mental illness which caused her to misperceive and misinterpret the words and actions of others, and which might seriously affect her ability "to know, comprehend and relate the truth," United States v. Partin, 494 F.2d at 769. More particularly, it hints that the witness' reactions to doctors and lawyers, and thus to the proprietors of Bay Therapy, might be especially distorted. The treatment received, the observations made, and the diagnoses were all relevant to the issue of the witness' mental condition.

When balanced against the great probative weight of the psychiatric records for the issues in this case, the district court's justifications of cumulativeness and remoteness are insubstantial. These records contrast sharply to those proffered in United States v. Diecidue, 603 F.2d 535 (5th Cir.1979), cert. denied, 445 U.S. 946, 100 S. Ct. 1345, 63 L. Ed. 2d 781 (1980). In Diecidue, defendants sought to admit records of an isolated and nonrecurrent episode of psychiatric illness occurring twelve years prior to the events at issue in the case. The witness had been voluntarily hospitalized for four months and never subsequently treated for mental illness. On appeal, the court, citing Partin, held the evidence was inadmissible because it was temporally remote from, and therefore not probatively related to, the time period about which the witness was testifying.

By contrast, the court's exhibits in this case reveal presence and treatment of a continuing mental illness embracing the time period of the alleged conspiracy. The indictment charged a period of conspiracy and scheme to defraud from 1976 to 1981. Court's Exhibits 3 and 4 include entries dated 1979 and 1980, periods during which the witness was assisting the government to uncover and prosecute the case. The relevance of these records is brought home by the several references in them to the patient's involvement with the Bay Therapy investigation. Although some of the exhibits include events occurring four or five years prior to the witness' involvement in Bay Therapy, they suggest a continuing symptomatology, including suicide threats and attempts, beginning before and continuing during the events at issue. While Exhibit 2 and parts of Exhibit 1 repeat some of the psychiatric evidence disclosed to appellants at trial, the records of Doctors DeMinico and Rubio suggest that the mental health problems of the witness were not isolated psychiatric episodes but parts of a continuing mental illness, at times requiring confinement and treatment. The depth of the psychiatric record suggests that the witness' symptoms were not merely reactive to the stress of the investigation and trial and could help to dispel the jury's feeling that "defense counsel was engaged in a speculative and baseless attack on the credibility of an apparently blameless witness." Davis v. Alaska, 415 U.S. 308, 318, 94 S. Ct. 1105, 1111, 39 L. Ed. 2d 347 (1974).

The government strenuously argues that the privacy interest of a patient in the confidentiality of her medical records and the societal interest in encouraging the free flow of information between patient and psychotherapist require affirmance of the trial court's rulings.9 While we recognize the general validity of those interests, they are not absolute and, in the context of this criminal trial, must "yield to the paramount right of the defense to cross-examine effectively the witness in a criminal case." United States v. Society of Independent Gasoline Marketers of America, 624 F.2d at 469.

Although a trial court should seek to prevent the disclosure of embarrassing, irrelevant information concerning a witness, it is an abuse of discretion to preclude defense counsel from obtaining relevant information, and the witness' privacy must yield to the paramount right of the defense to cross-examine effectively the witness in a criminal case.

Broad-brushed assertions of the societal interest in protecting the confidentiality of such information cannot justify the denial of these defendants' right to examine and use this psychiatric information to attack the credibility of a key government witness. A desire to spare a witness embarrassment which disclosure of medical records might entail is insufficient justification for withholding such records from criminal defendants on trial for their liberty. In Davis v. Alaska, 415 U.S. at 319, 94 S.Ct. at 1111, the Supreme Court rejected Alaska's attempt to utilize a similar confidentiality (the anonymity of a juvenile offender) to avoid confrontation and cross-examination:

In this setting we conclude that the right of confrontation is paramount to the State's policy of protecting a juvenile offender. Whatever temporary embarrassment might result to [the witness] or his family by disclosure of his juvenile record--if the prosecution insisted on using him to make its case--is outweighed by petitioner's right to probe into the influence of possible bias in the testimony of a crucial identification witness.

The government further argued that the excluded evidence would have confused the jury. Such a notion, that juries misconstrue evidence of mental illness because of passion and prejudice, might have been appropriate in the last century. But juries nowadays regularly deal with this sort of evidence.

We "cannot speculate as to whether the jury, as sole judge of the credibility of a witness, would have accepted this line of reasoning had counsel been permitted to fully present it. But we do conclude that the jurors were entitled to have the benefit of the defense theory before them so that they could make an informed judgment ...." Davis v. Alaska, 415 U.S. at 317, 94 S.Ct. at 1111. We hold that the jury was denied evidence necessary for it to make an informed determination of whether the witness' testimony was based on historical facts as she perceived them or whether it was the product of a psychotic hallucination. The jury was denied any evidence on whether this key witness was a schizophrenic, what schizophrenia means and whether it affects one's perceptions of external reality. The jury was denied any evidence of whether the witness was capable of distinguishing reality from hallucinations. Such denial was reversible error.

REVERSED and REMANDED.