The Veterans Crisis Line: Relations among immediate call outcomes and treatment contact and utilization following the call.

OBJECTIVE: Crisis line callers experience reductions in distress and suicidal ideation and utilize more health care following calls. The purpose of this study was to determine whether changes in distress and suicidal ideation during a call are associated with later healthcare contact and utilization. METHOD: Veterans Crisis Line calls from 599 veterans were extracted with call dates between 12/1/2018 and 11/30/2019. Calls were coded for changes in distress and suicidal ideation and linked with VA medical records to obtain healthcare data. Generalized Linear Mixed Modeling was used to examine the associations of changes in distress and suicidal ideation with healthcare contact (yes/no) and utilization (days of treatment) in the month (30 days) following the call. RESULTS: Reductions in distress were associated with behavioral (i.e., mental and substance use) healthcare utilization, F(1, 596) = 4.52, p = 0.03, and reductions in suicidal ideation were associated with any healthcare utilization, F(1, 596) = 6.45, p = 0.01. Changes in distress and suicidal ideation were not associated with healthcare contact. CONCLUSION: Responders need to help resolve distress and suicidal ideation and link callers with treatment. Unresolved distress and suicidal thoughts may signify later problems with treatment utilization. Research is needed to determine causality.

A randomized clinical trial to assess feasibility, acceptability and preliminary effects of telehealth-delivered cognitive-behavioral therapy for perceived social isolation among individuals with opioid use disorders.

BACKGROUND: The purpose of this study was to report on feasibility, acceptability, and initial efficacy data cognitive-behavioral therapy for perceived isolation (CBT-PSI) compared to health education among individuals with opioid use disorders (OUD) reporting elevated loneliness. METHODS: Participants (n = 125) with OUD reporting elevated loneliness were recruited using online advertising to participate in a telehealth-delivered randomized clinical trial. Participants received either a 6-session CBT-PSI (n = 63) or health education (n = 62). Measures assessing loneliness, quantity of social interactions, perceived social support, substance use, substance use consequences, and treatment engagement among others, were completed at baseline, post-intervention, and 1- and 2-months post-intervention. Participants also reported on treatment acceptability for both conditions. RESULTS: Target enrollment was met with loneliness, social disconnectedness measures, and substance use reflecting a clinically severe sample. Retention was high (> 80%) for both conditions. Participants rated both Health Education and CBT-PSI as acceptable, helpful, and useful interventions to address loneliness and opioid use. Loneliness was reduced and quantity of social interactions and perceived social support were increased to the same extent for both conditions and across the follow-up assessments. Opioid use and overall substance use were reduced in both conditions; however, the reductions among participants received CBT-PSI were significantly greater compared to Health Education. CONCLUSIONS: This study supports the feasibility and acceptability of CBT-PSI. CBT-PSI findings related to loneliness, substance use, and other social connectedness outcomes are encouraging. Additional testing of CBT-PSI in a fully-powered trial is warranted.

Deaf perceptions about treatment for alcohol use and mental health.

INTRODUCTION: Recent research suggests that alcohol use disorder may be more prevalent in the Deaf community, a diverse sociolinguistic minority group. However, rates of treatment-seeking among Deaf individuals are even lower than in the general society. This study used the Theory of Planned Behavior to identify Deaf adults' beliefs about treatment that may prevent their treatment-seeking behaviors. METHODS: This study conducted elicitation interviews with 16 Deaf adults. The study team recruited participants from across the U.S. and conducted interviews on Zoom. Participant ages ranged from 27 to 67 years (M = 40, SD =10.8). Seventy-five percent of the sample was male, 75 % were White, and 12.5 % were Hispanic/Latine. The study conducted interviews in American Sign Language, subsequently interpreted into English by a nationally certified interpreter, and transcribed for data analyses. The study analyzed transcripts using the Framework Method. The study team coded the interviews in groups and assessed for saturation (≤ 5 % new themes) of themes throughout the analysis. This study reached saturation in the third group (six total groups). RESULTS: Identified themes followed the Theory of Planned Behavior constructs. The study identified nine Behavioral Beliefs with four advantages and five disadvantages of seeking treatment, four Normative Beliefs with one support and three oppositions to seeking treatment, and thirteen Control Beliefs with five facilitators and eight barriers to seeking treatment. Overall, the Deaf participants reported several unique beliefs based on their cultural and linguistic perspectives, including a concern about unqualified providers, experiencing stress in treatment with hearing providers, stigma within the Deaf community, less access to cultural information about alcohol and mental health, less encouragement of traditional treatment in marginalized communities, and additional barriers (e.g., communication, limited Deaf treatment options, discrimination, etc.). CONCLUSIONS: A thorough understanding of individual beliefs about treatment is necessary to develop interventions that may increase treatment-seeking behaviors. Previous research has demonstrated that individual beliefs may be modified using Cognitive Behavioral Therapy techniques to increase treatment-seeking behaviors among hearing individuals. Similar interventions may be useful with Deaf individuals; however, they must consider the unique cultural and linguistic perspectives of the community.

Brief, cognitive-behavioral intervention to promote treatment seeking in adults with severe alcohol use disorder: A randomized controlled trial.

BACKGROUND AND AIMS: There is little RCT evidence that brief interventions improve treatment seeking in individuals with severe alcohol use disorder (AUD) or treatment seeking reduces alcohol use. The aim was to test the efficacy of a brief intervention to increase treatment seeking in treatment naïve adults with severe AUD and measure its effects on alcohol use. DESIGN: Parallel group, non-pharmacologic RCT with intervention (n = 197) and active control (n = 203) conditions, with blinded assessors conducting follow-ups at 1, 3 and 6 months. SETTING: Online recruitment in a 17-county region of upstate New York, USA. PARTICIPANTS: Inclusion criteria consisted of ages ≥18 years, Alcohol Use Disorders Identification Test score ≥16, exceeds recommended limits for alcohol use and no history of AUD treatment. n = 400; 50% female; 79% white; mean age, 40.7; mean education, 14.6 years. INTERVENTION AND COMPARATOR: One-session telephone-delivered interventions: Cognitive-Behavioral Therapy for Treatment Seeking (CBT-TS; intervention), review of a National Institute on Alcohol Abuse and Alcoholism pamphlet on AUD treatment (control). MEASUREMENTS: Self-report of any AUD treatment use over 3 months (primary outcome) and two standard measures of alcohol use over 6 months (secondary outcomes). FINDINGS: Intent-to-treat analyses were used. Assessment follow-up rates were ≥93%. Any alcohol-related treatment use over 3-month follow-up was obtained by 38 (19%) intervention participants and 36 (18%) control participants, a non-significant difference, χ2 [1] = 0.16, P = 0.689. Secondary analysis showed a significant interaction term between sex and intervention assignment (ß = -1.197, P = 0.027). The interaction suggested CBT-TS was effective in men (22% vs 13%), although the evidence was somewhat weak (P = 0.071), and it was not effective in women (17% vs 24%). CONCLUSIONS: A one-session cognitive-behavioral therapy intervention to increase treatment seeking in treatment naïve adults with severe alcohol use disorder did not increase treatment seeking.

Efficacy of CBT for Treatment Seeking (CBT-TS) in Untreated Veterans and Service Members at Risk for Suicidal Behavior.

OBJECTIVE: Military members and Veterans at-risk for suicide are often unlikely to seek behavioral health treatment. The primary aim of this study was to test the efficacy of brief CBT for Treatment Seeking (CBT-TS) to improve behavioral health treatment utilization among U.S. military service members and Veterans at-risk for suicide. METHODS: A total of 841 participants who served in the U.S. military since 9/11 and who reported suicidality but were not in behavioral health treatment were recruited to participate in this trial. Participants were randomly assigned to either brief CBT-TS delivered by phone or an assessment-only control condition. Follow-up assessments were conducted at baseline and months 1, 3, 6, and 12 to track treatment utilization and symptoms. RESULTS: CBT-TS resulted in significantly greater behavioral health treatment initiation within 1 month compared to the control condition (B = .93, p < .001); and the higher treatment initiation persisted for 12 months post intervention. CONCLUSIONS: This study employed a low-cost, easily implementable one-session intervention administered by phone. The study provides evidence that CBT-TS is efficacious in promoting behavioral health treatment initiation in an adult population at risk for suicidal behavior and showed enduring benefits for 6-12 months. CBT-TS provides a unique strategy for treatment engagement for at-risk adults unlikely to seek treatment. TRIAL REGISTRATION: Clinicaltrials.gov NCT05077514.

Veterans Crisis Line Call Outcomes: Treatment Contact and Utilization.

INTRODUCTION: Crisis lines are a central component of suicide prevention strategies in the U.S. and for the Department of Veterans Affairs. The purpose of this study is to evaluate the impact of calling the Veterans Crisis Line on treatment contact and utilization. METHODS: Call records from 599 veterans who called in 2019 were linked with medical records and analyzed in 2020. Multilevel generalized linear modeling examined pre-post changes in treatment contact (yes/no) and utilization (number of days of care). RESULTS: In the month after the call, 85% of callers made contact with health care, and 79% made contact with behavioral health care. Callers were more likely to make contact with health care in the month after the call than in the preceding month (AOR=6.27, 95% CI=4.22, 9.32) and more likely to make contact with behavioral health care (AOR=10.21, 95% CI=6.66, 15.67). Days of health care nearly doubled to 4.82, and days of behavioral health care more than doubled to 3.52. CONCLUSIONS: Among veteran callers who are linked to medical records, calling the Veterans Crisis Line may increase contact and utilization of health care and behavioral health care. These findings support crisis lines that are linked with healthcare systems in public health strategies for suicide prevention.

Risk factors of suicidal behaviors in a high-risk longitudinal veteran sample: A network analysis.

INTRODUCTION: Suicide is a substantial public health burden, particularly among veterans. Risk factors have been delineated for suicide; however, the dynamic interrelations between risk factors have not been fully examined. Such research has the potential to elucidate processes that contribute to suicide risk between individuals with a past suicide attempt (attempters) and those without a past suicide attempt (nonattempters). METHODS: In the current study, network analysis was used to compare networks between attempters and nonattempters in a high-risk veteran sample (N = 770; Mage  = 32.3 years, SD = 6.8; 326 with a past suicide attempt) who were followed over 1 year. Networks were estimated to examine (1) concurrent relations of suicide risk factors at baseline and (2) predictability of prospective suicidal behavior (SB). RESULTS: There were no differences in the overall connectivity of attempter and nonattempter networks. Perceived burdensomeness and posttraumatic stress disorder (PTSD) symptoms were most central in the attempters' network, whereas PTSD symptoms and insomnia were most central in the nonattempters' network. The risk factors prospective SB in either network. However, attempters were more likely to engage in SB over the course of the study. CONCLUSION: These findings highlight the difficulty in predicting who will attempt suicide.

Alcohol use during COVID-19 in adults with severe untreated AUD.

Evidence demonstrating increased alcohol use during COVID-19 comes from low- to moderate-alcohol use samples and has yet to use adults with severe but untreated AUD. Using a community sample of adults with severe AUD, this exploratory, cross-sectional study examined associations of COVID-19 alcohol use. Participants were recruited for a phase-II RCT. Only baseline measures, completed prior to randomization, were analyzed in the present study. Key variables were alcohol consumption, COVID-19-related worries and experiences, and qualitative responses of 1) alcohol use and 2) positive changes during COVID-19. 176 pariticpants recruited since COVID-19 were on average 41.4 years old, 49.1% female, and 79% White. Participants drank alcohol nearly 23 of the past 30 days, consumed 7 standard drinks per drinking day, and nearly 90% reported increased alcohol use. More heavy episodic drinking was reported in the first six-months of COVID-19 and more COVID-related concerns in the most recent six-months. Participants reported drinking increased due to "more time on their hands", but the pandemic also "strengthened relationships". Results affirm an increase in alcohol use during COVID-19 in adults with severe, untreated AUD. Findings underscore the need to understand how alcohol use and pandemic-related circumstances may influence one another for adults with severe AUD.

Heterogeneity in Alcohol-Related Severity and Interests in Going to Treatment in Community Adults with Alcohol Use Disorder (AUD).

RATIONALE: Alcohol use disorder (AUD) is highly prevalent among adults in the US and is associated with substantial personal and societal costs. Yet only a small percentage of adults with AUD initiate treatment, including those with severe AUD symptoms who are most in need of treatment. In this paper we use latent profile analysis (LPA) to describe differences in symptoms of AUD severity and alcohol-related consequences among untreated adults with severe AUD symptoms who were recruited from the community. Identification of such differences will result in better understanding of this population and will improve methods of treatment outreach. METHOD: The LPAs were conducted on the baseline data of 403 (50% male, 77% White, M age = 40.74 with severe AUD recruited from the community and enrolled in a trial of an intervention designed to encourage treatment initiation. Participants reported no prior AUD treatment history. The LPAs were based on indicators of AUD severity, alcohol-related negative consequences, and self-reported intention to initiate AUD treatment. RESULTS: The LPAs revealed 4 profiles. Profile membership was predicted by baseline participant sex and whether they were living with a partner and was associated with baseline past 30-days alcohol consumption. CONCLUSIONS: The findings characterize and describe the variability among adults in the community with untreated severe AUD on variables that tend to motivate such individuals to seek treatment, which could significantly advance treatment outreach efforts.

Alcohol-related consequences and the intention to seek care in treatment naïve women and men with severe alcohol use disorder.

INTRODUCTION: Research on treatment utilization for alcohol use disorder (AUD) is based primarily on clinical samples and community samples of low AUD severity that may not need formal care. Using a community sample of adults with untreated but severe AUD symptoms, we tested the hypothesis that alcohol-related consequences, but not alcohol consumption levels, are associated with the intention to seek AUD treatment, examined associations of specific types of alcohol-related consequences with intention, and explored sex differences in these associations. METHODS: The sample was recruited using social media ads for a randomized controlled trial to test a brief intervention to promote AUD treatment seeking. This report is based on analysis of baseline data collected prior to treatment intervention. Multiple linear regressions examined associations of measures of alcohol consumption, alcohol-related consequences broadly, and specific alcohol-related consequences with the intention to seek treatment. Moderating effects of sex on associations were explored. RESULTS: Subjects (n = 349) averaged 41 years of age, 48% were female, 6% were Latinx, 80% were white, 15% were Black, and 92% met criteria for severe AUD. Alcohol consumption measures were not associated with intention to seek treatment whereas interpersonal- and intrapersonal- consequences were associated with intention. Sex served as a moderator, with intrapersonal consequences (e.g., sad mood) showing a stronger association with intention in women and social responsibility consequences (particularly financial) associated with intention in men. CONCLUSION: Select alcohol-related consequences may be keys to understanding increased intention to seek AUD treatment including intrapersonal consequences in women and financial consequences in men.

Veterans Crisis Line Call Outcomes: Distress, Suicidal Ideation, and Suicidal Urgency.

INTRODUCTION: This study evaluates the effectiveness of the Veterans Crisis Line on immediate call outcomes (i.e., caller distress, suicidal ideation, and suicidal urgency) for veterans who provide identifying information. METHODS: Coders rated pre- and post-distress, suicidal ideation, and suicidal urgency for 647 calls from 2019 veteran callers. Intraclass correlation coefficients examined inter-rater reliability. Multilevel generalized linear modeling examined pre-post changes. RESULTS: Inter-rater reliability was good for distress, excellent for suicidal ideation, and fair for urgency. Callers had 5 times greater odds of a reduction in distress (AOR=5.03, 95% CI=3.98, 6.49), almost 5 times greater odds of a reduction in suicidal ideation (AOR=4.92, 95% CI=3.49, 6.94), and 11 times greater odds of a reduction in suicidal urgency (AOR=11.01, 95% CI=2.72, 44.50) at the end of calls than at the beginning. CONCLUSIONS: Veterans Crisis Line callers who provide identifying information experience reductions in distress and suicidal ideation during the call. Research is needed to examine the reduction in suicidal urgency because of fair reliability, generalizability of results to other callers, post-call treatment contact and engagement, and risk for suicide attempts and death.

Perceptions of Public Messaging to Facilitate Help Seeking during Crisis among U.S. Veterans at Risk for Suicide.

OBJECTIVE: The primary objectives of this study were to (1) examine how veterans at risk for suicide understand and interpret suicide prevention messages, and (2) identify message features that support (or hinder) help seeking behaviors. METHOD: Individual virtual interviews (N = 40) were conducted from August 2018 to April 2019 with a nationwide sample of veterans who had a recent non-fatal suicide attempt. Participants were exposed to three messages in public circulation that promote help seeking during crisis and an interview guide steered open-ended conversations on the mechanisms of persuasive communication. Interview transcripts were analyzed using a constant comparison analytic strategy in Atlas.ti. RESULTS: Preferences for particular kinds of messages and features emerged during interviews based on a need for novel, emotionally intense stimuli among veterans. Specifically, messages with solemn themes and darker, more provocative imagery were believed to be particularly potent for engaging those at high risk. Although the goal of promoting help seeking was discernable, actionable steps (crisis line use) were not clearly communicated potentially preventing messages from increasing help seeking behaviors. CONCLUSION: While messaging was perceived as capable of intervening to promote help seeking, participants reported distinct communication preferences and needs during periods of high risk. Findings underscore the significance of involving those with lived experience to inform the effective design and use of help seeking messaging targeting veterans at risk for suicide.HighlightsPublic messaging (campaigns) is a portable suicide prevention interventionCareful study is needed to effectively communicate help seeking messagesFindings inform the effective use of campaigns with veterans at risk for suicide.

Pain intensity trajectories among veterans seeking mental health treatment: Association with mental health symptoms and suicidal thoughts and behaviors.

BACKGROUND: The purpose of this study was to identify classes of pain intensity trajectories among veterans reporting suicidal ideation with no prior mental health treatment experience. We were interested in also assessing factors associated with the pain trajectory classes. METHODS: A total 747 participants completed measures of pain, depression, suicide ideation and behaviors, insomnia, substance use, and PTSD. Follow-up assessments were completed at 1-, 3-, 6-, and 12-months post-baseline. Growth mixture modeling was conducted, and pain trajectories were modeled from baseline to month 12. RESULTS: Three classes were identified: mild pain intensity that increased over time to severe pain intensity (Increasing-Severe; n = 9), low pain intensity that decreased over time (Mild-Decreasing; n = 172), and moderate pain intensity that remained relatively stable over time (Moderate-Stable; n = 566). The Moderate-Stable trajectory had more severe PTSD symptoms, more frequent depression symptoms, and more severe insomnia. The odds of endorsing suicide ideation at month 12 were significantly higher in the Moderate-Stable trajectory compared to the Mild-Decreasing trajectory. CONCLUSIONS: This was the first study to assess classes of pain intensity trajectories among individuals who were treatment naïve for mental health issues. The findings suggest that a moderate stable trajectory of pain intensity over the course of 12 months is common and associated with a more severe clinical profile, including suicide ideation. This study underscores the importance of addressing pain intensity among individuals seeking mental health treatment, particularly for those with pain intensity that is moderate and stable over time.

Opioid use disorder and its association with self-reported difficulties participating in social activities.

BACKGROUND AND OBJECTIVES: Difficulties participating in social activities are associated with increased mortality and are underemphasized in addressing the opioid epidemic. This study assessed the association of difficulties participating in social activities and opioid use disorder (OUD) and suicidal ideation and suicide attempts and difficulties participating in social activities among individuals with OUD. METHODS: Data on OUD, difficulties participating in social activities, suicidal ideation, suicide attempts, and other characteristics were assessed in 398,962 respondents from the 2008-2017 National Survey of Drug Use and Health. Logistic regressions examined the association of difficulties participating in social activities and OUD, and then among only respondents with OUD, difficulties participating in social activities and suicidal ideation and suicide attempts. RESULTS: Respondents with OUD reported higher levels of difficulties participating in social activities compared with respondents without OUD (severe difficulties: odds ratio [OR] = 4.10, 95% confidence interval [CI] = 3.40-4.93). Among those with OUD, difficulties participating in social activities were associated with suicidal ideation (severe difficulties: OR = 2.45, 95% CI = 1.77-3.38), but not attempts. CONCLUSION AND SCIENTIFIC SIGNIFICANCE: The findings indicate that people with OUD experience difficulties participating in social activities and these difficulties are associated with suicidal ideation. It may be important to address difficulties participating in social activities within the context of OUD treatment, potentially as it relates to suicide prevention. This is the first study utilizing 10 years of nationally representative data to assess difficulties participating in social activities, OUD, and suicidal ideation and suicide attempts. Difficulties participating in social activities represent an under-addressed, potentially important therapeutic target to address OUD.

Lonely, anxious, and uncertain: Critical risk factors for suicidal desire during the COVID-19 pandemic.

Public health measures enacted early in response to the COVID-19 pandemic resulted in unprecedented physical isolation. Social isolation, or the objective experience of being alone, and loneliness, the subjective feeling of being lonely, are both implicated in suicidal ideation. Anxiety sensitivity (i.e., fear of somatic anxiety) and intolerance of uncertainty (distress due to uncertainty), may also be heightened in response to the pandemic increasing risk for suicidal ideation in response to social isolation and loneliness. The direct and interactive relations loneliness, anxiety sensitivity, and intolerance of uncertainty shared with suicidal ideation were examined using structural equation modeling across two samples. Sample 1 comprised 635 people (M age = 38.52, SD = 10.00; 49.0% female) recruited using Mechanical Turk in May 2020. Sample 2 comprised 435 people (M age = 34.92, SD = 14.98; 76.2% female) recruited from faculty, staff, and students at a midwestern university in June 2020. Loneliness and anxiety sensitivity were positively, uniquely associated with suicidal ideation across samples. Results of this study were cross-sectional and included only self-report measures. These findings highlight loneliness and anxiety sensitivity as important correlates of suicidal ideation. Modular treatments should be employed to target these mechanisms to reduce COVID-19-related suicidal ideation.

Perceived burdensomeness, but not thwarted belongingness, mediates the impact of PTSD symptom clusters on suicidal ideation modeled longitudinally.

BACKGROUND: Studies examining Posttraumatic Stress Disorder symptom clusters, suicidal ideation, and variables described in the Joiner's interpersonal theory of suicidal behavior were cross-sectional. METHODS: We tested whether the relation between the Posttraumatic Stress Disorder symptom clusters and suicidal ideation is mediated by perceived burdensomeness and thwarted belongingness (variables of the interpersonal theory of suicidal behavior) in military personnel with current suicidal ideation using longitudinal design. DSM-IV model with reexperiencing, avoidance, numbing, and hyperarousal clusters was used. Structural equation modelling was used to test separate models for each symptom cluster with a symptom cluster at baseline, month 1 perceived burdensomeness and thwarted belongingness, and month 3 suicidal ideation, controlling for baseline values of the month 1 and month 3 variables. RESULTS: Analysis of direct non-mediation models showed that baseline reexperiencing (p = .08) and avoidance (p = .07) symptom clusters marginally predicted month 3 suicidal ideation. The mediation analyses showed indirect effects from baseline reexperiencing, avoidance, and hyperarousal to suicidal ideation at month 3 through perceived burdensomeness at month 1. Thwarted belongingness did not mediate the relations between symptom clusters and suicidal ideation. LIMITATIONS: The mechanism driving development of suicidal ideation in military personnel with PTSD may be different from the mechanism in other samples. Doing a phone interview limited us to use questionnaires instead of a clinical interview. CONCLUSIONS: Our findings suggest that interventions targeting perceived burdensomeness in military personnel with reexperiencing, avoidance, and hyperarousal symptoms may be beneficial to reduce suicidal ideation.

Veteran Perspectives of Barriers and Facilitators to Campaigns Promoting Help Seeking During Crisis.

The objective of this paper is to characterize barriers and facilitators reported by U.S. military veterans related to campaigns promoting help seeking during suicidal crisis. Individual telephone interviews (N = 40) were conducted from August 2018-April 2019 with a sample of veterans who had a recent non-fatal suicide attempt. Interview transcripts were analyzed using a constant comparison analytic strategy. Participants reported the four facilitators to message effectiveness: (a) potential reach of specific channels; (b) interruption of suicidal thoughts; (c) normalizing the suicidal experience and help seeking; and (d) modeling desired behavior change. Barriers that hindered campaigns were also identified and include (a) broad messages, (b) challenges in cognitive processing, (c) media avoidance and (d) a boomerang effect. This study underscores the significance of involving those with lived experience to identify factors that may improve or hinder message effectiveness.

Correlates of Suicidal Ideation and Behaviors Among Former Military Personnel Not Enrolled in the Veterans Health Administration.

The current study sought to explore suicidal concomitants, both demographic and psychological, among former military personal. The sample included 645 veterans who are at increased risk for suicide but have not yet pursued Veterans Health Administration (VHA) services. Descriptive statistics revealed that these veterans are primarily young Caucasian males who served in the U.S. Army. In terms of psychological characteristics, the current sample reported clinically significant levels of depression, post-traumatic stress, and insomnia. Furthermore, respondents acknowledged use of various substances and high levels of perceived burdensomeness and thwarted belongingness. The demographic and psychological makeup of our sample was somewhat similar to that of VHA-connected veterans except that our sample was slightly more educated and reported less physical pain.

Identifying and Predicting Intentional Self-Harm in Electronic Health Record Clinical Notes: Deep Learning Approach.

BACKGROUND: Suicide is an important public health concern in the United States and around the world. There has been significant work examining machine learning approaches to identify and predict intentional self-harm and suicide using existing data sets. With recent advances in computing, deep learning applications in health care are gaining momentum. OBJECTIVE: This study aimed to leverage the information in clinical notes using deep neural networks (DNNs) to (1) improve the identification of patients treated for intentional self-harm and (2) predict future self-harm events. METHODS: We extracted clinical text notes from electronic health records (EHRs) of 835 patients with International Classification of Diseases (ICD) codes for intentional self-harm and 1670 matched controls who never had any intentional self-harm ICD codes. The data were divided into training and holdout test sets. We tested a number of algorithms on clinical notes associated with the intentional self-harm codes using the training set, including several traditional bag-of-words-based models and 2 DNN models: a convolutional neural network (CNN) and a long short-term memory model. We also evaluated the predictive performance of the DNNs on a subset of patients who had clinical notes 1 to 6 months before the first intentional self-harm event. Finally, we evaluated the impact of a pretrained model using Word2vec (W2V) on performance. RESULTS: The area under the receiver operating characteristic curve (AUC) for the CNN on the phenotyping task, that is, the detection of intentional self-harm in clinical notes concurrent with the events was 0.999, with an F1 score of 0.985. In the predictive task, the CNN achieved the highest performance with an AUC of 0.882 and an F1 score of 0.769. Although pretraining with W2V shortened the DNN training time, it did not improve performance. CONCLUSIONS: The strong performance on the first task, namely, phenotyping based on clinical notes, suggests that such models could be used effectively for surveillance of intentional self-harm in clinical text in an EHR. The modest performance on the predictive task notwithstanding, the results using DNN models on clinical text alone are competitive with other reports in the literature using risk factors from structured EHR data.

Profiles of Risk for Suicidal Behavior in Past and Current United States Military Personnel: Latent Profile Analysis of Current Risk Factors.

Person-centered approaches are underutilized to identify people with shared risk profiles. In this study, an at-risk sample of 773 past/current military personnel (Mage = 31.3 years, SD = 6.8) with current ideation (90.6%) and/or a prior suicide attempt (43.9%) were assessed using latent profile analysis. Variables included prior suicide attempts, suicidal ideation, alcohol/drug use, insomnia, depression, belongingness, burdensomeness, and post-traumatic stress disorder symptoms. Three classes emerged: Low Symptoms (N = 502), Elevated Suicidality (N = 176), and Elevated Substance Use (N = 95). At 1-month follow-up, the Elevated Suicidality and Elevated Substance Use classes had the highest odds of suicidal behavior. The finding concerning the Elevated Substance Use class suggests it may represent a distinct short-term risk group in military personnel.