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.

The impact of scaffolded and non-scaffolded suicidal virtual human interaction training on clinician emotional self-awareness, empathic communication, and clinical efficacy.

BACKGROUND: Clinicians working with patients at risk of suicide often experience high stress, which can result in negative emotional responses (NERs). Such negative emotional responses may lead to less empathic communication (EC) and unintentional rejection of the patient, potentially damaging the therapeutic alliance and adversely impacting suicidal outcomes. Therefore, clinicians need training to effectively manage negative emotions toward suicidal patients to improve suicidal outcomes. METHODS: This study investigated the impact of virtual human interaction (VHI) training on clinicians' self-awareness of their negative emotional responses, assessed by the Therapist Response Questionnaire Suicide Form, clinicians' verbal empathic communication assessed by the Empathic Communication and Coding System, and clinical efficacy (CE). Clinical efficacy was assessed by the likelihood of subsequent appointments, perceived helpfulness, and overall interaction satisfaction as rated by individuals with lived experience of suicide attempts. Two conditions of virtual human interactions were used: one with instructions on verbal empathic communication and reminders to report negative emotional responses during the interaction (scaffolded); and the other with no such instructions or reminders (non-scaffolded). Both conditions provided pre-interaction instructions and post-interaction feedback aimed at improving clinicians' empathic communication and management of negative emotions. Sixty-two clinicians participated in three virtual human interaction sessions under one of the two conditions. Linear mixed models were utilized to evaluate the impact on clinicians' negative emotional responses, verbal empathic communication, and clinical efficacy; and to determine changes in these outcomes over time, as moderated by the training conditions. RESULTS: Clinician participants' negative emotional responses decreased after two training sessions with virtual human interactions in both conditions. Participants in the scaffolded condition exhibited enhanced empathic communication after one training session, while two sessions were required for participants in the non-scaffolded condition. Surprisingly, after two training sessions, clinical efficacy was improved in the non-scaffolded group, while no similar improvements were observed in the scaffolded group. CONCLUSION: Lower clinical efficacy after virtual human interaction training in clinicians with higher verbal empathic communication suggests that nonverbal expressions of empathy are critical when interacting with suicidal patients. Future work should explore virtual human interaction training in both nonverbal and verbal empathic communication.

Using Virtual Patients to Assess and Improve Clinicians' Emotional Self-awareness: a Randomized Controlled Study.

OBJECTIVE: Working with suicidal patients can elicit negative emotional responses that can impede clinicians' empathy and affect clinical outcomes. Virtual human interactions represent a promising tool to train clinicians. The present study investigated the impact of virtual human interaction training to enhance clinicians' emotional self-awareness and empathy when working with suicidal patients. METHODS: Clinicians were randomly assigned into two groups. Both groups interviewed a virtual patient presenting with a suicidal crisis; clinicians in the intervention condition (n = 31) received immediate feedback about negative emotional responses and empathic communication, whereas those in the control condition (n = 33) did not receive any feedback. All clinicians interviewed a second virtual patient 1 week later. Clinicians' emotional response to the two virtual patients and their empathic communication with each of them were assessed immediately after each interaction. Linear mixed models were used to assess change in clinicians' emotional response and verbal empathy between the two interactions across conditions. RESULTS: Clinicians' emotional responses toward the suicidal virtual patients were unchanged in both conditions. Clinicians in the intervention condition presenting low empathy level with the first virtual patient showed higher empathy level with the second virtual patient than with the first (B = 1.15, SE = 0.25, p < 0.001, 95% CI [0.42, 1.89]). CONCLUSIONS: This work demonstrates the feasibility of using virtual human interactions to improve empathic communication skills in clinicians with poor empathy skills. Further refinement of this methodology is needed to create effective training modules for a broader array of clinicians.

Stressful Life Events and Near-term Suicidal Risk in a Clinical Population.

The present study examined 22 specific stressful life events (SLEs) in relation to recent and prospective suicidal thoughts and behaviors (STBs). The effect of assessment method (self-report vs. chart-based ratings) and inpatient/outpatient status was also investigated. Past 3-month STBs and SLEs were assessed for 1,058 psychiatric patients; 696 completed one-month follow-up assessments. SLEs were common, with 684 participants (64.7%) reporting at least one. Total number of SLEs correlated with recent and prospective STB. A higher incidence of SLE's was found with self-report vs. chart-based measures (on 20 SLEs) and inpatients vs. outpatients (on 7 SLEs). SLEs of interpersonal rejection and loss, homelessness and academic failure offered elevated risk. In sum, SLEs are common and associated with STBs in psychiatric patients. SLEs of interpersonal rejection and loss, homelessness and academic failure may merit increased clinical attention.

Factor Structure and Validation of the Revised Suicide Crisis Inventory in a Korean Population.

OBJECTIVE: Because of the exceptionally high suicide rates in South Korea, new assessment methods are needed to improve suicide prevention. The current study aims to validate the revised Suicide Crisis Inventory-2 (SCI-2), a self-report measure that assesses a cognitiveaffective pre-suicidal state in a Korean sample. METHODS: With data from 1,061 community adults in South Korea, confirmatory factor analyses were first conducted to test the proposed one-factor and five-factor structures of the SCI-2. Also, an exploratory factor analysis (EFA) was performed to examine possible alternative factor structure of the inventory. RESULTS: The one-factor model of the SCI-2 resulted in good model fit and similarly, the five-factor model also exhibited strong fit. Comparing the two models, the five-factor was evaluated as the superior model fit. An alternative 4-factor model derived from EFA exhibited a comparable model fit. The Korean version of the SCI-2 had high internal consistency and strong concurrent validity in relation to symptoms of suicidal ideation, depression, and anxiety. CONCLUSION: The SCI-2 is an appropriate and a valid tool for measuring one's proximity to imminent suicide risk. However, the exact factor structure of the SCI-2 may be culture-sensitive and warrants further study.

Renegotiating tasks or goals as rupture repair: A task analysis in a cognitive-behavioral therapy for personality disorder.

OBJECTIVE: The purpose of this study was to provide some definition of rupture repair in a cognitive-behavioral therapy (CBT) for personality disorders, specifically how treatment tasks or goals are renegotiated. METHOD: Following a task analysis, a rational model was developed with the support of an expert panel. An empirical analysis was conducted on six CBT cases sampled from a clinical trial that included personality disordered patients and treatment adherent therapists. Two sessions from each case indicating rupture repair were selected, based on patient and therapist ratings of the Working Alliance Inventory-12 item version (WAI-12) and rupture presence. A qualitative analysis of the sessions was conducted with the support of the observer-based Rupture Resolution Rating System (3RS-2022). RESULTS: The empirical analysis provided some support for many of the stages defined in the rational model, but less support for the hypothesized sequences of stages. A rational-empirical synthesis yielded a revised model that suggested therapists combine various strategies in rupture repair in a variety of ways, not necessarily in consistent sequences. CONCLUSIONS: The renegotiation of tasks and goals in this CBT sample was variable. The importance of responsiveness and the need to validate the rational-empirical model were highlighted.

Who are the patients who deny suicidal intent? Exploring patients' characteristics associated with self-disclosure and denial of suicidal intent.

BACKGROUND: Patients' non-disclosure of suicidal ideation and intent concealment represent a major obstacle to the effective assessment of suicide risk and to the delivery of suicide prevention treatments. The present study aimed to investigate this phenomenon and to assess (1) if outpatient psychiatric patients are more or less likely to disclose intent to mental health clinicians in the context of psychiatric/psychological treatment than they are to in the context of research interviews with non-clinicians; and (2) if certain demographic, trait-like, and state-like characteristics may predict such disclosure differences. METHODS: A total of 780 psychiatric outpatient participants aged 18 to 84 and 193 clinician participants aged 25 to 54 were included in the study. The proportion of patients who disclosed to clinicians only, to research assistants (RAs) only, to both, and to none, was compared using a z-test. Univariate analyses were used to compare the participants' variables across disclosure groups, and significant individual predictors were included in multilevel regression analyses. RESULTS: Participants were more significantly more likely to disclose to RAs (10.4%) than to clinicians (5.6%), p < 0.001. Neuroticism and trait anxiety predicted disclosure to RAs vs no disclosure; low extraversion predicted disclosure to clinician versus no disclosure; and extraversion and trait anxiety predicted disclosure to RAs versus to clinicians. DISCUSSION: Patients' disclosure patterns, the personality variables predicting them, and their clinical implications were discussed in the context of the extant literature on patients' reasons for concealing suicidal ideation and intent.

Differential disclosure of suicidal intent to clinicians versus researchers: Associations with concurrent suicide crisis syndrome and prospective suicidal ideation and attempts.

Several patient and setting characteristics have been found to predict disclosure of suicidality to clinicians versus researchers. Less understood, however, is whether differential disclosure of suicidality predicts concurrent indirect indicators of suicide risk and future suicide-related outcomes. The present study examined differential disclosure of suicidal intent in clinical versus research settings as a predictor of (1) concurrent symptoms of the Suicide Crisis Syndrome (SCS); and (2) suicidal ideation and attempts within one month in patients (n = 1039) and their clinicians (n = 144), who completed a battery of self-report and interview measures at baseline. Patients who reported suicidal intent to anyone had higher concurrent SCS symptoms than those who denied suicidal intent, with no differences between those who reported intent to clinicians versus researchers only. Severity of suicidal ideation and rates of suicide attempts at one-month follow-up were higher among those who disclosed suicidal intent to a research assistant than among those who did not-regardless of whether suicidal intent was disclosed to their clinician. Overall, an improved understanding of the factors contributing to differential disclosure will improve both scientific inquiry and patient safety.

The Narrative-Crisis Model of suicide and its prediction of near-term suicide risk.

Despite decades of research, much remains unknown about the transition from chronic to imminent suicidal risk. In the context of COVID-19, this question is even more urgent. The present study tests a novel, stepwise model of this transition, termed the Narrative-Crisis Model. This model proposes that, in people with chronic risk factors, stressful life events can trigger a specific progression of cognitive-affective responses (the suicidal narrative and the suicide crisis syndrome), resulting in increased near-term risk of suicidal thoughts and behavior (STB). Identification of each step in this progression provides opportunities for more precise interventions. Concurrent validity was tested with 732 psychiatric patients and predictive validity with 524 participants, assessed one to two months later. Chronic risk factors were measured with the Childhood Trauma Questionnaire, Relationship Styles Questionnaire, and UPPS-P Impulsive Behavior Scale; acute risk factors with the Stressful Life Events Questionnaire, Suicide Narrative Inventory, and Suicide Crisis Inventory. The Columbia Suicide Severity Rating Scale was administered at the initial research assessment and follow-up. Indirect effects were significant for the full model and most pathways, in both concurrent and prospective analyses. In sum, this study provides empirical support for a novel, stepwise model of the progression from chronic to near-term suicidal risk.

Alliance-focused safety planning and suicide risk management.

Patients' interpersonal vulnerabilities and problems represent major distal and proximal risk factors in the etiology of suicide. These can be triggered by the interpersonal demands of therapy and safety planning and impede the development of a strong therapeutic alliance, and thus, the effectiveness of safety planning. This article proposes that the principles put forth by the Alliance-Focused Training (AFT; Eubanks-Carter et al., 2015; Muran & Eubanks, 2020), which view the therapeutic alliance both as a precondition to therapy and as an active change mechanism, present an ideal framework for addressing patients' interpersonal challenges in the context of safety planning and suicide risk management. After discussing the relevance of AFT principles to safety planning, we will propose an approach to incorporate AFT techniques into evidence-based safety planning interventions and to monitor their impact on the therapeutic alliance and treatment outcome. Last, we will provide a brief clinical report to illustrate the principles and techniques described in the article. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Associations between clinicians' emotion regulation, treatment recommendations, and patient suicidal ideation.

OBJECTIVE: This study examines how clinicians' emotional responses to suicidal patients and their emotion regulation abilities are related to their treatment recommendations for these patients and to patients' concurrent suicidal ideation and at one-month follow-up. METHODS: Adult psychiatric outpatients (N = 361) and the mental health professionals evaluating them for treatment (N = 43) completed self-report assessments following their first clinical meeting. Clinician emotion regulation traits, emotional responses to individual patients, and the recommended intensity of treatment were assessed. Patients were assessed for suicidal ideation immediately following the initial meeting and at a one-month follow-up. Moderation and mediation analyses were performed to examine the relationships between study variables. RESULTS: Patient suicidal ideation at the initial clinical encounter was associated with increased negative emotions in clinicians with lower emotion regulation. Further, recommended treatment intensity was associated with clinicians' negative emotional responses but not with patient suicidal ideation among clinicians with lower emotion regulation. CONCLUSIONS: Treatment intensification is related to clinicians' emotion regulation abilities. Clinicians' attention to their emotional responses may facilitate improved treatment process and ultimately may improve suicidal outcomes.

The beginning of the end: A comparison of treatment completers and early dropouts in trainee-provided time-limited Cognitive Behavioral Therapy.

About one in five clients drops out of treatment prematurely. Premature termination has been found to correlate with patient, therapist, and treatment factors, as well as complex interpersonal processes, including ruptures in the therapeutic alliance. This study examines the therapeutic alliance using a qualitative approach to patient-, therapist-, and observer-based data. The sample includes five trainee therapists, each of whom worked with one patient who terminated after the first or second session, and one who completed a cognitive-behavioral therapy protocol. The session(s) preceding premature termination in the drop case and the corresponding session(s) in the completer case were examined. Rupture resolution process was prevalent in both groups, though confrontation ruptures seemed more prevalent with completers and withdrawal ruptures were more clinically impactful with dropouts. Therapist awareness of process and responsiveness or contribution to rupture were identified as potential factors contributing to patient retention.

The revised suicide crisis inventory (SCI-2): Validation and assessment of prospective suicidal outcomes at one month follow-up.

BACKGROUND AND AIM: The Suicide Crisis Syndrome (SCS) is an evidence-based pre-suicidal cognitive and affective state predictive of short-term suicide risk. The most recent SCS formulation, proposed as a suicide-specific DSM diagnosis, features a feeling of Entrapment accompanied by four additional symptom clusters: Affective Disturbance; Loss of Cognitive Control; Hyperarousal; and Social Withdrawal. The aim of the present study was to revise the Suicide Crisis Inventory (SCI; Barzilay et al., 2020), a self-report measure assessing the presence of the SCS,in accordance with the current SCS formulation, as well as to assess the psychometric properties and clinical utility of its revised version, the Suicide Crisis Inventory-2 (SCI-2). METHODS: The SCI-2, a 61-item self-report questionnaire, was administered to 421 psychiatric inpatients and outpatients at baseline. Prospective suicidal outcomes including suicidal ideation, preparatory acts, and suicidal attempts were assessed after one month. Internal structure and consistency were assessed with confirmatory factor analysis, convergent, discriminant, and current criterion validity. Receiver-operating characteristic (ROC) curves with Area under the Curve (AUC) were used to examine the predictive validity of the SCI-2 to prospective outcomes. Exploratory analyses assessed the predictive validity of the five SCI-2 dimensions. RESULTS: The SCI-2 demonstrated excellent internal consistency (Cronbach's α = 0.971), good convergent, discriminant, and current criterion validity. The SCI-2 significantly predicted all three outcomes, and was the only significant predictor of suicidal attempts with AUC = 0.883. DISCUSSION: The results of this study indicate that the SCI-2 is a valid and reliable tool to assess the presence and intensity of the Suicide Crisis Syndrome and to predict short-term prospective suicidal behaviors and attempts among psychiatric outpatients and inpatients regardless of patients' readiness to disclose suicidal ideation.

The role of suicide ideation in assessing near-term suicide risk: A machine learning approach.

BACKGROUND: The majority of suicide attempters do not disclose suicide ideation (SI) prior to making an attempt. When reported, SI is nevertheless associated with increased risk of suicide. This paper implemented machine learning (ML) approaches to assess the degree to which current and lifetime SI affect the predictive validity of the Suicide Crisis Syndrome (SCS), an acute condition indicative of imminent risk, for near-term suicidal behaviors (SB ). METHODS: In a sample of 591 high-risk inpatient participants, SCS and SI were respectively assessed using the Suicide Crisis Inventory (SCI) and the Columbia Suicide Severity Rating Scale (C-SSRS). Two ML predictive algorithms, Random Forest and XGBoost, were implemented and framed using optimism adjusted bootstrapping. Metrics collected included AUPRC, AUROC, classification accuracy, balanced accuracy, precision, recall, and brier score. AUROC metrics were compared by computing a z-score. RESULTS: The combination of current SI and SCI showed slightly higher predictive validity for near-term SB as evidenced by AUROC metrics than the SCI alone, but the difference was not significant (p<0.05). Current SI scored the highest amongst a chi square distribution in regards to predictors of near-term SB. CONCLUSION: The addition of SI to the SCS does not materially improve the model's predictive validity for near-term SB, suggesting that patient self-reported SI should not be a requirement for the diagnosis of SCS.

Validity of the modular assessment of risk for imminent suicide in predicting short-term suicidality.

OBJECTIVE: The Modular Assessment of Risk for Imminent Suicide (MARIS) is a clinical assessment tool, consisting of four modules assessing (1) a pre-suicidal cognitive-affective state (Module 1); (2) patients' attitudes toward suicide (Module 2); (3) clinicians' assessment of suicide risk factors (Module 3); and (4) clinicians' emotional responses to patients (Module 4) that assesses short-term suicide risk. Initial evidence provided evidence for its reliability and concurrent validity. The present study extended these findings by examining the MARIS's predictive validity in relation to suicidal thoughts and behaviors at one-month follow-up. METHODS: A sample of 1039 psychiatric patients (378 inpatients, 661 outpatients) and their clinicians (N = 144) completed a battery of measures at baseline; 670 patients completed the one-month follow-up assessment. RESULTS: MARIS total scores predicted suicidal thoughts and behaviors at one-month follow-up, even after controlling for baseline suicidal thoughts and behaviors. Moreover, both Module 1 and the Distress subscale of Module 4 were uniquely associated with suicidal thoughts and behaviors at one-month follow-up, controlling for baseline suicidal thoughts and behaviors. Modules 2 and 3, on the other hand, exhibited poor internal consistency. CONCLUSION: Overall, both patient- and clinician-rated indices are uniquely predictive of suicidal thoughts and behaviors at one-month follow-up, highlighting the need for integrating clinicians' emotional responses into suicide risk assessment. Pending replication and extension of these findings in external samples, a briefer, two-module version of MARIS (MARIS-2) may be such an integrative, psychometrically sound, and clinically useful instrument that can be utilized to assess short-term suicide risk.

How do distal and proximal risk factors combine to predict suicidal ideation and behaviors? A prospective study of the narrative crisis model of suicide.

AIM: The Narrative Crisis Model of suicide posits that when individuals with trait vulnerabilities for suicide face stressful life events, they may develop distorted perceptions of themselves and society that culminate in a sense of no future. Referred to as the suicide narrative, these perceptions makes them more likely to experience the Suicidal Crisis Syndrome, an acute affective condition that increases the risk of engaging in suicidal ideation behaviors. The goal of this study was to assess the stage components of this model. METHODS: The stage components of the NCM were assessed among adult psychiatric inpatients (N = 223; listwise N = 85) aged 18-65 years old and admitted for suicidal ideation or attempts. Suicidal outcomes were assessed at one month follow-up. Structural equation modeling (SEM) was used to assess the model and its prediction of prospective suicidal outcomes. RESULTS: The model was supported by the SEM and proved to be a good fit for the data. Each temporal stage was significantly predicted by the precedent stage in the model and 13% of the variance in suicidal ideation and behaviors (when assessed conjointly) were explained by the model. When suicidal ideation and attempts were assessed separately, the amount of variance explained was 10.8% for suicidal ideation and 40.7% for suicidal attempts. DISCUSSION: The progression from trait vulnerabilities to suicidal outcomes proposed by the NCM was supported by our findings. These findings have clinical implications in the assessment and treatment of suicide risk and will need replication with larger samples.

Assessment of near-term risk for suicide attempts using the suicide crisis inventory.

BACKGROUND: The Suicide Crisis Syndrome (SCS) describes a pre-suicidal mental state marked by entrapment accompanied by affective disturbances, loss of cognitive control, hyperarousal and social withdrawal. This study tested the consistency and validity of the Suicide Crisis Inventory (SCI), a proposed measure of SCS severity, amongst a large, heterogeneous patient sample. METHODS: The SCI was used to assess 867 adult psychiatric inpatients and outpatients. Confirmatory factor analysis, logistic regressions and area under the curve analyses (AUC) were used to examine internal structure, construct validity and predictive validity for suicide ideation, plan and attempt one-month post-assessment. RESULTS: The five-factor model of the SCS demonstrated good fit and excellent internal consistency. SCI scores indicated significant associations but non-redundancy with depression, anxiety, and independence from other dimensions of psychiatric distress. SCI scores specifically predicted suicide attempts with an AUC of 0.733 and odds ratio=8.62 (p<0.001) at optimal cut-off point. SCI incremental predictive validity over and beyond suicidal ideation and attempts history reported at baseline was supported for predicting suicide attempts (ß= 0.012, S.E = 0.006; p=0.046). LIMITATIONS: The SCI is subject to self-report bias and does not include the SCS social withdrawal component. Follow-up assessment retention was partial (68%, n=591). CONCLUSION: The SCI is validated as a tool for the assessment of the SCS intensity and of imminent suicidal behavior. The SCI is suggested as a tool that could aid both researchers and clinicians in comprehensive assessment of a pre-suicidal mental state within moderate to high-risk populations, regardless of self-report on suicidal intent.

The suicide crisis syndrome: A network analysis.

Recent studies introduced the suicide crisis syndrome (SCS), a condition associated with imminent suicidal behavior and characterized by (a) a pervasive feeling of entrapment in which the escape from an unbearable life situation is perceived as both urgent and impossible (Criterion A) and (b) affective disturbance, loss of cognitive control, hyperarousal, and social withdrawal (Criterion B). The goal of the present study was to use some of the analytic tools provided by network analyses to further the understanding of the psychological, emotional, cognitive, behavioral, and physiological processes involved in the SCS by testing (a) whether the different symptoms of the proposed syndrome are related to each other, (b) whether symptoms form meaningful clusters, and (c) whether certain symptoms are more central than others. The study included 500 outpatient and 223 inpatient participants. A network analysis of the participants' scores on the various symptoms of the SCS was conducted. The network analysis suggested that most SCS symptoms are linked by strong connections and that entrapment and ruminative flooding are highly correlated with the other SCS symptoms. Three clusters of symptoms were identified, suggesting the existence of several interdependent psychological processes potentially involved in SCS phenomenology. Our findings support both the suggested symptoms of the SCS and the central role of entrapment in the proposed criteria for the syndrome. Emotional pain appears to be closely linked to entrapment and may belong in Criterion A. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Associations between clinicians' emotional responses, therapeutic alliance, and patient suicidal ideation.

BACKGROUND: Mental health clinicians frequently experience intense negative emotional responses to suicidal patients, which have been related to treatment outcome. This study examines the therapeutic alliance as a mediator of the relationship between clinicians' negative emotional responses at the initial encounter and patients' suicidal ideation (SI) concurrently and 1 month later. METHODS: We assessed 378 adult psychiatric outpatients (62.7% female; mean age = 39.1 ± 14.6 years) and their 61 treating clinicians. Following the initial encounter, self-report questionnaires assessed clinicians' emotional responses to their patients, patients' and clinicians' perception of the therapeutic alliance, and patients' SI. The SI was reassessed 1 month after the initial visit. Multilevel mediation analyses were performed. RESULTS: Patients' (but not clinicians') perception of the therapeutic alliance mediated the relationship between clinicians' negative emotional responses to patients and patients' SI 1 month following the initial visit (indirect effect estimate = 0.015; p < .001). CONCLUSIONS: The association between clinicians' negative emotional response and patients' prospective SI appears to be transmitted, at least partly, through the patients' perception of the poorer early quality of the therapeutic alliance. Thus, clinicians' awareness and management of their emotional states appear essential both for the identification of suicidal risk and to enhance therapeutic alliance and treatment outcomes.