Suicide Risk Assessment: Evidence-Based Screening, Risk Factors, Neurobiology, Safety Planning, and Means Restriction

Suicide is a leading cause of death worldwide, claiming approximately 700,000 lives annually according to the World Health Organization, with an estimated 12.2 million adults in the United States experiencing serious suicidal ideation in a given year (SAMHSA, 2022). In the U.S., suicide was the 11th leading cause of death overall and the 2nd leading cause among individuals aged 10–34 in 2022, with 49,449 deaths recorded by the CDC. The age-adjusted suicide rate was approximately 14.2 per 100,000 in 2022, reflecting a persistent public health crisis despite decades of prevention efforts.

Suicide risk assessment occupies a unique position in clinical practice: it is one of the most consequential evaluations a clinician performs, yet it remains one of the most difficult to standardize. No single risk factor, screening instrument, or clinical sign reliably predicts imminent suicidal behavior at the individual level. Meta-analytic evidence consistently demonstrates that our ability to predict suicide in any given individual is only modestly better than chance — a humbling reality that demands both rigorous evidence-based practice and clinical humility.

This article provides a detailed, research-informed review of suicide risk assessment, encompassing epidemiology, neurobiology, validated screening tools, risk and protective factors, the comparative effectiveness of interventions, safety planning, and means restriction. The goal is to equip clinicians, trainees, and informed readers with the depth of knowledge needed to approach this critical clinical task with competence, nuance, and compassion.

Understanding the epidemiology of suicidal behavior requires distinguishing among several related but distinct phenomena: suicidal ideation (thoughts about ending one's life), suicide planning (formulating a method and intent), suicide attempt (non-fatal self-directed behavior with intent to die), and suicide death (fatal self-directed injurious behavior with intent to die). These phenomena occur at very different rates and are influenced by overlapping but distinct risk factors.

United States Data

• Suicidal Ideation: Approximately 4.8% of U.S. adults (12.2 million) reported serious suicidal ideation in 2021 (SAMHSA National Survey on Drug Use and Health).
• Suicide Plans: About 3.5 million adults made a suicide plan in the same period.
• Suicide Attempts: An estimated 1.7 million adults attempted suicide, with attempts being substantially more common among younger adults (18–25) at a rate of approximately 3.8%.
• Suicide Deaths: 49,449 deaths in 2022 (CDC WISQARS). The rate is approximately 14.2 per 100,000.

Demographic Patterns

Sex: Males die by suicide at approximately 3.5–4 times the rate of females in the U.S. (approximately 23.0 vs. 5.9 per 100,000 in 2022), a disparity largely attributed to the use of more lethal means (particularly firearms, which account for over 50% of U.S. suicide deaths). However, females attempt suicide at approximately 1.5–2 times the rate of males — a phenomenon termed the gender paradox of suicidal behavior.

Age: The highest suicide rate in the U.S. is among middle-aged and older adults, with men aged 75+ having rates exceeding 40 per 100,000. However, suicide among youth and young adults has increased dramatically — the suicide rate among 10–24 year-olds increased approximately 56% between 2007 and 2017 before stabilizing somewhat.

Race/Ethnicity: American Indian/Alaska Native populations have the highest age-adjusted suicide rate (approximately 28.1 per 100,000), roughly double the national average. White males historically have the highest absolute number of suicide deaths. Recent years have shown alarming increases among Black youth, with the suicide rate among Black children aged 5–12 approximately doubling between 1993 and 2019.

Occupation: Certain professions carry elevated risk, including physicians (particularly female physicians, whose suicide rate is 2.3 times that of the general female population), veterinarians, farmers, and construction workers — often related to occupational stress, access to lethal means, and cultural barriers to help-seeking.

Global Patterns

Globally, approximately 77% of suicides occur in low- and middle-income countries (WHO, 2021). The highest national rates are found in Lesotho, Guyana, Eswatini, South Korea, and several Eastern European nations. Methods vary substantially by region: pesticide self-poisoning is the most common method in agrarian low-income countries, while firearms predominate in the United States and hanging is the most common method globally.

Suicidal behavior is not merely a consequence of psychiatric illness — it involves specific neurobiological vulnerabilities that are partially independent of the underlying disorder. Research over the past three decades has identified several convergent neurobiological systems implicated in suicide risk.

The Serotonergic System

The most extensively studied neurobiological finding in suicide research is serotonergic dysfunction. Landmark postmortem studies by J. John Mann and colleagues at Columbia University demonstrated reduced serotonin (5-HT) and its metabolite 5-hydroxyindoleacetic acid (5-HIAA) in the brainstem of suicide completers, regardless of psychiatric diagnosis. Reduced 5-HIAA in cerebrospinal fluid (CSF) has been prospectively linked to future suicidal behavior, with one seminal study by Åsberg and colleagues (1976) showing that patients with CSF 5-HIAA below the median were approximately 10 times more likely to die by suicide.

Postmortem studies consistently find alterations in serotonin receptor binding in suicide completers, including upregulation of 5-HT_2A receptors in the prefrontal cortex (particularly Brodmann areas 8 and 9) — interpreted as a compensatory response to reduced serotonergic transmission. The serotonin transporter (5-HTT) shows reduced binding in the ventromedial prefrontal cortex of suicide completers.

The Hypothalamic-Pituitary-Adrenal (HPA) Axis

Dysregulation of the stress response system is a robust finding in suicidal individuals. Elevated cortisol levels, non-suppression on the dexamethasone suppression test (DST), and elevated corticotropin-releasing hormone (CRH) in CSF have all been associated with suicidal behavior. A landmark prospective study by Coryell and Schlesser (2001) followed 78 patients with major depression and found that DST non-suppression predicted eventual suicide with an odds ratio of approximately 14.0 — one of the strongest biological predictors identified to date, though with limited sensitivity.

Prefrontal Cortex and Decision-Making Circuits

Neuroimaging studies consistently implicate dysfunction in the ventromedial and dorsolateral prefrontal cortex (vmPFC and dlPFC) in suicidal behavior. These regions are critical for impulse control, decision-making, and cognitive flexibility. Suicide attempters demonstrate impaired performance on neuropsychological tasks measuring executive function, including the Iowa Gambling Task and the Stroop task, suggesting that suicidal behavior partly reflects a deficit in the ability to generate and evaluate alternative solutions to distressing situations.

Functional neuroimaging studies show reduced activation of the dlPFC during decision-making tasks in suicide attempters and increased amygdala reactivity to negative emotional stimuli, suggesting a pattern of impaired top-down regulation of emotional responses — consistent with the diathesis-stress model proposed by Mann and colleagues.

The Inflammatory-Immune System

An emerging body of evidence links neuroinflammation to suicidal behavior. Meta-analytic data show elevated peripheral inflammatory markers — including interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-α), and C-reactive protein (CRP) — in individuals with suicidal ideation and suicide attempters. Postmortem studies of suicide completers reveal activated microglia and elevated inflammatory gene expression in the prefrontal cortex, particularly in the anterior cingulate cortex. This neuroinflammatory signature appears at least partially independent of depression severity, suggesting a specific role in suicidal vulnerability.

Genetic and Epigenetic Factors

Twin studies indicate that suicidal behavior has a heritable component, with estimated heritability of approximately 30–55% for suicidal behavior and approximately 43% for serious suicide attempts. This heritability is partially independent of the heritability of psychiatric disorders. Family studies show a 4- to 6-fold increased risk of suicidal behavior among first-degree relatives of suicide completers.

Candidate gene studies have implicated polymorphisms in genes regulating serotonergic function (e.g., the serotonin transporter gene SLC6A4, particularly the short allele of the 5-HTTLPR polymorphism), brain-derived neurotrophic factor (BDNF Val66Met polymorphism), and genes involved in HPA axis regulation. However, genome-wide association studies (GWAS) — including the landmark International Suicide Genetics Consortium analysis — have found that suicidal behavior, like other complex traits, is highly polygenic, with individual variants contributing very small effects.

Epigenetic modifications, particularly DNA methylation at the glucocorticoid receptor gene (NR3C1) promoter, have been linked to early-life adversity and subsequent suicidal behavior — providing a molecular mechanism by which childhood trauma may become biologically embedded and increase suicide risk decades later.

The Opioid System

More recent research has drawn attention to the endogenous opioid system. Reduced mu-opioid receptor binding has been observed in suicide attempters, and psychological pain (psychache) — a construct central to Edwin Shneidman's theory of suicide — may be mediated in part by opioid system dysfunction. The controversial use of buprenorphine as an anti-suicidal agent is grounded in this neurobiology, though clinical evidence remains preliminary.

A critical distinction must be drawn between screening (brief identification of individuals who may be at risk) and comprehensive risk assessment (in-depth clinical evaluation of risk level, acuity, and formulation). No instrument can replace clinical judgment, but validated tools significantly improve the consistency and documentation of risk assessment.

Screening Instruments

Columbia-Suicide Severity Rating Scale (C-SSRS): Developed by Kelly Posner and colleagues at Columbia University, the C-SSRS is the most widely used suicide screening instrument globally and has been adopted by the FDA, the U.S. military, and numerous health systems. It categorizes suicidal ideation on a 5-point ordinal scale (from passive wish for death to active ideation with specific plan and intent) and separately assesses suicidal behavior. The C-SSRS has demonstrated good sensitivity (approximately 67–100% across studies) and specificity (approximately 76–98%) depending on the population and threshold used. It is available in over 140 languages.

Patient Health Questionnaire-9, Item 9 (PHQ-9 Item 9): This single item ('Thoughts that you would be better off dead, or of hurting yourself in some way') is embedded in the most widely used depression screening tool. A positive response (any frequency) has a sensitivity of approximately 69–83% for suicidal ideation but low positive predictive value for suicide attempt or death, given the high base rate of transient death wishes in depressed populations. It is best used as a gateway to more detailed assessment rather than a standalone screen.

Ask Suicide-Screening Questions (ASQ): Developed at the National Institute of Mental Health, the ASQ is a brief 4-item screening tool validated for medical settings including emergency departments, inpatient medical/surgical units, and primary care. It has demonstrated sensitivity of approximately 97% and negative predictive value of 99.7% for identifying suicide risk in pediatric and adult emergency department samples.

Suicide Behaviors Questionnaire-Revised (SBQ-R): A 4-item self-report measure assessing lifetime ideation, past-year frequency, communication of intent, and estimated future likelihood of attempt. A score ≥7 in the general population and ≥8 in clinical populations is typically used as the clinical cutoff. Internal consistency is adequate (Cronbach's α ≈ 0.76–0.88).

Comprehensive Assessment Instruments

Beck Scale for Suicidal Ideation (BSS): A 21-item clinician-administered or self-report instrument developed by Aaron Beck that assesses the intensity of suicidal ideation, including specific wish to die, desire for an active or passive attempt, duration and frequency of ideation, deterrents, and preparatory behaviors. Prospective studies have shown that a BSS score ≥2 during the worst point of ideation is associated with an approximately 7-fold increase in eventual suicide death.

Suicide Intent Scale (SIS): Also developed by Beck, this instrument is used after a suicide attempt to assess the degree of intent, including preparation, precautions against discovery, and expectations of lethality. Higher scores predict eventual suicide completion.

The Prediction Problem

A landmark meta-analysis by Franklin et al. (2017), published in Psychological Bulletin, aggregated data from 365 studies spanning 50 years and found that the ability of any risk factor or combination of risk factors to predict suicidal thoughts and behaviors was only slightly better than chance (weighted OR ≈ 1.56 for ideation, 1.36 for attempt). Critically, prediction accuracy had not improved over 50 years. This finding does not mean risk assessment is futile — rather, it underscores that the purpose of risk assessment is to inform clinical decision-making and intervention planning, not to predict with certainty which individuals will die by suicide.

Machine learning approaches represent a frontier in risk stratification. Models using electronic health record data — such as those developed by Kessler, Bossarte, and colleagues using Army STARRS data — have achieved area-under-the-curve (AUC) values of approximately 0.84 for predicting suicide attempts in military populations, significantly outperforming traditional actuarial instruments. However, the positive predictive value remains low due to the low base rate of suicide, and these models raise important implementation and ethical questions.

Effective suicide risk assessment requires systematic consideration of risk factors (characteristics that increase probability), warning signs (observable behaviors signaling acute risk), and protective factors (characteristics that buffer against risk). The distinction between chronic/static risk factors and acute/dynamic risk factors is clinically essential — the former inform long-term risk stratification while the latter guide immediate clinical decisions.

Chronic/Static Risk Factors

• Prior suicide attempt: The single strongest predictor of future suicidal behavior. Approximately 25–30% of individuals who attempt suicide will make another attempt, and 3–10% will eventually die by suicide within 10 years (Owens et al., 2002, meta-analysis in British Journal of Psychiatry). The risk is highest in the first 3–12 months post-attempt.
• Psychiatric diagnosis: Over 90% of suicide completers had a diagnosable psychiatric disorder at the time of death in psychological autopsy studies. Major depressive disorder (OR ≈ 20), bipolar disorder (OR ≈ 15–30), schizophrenia (OR ≈ 8.5), borderline personality disorder (lifetime suicide rate ≈ 8–10%), anorexia nervosa (standardized mortality ratio ≈ 5.9 for suicide), and substance use disorders (OR ≈ 6) all carry substantially elevated risk.
• Family history of suicide: Confers approximately 2- to 4-fold increased risk, reflecting both genetic vulnerability and exposure to traumatic loss.
• Childhood adversity: Physical abuse, sexual abuse, emotional neglect, and other adverse childhood experiences (ACEs) show a dose-response relationship with adult suicidal behavior. Individuals with ≥4 ACEs have approximately 12 times the odds of attempting suicide compared to those with none (Felitti et al., 1998, ACE Study).
• Male sex: In most Western countries, males die by suicide at 3–4 times the rate of females.
• Access to lethal means: Firearm access roughly doubles suicide risk; living in a household with a firearm is associated with a 3- to 5-fold increase in suicide risk (meta-analytic data).

Acute/Dynamic Risk Factors and Warning Signs

• Recent psychiatric hospitalization/discharge: The weeks immediately following discharge from psychiatric hospitalization represent a period of extraordinarily elevated risk. A meta-analysis by Chung et al. (2017) found that the suicide rate in the first 3 months post-discharge was approximately 100 times the general population rate (approximately 1,000 per 100,000 person-years).
• Hopelessness: Aaron Beck's research demonstrated that hopelessness is a stronger predictor of suicide than depression severity alone. The Beck Hopelessness Scale score ≥9 prospectively predicted suicide with a sensitivity of approximately 80% in psychiatric outpatients.
• Agitation and insomnia: Psychomotor agitation and severe insomnia, particularly in the context of depression, are acute risk amplifiers. These symptoms may reflect noradrenergic hyperactivity and HPA axis activation.
• Acute intoxication: Alcohol is involved in approximately 25–50% of suicides. Acute intoxication increases impulsivity, narrows cognitive focus, diminishes pain perception, and impairs problem-solving.
• Social isolation and perceived burdensomeness: Central to Thomas Joiner's Interpersonal Theory of Suicide, thwarted belongingness and perceived burdensomeness are key proximal risk factors. When combined with acquired capability for self-harm (e.g., through prior attempts, self-harm, trauma, or occupational exposure to pain/death), these create conditions for lethal suicidal behavior.
• Recent loss or humiliation: Relationship breakup, financial collapse, legal problems, public disgrace, or loss of autonomy (particularly in older adults) are common precipitants.
• Giving away possessions, saying goodbye, sudden calm after distress: These behavioral warning signs may indicate that a decision to die has been made and preparatory actions are underway.

Protective Factors

• Social connectedness and reasons for living: Robust social support, family cohesion, and engagement in meaningful relationships are among the most consistently identified protective factors. Marvin Linehan's Reasons for Living Inventory operationalizes this construct.
• Children in the home: Having dependent children, particularly for women, is associated with reduced suicide risk.
• Religious or spiritual engagement: Regular religious attendance is associated with approximately 5-fold reduced risk of suicide (VanderWeele et al., 2016, JAMA Psychiatry).
• Effective psychiatric treatment: Lithium maintenance therapy (discussed below) and ongoing psychotherapy reduce risk.
• Restricted access to lethal means: Perhaps the most modifiable protective factor in acute risk situations.

Suicidal behavior occurs across virtually all psychiatric diagnoses, but certain comorbidity patterns carry multiplicatively elevated risk.

Depression and Substance Use Disorders

The co-occurrence of major depressive disorder and an alcohol or substance use disorder is one of the highest-risk comorbidity profiles. Each condition roughly doubles the suicide risk conferred by the other alone. In the National Comorbidity Survey Replication, individuals with both disorders had approximately 9 times the odds of a lifetime suicide attempt compared to those with neither diagnosis. Alcohol use disorder is present in approximately 25–50% of all suicide deaths, and acute intoxication at the time of death is present in approximately one-third of cases.

Borderline Personality Disorder

BPD deserves particular attention. The lifetime suicide rate is approximately 8–10% (though newer cohort studies suggest lower estimates of 3–6%), and recurrent suicidal behavior and non-suicidal self-injury (NSSI) are defining features of the disorder (DSM-5-TR Criterion 5). A critical clinical challenge is distinguishing between chronic suicidal ideation that is part of the BPD presentation, acute suicidal crises, and escalating risk. The presence of comorbid major depression, substance use, or psychotic symptoms in a person with BPD substantially increases the probability of a lethal attempt.

Schizophrenia Spectrum Disorders

Approximately 5% of individuals with schizophrenia die by suicide, with the highest risk occurring early in the illness course, particularly in young males with preserved insight, good premorbid functioning, and comorbid depression or substance use. Command auditory hallucinations directing self-harm occur in a minority of cases but are clinically significant and require careful assessment.

Bipolar Disorder

Bipolar disorder carries among the highest suicide risk of any psychiatric condition, with an estimated standardized mortality ratio (SMR) for suicide of 20–30. Risk is highest during mixed episodes and depressive episodes, not during pure mania. The transition from elevated or mixed states to depression is a particularly dangerous period.

PTSD and Trauma-Related Disorders

PTSD approximately doubles the risk of suicidal ideation and attempt, with the association partially mediated by comorbid depression and substance use but remaining significant after controlling for these factors. Among military veterans, PTSD with comorbid TBI (traumatic brain injury) is associated with further elevated risk.

Eating Disorders

Anorexia nervosa has the highest mortality rate of any psychiatric disorder, and a substantial proportion of these deaths are by suicide (approximately 20–30% of premature deaths in AN). The standardized mortality ratio for suicide in AN is approximately 5.9. Bulimia nervosa also carries elevated risk, particularly when comorbid with BPD traits, substance use, or impulsivity.

A critical but often insufficiently appreciated distinction must be made between treatments for the psychiatric disorders that underlie suicidal behavior and treatments that directly target suicidal behavior itself. Many effective antidepressants, for example, reduce depressive symptoms but have limited direct evidence for reducing suicide rates. The treatments with the strongest direct anti-suicidal evidence are surprisingly few.

Pharmacological Interventions

Lithium: Lithium is the medication with the most robust evidence for reducing suicide specifically. A landmark meta-analysis by Cipriani et al. (2005) and an updated Cochrane review by Cipriani et al. (2013) demonstrated that lithium reduces the risk of suicide and deliberate self-harm by approximately 60% in mood disorders (OR ≈ 0.36 for suicide; OR ≈ 0.36 for deliberate self-harm), corresponding to an NNT of approximately 20–50 depending on the baseline risk of the population. This anti-suicidal effect appears to be at least partially independent of its mood-stabilizing properties and may be mediated by serotonergic enhancement and impulse-reducing properties.

Clozapine: Clozapine is the only medication FDA-approved specifically for reducing suicidal behavior — in treatment-resistant schizophrenia. The InterSePT trial (Meltzer et al., 2003) demonstrated that clozapine reduced suicidal behavior by approximately 26% compared to olanzapine in schizophrenia/schizoaffective disorder patients at high risk for suicide (hazard ratio 0.76). NNT was approximately 12 over two years. The mechanism is thought to involve serotonergic modulation (5-HT_2A/2C antagonism), anti-impulsivity effects, and possibly the structure imposed by mandatory blood monitoring.

Ketamine and Esketamine: Intravenous ketamine (0.5 mg/kg over 40 minutes) produces rapid reductions in suicidal ideation, often within hours, in multiple randomized controlled trials. A meta-analysis by Witt et al. (2020) showed significant reduction in suicidal ideation scores within 24 hours (SMD ≈ −0.48). Intranasal esketamine (Spravato) received FDA approval for depressive symptoms in treatment-resistant depression and major depression with suicidal ideation or behavior. However, it is essential to note that neither ketamine nor esketamine has yet been shown to reduce suicide attempts or deaths — the evidence is limited to ideation reduction, and the durability of the anti-ideation effect is typically days to weeks.

Antidepressants: The relationship between antidepressants and suicide is complex. The FDA black box warning, implemented in 2004, was based on meta-analytic evidence of increased suicidal ideation and behavior (but not completed suicide) in youth aged <25 during the first weeks of SSRI treatment. However, epidemiological data consistently show that higher rates of antidepressant prescribing at the population level are associated with lower suicide rates, particularly in older adults. A large meta-analysis by Gibbons et al. (2012) supported the protective effect of antidepressants against suicidality in adults ≥25 years. The clinical implication is that antidepressants should be initiated with appropriate monitoring (particularly in younger patients) but should not be withheld due to suicide risk — untreated depression is far more dangerous than the small, time-limited increase in activation-related suicidality.

Psychotherapeutic Interventions

Cognitive Behavioral Therapy for Suicide Prevention (CBT-SP): A randomized controlled trial by Brown et al. (2005) demonstrated that a brief CBT protocol specifically targeting suicidal behavior (10 sessions) reduced the probability of a repeat suicide attempt by approximately 50% over 18 months compared to usual care (OR ≈ 0.51). This protocol directly targets suicidal cognitions, develops safety plans, and builds coping skills for suicidal crises.

Dialectical Behavior Therapy (DBT): Originally developed by Marsha Linehan for chronically suicidal individuals with BPD, DBT has the most extensive evidence base of any psychotherapy for reducing suicidal behavior. The original RCT by Linehan et al. (1991) and the replication by Linehan et al. (2006) demonstrated approximately 50% reduction in suicide attempts and self-harm over 12 months compared to community treatment by experts. A meta-analysis by DeCou et al. (2019) supported DBT's effectiveness for reducing self-harm (Hedges' g ≈ 0.32) and suicidal ideation (Hedges' g ≈ 0.23), with effects being most robust for BPD populations.

Collaborative Assessment and Management of Suicidality (CAMS): Developed by David Jobes, CAMS is a therapeutic framework that treats suicidality as the primary clinical focus rather than viewing it solely as a symptom of an underlying disorder. Multiple RCTs have demonstrated that CAMS reduces suicidal ideation and overall distress more rapidly than treatment as usual, with resolution of suicidal ideation occurring in approximately 8 sessions on average.

Safety Planning Intervention (SPI): Developed by Barbara Stanley and Gregory Brown, the Safety Planning Intervention is a brief, structured intervention that creates a personalized, hierarchical list of coping strategies and support resources for suicidal crises. A landmark study by Stanley et al. (2018), published in JAMA Psychiatry, demonstrated that SPI combined with structured follow-up phone calls reduced suicidal behavior by approximately 45% over 6 months compared to treatment as usual in Veterans Affairs emergency department patients (OR ≈ 0.44, NNT ≈ 16). The SPI is now considered a standard-of-care intervention for any individual presenting with acute suicidal ideation or behavior.

Brief Contact Interventions and Follow-Up

A surprising body of evidence supports the effectiveness of very low-intensity interventions. The landmark study by Motto and Bostrom (2001) randomized patients who refused treatment after a suicide-related hospitalization to receive brief caring letters versus no contact over 5 years and found a significant reduction in suicide deaths in the letter group. Subsequent studies of follow-up caring contacts (phone calls, texts, postcards) have shown mixed but generally favorable results, with a Cochrane review suggesting a modest reduction in repeat self-harm (RR ≈ 0.83).

The Safety Planning Intervention (SPI), as described by Stanley and Brown (2012), involves six specific steps, constructed collaboratively with the patient during or immediately after a suicidal crisis. The clinician's role is to help the patient generate personalized, specific, and actionable items for each step.

The Six Steps of Safety Planning

1. Warning signs: Identifying the internal states (thoughts, images, moods, situations, behaviors) that signal an approaching suicidal crisis. These serve as triggers for activating the safety plan.
2. Internal coping strategies: Activities the individual can use independently to manage the crisis without contacting another person — such as distraction techniques, relaxation exercises, physical activity, or mindfulness practices.
3. Social contacts and social settings that provide distraction: People and places that can help take the individual's mind off the crisis, without necessarily disclosing suicidal thoughts.
4. People to contact for help: Specific family members, friends, or other trusted individuals who can be called during a crisis. This step includes names and phone numbers.
5. Professionals and agencies to contact: Therapists, crisis lines (988 Suicide and Crisis Lifeline), emergency services, crisis text lines (text HOME to 741741), and relevant local crisis services.
6. Making the environment safe (means restriction): Steps to reduce access to lethal means, discussed in detail in the next section.

The safety plan is not a no-suicide contract (also called a contract for safety), which is an older practice that has no evidence of effectiveness and may create a false sense of security. A safety plan, by contrast, is a cognitive-behavioral tool that provides concrete steps to interrupt the progression from ideation to action.

Implementation research shows that the SPI is most effective when (1) it is created collaboratively during a clinical encounter, (2) it is reviewed and revised over time as the patient's situation changes, (3) it is easily accessible (e.g., on a wallet card, phone, or refrigerator), and (4) it is accompanied by structured follow-up contact. The SPI should be treated as a living document, not a completed form.

Means restriction — reducing access to the methods used for suicide — is the single most consistently effective population-level suicide prevention strategy, supported by natural experiments, quasi-experimental studies, and international comparisons. The rationale rests on two critical empirical findings.

The Suicidal Crisis Is Often Brief and Impulsive

A study by Deisenhammer et al. (2009) found that among survivors of near-lethal suicide attempts, approximately 48% reported that the time between the decision to attempt and the actual attempt was less than 10 minutes. Simon et al. (2001) found that approximately 24% of attempt survivors reported less than 5 minutes of deliberation. This means that for a substantial proportion of suicide attempts, removing access to immediately available lethal means can interrupt the process during a brief, high-intensity but time-limited crisis.

Means Substitution Does Not Fully Compensate

A common objection to means restriction is that individuals will simply switch to another method. However, the evidence strongly contradicts this. The classic natural experiment is the detoxification of domestic gas in the United Kingdom in the 1960s–1970s: when coal gas (which contained lethal concentrations of carbon monoxide) was replaced by natural gas, suicide by gassing dropped to near zero, and the overall national suicide rate dropped by approximately one-third — with no proportional increase in other methods. Similarly, studies of bridge barriers (e.g., at the Bloor Street Viaduct in Toronto and the Bern Muenster Terrace) show that installation of barriers virtually eliminates suicide at those sites, with limited substitution to nearby sites.

Firearm Means Restriction

Firearms are involved in approximately 55% of U.S. suicide deaths and have a case fatality rate of approximately 85–90%, compared to approximately 1–2% for drug overdose. This enormous lethality differential means that firearm access is a particularly potent risk amplifier. Ecological studies consistently show that states with higher rates of firearm ownership have higher suicide rates, driven entirely by firearm suicides, with non-firearm suicide rates being roughly equal across states — strong evidence against the means substitution argument.

Clinically, counseling about lethal means restriction — particularly secure firearm storage (e.g., using gun safes, trigger locks, temporary off-site storage, or voluntary transfer to a trusted person) — is a critical and evidence-based component of suicide risk management. Research by Barber and Miller (2014) and the Means Matter campaign at the Harvard T.H. Chan School of Public Health has provided frameworks for clinicians to conduct these conversations in a non-judgmental, collaborative manner. The goal is temporary separation of the at-risk individual from the most lethal means during the acute crisis period.

Medication Means Restriction

For individuals at risk of overdose, clinicians should limit the quantity of prescribed medications (particularly tricyclic antidepressants, opioids, and benzodiazepines, which are far more lethal in overdose than SSRIs), prescribe medications with lower lethality profiles when possible, and counsel patients or family members about securely storing medications. The United Kingdom's decision in 1998 to limit paracetamol (acetaminophen) pack sizes was followed by a significant reduction in paracetamol-related suicide deaths — approximately 43% fewer deaths in the 11 years following the legislation (Hawton et al., 2013).

Identifying prognostic factors is essential for clinical decision-making, resource allocation, and treatment planning. Outcomes in suicidality research include resolution of suicidal ideation, time to reattempt, and suicide death.

Factors Associated with Poorer Prognosis

• Multiple prior attempts: Each successive attempt increases the risk of eventual death. The method-specific case fatality rate also tends to increase across attempts.
• High lethality of prior attempt(s): Higher medical lethality of the index attempt (as measured by instruments such as the Lethality Rating Scale) predicts subsequent suicide death independent of stated intent.
• Comorbid personality disorder: Particularly BPD and antisocial personality disorder, which are associated with chronic instability, impulsivity, and treatment nonadherence.
• Chronic substance use: Active alcohol or drug use dramatically complicates treatment and increases impulsivity during crises.
• Social isolation and absence of reasons for living: Individuals with weak social networks, no dependents, unemployment, and no identified reasons for living have poorer outcomes.
• Treatment nonadherence and dropout: Discontinuation of psychiatric medication and therapy is a major risk factor for relapse into suicidal crisis, particularly in the first 1–3 months after treatment initiation or after hospital discharge.
• Male sex and older age: Older white males represent the highest-risk demographic for suicide death in the U.S., and their attempts are disproportionately lethal due to greater planning, greater use of firearms, lower probability of rescue, and lower rates of help-seeking.

Factors Associated with Better Prognosis

• Engagement in evidence-based treatment: Individuals who complete a full course of DBT, CBT-SP, or CAMS have substantially lower rates of reattempt.
• Strong therapeutic alliance: Across modalities, the quality of the therapeutic relationship is a key predictor of treatment retention and outcome in suicidal patients.
• Effective means restriction: Successful separation from lethal means during acute crises buys time for the crisis to pass and treatment to take effect.
• Social connectedness and instrumental support: Presence of a caring, stable family member or partner who is actively involved in safety planning is a strong positive prognostic factor.
• Identifiable precipitant: When suicidal behavior is clearly linked to a specific, potentially resolvable stressor (e.g., a relationship crisis) rather than chronic and pervasive hopelessness, the prognosis for resolution of the acute crisis is generally better.

Despite significant advances, the field of suicidology faces several persistent methodological challenges and knowledge gaps that constrain clinical practice.

The Prediction Problem Remains Unsolved

As demonstrated by the Franklin et al. (2017) meta-analysis, prediction of suicidal behavior at the individual level remains extremely difficult. The fundamental challenge is that suicide is a statistically rare event (even in high-risk populations), which means that even moderately accurate predictive models generate an unacceptably high number of false positives. Efforts to improve prediction through machine learning, natural language processing of clinical notes, and passive sensing of digital biomarkers (e.g., smartphone usage patterns, social media language) represent active research frontiers but remain far from clinical deployment.

Real-Time Risk Monitoring

Ecological momentary assessment (EMA) — using smartphones to collect real-time data on mood, ideation, and contextual factors — offers the potential to detect short-term fluctuations in suicidal ideation that are missed by traditional clinical assessments. Studies by Kleiman et al. (2017) using intensive longitudinal designs have revealed that suicidal ideation fluctuates dramatically over hours and that the between-person risk factors identified in epidemiological studies may not accurately characterize within-person risk dynamics.

Rapid-Acting Anti-Suicidal Treatments

Beyond ketamine and esketamine, ongoing research is evaluating psilocybin-assisted therapy, intravenous buprenorphine, and high-dose intravenous lithium as potential rapid-acting interventions for acute suicidal crises. The challenge is that most psychiatric medications require weeks to achieve therapeutic effect — a dangerous lag time for actively suicidal individuals.

Biomarker Development

Researchers are pursuing blood-based biomarker panels for suicide risk, incorporating markers of inflammation (IL-6, CRP, TNF-α), HPA axis dysfunction (cortisol, ACTH), serotonergic function (platelet serotonin levels, tryptophan metabolism), and epigenetic markers. Alexander Niculescu and colleagues have developed a polygenic panel combining gene expression and clinical data that showed promising predictive accuracy (AUC ≈ 0.78–0.90) for suicidal ideation and hospitalization in validation cohorts. However, no biomarker panel has yet achieved sufficient sensitivity and specificity for clinical use.

Zero Suicide as an Organizational Framework

The Zero Suicide framework, developed by the Suicide Prevention Resource Center and endorsed by the National Action Alliance for Suicide Prevention, is a system-wide approach to suicide prevention within health care organizations. Based on the premise that suicide deaths among individuals in active contact with health and behavioral health systems are preventable, Zero Suicide emphasizes universal screening, risk assessment and stratification, evidence-based treatment, transition management, and continuous quality improvement. Early implementation data from the Henry Ford Health System showed a dramatic reduction in patient suicide deaths after implementing a perfect depression care model — from an average of approximately 89 per 100,000 to near zero among enrolled patients. However, these results have been difficult to replicate uniformly, and questions remain about generalizability and sustainability.

Limitations of the Evidence Base

Key limitations include: (1) the exclusion of actively suicidal individuals from most RCTs for ethical and safety reasons, severely limiting the evidence base for the very patients we most need to treat; (2) the predominance of research conducted in high-income Western countries, limiting generalizability to global populations; (3) insufficient research on suicidal behavior in underrepresented groups, including LGBTQ+ youth (who have 3–4 times higher rates of suicide attempts), transgender individuals, and racial/ethnic minority populations; and (4) the use of suicidal ideation rather than attempts or deaths as the primary outcome in most intervention trials, which may not capture the most clinically significant effects.

Effective suicide risk assessment and management requires integrating epidemiological knowledge, neurobiological understanding, validated assessment tools, evidence-based interventions, and strong clinical skills. The following principles synthesize the evidence reviewed above.

• Use validated screening tools systematically — the C-SSRS, ASQ, and PHQ-9 Item 9 all have adequate psychometric properties for initial screening. Positive screens must trigger comprehensive clinical assessment.
• Assess both chronic and acute risk factors — a structured approach that considers prior attempt history, psychiatric diagnosis, substance use, and social factors alongside acute indicators (hopelessness, agitation, insomnia, recent loss, access to means) provides the most clinically useful risk formulation.
• Create and update safety plans collaboratively — the Stanley-Brown Safety Planning Intervention is an evidence-based, brief, and scalable intervention that should be standard practice for any individual identified as at risk.
• Address lethal means directly — means counseling, particularly regarding firearms and medication storage, is one of the most actionable components of risk management and has strong population-level evidence supporting its effectiveness.
• Prioritize evidence-based treatments that directly target suicidal behavior — including DBT, CBT-SP, CAMS, and pharmacotherapy with lithium (for mood disorders) or clozapine (for schizophrenia-spectrum disorders). Treat suicidal behavior as a clinical target in its own right, not merely as a symptom of a psychiatric disorder.
• Manage transitions of care — the period immediately following psychiatric hospitalization discharge, emergency department visits, and treatment changes is the highest-risk period. Structured follow-up contacts within 24–72 hours of discharge are a critical and evidence-supported practice.
• Document thoroughly — risk assessment documentation should reflect the clinician's reasoning, including risk factors considered, protective factors identified, the patient's current presentation, the interventions implemented, and the rationale for the level of care determined. Good documentation protects patients through continuity of care and protects clinicians through clear reasoning.
• Maintain clinical humility — no clinician can predict suicide with certainty. The goal is not prediction but risk reduction through systematic, evidence-informed intervention and compassionate clinical engagement.