Suicide risk assessment documentation: what a defensible, clinically useful note contains

By Selam Koch, PMHNP-BC, founder of OneStep Scribe and practicing psychiatric nurse practitioner. Published July 2026.

Here is the professional truth, stated plainly: in a bad outcome review, the note is the assessment. A reviewer cannot see the careful questions you asked or the clinical judgment running in your head, only what you wrote. An excellent assessment that was never documented does not exist as far as anyone reading the chart can tell. The encouraging flip side: most clinicians assess suicide risk far better than they document it. The gap is translation, not clinical skill. This guide covers what a complete risk assessment note contains, how to document safety planning, and two fictional examples of documenting suicidal ideation, one routine and one elevated.

What a complete risk assessment documents

A defensible note answers the questions a future reader will ask, in roughly this order:

Ideation. Present or absent, and if present, passive ("wishing I would not wake up") versus active (thoughts of ending one's life), plus frequency and duration. "Denies SI" is a conclusion. "Denies passive and active suicidal ideation when asked directly" is documentation.
Plan. Specific or absent. If a plan exists, what it is, in the patient's words where possible.
Intent. Distinct from plan. A patient can describe a method with no intent to act, or express intent with no formed plan. Document each separately.
Means and access. What could the patient use, and can they reach it? Ask about firearms and document the answer, whatever it is. Medication stockpiles count too.
History. Prior attempts, non-suicidal self-injury, and psychiatric hospitalizations. A prior attempt is among the strongest known predictors; the absence is worth stating too.
Current stressors and warning signs. What is happening in this patient's life right now: losses, legal or financial events, relationship rupture, recent discharge, anniversary dates.
Risk factors, split into modifiable and static. Modifiable: untreated or undertreated symptoms, insomnia, active substance use, access to means. Static: prior attempts, family history of suicide, and demographic factors such as male sex and middle or older age. The split matters because the plan should target the modifiable column.
Protective factors. Named and real for this patient: the daughter she is raising, the faith community he attends weekly, stated future plans. "Family support" pasted into every chart protects no one, including you.
Clinical synthesis with a stated risk level. A sentence or two in your own words pulling the data into a judgment: low, moderate, high, or elevated relative to baseline, with the reasoning visible.
A plan that matches the stated level. This is the heart of defensibility. A note that says "moderate risk" and then "return in three months" contradicts itself on its own page. Whatever level you state, the plan must be one a reasonable clinician would pair with it.

The synthesis plus the matching plan is the pairing reviewers weigh most heavily. Data without judgment looks like a checklist. Judgment without a matching response looks worse.

Frameworks and tools

Structured tools are widely used and some settings require them. The Columbia protocol (C-SSRS) is widely used, and many hospital systems and accreditation programs have built it into workflows. The SAFE-T style approach offers a common organizing frame for the note itself: identify risk factors, identify protective factors, conduct the suicide inquiry, determine the response, and document. This article describes categories only and does not reproduce the items of any instrument; use licensed instruments as their publishers direct.

Tool choice varies by setting and payer. The documentation principles do not. Structured protocol, template, or free text, the finished note still needs ideation, plan, intent, means, history, risk and protective factors, a synthesis, and a matching plan.

Safety planning, not contracts

No-suicide contracts, in which the patient promises not to harm themselves, are widely discouraged in the literature. The evidence does not show they reduce suicide, and they are generally understood to offer no meaningful protection, clinical or legal. A signed promise is not an intervention.

The standard alternative is collaborative safety planning: working with the patient to write down their personal warning signs, internal coping strategies, people and places that provide distraction, contacts they can ask for help, steps to restrict access to lethal means, and crisis resources including the 988 Suicide and Crisis Lifeline. The plan belongs to the patient; your note documents that it exists and, just as importantly, how the patient engaged with it. "Safety plan reviewed, patient identified her sister and her morning walks as top coping steps, copy given" reads very differently from "safety plan in place."

If the patient declines to participate, document the refusal, your read of their capacity and current risk in light of it, and what you did instead.

Two documented examples

Both patients are fictional. The first is routine follow-up documentation; the second is elevated risk documented well.

Example 1: routine low-risk documentation at follow-up

Risk assessment: Asked directly about suicidal thoughts. Denies passive and active suicidal ideation, denies plan and intent. No history of suicide attempts, self-harm, or psychiatric hospitalization. No firearms in the home, asked again today, unchanged. Current stressors limited to routine work demands. Protective factors: engaged in treatment, close daily contact with her sister, cares for two school-age children, describes concrete plans for a family trip next month. No new warning signs since last visit.

Synthesis: Risk assessed as low, consistent with her baseline across the past year of treatment.

Plan: Continue standard monthly follow-up and current treatment. Patient will call the office or 988 if thoughts of self-harm emerge. Routine screening to continue at each visit.

Why this holds up: it takes about ninety seconds to write and converts "denies SI" into evidence that a real inquiry happened: passive and active ideation asked separately, firearms addressed, history stated, protective factors specific to this patient. It supports the clinician's defense because the low risk conclusion sits on visible data, and it serves the next clinician because the baseline is on the page; any future change is measurable against it.

Example 2: elevated-risk documentation done well

Risk assessment: Patient reports new passive suicidal ideation over the past two weeks, stating "some mornings I think it would be easier if I just didn't wake up." Thoughts occur a few times per week, last minutes, and are distressing to him. Denies active ideation, denies any plan, denies intent: "I would never do it, I have my kids." No prior attempts or self-harm. Asked directly about firearms: none in the home. Medications discussed; wife, still in the home during the separation, will hold the sertraline supply for now, patient agreeable. Stressors: separation initiated by spouse last month, sleeping 4 to 5 hours. Modifiable risk factors: worsening depressive symptoms, insomnia. Static factors: male, age 52, father died by suicide. Protective factors: two children at home, no substance use, engaged and forthcoming in session.

Synthesis: Risk elevated above his baseline given new passive ideation, acute stressor, insomnia, and family history, but without plan, intent, or access concerns. Assessed as moderate; outpatient management appropriate with tightened follow-up.

Plan: Safety plan updated collaboratively in session; patient named his warning signs and chose his brother as first contact. Means counseling completed as above. Follow-up shortened from 4 weeks to 1 week, telephone check-in at 72 hours. Crisis resources reviewed including 988; patient saved the number in his phone during the visit. Wife joined the last ten minutes with patient consent and is aware of the plan. Addressing insomnia at this visit as a modifiable target.

Why this holds up: the patient's own words anchor the finding to this session. Plan, intent, and means are addressed separately rather than compressed into one denial. The synthesis states a level and shows the reasoning, and each element of the plan maps to the assessment: shortened interval matches elevated risk, means counseling matches the medication discussion, collateral involvement matches the acute stressor. A reviewer can trace each decision, and the clinician who sees him next week knows exactly what to re-ask.

Five documentation failures

The "denies SI/HI" macro pasted regardless of what was asked. Identical language across every visit and every patient signals a template, not an inquiry. Cloned text is a chart-wide problem; see the progress note guide.
A positive PHQ-9 item 9 with no risk assessment the same visit. A score flagging thoughts of self-harm followed by silence is one of the most indefensible patterns in outpatient records; see the PHQ-9 and GAD-7 guide.
A stated risk level with a plan that does not match it. "Moderate risk, return in 12 weeks" impeaches itself.
No means or access documentation when ideation is present. Once ideation exists in the note, the absence of a firearms and access discussion becomes the loudest thing on the page.
Assessment documented once at intake and never revisited. Risk is dynamic. A thorough intake assessment followed by two years of notes that never mention risk again documents that nobody looked.

Frequently asked questions

How often should risk be assessed and documented?

A brief risk statement at every visit, and a full assessment when indicated: new or changed ideation, a major stressor, and care transitions such as hospital discharge or transfer between clinicians. Exact expectations vary by setting, but this cadence is widely expected practice in outpatient behavioral health.

Do I have to use the C-SSRS?

No single tool is universally required. Settings and payers vary; some mandate specific instruments, most do not. Use what your setting requires, but a completed screen is not a completed note. The principles in this article apply regardless of tool.

What if the patient refuses safety planning?

Document the refusal in the patient's words if possible, your assessment of their capacity and current risk given the refusal, and what you did instead, such as increased contact, collateral involvement with consent, or referral to a higher level of care. A documented refusal with a documented response is defensible. A silent gap is not.

Does telehealth change the assessment?

The assessment is the same. The logistics change: documenting the patient's location and an emergency contact at session start is widely recommended for telehealth, because if a crisis emerges mid-session you need to know where to send help.

Can an AI scribe document risk assessments?

OneStep Scribe drafts what was said in the visit, including the risk discussion that happened: the questions asked, the patient's words, the safety planning conversation. The assessment, the synthesis, and the risk level are always the clinician's, reviewed and signed by the clinician. A good draft means the inquiry you actually conducted makes it into the chart instead of collapsing into a macro at 9 pm.

Your risk assessments are better than your notes show. Close the gap.

OneStep Scribe is an AI scribe for mental health clinicians. It drafts the full visit note, including the risk discussion in the patient's own words, for your review and signature. Every account is NPI-verified.

Start a 14-day free trial

This article is documentation education for licensed clinicians, not crisis guidance, and it reflects one clinician's documentation approach. It is not clinical, legal, or billing advice for any specific patient, and it does not guarantee any outcome. Clinical decisions about suicide risk belong to the treating clinician. The patients described are fictional. Verify requirements with current guidelines, your payers, and your compliance advisors. If you or someone you know is in crisis in the US, call or text 988.