SAFE-T Suicide Risk Assessment: Five Steps and Triage

The SAFE-T (Suicide Assessment Five-step Evaluation and Triage) is a clinical framework that walks a clinician from risk and protective factors, through a direct suicide inquiry, to a risk level, an intervention and a documented rationale ¹. It is a structure for clinical judgement, not a scored scale.

The card was originally conceived by Douglas Jacobs, MD, developed by Screening for Mental Health, Inc. and the Suicide Prevention Resource Center with SAMHSA support, and drew on the American Psychiatric Association practice guideline for the assessment and treatment of suicidal behaviours ¹.

What does the SAFE-T measure?

The SAFE-T organizes a suicide risk formulation. It does not measure a single construct; it gathers the information a clinician needs to judge current risk and choose a response. Its five steps are ¹:

1. Identify risk factors, and note those that can be modified to reduce risk. The card groups them as past suicidal behaviour, current or past psychiatric conditions, key symptoms (such as hopelessness, anhedonia, impulsivity, insomnia and command hallucinations), family history, precipitants and stressors, changes in treatment, and access to firearms.
2. Identify protective factors, and note those that can be enhanced: internal (such as coping skills and frustration tolerance) and external (such as responsibility to children, a positive therapeutic relationship and social supports). The card cautions that protective factors, even when present, may not counteract significant acute risk.
3. Conduct a suicide inquiry: specific questions about suicidal thoughts (frequency, intensity and duration, in the last 48 hours, the past month and at their worst), plans, behaviours and intent, including the person's reasons for living and for dying. For youth, the card adds asking a parent or guardian about signs of suicidal thoughts, plans or behaviour.
4. Determine the risk level and intervention, based on clinical judgement after steps 1-3, and reassess as circumstances change.
5. Document the risk level and rationale, the treatment plan to reduce current risk, firearm instructions where relevant, and the follow-up plan. For youth, the plan includes roles for the parent or guardian.

How do you give the SAFE-T?

• Who. The card is written for mental health professionals, and SAFE-T resources are meant for clinicians trained to work with people at risk of suicide ¹.
• When. The card recommends an assessment at first contact, with any later suicidal behaviour, increased ideation or relevant clinical change, and on inpatient units before increasing privileges and at discharge ¹.
• How. It is a clinical interview guided by the five steps, drawing on the person's account, the clinical record and, where available, collateral information.
• With the C-SSRS. The Columbia Lighthouse Project publishes a SAFE-T protocol that uses C-SSRS questions in steps 1 and 3 ². Ideation severity (wish to be dead through intent with a plan) is asked for the past month, and suicidal behaviour for lifetime and the past 3 months. If the ideation questions are all no, the intensity questions are skipped; otherwise the clinician rates the intensity of the most severe ideation for frequency, duration, controllability, deterrents and reasons ². The C-SSRS itself is described on the C-SSRS page.

How do you score and interpret the SAFE-T?

The SAFE-T has no numeric cutoffs. The pocket card is unscored. The Columbia SAFE-T with C-SSRS totals the five ideation intensity ratings as a subscale ², but the risk level comes from its stratification rules, not from a cutoff on that total. In both versions the risk level is a clinical judgement, and each gives a table to guide it.

The SAMHSA pocket card

The card describes three levels and possible interventions ¹:

The card states that the chart represents a range of risk levels and interventions, not actual determinations ¹.

The SAFE-T with C-SSRS (Columbia)

The Columbia version anchors the levels to C-SSRS answers ²:

The two versions do not draw the moderate line in the same place: the pocket card places ideation with a plan (but no intent) at moderate, while the Columbia version places any intent or plan in the past month at high ^{1 2}. Use one version consistently within a service, and record which one was used.

The Columbia form quotes the APA practice guideline: estimating suicide risk is the quintessential clinical judgement, because no single risk factor or set of risk factors has been shown to predict suicide specifically ².

When risk is high

• Do not leave the person alone; keep the environment safe and restrict access to means.
• Call or text 988 (the Suicide & Crisis Lifeline in the US and Canada), or contact emergency services (911).
• Complete safety planning, involve the person's supports, and document the rationale and actions.

Older printings of the SAFE-T card and forms list a former 1-800 Lifeline number; in the US and Canada, 988 is the current number. See crisis resources.

Psychometric properties

The SAFE-T is a framework rather than a scored scale, so it has no reliability or validity statistics of its own in the way a questionnaire does. The evidence comes from its components and from studies of SAFE-T-based assessment in practice.

• Clinician SAFE-T assessments in a health system. In 89,957 people assessed by 2,577 clinicians at 12 hospitals, using a structured assessment based on the SAFE-T framework, clinicians' single overall risk rating predicted a suicide attempt within 90 days better than chance: area under the curve 0.77 after outpatient visits, 0.64 after inpatient stays and 0.60 after emergency department visits. A model using all 87 documented assessment items raised these to 0.87, 0.79 and 0.76 ⁴.
• Emergency department disposition. In 276 people with suicidal ideation in an emergency department, the SAFE-T, the modified SAD PERSONS scale and the C-SSRS each showed only modest negative predictive value (0.66-0.73) for disposition. The authors concluded that screening tools should not be used in isolation to guide disposition ⁵.
• C-SSRS component. The C-SSRS questions used in the Columbia version showed good convergent and divergent validity and sensitivity to change in their initial validation ³.

Limitations

• Judgement, not a score. The risk level depends on the clinician's training and the information available. Two clinicians can reach different levels from the same facts.
• Modest prediction. SAFE-T-based ratings predicted suicide attempts better than chance but imperfectly, and least well in the emergency department ⁴. No framework identifies who will attempt suicide with certainty.
• Version differences. The pocket card and the SAFE-T with C-SSRS define the moderate and high levels differently; results are only comparable within one version.
• Interventions are examples. The intervention column shows a range of options, not prescribed actions; local protocols and the person's circumstances decide the response ¹.
• Not for self-report. The SAFE-T is designed for clinicians; it is not a questionnaire for clients to complete alone.

How does the SAFE-T fit with other suicide risk measures?

• Screen first. A brief screen such as the ASQ or the C-SSRS screener identifies who needs a SAFE-T-style assessment. A positive response to PHQ-9 item 9 is another common trigger.
• Monitor ideation. The SIDAS tracks how severe suicidal thoughts have been over the past month between assessments.
• Avoid checklists alone. Risk-factor checklists such as SAD PERSONS have not predicted suicidal behaviour well in later studies; a structured inquiry like the SAFE-T asks about current thoughts, plans, behaviour and intent directly.

If you or someone you know is in crisis, call or text 988 or contact local emergency services. Help is available 24/7.