What Doesn’t Work in Suicide Prevention

This is the second installment in a three-part series. Links to Parts I and III are available at the end of the story.

***

Let’s be honest about something tender and difficult: many of the practices we now call “gold standards” in suicide prevention were built inside systems shaped by fear.

Fear of losing a client. Fear of missing something. Fear of being blamed. Fear of a board complaint, a lawsuit, a chart review, an inquest, a colleague’s judgment, or the unbearable question: Should I have done more?

These fears aren’t irrational. Anyone who’s sat across from a person in suicidal agony knows how high the stakes feel. But fear, when left unexamined, can quietly take over the work and move too quickly from curiosity to categorization, from presence to procedure, from relationship to risk management. It can lead us to reach for tools that promise certainty in a clinical reality that is, by its nature, uncertain. And over time, some of those tools can begin to feel less like supports for care and more like substitutes for it.

This is where our field gets into trouble.

Risk assessments, safety protocols, documentation standards, and higher levels of care all have a place. The problem isn’t that we use them; it’s that we tend to rely on them as if they can protect us from helplessness, protect clients from all possible outcomes, or replace the deeper clinical work of staying connected to a person who’s terrified, ashamed, ambivalent, and in pain.

Three practices have become so embedded in standard suicide care that questioning them can feel almost heretical: predictive risk assessment, no-suicide contracts, and reflexive forced hospitalization. Each can emerge from an understandable desire to keep people alive, but each can also drift into something defensive, procedural, and disconnected from the person we’re trying to help.

To reclaim compassionate suicide care, we must be willing to ask not simply, Did we follow the protocol? but Did the protocol help us stay connected, collaborative, and clinically brave?

When “Risk Assessment” Becomes an Interrogation

Risk assessment is not the enemy of compassionate suicide care. Done well, it can open the door to lifesaving conversation. Asking directly about suicide, understanding a person’s pain, exploring access to lethal means, and clarifying what’s helped them survive so far are all essential parts of good clinical practice.

The problem begins when assessment shifts from conversation to prediction.

This is the core tension for therapists: we’ve been tasked with trying to predict imminent suicide risk, even though the tools available to us aren’t adequate for that job. We’re asked to make high-stakes determinations about what a person might do in the next hour, day, or week, while relying on instruments and categories that can’t reliably forecast something as complex, dynamic, and context-dependent as suicidal behavior.

Somewhere along the way, our field began to treat structured risk assessments as if they could tell us who will die by suicide and who will not—as if the right checklist, completed carefully enough, could offer certainty in one of the most uncertain areas of clinical work. It can’t.

Newer algorithms may help us see patterns our clipboards miss, and some predictive models are certainly better built than others. Researchers are right to call for more rigorous methods, better validation, calibration, transparent reporting, and real-world impact studies before dismissing prediction models altogether. But prediction is still not the same as care. The real clinical task is asking, How do we help this person find a path to a passion for living, deep connection, and a sense of worth?

Suicide risk is dynamic. It can change quickly, sometimes over hours or even minutes, shaped by humiliation, loss, intoxication, pain, insomnia, conflict, connection, shame, relief, or one unexpected moment of profound aloneness. A client may answer every question “correctly,” be categorized as low risk, and still be in life-and-death danger later that day. Another may endorse frightening thoughts and still have strong reasons for living, deep ambivalence, and a genuine wish to be helped.

This doesn’t mean assessment is useless. It means assessment must be humble.

Rather than using risk assessment primarily as a prediction exercise, we can use it as a conversation for risk formulation and treatment planning. The goal is to understand the story: What has brought suicide into the room? What pain has become unbearable? What has changed recently? What has helped before? What makes death feel like relief, and what keeps the person tethered to life?

That kind of storytelling inquiry does more than gather data. It builds trust, helps the client feel seen as a whole person rather than reduced to a set of risk factors, gives the clinician more clinically useful information, and it turns assessment into the beginning of collaboration rather than the end of curiosity.

When we over-rely on forms, scores, categories, or clinical shorthand, we risk two harms at once. First, we may create a false sense of security when the answers look reassuring. Second, we may damage the very relationship that allows clients to tell us the truth. A person in suicidal pain can quickly sense when we’re listening for liability rather than listening for understanding.

What begins as care can start to feel like interrogation—or what I’ve come to think of as clipboard counseling. The assessment can still happen, but in service of connection, shared understanding, and a plan that fits the person sitting in front of us.

The “No-Suicide” Contract

For decades, many clinicians were taught to ask suicidal clients to sign or verbally agree to a “no-suicide” or “no-harm” contract: a promise that they would not kill or harm themselves before the next session.

I understand why this practice took hold. In the terror of suicidal crisis, a contract can feel like something solid to grab onto. It can appear to offer reassurance, structure, accountability, and documentation. Many of us were taught, explicitly or implicitly, that it might also protect us legally if the worst happened. But the evidence has not been kind to this practice.

Reviews of the literature have found little to no empirical support that no-suicide contracts prevent suicide or self-harm, and researchers have raised significant ethical and clinical concerns about their use. Another review similarly concluded that no-harm contracts aren’t an effective method for preventing suicide and don’t protect clinicians from malpractice litigation after a client suicide. In fact, when used as a standalone intervention—or when they allow the clinician or system to feel falsely reassured—they may create more danger, not less.

The problem isn’t simply that the contract may fail; it’s what it may ask of the client. A person in suicidal crisis is often overwhelmed, ashamed, frightened, ambivalent, and desperate for relief. Asking that person to sign a promise not to die can unintentionally shift the emotional burden in the wrong direction. It can sound like: I need you to protect me from my fear of losing you.

That is not collaboration, or containment, and it’s not the same as safety.

For some clients, especially those who already feel trapped, powerless, or afraid of being hospitalized, a no-suicide contract may feel coercive. They might sign it because they mean it in the moment or because they don’t want to disappoint us. Either way, the signature can tell us far less than we think it does.

The word contract is especially misleading. A legal contract assumes capacity, freedom, informed agreement, and meaningful choice. Suicidal crisis often compromises all four. And the literature is clear that relying on a no-suicide contract does not shield the clinician from liability. Careful assessment, sound clinical judgment, appropriate consultation, documentation of reasoning, and a collaborative plan for safety matter far more than a promise extracted in a moment of despair.

So no, a no-suicide contract is not a therapeutic intervention. At best, it’s a weak substitute for a real safety plan. At worst, it’s a liability ritual dressed up as care.

The good news is that we have a much better alternative: collaborative safety planning. It’s the difference between asking someone to promise they will not die and partnering with them to build a practical, personalized plan for how they will survive. It’s the difference between handing someone a form and sitting beside them to ask: When the next wave comes, what will help you get through it? Who can we bring closer? What can we put between you and lethal means? What has helped you hold on before?

A contract asks for compliance. A safety plan builds capacity. And in suicidal crisis, capacity is what we are trying to restore.

Forced Hospitalization

Forced hospitalization often emerges from terror. When a clinician is convinced a client may die, the impulse to do something—to lock the person away somewhere “safe”—is profoundly human. It’s also profoundly risky.

Sometimes hospitalization is exactly what’s needed. In acute circumstances, especially when a person can’t collaborate around safety, is highly intoxicated or psychotic, has imminent intent, or has access to lethal means they cannot or will not step away from, hospitalization can be lifesaving. But the research tells us that safety and hospitalization are not synonymous.

The period after discharge from psychiatric hospitalization is one of the highest-risk windows we know of in suicide prevention. A meta-analysis of 100 studies, published in a 2017 issue of JAMA Psychiatry, found that suicide rates after discharge were approximately 100 times the global suicide rate during the first three months, and nearly 200 times the global rate among people admitted with suicidal thoughts or behaviors. Even years after discharge, suicide rates among formerly hospitalized psychiatric patients remained roughly 30 times higher than global rates, according to a study published in a 2020 issue of the journal Suicide and Life-Threatening Behavior. These numbers don’t prove that hospitalization causes suicide, but they do demolish the assumption that it prevents it.

The ethical picture is equally complex. Ethicists argue that the field has often underweighted the harms of involuntary hospitalization, including loss of autonomy, coercion, damage to trust, and the possibility that some suicidal patients may benefit more from interventions that maximize collaboration, dignity, and personal agency. A qualitative review of patients’ experiences of psychiatric hospitalization found that feelings of coercion were shaped not only by legal admission status, but also by whether patients felt heard, respected, involved in decisions, treated fairly, and safe in relationship with staff.

A person can be “voluntary” on paper and still feel coerced. A person can be kept alive by hospitalization and still leave feeling ashamed, frightened, dehumanized, or less likely to tell the truth next time.

One of my closest friends, a global suicide prevention advocate and two-time attempt survivor, was coerced into hospitalization. He remembers the sound of the door locking. He told me he felt “like a caged animal.” Then he asked me, “So, what’s more dignified—to be locked away in the back of a psych unit as a mental patient, or to kill yourself?”

Another scenario to consider is the highly educated young woman who called a crisis hotline during a panic attack. Police were called. She was handcuffed, transported to the ER, had her phone confiscated so she couldn’t contact family, and was left restrained to a chair for hours before being forcibly committed for five days. When she leaves the hospital, she won’t call a hotline again. She won’t reach out. She will suffer alone.

There’s a phrase I’ve started using to describe what I see happen too often in our field: hot potato-ing. A clinician gets scared and sends the client to the emergency department. The emergency department sends them to inpatient care. Inpatient stabilizes and discharges. Outpatient care is told to resume. Each provider may be acting in good faith, trying to reduce immediate danger and meet their duty of care. But the client often doesn’t experience good faith—they experience being passed like a problem to the next person down the line. Somewhere in that chain of handoffs, the actual human being in pain can get lost. And what they experience isn’t safety, but abandonment in installments.

I’m not arguing against hospitalization, just against reflexive hospitalization, coercion without humility, and crisis systems that prioritize transfer of liability over continuity of care.

If hospitalization happens—and sometimes it must—we have a moral obligation to make it less traumatic. We can explain why we’re concerned. We can acknowledge the loss of control. We can involve the client as much as possible in the decision. We can reduce shame. We can plan for discharge before discharge happens. We can stay connected rather than disappearing once the ambulance arrives or the intake paperwork begins.

The field has a blind spot here. We have medicalized the decision to hospitalize, treating it as a clinical necessity rather than what it often is: a legal action that removes a person’s freedom, restructures their identity as a “psychiatric patient,” and frequently traumatizes them in the name of safety.

Safety is not just a locked door. Safety is not just what we do to a person. Safety is what we do with them. Safety is relationship, continuity, and dignity, which are the actual mechanisms of survival.

Read Part I of this series: Rethinking Suicide Prevention: Confronting an Unspoken Hazard of Our Work

Read Part III of this series: What Actually Works in Suicide Prevention: Reclaiming the Essential Skills We’ve Forgotten