Psychometric Properties of Four Commonly Used Suicide Risk Assessment Measures: Applicability to Military Treatment Settings

Adaptation and validation of a suicide-focused Word Sentence Association Paradigm to assess suicide-specific interpretation biases

The cognitive model of suicide proposes that biased cognitive processes contribute to suicidal thoughts and behaviors, and initial evidence suggests that attentional bias to suicide-congruent cues may predict suicidal behavior beyond established clinical risk factors. To date, much less work has explored the potential role of interpretation biases in the development and prediction of suicide risk. The present study assessed the psychometric properties, including reliability, concurrent validity, and predictive validity, of a novel adaptation of the Word Sentence Association Paradigm (WSAP) to assess suicide-suicide interpretation biases. In a sample of 239 psychiatric inpatients, including patients with no recent suicidal ideation or lifetime attempts (n = 35), those with recent suicide attempts (n = 29), and those with recent suicidal ideation (n = 174), participants completed the modified WSAP as well as self-reported suicidal ideation and attempts for the subsequent 6 months. Although the WSAP demonstrated good reliability, evidence of concurrent and prospective validity (in terms of self-reported suicidal ideation and attempts) was limited. Specifically, the clinical groups did not differ from one another on any measure of interpretation bias, nor did suicide-specific interpretation endorsements predict concurrent or future suicidal ideation when controlling for dysphoric interpretation bias. However, suicide-specific interpretation biases were uniquely associated with a history of a lifetime suicide attempt. Future work should further clarify the extent and specificity of the relationship between suicide-specific interpretation biases and clinical outcomes.

Military experiences, connection to military identity, and time since military discharge as predictors of United States veteran suicide risk

Veterans navigating the military-to-civilian transition appear at elevated risk for suicide. However, research on the transition-suicide association often fails to consider co-occurring risk factors. The independent association of time since military discharge and suicide among veterans therefore remains unclear. Data from 1,495 post-Vietnam community veterans provided estimates of suicide risk, military-based stressful experiences, connection to military identity, and recency of military discharge. Hierarchical regression analyses examined independent, incremental utility of factors associated with suicide risk after controlling for quality of life, age, and duration of military service among the total veteran sample and a subsample discharged from military service within five years prior. The resulting model explained 41% of variance in suicide risk in the total veteran sample and 51% of variance in suicide risk in the recently discharged subsample. Recency of discharge, combat exposure, moral injury, poor quality of life, and poor psychological wellness showed statistically significant, independent associations with suicide risk, whereas connection to military identity did not show significant, independent associations. Results highlight the salience of the military-to-civilian transition as an independent risk factor for veteran suicide even after controlling for military-based stressful experiences, military identity, quality of life, age, and service duration.

Machine learning approach for the development of a crucial tool in suicide prevention: The Suicide Crisis Inventory-2 (SCI-2) Short Form

The Suicide Crisis Syndrome (SCS) describes a suicidal mental state marked by entrapment, affective disturbance, loss of cognitive control, hyperarousal, and social withdrawal that has predictive capacity for near-term suicidal behavior. The Suicide Crisis Inventory-2 (SCI-2), a reliable clinical tool that assesses SCS, lacks a short form for use in clinical settings which we sought to address with statistical analysis. To address this need, a community sample of 10,357 participants responded to an anonymous survey after which predictive performance for suicidal ideation (SI) and SI with preparatory behavior (SI-P) was measured using logistic regression, random forest, and gradient boosting algorithms. Four-fold cross-validation was used to split the dataset in 1,000 iterations. We compared rankings to the SCI-Short Form to inform the short form of the SCI-2. Logistic regression performed best in every analysis. The SI results were used to build the SCI-2-Short Form (SCI-2-SF) utilizing the two top ranking items from each SCS criterion. SHAP analysis of the SCI-2 resulted in meaningful rankings of its items. The SCI-2-SF, derived from these rankings, will be tested for predictive validity and utility in future studies.

On the Selection of Item Scores or Composite Scores for Clinical Prediction

Recent shifts to prioritize prediction, rather than explanation, in psychological science have increased applications of predictive modeling methods. However, composite predictors, such as sum scores, are still commonly used in practice. The motivations behind composite test scores are largely intertwined with reducing the influence of measurement error in answering explanatory questions. But this may be detrimental for predictive aims. The present paper examines the impact of utilizing composite or item-level predictors in linear regression. A mathematical examination of the bias-variance decomposition of prediction error in the presence of measurement error is provided. It is shown that prediction bias, which may be exacerbated by composite scoring, drives prediction error for linear regression. This may be particularly salient when composite scores are comprised of heterogeneous items such as in clinical scales where items correspond to symptoms. With sufficiently large training samples, the increased prediction variance associated with item scores becomes negligible even when composite scores are sufficient. Practical implications of predictor scoring are examined in an empirical example predicting suicidal ideation from various depression scales. Results show that item scores can markedly improve prediction particularly for symptom-based scales. Cross-validation methods can be used to empirically justify predictor scoring decisions.

Male sex and hazardous alcohol use following military sexual assault increase suicide risk among US service members and veterans

Background: Higher alcohol use and military sexual assault (MSA) are associated with increased risk of death by suicide. Risk for death by suicide is rapidly increasing among females, who report higher rates of MSA, yet actual death by suicide and alcohol use are higher among males. It is not well understood whether higher alcohol use confers greater suicide risk in male or female service members and veterans who have experienced MSA.Objective: To determine whether the association between alcohol misuse and suicide risk was moderated by biological sex in a sample of male and female service members (N = 400, 50% female) who reported MSA.Method: Participants completed surveys of alcohol use and suicide risk as well as a demographic inventory. Linear regression with an interaction term was used to determine if suicide risk differed by sex and alcohol use severity after accounting for discharge status, sexual orientation, and age.Results: Average scores on the suicide risk measure were consistent with an inpatient psychiatric sample and scores on the AUDIT-C were indicative of a probable positive screen for alcohol misuse. Suicide risk was most pronounced among males who reported higher levels of hazardous alcohol use. A sensitivity analysis examining suicide risk by sex and screening results for alcohol misuse (positive/negative) showed that men with a probable positive screen had higher suicide risk.Discussion: The current study provides novel findings on suicide risk among survivors of military sexual violence by including both male and female survivors. Interventions to decrease suicide risk following MSA may consider alcohol reduction strategies, and optimizing these interventions in males. Engaging military culture at both the US Departments of Defense and Veterans Affairs to encourage more healthy alcohol consumption may mitigate this public health concern. Future research may consider how country of origin relates to these associations.

Uncontrollability of suicidal ideation adds incremental explanatory power in prediction of later suicidal ideation

OBJECTIVE

Suicidal ideation and suicidal behaviors are major public health concerns in the United States and are difficult to treat and predict. Risk factors that are incrementally informative are needed to improve prediction and inform prevention of suicidal thoughts and behaviors. Uncontrollability of suicidal ideation, one parameter of suicidal ideation, is one such candidate.

METHOD

In the current study, we assessed the predictive power of uncontrollability of suicidal ideation, over and above overall suicidal ideation, for future suicidal ideation in a large sample of active-duty service members. A total of 1044 suicidal military service members completed baseline assessments, of whom 664 (63.6%) completed 3-month follow-up assessments.

RESULTS

While baseline overall suicidal ideation itself was the strongest predictor of future suicidal ideation, uncontrollability of suicidal ideation added some incremental explanatory power.

CONCLUSION

Further study of uncontrollability of suicidal thought is needed to elucidate its impact on suicidal outcomes.

Psychometric Properties of the Self-Injurious Thoughts and Behaviors Interview-Short Form Among U.S. Active Duty Military Service Members and Veterans

We assessed the interrater reliability, convergent validity, and discriminant validity of the Self-Injurious Thoughts and Behaviors Interview-Short Form (SITBI-SF) in a sample of 1,944 active duty service members and veterans seeking services for posttraumatic stress disorder (PTSD) and related conditions. The SITBI-SF demonstrated high interrater reliability and good convergent and discriminant validity. The measurement properties of the SITBI-SF were comparable across service members and veterans. Approximately 8% of participants who denied a history of suicidal ideation on the SITBI-SF reported suicidal ideation on a separate self-report questionnaire (i.e., discordant responders). Discordant responders reported significantly higher levels of PTSD symptoms than those who denied suicidal ideation on both response formats. Findings suggest that the SITBI-SF is a reliable and valid interview-based measure of suicide-related thoughts and behaviors for use with military service members and veterans. Suicide risk assessment might be optimized if the SITBI-SF interview is combined with a self-report measure of related constructs.

Aberrant connectivity in the right amygdala and right middle temporal gyrus before and after a suicide attempt: Examining markers of suicide risk

Functional neuroimaging has the potential to help identify those at risk for self-injurious thoughts and behaviors, as well as inform neurobiological mechanisms that contribute to suicide. Based on whole-brain patterns of functional connectivity, our previous work identified right amygdala and right middle temporal gyrus (MTG) connectivity patterns that differentiated Veterans with a history of a suicide attempt (SA) from a Veteran control group. In this study, we aimed to replicate and extend our previous findings by examining whether this aberrant connectivity was present prior to and after a SA. In a trauma-exposed Veteran sample (92 % male, mean age = 34), we characterized if the right amygdala and right MTG connectivity differed between a psychiatric control sample (n = 56) and an independent sample of Veterans with a history of SA (n = 17), using fMRI data before and after the SA. Right MTG and amygdala connectivity differed between Veterans with and without a history of SA (replication), while MTG connectivity also distinguished Veterans prior to engaging in a SA (extension). In a second study, neither MTG or amygdala connectivity differed between those with current suicidal ideation (n = 27) relative to matched psychiatric controls (n = 27). These results indicate a potential stable marker of suicide risk (right MTG connectivity) as well as a potential marker of acute risk of or recent SA (right amygdala connectivity) that are independent of current ideation.

Rapid intensification of suicide risk preceding suicidal behavior among primary care patients

BACKGROUND

Approximately half of those who attempt suicide report experiencing suicidal ideation and suicidal planning in advance; others deny these experiences. Some researchers have hypothesized that rapid intensification is due to past suicidal ideation and/or behaviors that are "mentally shelved" but remain available for rapid access later.

METHOD

To evaluate this hypothesis, we examined (a) temporal sequencing of suicidal ideation, suicidal planning, and suicidal behavior, and (b) speed of emergence of suicidal behavior in a prospective cohort study of 2744 primary care patients.

RESULTS

Of 52 patients reporting suicidal behavior during follow-up, 20 (38.5%) reported suicidal ideation and planning prior to their suicidal behavior, 23 (44.2%) reported suicidal ideation but not planning, and nine (17.3%) denied both suicidal ideation and planning. Over half (n = 30, 57.7%) reported the onset of suicidal ideation and/or planning on the same day as or after their suicidal behavior (i.e., rapid intensification). Rapid intensification was not associated with increased likelihood of reporting recent or past suicidal ideation, planning, or behaviors, suggesting rapid intensification does not depend on prior experience with suicidal ideation and/or behaviors.

CONCLUSION

Detecting primary care patients at risk for this form of suicidal behavior may be limited even with universal suicide risk screening.

Sexual compulsivity, erectile dysfunction, and suicidality among male survivors of military sexual violence

Although military sexual trauma (MST) is associated with an increased risk of suicide, suicide attempts, and suicidal ideation among service members and veterans, there is limited knowledge regarding the mechanisms of MST and suicidality among men. The current study examined whether MST was associated with sexual compulsivity and/or erectile dysfunction and if these, in turn, explained elevated suicidal thoughts and the likelihood of engaging in future suicidal behavior after accounting for mental health, military, and demographic characteristics. Service members and veterans who reported their gender as male (N = 508) were recruited via social media and completed online self-report measures assessing MST, erectile dysfunction, sexual compulsivity, suicidal ideation frequency, and the likelihood of engaging in future suicidal behavior. Path analysis was used to examine the study hypotheses. In total, 67 participants (13.2%) reported a history of MST; of these individuals, 27 (40.3%) reported suicidal ideation in the past 12 months, and 29 (43.9%) reported an increased likelihood of engaging in future suicidal behavior. MST was associated with increased sexual compulsivity, which, in turn, predicted more frequent suicidal ideation as well as a higher self-reported likelihood of engaging in future suicidal behavior. MST was associated with higher levels of erectile dysfunction, but erectile dysfunction was not associated with suicidal ideation in the adjusted model. Although the data were cross-sectional, precluding determinations of causality, the results support assessing and intervening with regard to sexual compulsivity to mitigate the risk for suicide-related outcomes among men who experience MST.

Comparative clinical utility of screening for Suicide Crisis Syndrome versus suicidal ideation in relation to suicidal ideation and attempts at one-month follow-up

INTRODUCTION

Suicidal ideation (SI) has numerous limitations in predicting suicidal behavior. The Suicide Crisis Syndrome (SCS) is proposed as an alternative method of detecting risk. This study compares the relative utility of SI and SCS in statistically predicting SI and behaviors at one-month follow-up.

METHODS

382 psychiatric patients (98 inpatients, 284 outpatients) completed baseline measures and provided information about suicide-related outcomes one month later. Participants were grouped based on responses to measures assessing SCS and SI.

RESULTS

Rates of follow-up suicidal behavior were significantly higher among those reporting both SCS and SI (22.2%) than those reporting SI alone (6.0%) or neither SCS nor SI (0.9%). SCS alone (8.3%) had descriptively, but not statistically, higher rates of suicidal behavior than those with neither SCS nor SI, and did not differ from SI alone and the combination of SCS and SI. Those reporting SI-with and without SCS-had higher levels of follow-up suicidal thoughts than those without SI.

CONCLUSION

The SCS was equivalent to SI, and incrementally informative alongside SI, in detecting individuals at risk of future suicidal behavior, whereas SI was more strongly related to future SI than SCS. The combination of SCS and SI may be clinically useful in detecting individuals who are at risk for suicide.

Non-response to questions about suicide ideation and attempts among veterans: Results from the National Health and Resilience in Veterans Study

OBJECTIVES

Veterans who decline to provide information on their history of self-injurious thoughts and behaviors (SITBs) may be at elevated suicide risk. We examined associations between non-response to a question assessing lifetime SITBs and proxy variables of suicide risk.

METHODS

In this population-based cross-sectional study of 4069 US veterans, responses to the Suicidal Behaviors Questionnaire-Revised were examined to group veterans into one of three categories: (1) denied lifetime SITBs, (2) reported lifetime SITBs, or (3) declined to respond.

RESULTS

Overall, 69.5% of veterans denied a SITB history, 29.5% reported a SITB history, and 1.0% declined to provide information regarding SITBs. In adjusted analyses, veterans who declined to provide information on SITBs were significantly more likely than veterans who explicitly denied lifetime SITBs to screen positive for posttraumatic stress disorder; report lifetime non-suicidal self-injury; and report elevated levels of total trauma burden, externalizing behaviors, loneliness, received social support, and provided social support. Across these constructs, veterans who declined to provide SITB information were statistically indistinguishable from veterans who explicitly reported lifetime SITBs.

CONCLUSIONS

Veterans who decline to provide information about suicidal thoughts and behaviors may represent a covert group at elevated risk for suicide.

Optimizing the Beck Scale for Suicide Ideation: An Item Response Theory Approach Among U.S. Military Personnel

The Beck Scale for Suicide Ideation (BSS) is one of the most used and empirically supported suicide risk assessment measures for behavioral health clinicians and researchers. However, the 19-item BSS is a relatively long measure and can take 5 to 10 minutes to administer. This study used Item Response Theory (IRT) techniques across two samples of mostly U.S. military service members to first identify (n₁ = 1,899) and then validate (n₂ = 757) an optimized set of the most informative BSS items. Results indicated that Items 1, 2, 4, 6, and 15 provided a similar-shaped test information curve across the same range of the latent trait as the full-length BSS and showed reliable item functioning across participant characteristics. The sum score of these five items showed a linear score linkage with the full-scale score, ρ > 0.87, and was equally as sensitive as the full scale for prospectively predicting near-term suicidal behavior at 74% with a cut score ≥1 (equivalent to full-scale score ≥6). Results are consistent with those from civilian samples. In time- or length-limited assessments, using these five BSS items may improve administration efficiency over the full BSS, while maintaining classification sensitivity.This study suggests that summing Items 1, 2, 4, 6, and 15 of the Beck Scale for Suicide Ideation (BSS) is an acceptable approach for shortening the full-length measure.

Pilot study of the Collaborative Assessment and Management of Suicidality-Group

INTRODUCTION

Few evidence-based options exist for outpatient treatment of patients at risk of suicide, and to-date almost all research has focused on individually delivered psychotherapy. Group therapy for veterans at risk of suicide is a promising alternative.

METHODS

Thirty veterans receiving care at an urban Veterans Affairs Medical Center in the southern United States were randomized to either care as usual (CAU) or to CAU plus the Collaborative Assessment and Management of Suicide-Group (CAMS-G). Veterans were assessed prior to randomization to condition and at 1, 3, and 6 months post-randomization on a range of suicide-specific measures, burdensomeness, belonging, treatment satisfaction, and group cohesion.

RESULTS

Across measures and follow-up assessments, veterans in CAMS-G reported good satisfaction with the intervention, a sense of cohesion with other members of the group, and reduced symptom distress. Veterans in both conditions reported decreases in suicidal ideation and behavior, with CAMS-G participants potentially improving slightly faster.

CONCLUSION

This description of CAMS-G for veterans adds to the growing literature on suicide-specific interventions and supports the need for additional research to determine if wide-spread rollout is justifiable.

Examining suicide assessment measures for research use: Using item response theory to optimize psychometric assessment for research on suicidal ideation in major depressive disorder

INTRODUCTION

Progress reducing suicide death will require randomized clinical trials (RCTs) specifically targeting suicide risk. Even large RCTs may not stipulate suicide death as the primary outcome, as suicide death is relatively uncommon. Therefore, RCTs may need to specify suicidal ideation as a proxy indicator of risk. There is no consensus on the best tool for measuring suicidal ideation within RCTs. We contrasted the psychometric performance of three suicidal ideation measures to address this need.

METHODS

We applied item response theory to the Beck Scale for Suicide Ideation (BSSI), the Columbia-Suicide Severity Rating Scale (C-SSRS), and the suicide item of the Hamilton Rating Scale for Depression (HRSD) for 101 outpatients with depression and suicidal ideation participating in a RCT with suicidal ideation as the primary outcome.

RESULTS

All measures of suicidal ideation were equally able to detect low and very high levels of suicidal ideation.

CONCLUSIONS

The choice of the specific measure of suicidal ideation in a clinical trial may be dictated by time and financial resources versus the need for granularity in the interpretation of the scores.

Suicide Notes, Attempts, and Attempt Lethality During Episodes of Ideation Among Suicidal Soldiers and Marines

OBJECTIVE

Suicide remains a significant public health problem among military personnel despite expanded suicide prevention efforts over the last two decades. It is important to understand the behavioral antecedents of suicide, including the writing of a suicide note, to inform efforts to identify imminent risk. However, the completion of a suicide note increasing the likelihood of making a suicide attempt (SA) and predicting a higher lethality SA during episodes of suicidality have not been evaluated.

METHOD

To determine whether or not the completion of a suicide note increased the likelihood of making a SA during a given episode of suicidal ideation (current or worst) and predicted a higher lethality SA, we conducted secondary data analysis with a sample of 657 help-seeking, active-duty U.S. Soldiers and Marines. We hypothesized that service members who completed a suicide note would be more likely to make a SA during that given episode of suicidality and make a higher lethality SA.

RESULTS

Completion of a suicide note increased the likelihood of making a SA in both current and worst episodes of suicidal ideation. Additionally, writing a suicide note predicted making a higher lethality SA during a service member's current episode of ideation but not their worst episode.

CONCLUSIONS

This is the first study to examine note-writing behavior during episodes of suicidal ideation rather than following a suicide death or attempt, demonstrating a non-trivial number (17%) had written a suicide note and this increased the likelihood of making a SA and a higher lethality SA.HIGHLIGHTSThe first study of suicide notes during periods of ideation regardless of attempt.A suicide note written during an episode of ideation predicted making an attempt.A suicide note predicted making a more lethal suicide attempt.

More frequent negative social exchanges are associated with higher suicide ideation and risk in men service members/veterans

INTRODUCTION

Suicide is a leading cause of death in military service members/veterans (SM/Vs). Positive social exchanges (e.g., emotional support) buffer against suicide risk but the influence of negative social exchanges (e.g., unwanted advice) is understudied.

METHOD

The current study explored (1) the contributions of positive and negative social exchanges as simultaneous correlates of suicide ideation and risk, (2) whether positive social exchanges mitigated the association of negative social exchanges and suicide ideation and risk, and (3) if facets of social exchanges were uniquely related to suicide ideation and risk. SM/Vs men (N = 508) completed self-report questionnaires.

RESULTS

Roughly 27% endorsed suicide ideation. Between 23.62 and 41.14% reported elevated suicide risk. Adjusted analyses revealed higher negative social exchanges, particularly insensitive behaviors, and lower positive social exchanges were associated with higher suicide ideation and risk. The magnitude of positive and negative social exchanges with suicide ideation or risk was statistically similar. Higher positive social exchanges did not attenuate the association of negative social exchanges and suicide ideation or risk (p > 0.05).

CONCLUSION

Higher negative social exchanges are equally strong correlates of suicide ideation and risk as positive social exchanges. Decreasing insensitive behaviors may reduce suicide ideation and risk in men SM/Vs.

Connectome-based functional connectivity markers of suicide attempt

BACKGROUND

Functional brain markers of suicidality can help identify at-risk individuals and uncover underlying neurocognitive mechanism(s). Although some converging evidence has implicated dysfunction in several brain networks, suicide-related neuroimaging markers are inconsistent across studies, due to heterogeneity of neuroimaging approaches, clinical populations, and experimental methods.

METHODS

The current study aimed to address these limitations by examining resting-fMRI connectivity in a sample of post-9/11 veterans with a past suicide attempt (SA; n = 16) compared to a psychiatric control group (PC; n = 124) with no SA history but comparable past and present symptomatology, as well as a trauma control group (TC; n = 66) of trauma-exposed healthy controls. We used both a novel graph-analytic and seed-based approach to characterize SA-related connectivity differences across brain networks.

RESULTS

First, the graph-analytic approach identified the right amygdala and a region in the cognitive control network (right middle temporal gyrus; MTG) as regional SA-related hubs of dysfunction (HoD), or regions that exhibited a high number of SA-related connections. Aberrant SA-related connectivity between these hubs spanned multiple networks, including the cognitive control, default mode and visual networks. Second, the seed-based connectivity analysis that identifies SA-related differences in the strength of neural connections across the whole brain further implicated the right amygdala.

LIMITATIONS

Small sample size and potential underreporting of SA.

CONCLUSIONS

These two analytic approaches preliminarily suggest that the right amygdala and right MTG may be specific neural markers of SA that can be differentiated from neural markers of psychopathology more broadly.

C-SSRS performance in emergency department patients at high risk for suicide

OBJECTIVE

To evaluate the psychometric and predictive performance of the Columbia-Suicide Severity Rating Scale (C-SSRS) in emergency department (ED) patients with suicidal ideation or attempts (SI/SA).

METHODS

Participants (n = 1,376, mean age 36.8, 55% female, 76.8% white) completed the C-SSRS during the ED visit and were followed for one year. Reliability analyses, exploratory structural equation modeling, and prediction of future SA were explored.

RESULTS

Reliability of the Suicidal Ideation subscale was adequate, but was poor for the Intensity of Ideation and Suicidal Behavior subscales. Three empirically derived factors characterized the C-SSRS. Only Factor 1 (Suicidal Ideation and Attempts) was a reliable predictor of subsequent SA, though odds ratios were small (ORs: 1.09-1.10, CI95% : 1.04, 1.15). The original C-SSRS Suicidal Ideation and Suicidal Behavior subscales and the C-SSRS ED screen predicted subsequent SA, again with small odds ratios (ORs: 1.07-1.19, CI95% : 1.01, 1.29). In participants without a SA history, no C-SSRS subscale predicted subsequent SA. History of any SA (OR: 1.98, CI95% : 1.43, 2.75) was the strongest predictor of subsequent SA.

CONCLUSIONS

The psychometric evidence for the C-SSRS was mixed. History of a prior SA, as measured by the C-SSRS, provided the most parsimonious and powerful assessment for predicting future SA.