The Duty to Protect: Four Decades After Tarasoff

Since the time of Hippocrates, the extent of patients' right to confidentiality has been a topic of debate, with some arguing for total openness and others for absolute and unconditional secrecy (1). In Tarasoff v. Regents of the University of California (1976), the California Supreme Court held that mental health providers have an obligation to protect persons who could be harmed by a patient. The court's decision mandates that mental health professionals use "reasonable care" in informing authorities or warning potential victims, initially referred to as the "duty to warn," or by using whatever means deemed necessary, should they determine that a patient poses a threat to a third party (2). The duty to protect has proliferated widely and has been adapted in some form throughout the United States. Forty years after the Tarasoff ruling, the threshold of the duty to protect remains subjective, with no clear set of clinical guidelines regarding when a breach of confidentiality is warranted, which places mental health providers in a dubious position.

Confidentiality facilitates open communication by reassuring patients that the intimate details of their lives that they disclose to their health care providers will remain private. U.S. legislation emphasizes the importance of confidentiality, which is enforced through the Health Insurance Portability and Accountability Act (HIPAA). HIPAA ensures that communication (for the purpose of treatment) among health care providers about a patient is privileged. Granted, the exact scope of the patient protection (through HIPAA) varies, depending on the state and on the specific context. However, some form of patient protection (i.e., privilege) exists in most states and may be invoked in judicial or quasi-judicial proceedings, whether civil, criminal, or administrative in nature (3). The Tarasoff decision ultimately paved the way for the codification of the principle that confidentiality and, in turn, privilege are not absolute, especially when a patient communicates a seemingly legitimate threat that jeopardizes the safety of a third party (4).

The immediate dilemma created by the Tarasoff ruling is that of identifying the point at which "dangerousness" (typically, but not always, of an identifiable individual) outweighs protective privilege. According to HIPAA guidelines, mental health providers, similar to other health care professionals, are subject to liability for breaching provider-patient confidentiality. However, although the duty to protect, as delineated in the Tarasoff decision, is intended to relieve providers of such liability by mandating that they alert others of a possible threat from a patient, an incorrect reading of a situation could have the opposite effect. Specifically, in a situation in which a provider strongly feels that a particular circumstance justifies a breach of provider-patient confidentiality but is ultimately mistaken, the provider could then be held liable to the patient for the breach, irrespective of any good intention on the part of the provider. Conversely, a provider who favors confidentiality over the issuance of a warning could be subject to civil liability for negligence to any threatened third party (5).

In the years following the Tarasoff ruling, its effects on the mental health field have been substantial. Mental health providers, mindful of the duty they have to warn potential third-party victims, are more acutely aware of risk factors for violence (6). However, there remain some challenges involved in implementing the duty to protect. These challenges include clarifying expectations (regarding warning or protecting) for providers and establishing guidelines pertaining to the accurate prediction and assessment of dangerousness.

The Tarasoff decision, as it is presently interpreted, raises a set of questions that may be problematic from both medical and legal standpoints. Some have suggested that once a threat has been made, "there is generally little a victim can do unless the threat is imminent" and that "warning sometimes can inflame the situation and increase the danger" (7). This poses the question of whether there is any benefit from simply warning a third party. Part of the heterogeneity of the impact of the Tarasoff ruling is that different states have adopted different approaches to the implementation of the duty to warn or protect. Although some state legislation imposes a mandatory duty on mental health providers, other states have implemented a permissive duty (in that providers are not liable for breaching confidentiality and are not required to do so). Yet some states have not established a clear position on the implementation of Tarasoff-like decisions (either they do not have laws or have different laws for different types of mental health providers) (see box) (8).

One challenge in predicting dangerousness is that providers are often unclear about how to accurately prognosticate, because "prediction and assessment of violent behavior do not yet have reliable, clinically validated paradigms" (1). This is especially problematic because, in many instances, people do not always intend to act upon their threats (9). Although mental health providers have some tools for violence risk assessment, such tools are not foolproof, and thus mental health providers are vulnerable to malpractice lawsuits (10). For example, in California "psychotherapists must warn both the foreseeable victim and the police in order to enjoy protection from subsequent lawsuits" (11).

The Historical, Clinical and Risk Management-20 scales are used for violence risk assessment. In one study, this risk-assessment model was validated to predict violent behavior in an inpatient setting (12). Another risk-assessment measure is the Violence Risk Appraisal Guide, which was validated to predict violent behavior among patients charged with criminal offenses in a study conducted in Germany (13). A study conducted in the United Kingdom examined both the aforementioned risk-assessment models in a prison setting (14). The authors reported that neither model was sufficiently predictive in the assessment of persons with severe mental disorders and particularly ineffective in the evaluation of persons with personality disorders (14). The main limitation of the three aforementioned studies is that the validity of the measures assessed was not examined in an outpatient setting, which is the setting in which a duty to protect situation is most likely to occur.

We argue for an unambiguous and ubiquitous method for predicting danger and applying the duty to warn directive. It is noteworthy that the decision to warn is not necessarily harmful and has been shown to be beneficial to potential third-party victims, as well as to the therapeutic progress of patients (1). The duty to warn directive could be made more universal by establishing it as a federal law, or by implementation of federal guidelines to assist states in consistent application of the injunction, which would minimize the legal liability among mental health providers, because they would be able to measure their actions against a clearly defined objective standard.

One possible mechanism by which third parties could be warned is a clinical point-system scale capable of assisting in the evaluation of the probability of a patient carrying out a threat. To be effective, such a measure would need to be developed on the basis of current evidence and authorized by mental health professionals who are experts in the field. Furthermore, a national consensus on the guidelines pertaining to the duty to protect needs to be established.

Previous studies have reported risk factors for patient violence to include previous diagnosis of antisocial personality disorder or thought disorders, previous suicidal or homicidal ideation or attempts, lack of social support, access to weapons, and current treatment with antipsychotics or mood stabilizers (1, 15–17). Other factors, on the basis of our literature review, include a patient's previous treatment rapport with his or her psychiatrist, whether the patient's symptoms are responsive to treatment or therapy, whether the patient has identified a specific person to harm or a location to carry out an act of violence, and whether the patient has identified a single person or a group of persons.

Four decades have passed since the Tarasoff ruling, yet a clear and ubiquitous method for its application has not been established. Discrepancies and vagueness between states, as well as between providers, regarding how and when to apply the duty to protect still exist. Such variances affect both therapeutic alliances and providers' risk of legal liability. Development of more validated risk-assessment tools would assist mental health professions in their decision making, enabling preservation of the integrity of the provider-patient relationship and minimizing the risk of legal liability. Clinical judgment remains an invaluable addition to instruments for determining whether the duty to protect is warranted.

Key Points/Clinical Pearls