What to Do When Someone Tells You They're Suicidal: a Six-Step Plan That Cuts the Risk Nearly in Half.
By Courtney Turrin · September 28, 2026 · 12 min read read
Category: For Curious People
A six-step written safety plan can cut suicidal behavior nearly in half. Here’s what the evidence says about asking directly, staying present and planning ahead.
Three pieces in this series covered why people don't tell anyone they're suicidal. This one is for the person they finally do tell.
This is the fourth and final piece in a September series on suicide awareness, running through National Suicide Prevention Awareness Month. It discusses suicidal ideation directly. If you're struggling, dial 988 to reach the Suicide and Crisis Lifeline, free and confidential, 24 hours a day, by call or text.
Someone is going to tell you.
A child, a parent, a friend, a coworker who stays five minutes too long after a meeting. Getting a disclosure at all means something already went right, or wrong, to get there: shame that keeps people from naming the thought out loud, a healthcare system with gaps in how it handles a crisis, a growing number of people choosing a chat window over a person. Those same barriers are why the disclosure rarely comes on its own. Someone has to ask.
Most people get that first move wrong. They think even asking about suicidal thoughts is dangerous.
Asking About Suicide Directly Does Not Cause It
A 2014 review in Psychological Medicine and a 2018 meta-analysis of thirteen studies in Suicide and Life-Threatening Behavior, both covered in the first piece of this series, found no evidence that asking someone directly about suicide increases risk, and some evidence that it lowers distress. That finding contradicts what stops most responders before they start: the belief that saying the word out loud, or asking "are you thinking about killing yourself," might push someone over an edge they were only near before you spoke.
The evidence shows that this belief is unfounded, and the data is consistent across over a dozen studies. Silence doesn't protect anyone. It leaves the person carrying the thought alone.
Once that fear is gone, the next question is what a responder should do when someone affirms that they are experiencing suicidal thoughts.
A Good Responder Provides Scaffolding
Philosophers Giovanna Colombetti and Joel Krueger have spent the past decade developing an idea they call affective scaffolding: the claim, built on cognitive scientist Kim Sterelny's broader framework of the scaffolded mind, that human emotional regulation isn't purely internal. It runs through the material and social world around a person, constantly, the way a hiker's confidence on a steep trail runs through the grip of their boots rather than existing independently inside their mind. A person doesn't just feel their emotions. They regulate them by leaning on things outside themselves: routines, familiar spaces, other people's steady presence.
In a later paper applying this framework specifically to depression, Krueger and Colombetti found that severe depression doesn't remove a person's access to the practical use of their environment. Someone in a depressive episode can usually still have a conversation with their family member. What depression disrupts is access to the environment's regulative significance: the way ordinary things and ordinary people are supposed to steady a person's emotional state, and stop doing so. The family member is still there. They just feel more like a stranger because the person in a depressed state is disconnected emotionally from everything around them.
This changes how a responder should act, not what a responder is "for." The instinct is to try to fix the internal state directly, to find the right words that make the feeling go away. That isn't what the evidence supports, and it isn't what scaffolding does in any other context either. A crutch doesn't heal a broken leg. It changes what the leg has to do on its own. A responder's job, in this framing, is closer to rebuilding a piece of external structure the person's own mind has temporarily lost the ability to use: being a place the person's attention can rest, reliably, until their own regulative access comes back online.
The third piece in this series described the opposite failure mode. Philosophers Ryan Timms and David Spurrett coined the term hostile scaffolding to describe structures that use the exact same mechanisms as helpful scaffolding but are built to serve someone else's interest at the user's expense. The authors’ original example was a casino. Similarly, an AI chatbot engineered to maximize how long someone stays engaged, even while that person is disclosing suicidal thoughts, is hostile scaffolding cloaked in a supportive tone of voice. It keeps a hand on the person's back while quietly steering them toward the register. In contrast, a responder who shows up, stays present, and asks for nothing in return runs the same underlying mechanism toward the positive outcome: not extraction, but support that costs the person nothing to receive.
A Six-Step Plan, Built Before the Crisis, Cuts Suicidal Behavior Nearly in Half
The most widely used structured tool in this space is the Stanley-Brown Safety Planning Intervention (Figure 1), a six-step written plan built collaboratively with a person before or immediately after a crisis, meant to be used during the next one. It takes 20 to 45 minutes to build and covers:
Recognizing personal warning signs that a crisis may be building
Internal coping strategies the person can use alone
Specific people and places that offer distraction
Specific people to contact for direct help
Specific professionals and crisis services to contact if the crisis escalates
Reducing access to lethal means during a crisis
[datawidget key="fig1-safety-planning-steps"]
The evidence behind it is unusually solid for this field, and unlike some of the interventions that follow, the effect size is large. The original randomized trial of the intervention, combined with structured follow-up contact among 1,640 emergency department patients, found a 45% reduction in suicidal behavior, with one event prevented for every 44 patients who received it instead of usual care. A separate, later meta-analysis pooling six studies of similar safety-planning interventions, a broader and more varied set of patients than the original trial, found a 43% reduction, with one event prevented for every 16 patients. Neither study found a measurable effect on suicidal ideation itself, only on behavior. The plan doesn't appear to make the thoughts go away, but it can change what happens when the thoughts show up.
For a responder who isn't a clinician, the plan is still something you can help build. The value isn't in the document. It's in walking through the six questions with someone before they're in crisis, so the answers already exist the next time they are.
Hope Is a Skill, and Parents Can Measure and Build It Before a Crisis
Safety planning assumes a crisis is already imminent. Psychologist C.R. Snyder's hope theory operates further upstream, and it speaks most directly to parents rather than to responders in general.
Snyder's 1991 model, validated across college and patient samples in the Journal of Personality and Social Psychology, defines hope as a cognitive skill built from three parts: a goal, pathways thinking (the ability to generate a route toward it) and agency thinking (the belief that you have the energy to walk that route). Six years later, Snyder and colleagues adapted the framework into the Children's Hope Scale, a six-item measure validated for kids ages 8 to 16 that scores a child's pathways and agency thinking separately.
The construct has a genuine, though narrow, track record. Snyder's six-year longitudinal study of 213 entering college freshmen found that Hope Scale scores predicted cumulative GPA and graduation status years later, after controlling for ACT scores. The effect held, but it was modest: the correlation between hope and GPA in this literature runs around 0.21, independent of intelligence.
The suicide-specific evidence for hope theory comes from protective-factor research. Several studies across different populations, including Native American youth and Chinese college students, have found that higher hope, particularly the agency component, correlates with lower suicidal ideation. A 2024 study of nearly 5,000 Chinese college students found that hope agency was the only factor that separated students who moved from ideation to an actual plan from students who stayed at ideation alone. This is correlational evidence, gathered at a single point in time in most of these studies, and the field is still working out the mechanism behind it.
What hope theory offers parents is a way to build a protective factor into a child's daily life long before any disclosure happens, using a scale a pediatrician, school counselor, or therapist can administer directly. Snyder considered pathways and agency thinking learnable rather than fixed, present in rough form by age two and shaped continuously afterward by how caregivers respond when a child's first plan for something doesn't work. A parent who helps a child find a second route to a goal, rather than solving it for them or letting them give up, is practicing the behavior this research points to.
The Postcard Study Is Weaker Than Its Reputation
The intervention most people have actually heard of, if they've heard of any of this research, is a postcard.
In the 1970s, psychiatrist Jerome Motto identified 843 patients who had been hospitalized for suicidal crisis and who refused all offers of ongoing care after discharge. Motto and biostatistician Alan Bostrom randomly split them into two groups. One received nothing further. The other received a short, personal letter at least four times a year for five years, expressing concern and an open door, with no request for a response. The story built around this trial, that a human being remembering another human being exists and saying so on a schedule can save lives, is one of the most widely repeated findings in the suicide-prevention field. It also highlights an important caveat: a statistically significant result is not necessarily a large one.
Through the first two years of the trial, the letters group's suicide rate was under 2%, against the no-contact group's over 3%, a gap of roughly one percentage point. It cleared the bar for statistical significance, meaning it probably wasn't due to chance. It did not clear a much higher bar: whether the gap was big enough to matter, or durable enough to trust. A longer-term look at the same trial, described in later reviews in BMC Psychiatry and the British Journal of Psychiatry, found that by year fourteen there were 25 deaths among the 389 people who'd received letters (6.4%) and 26 among the 454 who hadn't (5.7%), a statistically indistinguishable difference.
A 2023 meta-analysis of newer, larger caring-contacts trials found a similar pattern: the messages reduced self-reported suicide attempts at one year, but showed no measurable effect on suicide mortality or hospitalization and emergency-department visits in that same window.
None of that means consistent, low-pressure contact from someone who means it is worthless. It means the evidence behind the postcard study specifically is thinner than its reputation.
Gratitude Journals and Vision Boards Have Never Been Tested Against Suicide at All
A lot of well-meaning advice fills the space around suicide prevention with practices borrowed from general wellness culture. Gratitude journaling, vision boards, gratitude apps, and the broader positive-psychology toolkit have research behind them for other outcomes, mood and general life satisfaction among them. None of it has been tested against a suicide-specific outcome. A responder should be cautious about recommending these types of interventions given the lack of data about their effectiveness in helping people experiencing severe depression.
Fewer Than 4 in 100 Die by Suicide Within Five Years of a Crisis
A 2014 meta-analysis in PLOS ONE, led by Robert Carroll, Chris Metcalfe, and David Gunnell, pooling 177 studies on what happens after a hospital-treated self-harm episode, found that 1.6% of people died by suicide within 12 months, and 3.9% within five years (Figure 2).
[datawidget key="fig2-five-year-mortality-stat"]
Each of those deaths was a person, not a statistic weighed against everyone who lived. But what the number does say is that in the period immediately following a disclosure or a crisis, the period a responder is most afraid of, most people go on living, many of them for the rest of their lives. That's a number for the responder to keep in mind to steady their own nervous system across a long, uncertain stretch of time, which is often what showing up for years requires.
Staying Present Is the Whole Job
Every piece in this series has aimed to close the same gap: the distance between when someone needed to be heard and when someone was heard. A parent who doesn't ask because the divorce made everything feel fragile already. A teenager who tells a chatbot instead of a friend because the chatbot can't tell anyone. A grieving mother who learns, months too late, that her daughter had been telling an AI everything for months, and no one else.
The fear underneath all of it is the same: saying the wrong thing, asking the wrong question, making it worse. The evidence shows that asking directly doesn't cause harm. And a plan built together, before the worst night rather than during it, cuts suicidal behavior nearly in half. Neither action requires a script.
A chatbot continues a conversation for exactly as long as its engagement design rewards it, then lets the thread go quiet right when a person needs it not to. A person doesn't come with that failure mode built in. Fewer than four in a hundred people who reach a crisis this severe will die by suicide within five years. Most are still here, still reachable, for a long time.
That's what showing up gets you: another chance to check in tomorrow, and the day after that.
If you or someone you know needs support, dial 988 to reach the Suicide and Crisis Lifeline by call or text, 24 hours a day. In the Richmond region, the Richmond Behavioral Health Authority also operates a crisis line at 804-819-4100.