Do I Have to Warn Someone My Patient Threatened? Tarasoff Rules by Timeline

12 min read
The Tarasoff Dilemma: Duty to Warn vs. Duty to Protect

A patient leans forward and says, quietly, "I know where she parks. I'm going tonight."

That's the moment. Not ten minutes later. Not after you finish the note. Not after you "see how they're feeling in a bit." Right there, the air changes. You're no longer just doing supportive listening or routine risk assessment. You may be standing in the narrow gap between a threat and a body.

I've seen clinicians freeze here. Good people. Smart people. They start bargaining with themselves: maybe he's venting, maybe she doesn't mean it, maybe if I keep talking this down I won't have to break confidentiality. That instinct is understandable. It's also dangerous.

Tarasoff situations are not about vague anger. They are not about every dramatic statement made in an ER, clinic, or psych unit. They're about something much tighter: a credible threat, a real risk, and a potential victim who may be identifiable enough that you now owe duties beyond the patient in front of you.

This article is for educational purposes only. It is not financial advice, not legal advice, not tax advice, and figures vary by jurisdiction and setting. Tarasoff duties are state-specific, so consult a qualified professional.

The Scenario: The Moment the Threat Emerges

Here's the classic setup: a patient discloses a specific, imminent threat against a named person. "I'm going to stab my ex-wife, Melissa, when she leaves work tomorrow." That's very different from "I'm so angry I could kill her." One is potentially actionable. The other may still be clinically serious, but it doesn't automatically trigger the same legal response.

At this point, you should do three things immediately:

  1. Stop treating this as casual disclosure

    • Don't glide past it.
    • Don't soften the language in your own head.
    • Repeat it back clearly: "You're telling me you plan to harm Melissa tomorrow after work?"
  2. Shift from empathy-only mode to risk-command mode

    • Stay calm.
    • Keep the patient engaged.
    • But internally, move into structured assessment. Fast.
  3. Separate noise from signal

    • General hostility: "People piss me off."
    • Fantasy without plan: "Sometimes I imagine hurting him."
    • Specific threat: named target, method, timing, access, intent.

That distinction matters. A lot. Clinicians get into trouble not because they cared too little, but because they documented a concrete threat as if it were emotional venting. That's sloppy, and sloppy is expensive.

Emotionally, this is hard because confidentiality is drilled into you as sacred. And it is. But it is not absolute. Once a patient presents a serious danger to another person, your job changes. Privacy still matters. Public safety now matters more.

Phase 1: Immediate Risk Assessment (The "Now")

This is the "now" phase. Minutes matter. Your task is to decide whether you're dealing with anger, fantasy, or a credible violent threat.

At this point, you should assess four things in order:

1. Imminence

Ask yourself: Is this likely to happen soon?

Red flags:

  • "Tonight"
  • "Tomorrow after court"
  • "As soon as I get discharged"
  • "I already bought the gun"

Imminence doesn't require the patient to be standing at the door with a weapon. It means the threat has moved from abstract to near-term.

2. Specificity

Ask: Is there an identifiable victim?

Higher concern:

  • Named individual
  • Specific location
  • Predictable routine
  • Known relationship conflict

Lower concern:

  • "Somebody's going to pay"
  • "People like that deserve to die"

Generalized threats still require treatment and risk management. But Tarasoff-type duties are strongest when the target is identifiable.

3. Intent

Ask directly. Yes, directly.

  • "Do you want to hurt her?"
  • "Are you planning to do it?"
  • "What have you done to prepare?"
  • "What would stop you?"

Don't dance around it. Euphemisms are useless here. People who are serious often become more specific when asked plainly.

4. Capacity and means

Ask:

  • Does the patient have access to weapons?
  • Are they intoxicated, psychotic, manic, or severely impulsive?
  • Can they physically carry it out?
  • Have they attempted violence before?

A threat from a frail patient with florid disorganization and no access may be less actionable than a threat from a sober, organized patient who has a key, a gun, and a timeline. Context changes everything.

At this point, you should also think operationally: how do I prevent immediate contact between patient and victim? That may mean:

  • keeping the patient in the clinic room,
  • initiating emergency psychiatric evaluation,
  • placing the patient on a hold if your jurisdiction and facts support it,
  • involving security,
  • restricting departure while following institutional policy and law.

The phrase "separate the patient from the potential victim" isn't poetic. It's practical. If they can walk out and act while you're still debating doctrine, you've already lost the timeline.

Phase 2: Applying the Tarasoff Standard (The "Should")

Now we move from clinical urgency to legal reasoning.

The landmark case, Tarasoff v. Regents of the University of California (1976), came out of a horrifying failure to act. Prosenjit Poddar told a therapist he intended to kill Tatiana Tarasoff. Police were notified briefly, but Tatiana herself was not warned. Poddar later killed her. The California Supreme Court held that mental health professionals may have a duty to protect identifiable third parties when a patient poses a serious danger.

That's the heart of Tarasoff. Not gossip. Not broad moral panic. A serious danger to another person.

At this point, you should ask the legal question this way:

Does this patient present a serious risk of violence toward an identifiable person such that my jurisdiction imposes a duty to warn, protect, or both?

That wording matters because states vary:

  • Some impose a duty to warn
  • Some impose a duty to protect
  • Some make disclosure permissive rather than mandatory
  • Some define who must be warned very specifically

This is where clinicians get burned by oversimplification. "Tarasoff means always warn the victim" is too crude. In some places, notifying law enforcement may satisfy the duty. In others, reasonable steps to protect may include hospitalization, warning the victim, notifying family, or some combination.

You also need judgment about exceptions and limiting factors. A threat may sound dramatic but lack legal force if:

  • the patient lacks actual intent,
  • the patient is too disorganized to form and execute a plan,
  • the statement is delusional without a realistic path to action,
  • the risk is not imminent or not directed toward an identifiable victim.

Still, don't misuse those factors as escape hatches. Clinicians love to tell themselves, "He was psychotic, so it wasn't a real plan." That can be disastrously wrong. Psychosis does not equal harmlessness. Neither does intoxication. Neither does "he calmed down later."

By this stage, you should have:

  • a documented violence risk assessment,
  • your jurisdiction's rule in mind,
  • consultation with supervising psychiatry, legal/risk management, or on-call leadership when available,
  • a clear rationale for either acting or not acting.

If you're guessing, slow down and get help fast. But don't use consultation as a delay tactic. The law respects reasoned action. It has very little patience for paralysis.

Phase 3: Execution of the Duty to Warn (The "Do")

The Chain of Notification

Once the threshold is met, act. This is not the phase for hand-wringing.

At this point, you should identify who needs protection and who must be notified under your state's rules and your institution's policy.

Who is the identifiable victim?

Usually:

  • a named person,
  • a clearly described person,
  • sometimes a small definable group.

Examples:

  • "My ex-wife, Melissa Green"
  • "My pediatrician"
  • "The night manager at the gas station on Pine Street"

That's enough in many settings. The victim does not need to be standing in your office.

Who do you notify?

Depending on jurisdiction:

  • local law enforcement,
  • the intended victim directly,
  • a parent or guardian if the victim is a minor,
  • hospital security or administration as part of immediate protection,
  • family members when necessary and legally appropriate.

How do you warn?

Use the method most likely to prevent harm quickly.

Practical order:

  1. Phone call now if immediate risk exists
  2. Law enforcement notification with concrete details
  3. Direct warning to victim if required or prudent
  4. Written follow-up/documented communication as institutionally appropriate

Verbal warnings are fast. Written records are defensible. You often need both.

Keep the warning factual:

  • patient identity if legally justified,
  • nature of the threat,
  • timeframe if known,
  • recommendations to take protective action and contact police.

Don't editorialize. Don't dump the whole chart. Disclose what is necessary to prevent harm.

And yes, if you're wondering whether this damages the therapeutic relationship, it might. That's unfortunate. A funeral is worse.

By the end of the shift, you should have a complete, dated, contemporaneous record. Not tomorrow. Not after you've slept. Same shift.

Your note should include:

  • The exact threat
    • Quote the patient when possible.
  • The target
    • Named victim, relationship, location.
  • Your risk assessment
    • imminence, specificity, intent, means, capacity, prior violence.
  • Mental status findings
    • psychosis, intoxication, agitation, impulse control.
  • Consultations
    • attending, psychiatry, legal, risk management, security.
  • Actions taken
    • hold initiated, police called, victim warned, family contacted.
  • Time stamps
    • who was called, when, and what was communicated.
  • Your rationale
    • why duty was or was not triggered.

I'm blunt about this: documentation is your shield. If your care was thoughtful but your chart is vague, the chart wins. Every time.

Bad note:

  • "Patient upset, made threats, staff aware."

Good note:

  • "At 19:42 patient stated, 'I am going to shoot my former supervisor, Daniel K., tomorrow at 7 a.m. in the employee parking lot.' Patient reports access to handgun in vehicle, endorses intent, denies deterrents. Due to specific, imminent threat to identifiable victim, hospital security notified at 19:47, police contacted at 19:49, on-call psychiatrist consulted at 19:52, emergency hold initiated at 19:55."

That's not paranoia. That's competence.

Summary: The Chronological Checklist

Medical Ethics Timeline Summary

Here's the clean timeline.

Phase 1: Now

  • Assess imminence
  • Identify specificity
  • Determine intent and means
  • Prevent immediate access to the victim

Phase 2: Should

  • Apply your state's Tarasoff-type standard
  • Decide whether there is a serious danger to an identifiable person
  • Consult quickly, but don't stall

Phase 3: Do

  • Notify the right people
  • Warn or protect as required
  • Use the fastest effective method first

Phase 4: Done

  • Document every step
  • Time-stamp actions
  • Record your reasoning clearly

Final point. The Tarasoff rule is not a loophole to abandon confidentiality casually, and it is not an excuse to hide behind privacy when someone may be killed. Your job is to balance both. In sequence. Under pressure. With a chart that shows you knew exactly what you were doing.

Key takeaways

  • The Tarasoff rule applies when a patient poses a specific, imminent threat to an identifiable person.
  • Your duty may be to warn, protect, or both, depending on state law.
  • Documentation isn't administrative clutter. It's legal armor.
  • At this point, you should act early, document clearly, and never confuse hesitation with ethics.

Keep reading

View more