Warning a Potential Victim of a Person's Dangerousness: Clinician's Duty or Victim's Right?

Cases of Duty to Warn or Protect

The practice of warning an identifiable victim of the risk of violence, adequately determined through clinical assessment, is the model that is discussed and promoted in the professional literature and is in greatest agreement with the Tarasoff principle itself. The Tarasoff principle does not require the clinician to warn the victim of the “verbal threat,” but rather of the “danger,” when victim warning is the appropriate protective measure. Neither is the warning necessarily triggered or legally required by a verbal threat expressed by the patient, the facts of the Tarasoff case notwithstanding. Rather, the duty to warn, or to take any appropriate protective measures arises, “[W]hen a therapist determines, or pursuant to the standards of his profession should determine, that his patient presents a serious danger of violence to another”(Ref. 1, p 426), according to the original Tarasoff principle. Regardless of the limited professional standards at the time of the Tarasoff event and the case for accurately predicting future acts of violence, the court's formulation made clear that some attempt to assess the seriousness of the danger was required; a simple knee‐jerk acceptance of any expressed threat would not do. By improving the accuracy of a determination of dangerousness, protective disclosures would be limited to situations wherein warnings are deemed necessary to prevent serious harm. Assessment and determination of risk further justifies violating confidentiality and serves to reduce the likelihood of unnecessary disclosures, which could result in adverse consequences from the disclosures themselves. According to the Tarasoff principle, the intended victim is to be warned of the “danger” (Ref. 1, p 426) posed by the patient, not simply of the patient's verbal threat.

Several early Tarasoff‐like cases did not involve a verbal threat against an individual, yet the court found a duty to warn. Obviously, the warning would not have been of a nonexistent threat, but of the presumed danger to the otherwise foreseeable victim. In McIntosh v. Milano¹⁹ for example, the Superior Court of New Jersey held that a psychiatrist had a duty to warn, even though his patient had not expressed a threat to harm the patient's homicide victim. According to the court, the question for the jury was whether the psychiatrist “knew or should have known” that his patient “presented a clear danger or threat” to the victim (Ref. 19, p 511). Here, the court must have been more concerned about a possible danger than a verbal expression for which no evidence was entered. In suggesting that the jury look for “retaliation fantasies,” the court implied that the psychiatrist should have considered this in assessing whether a clear danger existed. The court's emphasis on “retaliation fantasies,” can be faulted for failing to distinguish between idle fantasies and serious intentions. Was the court asking the jury to “fish” for surrogates of missing threats or intentions? Or, more favorably, perhaps the court was simply attempting to advise the jury how to determine whether dangerousness was present and properly assessed.

Especially on point is Jablonski v. United States²⁰ in which the United States Court of Appeals for the Ninth Circuit affirmed the trial court's finding that the psychiatrist had a duty to warn the future homicide victim, Ms. Kimball. Although the patient, Mr. Jablonski, had not expressedly threatened to kill Kimball, further assessment should have established the danger he posed to her. Specifically, the hospital should have “secure[d] Jablonski's prior records” (Ref. 20, p 397). This appears to support a duty to warn of the patient's danger or risk to the victim. Ms. Kimball told Dr. Kopiloff that she felt “insecure around Mr. Jablonski and was concerned about his unusual behavior” (Ref. 20, p 393). Dr. “Kopiloff recommended that she leave Jablonski at least while he was being evaluated. When Kimball responded ′I love him', [Dr.] Kopiloff did not warn her further because she would not listen to him” (Ref. 20, p 393). According to the Ninth Circuit's holding, the duty to warn Ms. Kimball of the risk was insufficiently fulfilled.

Dangerous Patient Exception to Privilege Cases

Dangerous patient exception to privilege cases do not contribute to the duty to warn jurisprudence. However, the genre of cases that invoke the Tarasoff exception to confidentiality provides another view of how some courts view the duty (i.e., whether it is a duty to communicate the patient's dangerousness or the patient's threat). Long before the Tarasoff decision, the patient's privilege prohibiting a psychiatrist or therapist from testifying in court could be violated to allow for expert testimony to support civil commitment. The Supreme Court of California in Tarasoff noted this exception to justify violating confidentiality for protective warnings. Justice Tobriner's famous quotation, “The protective privilege ends where the public peril begins” (Ref. 1, p 347) alludes to this comparison, even though conceptually and semantically conflating confidentiality with the evidentiary privilege.

Following Tarasoff, a series of cases established in California a “dangerous patient exception” to this privilege, which allows therapists to testify in court in the prosecution of their patient for a criminal offense. If a Tarasoff warning had been issued, the court could compel the therapist to testify about the warning and the basis for it to prove the elements of the crime. (For analysis of dangerous patient exception to privilege cases in California, the reader is referred to articles by Harris,²¹ Leong et al.,^22–25 and Weinstock et al.²⁶ In the present discussion, interest belongs with the court's view of the duty to warn, not the exception to privileges for which these cases are legally important.

In People v. Wharton²⁷ the Supreme Court of California held: [W]here a psychotherapist warns a potential victim … [the] statute permits the psychotherapist to reveal, in a later trial or proceeding, both the substance of the warning and the patient's statements made in therapy, which caused or triggered the warning [Ref. 27, p 314].

However, because the defendant/patient did not threaten to kill or murder the victim, there was no verbal threat to convey. The therapist was apparently alarmed by the patient's expressed fear of losing control of his anger. Thus, the therapist warned the victim not of a nonexistent verbal threat, but that “she was in danger” (Ref. 27, p 304).

Wharton is not the only case involving the dangerous patient exception to the therapist‐patient privilege that could be cited, wherein the court supported or referenced a duty to warn the potential victim, not of a threat but of an independently determined danger (e.g., Bradley v. Ray²⁸). If the duty is to warn of a danger and not of a specific threat, then a verbal threat may not be necessary to trigger the duty. Other findings can establish the potential for violence against a foreseeable victim.

How to Warn of the Danger

Not discussed in Tarasoff‐like cases, Tarasoff statutes, or even clinical writings is how warnings of danger ought to be expressed to an identifiable victim. Should the therapist say, for example: “After assessing my patient, who is your live‐in partner, I am convinced that he presents a substantial risk to your life. In other words, he could kill you.”? Would the potential victim be helped or unnecessarily shocked if the therapist quoted the patient's threats or other statements that substantiated the risk? Rather than just paraphrasing the patient's threat or giving an absolutistic warning (e.g., “Your life is endangered.”), conveying estimated degree of risk (high, medium, low) may be helpful. Should the therapist offer any advice about self‐protective measures? Or does it suffice simply to state, “I believe John Doe is a danger to you.” But then how should the therapist respond to questions evoked by such a chilling warning? The best practices for how clinicians should issue warnings is an important topic that warrants more discussion than current space allows, but now we shall see that warning the victim of the risk is by no means the universal standard for discharging a protective duty via a warning.

Definition of a Threat

Without a disambiguating definition, the word “threat” can carry multiple meanings. A threat can be “an expression of intention to inflict evil, injury, or damage” (Ref. 29, p 1224) or “an indication of something impending” (Ref. 29, p 1224), especially something “evil”³⁰ or dangerous. Apart from Tarasoff jurisprudence, the legal definition of a threat occurs in the context of criminal law: “A communicated intent to inflict harm or loss on another or on another's property, especially one that might diminish a person's freedom to act voluntarily or with lawful consent.”³¹ However, even the offense of a “criminal threat” need not involve an intent of actually carrying out the threat. The statement is intended to be taken as a threat and to convey “gravity of purpose,”³² regardless of whether the speaker actually intends to follow through with the threat. When applied to people, a threat usually denotes verbal or written expression rather than actual potential for harmful act(s). Unless otherwise defined or unless context indicates an alternative meaning, in Tarasoff jurisprudence and clinical practice, the term threat denotes a patient's verbally expressed intention to harm another person. A relatively recent development in California was an appellate court interpretation of “patient communication” to include communications from the patient to a family member who in turn conveys the communication to the therapist (Ewing v. Goldstein¹⁷), to be discussed later in the article. More typically, however, the threat is expressed to the clinician and not necessarily to the intended victim.

Tarasoff Statutes

Some clinicians warn intended victims of a patient's verbal threat to harm them without conducting a clinical assessment to determine whether the threat is serious or likely to be acted on. In actuality, the formulation of various Tarasoff statutes and appellate court decisions indicate that warnings are triggered by the threat itself. Granted, the threat‐triggered warning is typically qualified as “serious” (e.g., California³³ and Colorado³⁴), “actual” (e.g., Indiana,³⁵ Kentucky,³⁶ Montana,³⁷ and Washington³⁸), “immediate” (Louisiana³⁹), or “specific [and] serious” (Minnesota⁴⁰), suggesting at least some judgment about the nature of the threat. Follow‐up qualifiers further define the threat: for example, Kentucky's statute requires the threat to be one of “physical violence against a clearly identified or reasonably identifiable victim.”³⁶ Tarasoff statutes are typically mute on the matter of assessment and refer to threats as triggering events, not clinically established risks of violence. Furthermore, the warning option for protecting an identifiable victim is to “communicate the threat” (e.g., California,³³ Indiana,³⁵ Kentucky,³⁶ Minnesota,⁴⁰ New Hampshire,⁴¹ and Washington³⁸), not the assessed danger or risk, to the intended victim. Likewise, police are to be “notif[ied]” of the same threat, not of the risk or danger.

Although Tarasoff statutes typically allow or require warning of the patient's threat rather than danger, some statutes are exceptional and require communication of “communications” or “information” after dangerousness has been determined. Florida's permissive protective disclosure law, for example, permits communication of the “patient communication” to the potential victim or other, when the treating psychiatrist “makes a clinical judgment that the patient has the apparent capability to commit such an act and that it is more likely than not that in the near future the patient will carry out the threat.”⁴² Presumably a verbal threat per se is neither sufficient nor necessary to occasion a protective disclosure, but how the danger is to be determined is left to the clinician. Oregon's law allows for disclosure of neither a threat nor the risk but rather, “[i]nformation obtained in the course of diagnosis, evaluation or treatment of an individual that in the professional judgment of the health care services provider, indicates a clear and immediate danger to others or to society” (emphasis added).⁴³ Another variation is a warning not of the threat, the danger, or the information, but rather of the “conduct” that occasioned the concern (Almonte v. New York Medical College).⁴⁴

The first Tarasoff statute, sometimes referred to by its sponsor's name, McAlister, became effective in California in 1986. It was intended to provide California psychotherapists with more guidance and liability protection than the post‐Tarasoff case law did. Without affirmation or denial that a duty to warn or protect exists, the law stated that if such a duty exists, it exists only if “the patient has communicated to the therapist a serious threat of physical violence against a reasonably identifiable victim or victims” (§ 43.92 subdiv. (a)³³). The therapist fulfills any duty that may exist by making “reasonable efforts to communicate the threat to the victim or victims and to a law enforcement agency.”

Beyond its attempts to circumscribe and clarify any potential protection responsibilities, two specific departures from the Tarasoff principle should be noted. First, the therapist need not determine, by application of his profession standards, that “his patient presents a serious danger of violence to another.” If the patient has communicated a “serious threat of physical violence against a reasonably identifiable victim or victims” [emphasis added],³³ this alone is sufficient to elicit the duty. The qualifier “serious” could require some discernment or assessment. The original bill used the term “actual” (AB1133) rather than “serious.” However, the Attorney General expressed concern that “actual” could prevent liability where the threat is credible but conditional.⁴⁵ Even with the measured expansion of liability, any duty of the therapist hinges more on what the patient said (threat) and how he said it (serious), than whether or not the therapist conducted an assessment of risk.

Second, any protective duty is not discharged by warning of the “danger” or risk as required by the Tarasoff principle, but rather by attempting to “communicate the threat.” Incidentally, although the Tarasoff principle included family members by implication (i.e., warning “others likely to apprise the victim of the danger” could suffice), the McAlister law limits the sphere of warnings to the victim(s) and law enforcement and does not include family members.

Example of a Court's Interpretation of “Serious Threat” in a Tarasoff Statute

The majority of appellate courts in states with Tarasoff statutes¹⁸ surprisingly do not consider the relevant statute when considering a case of duty to warn or protect. In contrast to these other courts, the Second District, California Court of Appeals in Ewing v. Goldstein¹⁷ actually examined California's Tarasoff statute (i.e., § 43.92 of the California Civil Code) in arriving at its decision.

The patient, Mr. Colello, told his father over dinner that, “[h]e couldn't handle the fact that [his former girlfriend] was going with someone else” and that he “was considering causing harm to the young man that [his girlfriend] was seeing” (Ref. 17, p 867). Mr. Colello's father told Dr. David Goldstein, Mr. Colello's therapist, what Mr. Colello had said. Dr. Goldstein was instrumental in initiating Mr. Colello's hospitalization.

Upon hearing that Mr. Colello would soon be discharged, Dr. Goldstein shared with Dr. Levinson, Mr. Colello's hospital psychiatrist, his concerns and urged Dr. Levinson not to discharge Mr. Colello. Nonetheless, Mr. Colello was discharged and on the very next day murdered his former girlfriend's new boyfriend and then killed himself. The trial court granted summary judgment, because the California Tarasoff statute requires that, “the patient has communicated to the psychotherapist a serious threat of physical violence against a reasonably identifiable victim or victims” (emphasis added, § 43.92). Because the patient himself did not communicate the threat, as required by the statute, the trial court found that the immunity provided by the statute protected the therapist. Moreover, Dr. Goldstein's information did not constitute the “serious threat of physical violence” specified in the statute.

Rather than settle for the literal meaning of the statute, the appellate court looked for legislative intent and construed a “patient communication” to include a “communication from a patient's family member to the patient's therapist, made for the purpose of advancing the patient's therapy” (Ref. 17, p 868). Thus, Mr. Colello's father's report of what Mr. Colello said amounted to a “patient communication.” The California appellate court held that the trial court's finding of summary judgment was in error and that the communication from the patient's father “raised a triable issue concerning the therapist's duty to warn the victim” (Ref. 17, p 866). In discussion the court explained that the threat must actually lead the therapist, “to believe the patient poses a risk of grave bodily injury to another person” (Ref. 17, p 874). “… [A] threat to take another's life, if believed, is sufficient to trigger a therapist's duty to warn the intended victim and a law enforcement agency” (Ref. 17, p 874).

In the original Tarasoff case, the greatest error appears to have been committed by the campus police officers who failed to carry out the therapist's instruction and take patient Poddar to the hospital for admission. Dr. Moore, the therapist, was potentially liable because he tried to do the right thing. By initiating hospitalization he demonstrated his belief that the patient was dangerous and thereby established the possibility of his own liability. Similarly, in Ewing, if an error was committed, it was the hospital psychiatrist's failure to keep Mr. Colello in the hospital despite Dr. Goldstein's caution. By urging Dr. Levinson to keep Mr. Colello in the hospital, Dr. Goldstein demonstrated his belief that Mr. Colello was dangerous, thereby contributing to the clinician's own potential liability in this new, expanded version of the Tarasoff duty in California.

The appellate court did not find that the communication from Mr. Colello's father to Dr. Goldstein was sufficient alone to trigger the duty to warn. However, this communication should have been allowed for the jury to determine “whether the patient had communicated to the therapist a serious threat of physical violence to another” (Ref. 17, p 873). From this, one might conclude that this opinion suggests that the threat triggers a risk assessment and that the warning is to be issued only after the risk is determined to be serious. The opinion, however, does not say this. Rather, the second critical question in determining when a therapist has a duty to warn, is whether the threat “actually leads him or her to believe the patient poses a risk of grave bodily injury to another person” [emphasis added] (Ref. 17, p 874). Unlike the Tarasoff principle which required the therapist to use the standards of his profession to assess the risk of danger, Ewing, consistent with the dissenting opinion of Justice Mosk in Tarasoff, required only that the therapist believed the risk to be present.

Anti‐Tarasoff Court Decisions

As illustrated here, some court decisions require therapists to warn the victim of the danger posed by the patient,⁴⁶ whereas others specify that the victim is to be warned of the patient's threat, that is, the verbal threat that occasioned the protective measure. Even decisions contrary to Tarasoff refer to warnings of and based on threats, not dangers or risks established as a result of clinical assessment. For example, in the Texas Supreme Court's rejection of Tarasoff‐like duties in its decision in Thapar v. Zezulka⁴⁷ the court “decline[d] to impose a common law duty on mental health professionals to warn third parties of their patient's threats” [emphasis added] (Ref. 47, p 640).

Lyndall Zezulka, heir to the estate of Henry Zezulka, who was killed by her son, Freddy Ray Lilly, brought action against Dr. Thapar who had treated Lilly. She claimed that Dr. Thapar had been negligent in failing to warn her family that Lilly “contemplated killing” his stepfather, Henry Zezulka (Zezulka v. Thapar, Ref. 48, p 507). On appeal, the Texas Supreme Court articulated the complaint as a negligent failure “to warn of Lilly's threats toward Henry Zezulka” (Ref. 47, p 636). Therefore, the question was whether a mental health professional, such as Dr. Thapar, has a “duty to directly warn third parties of a patient's threats” [emphasis added] (Ref. 47, p 637), not of the patient's risk, danger, potential for violence, or even contemplations, but simply the threats themselves. (One could argue that since the court found no duty to convey the patient's threats and did not address whether a duty exists to notify the victim of the homicidal risk posed by the patient, the Texas court did not reject the Tarasoff principle after all.)

Dangerous Patient Exception to Privilege Cases

In the previous section, the California Supreme Court's Wharton case, concerning the dangerous patient exception to testimonial privilege, involved therapists who warned the identified victim of the danger that the patient posed to her safety. In a similar case (Menendez v. Superior Court, 1992),⁴⁹ the same court again acknowledged a qualified and limited “dangerous patient” exception to the therapist‐patient testimonial privilege. However, in this case, the facts described the Tarasoff‐type warning, not as a warning of the danger, but of the verbal threat. The psychotherapist had reason to believe that “(1) the patient [was] dangerous and (2) disclosure [was] necessary to prevent any harm,” suggesting some impressionistic weighing, if not assessment, of risk. In Menendez, however, the two patients expressed during therapy verbal threats to kill identifiable individuals, and the warnings are described as disclosures of “communications” made in the course of psychotherapy. This practice conforms to the California statutory duty to communicate the threat,²⁷ already expressed by the patient(s), but not to the Tarasoff duty “to apprise the victim of the danger” [emphasis added] (Ref. 1, p 426).

The Ninth Circuit Court of Appeals (United States v. Chase⁵⁰) considered whether a dangerous patient exception to a psychotherapist‐patient privilege should be acknowledged. Of particular relevance to the practice of warning of threats was the factual account of the interaction between patient Chase and his psychiatrist. The psychiatrist “warned Chase that if he told her specifics about whom he planned to kill, she would have a duty to alert those people” (Ref. 50, p 1021). This approach reflects a practice of fulfilling the duty to warn the potential victim of the threat, but not primarily to protect the victim by first evaluating the nature and seriousness of the risk. After consulting with her supervisor and obtaining more information from the patient, the psychiatrist, “disclosed to FBI agents the threatening statements Chase had made during the therapy sessions and described whom he had threatened” (Ref. 50, p 1021).

In a case (United States v. Glass⁵¹) permitting the “dangerous patient exception to testimonial privilege,” the Tenth Circuit Court of Appeals wrote: “[A] psychotherapist may testify about a threat made by a patient if the threat was serious when it was uttered and … its disclosure was the only means of averting harm …” [emphasis added] (Ref. 51, p 1360). The qualifier “serious” leaves open the question of whether the seriousness of a threat is self‐evident or requires clinical assessment. In any case, it is the threat and not the risk that is to be disclosed, although, by the second part of this formulation, the risk of harm would have been appropriately assessed. Thus, there is a subcategory of warning of verbal threats, only after the threat has been assessed as representing an actual danger. Emerging literature¹³ that advises clinicians how to assess the risk of violence when a patient has expressed a verbal threat further supports this practice. If only those verbal threats deemed to represent actual risk of violence are shared with the intended victim, this second step would limit disclosures and reduce unnecessary warnings with their potential for untoward consequences.

What percentage of court decisions that explicitly require clinicians to assess the degree of risk (e.g., Jablonski²⁰), rather than simply to convey a verbal threat, is an interesting empirical question. In any event, according to some court decisions, clinicians who warn of verbal threats based only on the threats themselves may be guided by duties articulated in case and statutory law.

An Unexpressed Ethic?

Why would a verbal threat trigger a warning without an assessment of the risk? Although it is admittedly speculative, one cannot help but wonder whether a powerful, unexpressed ethic is at work here. The threat can be regarded as either a red flag that signals further danger or as an essentially offensive act per se. An expression of harmful intent renders the object of the patient's resentment even more vulnerable if the person is unaware of the threat. The expressed intent can be regarded as a prelude to the grim performance.

However valid or invalid as a predictor of future violence in individual cases, a verbal threat can also be considered as an act of aggression in itself, even if not expressed directly to the potential victim in person. Regardless of whether physically attacked, the “victim” is already victimized in effect by the threat, somewhat as one can be victimized by libel or slander. In fact patients have been prosecuted for having expressed a threat, even when the threat was confined to the context of therapy (see Weiner⁵²). The tacit, underlying assumption, then, is that the object of the threat has a right to know of the act (viz, the threat) “already committed” against him or her. In a sense, the victim's right to know compares with the government's right to know of terrorists' plans to commit atrocities. From this perspective, the intended victim has a right to be warned of the threat regardless of how great or little the risk of violence in fact is.

Duty to Warn Cases

Though not involving clinical care, a pre‐Tarasoff California case, Morgan v. County of Yuba⁵³ illustrates this genre of potential liability based on a promise to warn. A police officer, who was aware that the accused had threatened to harm a complaining witness, allegedly promised the victim that he would warn her if and when the accused were to be released from jail custody. However, when the accused was released on bail, no such warning was issued. The accused fatally attacked the victim, whose estate sued, claiming she should have been warned as promised. The appellate court found that the defendants did not “induce reliance or lull (the victim) into a false sense of security” (Ref. 53, p 886), because no such promise had been made. Even this negative finding, however, suggests a possible legal duty to warn of the accused's release, if in fact such a reassuring promise had been made to the complainant.

The factual account of Nasser v. Parker,⁵⁴ a Virginia case, illustrates how knowledge of a patient's hospitalization and hospital discharge can affect a victim's actual or perceived vulnerability. George Edwards put a gun to the head of his rejecting girlfriend, Angela Lemon, and threatened to kill her. It is difficult to imagine a verbal warning from a therapist that could have been anymore informative of the danger than the patient's actual threat and conduct toward the victim. Indeed, in fear of her safety, Ms. Lemon left her home so Mr. Edwards would not find her. Five days after he threatened to kill Ms. Lemon, Mr. Edwards was hospitalized voluntarily. Feeling safe with the knowledge that he was hospitalized, Ms. Lemon returned to her home. However, Mr. Edwards was not on a secure unit and he left the hospital without authorization. Eleven days later, Mr. Edwards fatally shot Ms. Lemon and then himself.

The Virginia Supreme Court found no duty to warn or protect, and this case does not provide legal support for the practice of requested warnings. Regardless of whether the psychiatrist acted negligently or not, one can appreciate post hoc that the victim would have wanted the patient to remain hospitalized. Or, if he were released or allowed to leave without authorization, she would have wanted to know of this fact, so she could remain in hiding or do whatever was needed to protect herself. Though typically unsuccessful, other plaintiffs have claimed failure to warn of a patient's hospital release (e.g., Bishop v. S.C. Dept. of Mental Health).⁵⁵

Is there a Duty to Warn Victims Who Already Know of the Danger?

If a person has already been threatened or attacked by the subject, is it reasonable to assume that the person wants to be warned, even without an expressed request? Does the law require warnings in such cases? Or is there no duty, because the would‐be victim would not be told of something she or he does not already know? In Cantrell v. United States⁵⁶ the United States District Court Eastern District of North Carolina, Raleigh Division, found that the victim's prior knowledge of the subject's “violent tendencies” was further reason for not supporting a clinician's duty to warn her. The Supreme Court of Minnesota found that, “… if a duty to warn exists at all, it is a duty to warn of latent dangers [emphasis added]” (Cairl v. State, Ref. 57, p 26)—that is, when the victim is not already aware of the patient's violent disposition (Ref. 2, p 24). In contrast, in Jablonski,²⁰ the warning to Ms. Kimball that if she felt threatened, she should consider staying away from patient Jablonski, was unheeded and afterward found by the court to have been insufficient. Courts are divided on whether a duty to warn exists when the victim is otherwise already aware of the danger.