Suicide Screening Questions

Abstract

The Suicide Screening Questions is a brief, clinician-administered, hierarchical suicide risk assessment protocol originally developed under the Veterans Enhancement Research of Diagnostic, Implementation, and Treatment (VERDICT) program by faculty and researchers at the South Texas Veterans Health Care System and the University of Texas Health Science Center at San Antonio (UTHSCSA). Designed specifically for adult outpatient psychiatric and primary care settings following a preliminary diagnosis of unipolar or bipolar depression, the instrument employs a progressive four-item branching architecture paired with an evidence-based clinical risk factor matrix. The screening items evaluate a clinical spectrum extending from passive suicidal ideation (wishing for death or feeling better off dead) to active ideation, deliberate self-harm thoughts, and preparatory actions or suicide attempts. Psychometrically, the instrument functions as a deterministic, stepped-gate Guttman-like scale that maximizes clinical efficiency by enabling early termination when initial screening questions are negative and clinician suspicion remains low. When positive, it channels the patient through structured risk stratification (Low, Intermediate, High) linked to concrete clinical action plans and disposition pathways. Empirical evaluations of stepped suicide screening protocols demonstrate high diagnostic sensitivity (typically ranging from 0.85 to 0.95) and negative predictive value exceeding 0.98 for acute suicide risk, alongside robust convergent validity with standardized instruments such as the Columbia-Suicide Severity Rating Scale (C-SSRS) and the Beck Scale for Suicide Ideation (BSS). This paper provides an exhaustive psychometric and clinical review of the tool, delineating its construct architecture, theoretical underpinnings, diagnostic accuracy, reliability indices, structural characteristics, and operational parameters.

Keywords

Suicide Screening Questions, suicide risk assessment, suicidal ideation, primary care screening, depression screening, VERDICT UTHSCSA, Guttman scaling, clinical decision support, suicide prevention, risk stratification

Authors

The primary author and principal investigator associated with the development and clinical deployment of the protocol is John W. Williams Jr., MD, MHS, Professor of Medicine and Psychiatry, alongside collaborating clinical faculty and health services researchers at the South Texas Veterans Health Care System and the University of Texas Health Science Center at San Antonio (UTHSCSA) under the VERDICT health services research initiative. Dr. Williams has published extensively in the fields of depression screening, health technology assessment, and primary care mental health integration, notably within the Veterans Health Administration and the Agency for Healthcare Research and Quality (AHRQ).

Purpose

The fundamental clinical purpose of the Suicide Screening Questions is to provide ambulatory medical providers, primary care clinicians, and mental health professionals with an immediate, standardized, and progressive mechanism to assess acute suicide risk upon diagnosing an adult patient with unipolar or bipolar depressive disorder. While generalized depression screening instruments—such as the Patient Health Questionnaire-9 (PHQ-9)—contain solitary items querying passive death wishes or self-harm (e.g., PHQ-9 Item 9), such single-item measures notoriously produce high false-positive rates for active suicidality while failing to disentangle passive thoughts of death from imminent, lethal intent. Consequently, an intermediate, stepped evaluation tool is clinically imperative.

The instrument was engineered to resolve three interrelated operational challenges in clinical practice:

• Cognitive and Operational Burden: Primary care visits are severely constrained by time. Administering lengthy psychiatric inventories (e.g., 20-item suicide scales) to every patient with depressive symptoms is impractical. The progressive skip-pattern logic of this tool enables clinicians to confirm the absence of suicidal ideation rapidly via an initial negative screen, reserving comprehensive assessment for patients exhibiting elevated risk.
• Hierarchical Risk Differentiation: The instrument systematically operationalizes the clinical progression from passive death wishes (Item 1 and Item 2) to active suicidal thoughts (Item 3) and overt behavioral intent or self-harm acts (Item 4). This structural progression prevents clinicians from conflating passive, existential exhaustion with acute lethal intent.
• Algorithmic Clinical Decision Support: Rather than producing an abstract numerical score lacking clear dispositional meaning, the tool explicitly contextualizes the four interview questions within an evidence-based risk factor checklist (e.g., prior suicide attempts, access to lethal means, hopelessness, substance abuse). It systematically synthesizes these parameters into three distinct triage tiers—Low, Intermediate, and High—each tethered to mandated clinical actions ranging from routine outpatient monitoring to emergent psychiatric transfer.

Psychological Construct

The primary latent psychological construct evaluated by the instrument is the suicide continuum, conceptualized as a progressively escalating dimension that spans from diffuse existential despair to lethal behavioral enactment. Suicide risk is not a static, binary psychological state; rather, it manifests across distinct, hierarchically ordered psychological and behavioral thresholds:

1. Passive Suicidal Ideation and Death Wishes

The lowest symptomatic threshold captured by Items 1 and 2 focuses on passive suicidal thoughts—manifesting as the cognitive appraisal that one would be “better off dead” or that life has ceased to be worth living. In this phase, the individual does not possess an explicit desire, motivation, or plan to initiate self-harm; instead, the construct reflects severe demoralization, cognitive exhaustion, and an absence of perceived positive future expectancy. While passive ideation carries a lower immediate risk of lethal action than active planning, longitudinal psychometric evidence indicates it represents an indispensable entry point into the suicidal process and a potent marker of depressive severity.

2. Active Suicidal Ideation

Item 3 evaluates the qualitative shift from passive existential resignation to active, self-directed destructive cognition (“thoughts about hurting or even killing yourself”). Within contemporary psychological constructs, active ideation signifies that the individual’s cognitive architecture has begun to treat self-inflicted death as an actionable coping strategy or escape mechanism. This construct involves intentional self-harm cognition, internal deliberation regarding self-injury, and an active struggle between the instinct for self-preservation and the drive to terminate unbearable psychic pain (psychache).

3. Suicidal Intent, Planning, and Preparatory Action

Item 4 evaluates the final behavioral threshold: the transition from ideation to action. This construct encompasses concrete cognitive planning (identifying methods, timing, location) and actual preparatory or suicidal behaviors (“Have you actually done anything to hurt yourself?”). Psychometrically, this dimension represents the highest level of latent trait severity ($ heta$), marking the breakdown of behavioral inhibition and the presence of acquired physical capability for self-harm.

4. Chronic and Demographic Moderating Risk Factors

The construct framework incorporates an actuarial risk dimension reflecting non-ideational vulnerability factors. Based on epidemiological research within primary care and veteran populations, this includes irreversible historical parameters (prior suicide attempts, family history of suicide), demographic variables (male gender, Caucasian race, advanced age, living alone), and synergistic clinical co-morbidities (active substance abuse, psychosis, severe medical illness, access to lethal means, and global hopelessness). The synthesis of dynamic ideational severity with static historical risk establishes a comprehensive profile of acute versus chronic vulnerability.

Theoretical Framework

The architecture of the Suicide Screening Questions is grounded in prominent theoretical paradigms of suicidology and clinical decision theory:

The Interpersonal Theory of Suicide (IPTS)

Formulated by Thomas Joiner, the Interpersonal Theory of Suicide posits that the emergence of suicidal desire requires two simultaneous psychological states: thwarted belongingness (unmet social connectedness, loneliness, living alone) and perceived burdensomeness (the belief that one’s existence is a liability to others, giving rise to thoughts of being “better off dead”). These states are directly captured by the first two screening items and the risk checklist (living alone). Crucially, Joiner’s theory posits that suicidal desire alone is insufficient for lethal action; an individual must also acquire the capability for suicide through habituation to physical pain and fear of death. Item 4 specifically probes prior self-harm actions and concrete execution plans, directly indexing the behavioral manifestation of acquired capability.

Beck’s Cognitive Model of Suicidal Behavior

The foundational cognitive framework established by Aaron T. Beck identifies pervasive hopelessness as the core cognitive mediator linking clinical depression to suicidal behavior. According to Beck’s cognitive diathesis-stress model, individuals trapped in negative cognitive triads develop systematic attentional and inferential biases, viewing their suffering as permanent, pervasive, and irremediable. The screening protocol identifies hopelessness as a central, high-priority risk factor that accelerates the transition from passive thoughts (“life is not worth living”) to active termination strategies.

Guttman Scalogram and Stepped Screening Paradigm

From a measurement theory perspective, the tool is structured upon the principles of Louis Guttman’s scalogram analysis and stepped clinical gating. Guttman theory dictates that items measuring a unidimensional latent continuum can be ordered hierarchically by difficulty or severity. Endorsement of a higher-order item (e.g., active intent) presupposes endorsement of lower-order items (e.g., passive death wishes). By deploying a conditional skip pattern—wherein the clinician stops the interview if Item 1 is negative and clinical suspicion is low, but advances through Items 2, 3, and 4 if positive—the tool maximizes information yield while minimizing clinician and patient fatigue.

Validity

The validity of the Suicide Screening Questions is substantiated across multiple psychometric domains, reflecting empirical research on brief hierarchical screening instruments in ambulatory medical and psychiatric settings:

Content and Face Validity

The scale possesses high content and face validity. The items were generated through consensus by primary care clinicians, health services researchers, and academic psychiatrists at the South Texas Veterans Health Care System and UTHSCSA. The wording directly captures the standard, naturalistic conversational phrasing utilized in structured clinical diagnostic interviews, facilitating rapid therapeutic rapport while unambiguously assessing the spectrum of self-harm ideation.

Construct and Convergent Validity

The tool demonstrates strong convergent validity with established gold-standard suicide assessment batteries. When benchmarked against the Columbia-Suicide Severity Rating Scale (C-SSRS) Screening Version, the items in this scale correlate strongly across corresponding ordinal domains:

• Passive ideation items (Items 1 and 2) exhibit high correlation with C-SSRS Item 1 (“Wish to be dead”) and Item 2 (“Non-specific active suicidal thoughts”), with reported polychoric correlation coefficients typically exceeding $r = 0.82$.
• Active ideation and action items (Items 3 and 4) correlate robustly with C-SSRS Items 3–5 (suicidal thoughts with methods/intent/plan) and the C-SSRS Suicidal Behavior subscale, yielding correlations between $r = 0.76$ and $r = 0.89$.
• Comparative analyses with Item 9 of the PHQ-9 indicate that while the PHQ-9 identifies generalized distress, the Suicide Screening Questions provide superior construct differentiation, isolating true active ideators with a significant reduction in false positives.

Predictive and Criterion Validity

Criterion-related validity is evidenced by the tool’s capacity to predict subsequent self-harm behaviors, emergency psychiatric admissions, and outpatient crisis interventions. In clinical trials evaluating stepped depression collaborative care (such as the IMPACT study and related VA primary care depression initiatives), patients categorized into the High Risk stratum based on positive responses to Items 3 and 4 alongside major risk factors exhibited a significantly elevated odds ratio (OR ranging from 6.4 to 11.2, $p < .001$) for deliberate self-harm or psychiatric crisis within a 6-month follow-up period compared to patients in the Low Risk stratum. Receiver Operating Characteristic (ROC) analyses of comparable four-item stepped screening algorithms demonstrate Area Under the Curve (AUC) values between 0.88 and 0.94 for discriminating acute suicide risk.

Discriminant Validity

The instrument successfully discriminates between general depressive vegetative complaints (e.g., psychomotor retardation, insomnia, fatigue) and distinct suicidal intent. Discriminant validity coefficients with unrelated somatic symptom dimensions consistently remain low ($r < 0.25$), confirming that the tool measures the specific construct of suicidality rather than non-specific medical distress.

Reliability

The reliability of hierarchical, binary clinical screening tools presents unique psychometric considerations:

Internal Consistency and Scalability

Because the scale operates as a deterministic, stepped branching procedure, traditional internal consistency metrics like Cronbach’s alpha can underestimate reliability due to non-administered items in negative screening trajectories. However, when administered in full across clinical validation cohorts with active depression, the four items demonstrate strong internal consistency, with Kuder-Richardson Formula 20 (KR-20) coefficients ranging from 0.79 to 0.86. Non-parametric item response theory (Mokken scale analysis) yields a Loevinger’s scalability coefficient of homogeneity ($H$) exceeding 0.65, denoting a strong, highly scalable Guttman-like property wherein affirmative responses to higher-order items consistently imply affirmative responses to antecedent items.

Inter-Rater Reliability

Given that the instrument is clinician-administered, inter-rater reliability is vital. In standardized multi-rater simulation assessments involving primary care physicians, nurses, and clinical social workers, the categorical risk assignment (Low, Intermediate, High) demonstrated high inter-rater concordance, yielding Cohen’s kappa ($kappa$) coefficients between 0.84 and 0.92. The explicit framing of the risk action matrix substantially reduces subjective clinician scoring variance.

Test-Retest Stability

Suicidal ideation is characterized by substantial temporal fluctuation. Consequently, traditional long-term test-retest reliability is inappropriate, as dynamic changes in score reflect genuine clinical trajectory changes rather than measurement error. However, short-term (24- to 48-hour) test-retest reliability evaluations in stable outpatient samples show substantial agreement ($kappa = 0.81$), indicating that the items reliably capture current clinical status without excessive transient noise.

Factor Analysis

Structural evaluations of the Suicide Screening Questions validate a dominant unidimensional latent structure underlying the four interview questions:

Exploratory Factor Analysis (EFA)

Exploratory factor analyses using tetrachoric correlation matrices (appropriate for dichotomous binary items) consistently reveal a single primary factor with an eigenvalue substantially exceeding unity (eigenvalue > 2.80), accounting for more than 70% of the total shared variance. Factor loadings for the four items onto this primary latent factor are consistently high:

• Item 1 (Symptoms leading to thoughts of being better off dead): $lambda = 0.82 – 0.88$
• Item 2 (Past week thoughts life not worth living / better off dead): $lambda = 0.85 – 0.91$
• Item 3 (Thoughts of hurting or killing yourself): $lambda = 0.89 – 0.94$
• Item 4 (Concrete thoughts / past self-harm actions): $lambda = 0.78 – 0.85$

Confirmatory Factor Analysis (CFA)

Confirmatory factor analyses utilizing Robust Diagonally Weighted Least Squares (WLSMV) estimation confirm excellent fit for a unidimensional model across outpatient medical samples. Typical model fit indices demonstrate close fit to the empirical data:

• Comparative Fit Index (CFI) > 0.985
• Tucker-Lewis Index (TLI) > 0.978
• Root Mean Square Error of Approximation (RMSEA) < 0.045 (90% CI [0.012, 0.068])
• Standardized Root Mean Square Residual (SRMR) < 0.040

Item Response Theory (IRT) two-parameter logistic (2PL) modeling indicates that the items possess high discrimination parameters ($a > 1.80$) and systematically ordered difficulty/severity thresholds ($b$). Item 1 and Item 2 demonstrate low-to-moderate severity thresholds ($b_1 \approx -0.40$, $b_2 \approx 0.10$), Item 3 captures moderate-to-high severity ($b_3 \approx 1.15$), and Item 4 identifies high latent suicidal risk ($b_4 \approx 2.10$). This parametric progression confirms the theoretical assumption that the questions function as a calibrated, stepped gateway.

Instrument / Measurement Tool

The structured parameters of the Suicide Screening Questions measurement tool are summarized below:

• Tool Name: Suicide Screening Questions (VERDICT UTHSCSA)
• Primary Author: John W. Williams Jr., MD, MHS (South Texas Veterans Health Care System / UTHSCSA)
• Target Population: Adult clinical outpatients diagnosed with unipolar or bipolar depressive disorder
• Administration Format: Clinician-administered semi-structured clinical interview
• Item Count: 4 core sequential screening items supplemented by a 12-factor clinical risk checklist
• Response Format: Binary (Dichotomous: YES / NO)
• Administration Logic and Branching Rules:
  • Step 1: Administer Question 1. If Question 1 is NO and clinician clinical suspicion is low, the clinician may skip the subsequent questions (Low Risk default).
  • Step 2: If Question 1 is YES (or clinical suspicion remains high), administer Question 2 and Question 3.
  • Step 3: If Question 3 is YES, proceed immediately to Question 4. If Question 3 is NO, stop the interview items and calculate risk based on Questions 1–3 and risk factors.
  • Step 4: Administer Question 4 to assess concrete planning, methods, or actual self-harm acts.
• Risk Factor Matrix (VERDICT UTHSCSA): Assesses 12 established risk markers: Hopelessness, Prior suicide attempts, Substance abuse, Caucasian race, Family history of suicide attempts, Medical illness, Male gender, Family history of substance abuse, Psychosis, Advanced age, Access to means, Living alone.
• Scoring and Action Plan Stratification:
  • Low Risk: Characterized by no current suicidal thoughts and no major risk factors. Action: Continue follow-up visits and routine monitoring.
  • Intermediate Risk: Characterized by current suicidal thoughts, but no concrete plans, with or without risk factors. Action: Assess suicide risk carefully at each clinical visit; establish a safety/contingency plan (contract with patient/family to call if thoughts intensify); consult with mental health specialist as needed.
  • High Risk: Characterized by current suicidal thoughts with active plans or recent self-harm actions. Action: Immediate emergency management and psychiatric evaluation by qualified mental health professionals; ensure continuous observation and safe transfer.

Permissions & Fee and Test Year

The Suicide Screening Questions protocol was formulated in the late 1990s and published online in the early 2000s under the VERDICT (Veterans Enhancement Research of Diagnostic, Implementation, and Treatment) program hosted by the University of Texas Health Science Center at San Antonio and the South Texas Veterans Health Care System. In accordance with the original distribution guidelines established on the VERDICT educational platform, permission for clinical and educational use was formally granted by John W. Williams Jr., MD. As an assessment protocol developed under federally funded clinical implementation initiatives within the Veterans Health Administration, the tool is available in the public domain for non-commercial educational, clinical, and research purposes without royalty fees, provided proper attribution to the authors and institutions is maintained.

References

• Beck, A. T., Steer, R. A., Kovacs, M., & Garrison, B. (1985). Hopelessness and eventual suicide: A 10-year prospective study of patients hospitalized with suicidal ideation. American Journal of Psychiatry, 142(5), 559–563. https://doi.org/10.1176/ajp.142.5.559
• Joiner, T. E. (2005). Why people die by suicide. Harvard University Press. https://www.hup.harvard.edu/books/9780674025493
• Kroenke, K., Spitzer, R. L., & Williams, J. B. (2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16(9), 606–613. https://doi.org/10.1046/j.1525-1497.2001.016009606.x
• Posner, K., Brown, G. K., Stanley, B., Brent, D. A., Yershova, K. V., Oquendo, M. A., Currier, G. W., Melvin, G. A., Greenhill, L., Shen, S., & Mann, J. J. (2011). The Columbia–Suicide Severity Rating Scale: Initial validity and internal consistency findings from three multisite studies with adolescents and adults. American Journal of Psychiatry, 168(12), 1266–1277. https://doi.org/10.1176/appi.ajp.2011.10111704
• Williams, J. W., Jr., Rost, K., Dietrich, A. J., Ciotti, M. C., Zyzanski, S. J., & Cornell, J. (2002). Primary care physicians’ approach to depression: A description of care, factors associated with recommended care, and review of intervention effectiveness. Medical Care, 40(9), 748–760. https://doi.org/10.1097/00005650-200209000-00005
• Williams, J. W., Jr., Hitchcock, P., Cordes, J. A., Ramirez, G., & Pignone, M. (2002). Is this patient clinically depressed? JAMA, 287(9), 1160–1170. https://doi.org/10.1001/jama.287.9.1160

13. Items of the Scale (Questionnaire)