Decomposing implicit associations about life and death improves our understanding of suicidal behavior

How do explicit, implicit, and sociodemographic measures relate to concurrent suicidal ideation? A comparative machine learning approach

INTRODUCTION

Suicide is a leading cause of death, and decades of research have identified a range of risk factors, including demographics, past self-injury and suicide attempts, and explicit suicide cognitions. More recently, implicit self-harm and suicide cognitions have been proposed as risk factors for the prospective prediction of suicidal behavior. However, most studies have examined these implicit and explicit risk factors in isolation, and little is known about their combined effects and interactions in the prediction of concurrent suicidal ideation.

METHODS

In an online community sample of 6855 participants, we used different machine learning techniques to evaluate the utility of measuring implicit self-harm and suicide cognitions to predict concurrent desire to self-harm or die.

RESULTS

Desire to self-harm was best predicted using gradient boosting, achieving 83% accuracy. However, the most important predictors were mood, explicit associations, and past suicidal thoughts and behaviors; implicit measures provided little to no gain in predictive accuracy.

CONCLUSION

Considering our focus on the concurrent prediction of explicit suicidal ideation, we discuss the need for future studies to assess the utility of implicit suicide cognitions in the prospective prediction of suicidal behavior using machine learning approaches.

Predicting suicidal ideation in psychiatrically hospitalized veterans using the death/suicide Implicit Association Test: A prospective cohort study

INTRODUCTION

We investigated whether the Death/Suicide Implicit Association Test (D/S-IAT) predicted suicidal ideation (SI) in psychiatric inpatients.

METHODS

One hundred eighty veterans admitted for either SI or suicidal behavior (SB) (the primary sample) (N = 90) or alcohol detoxification (N = 90) completed the D/S-IAT and scales measuring SI. Correlation and regression coefficients were measured between the D/S-IAT (as a full-scale or dichotomized score [D > 0]) and self-reported current or imminent SI (over the next 1-3 days).

RESULTS

In the primary sample, the full-scale D/S-IAT was significantly correlated with the intensity of current SI (r = 0.22, p = 0.04) and especially with wishes to be dead (r = 0.35, p < 0.001). The intensity of imminent SI was significantly predicted by the full-scale (p = 0.02) and dichotomized D/S-IAT score (p = 0.05) in a multiple regression model. However, no significant associations were observed when both the D/S-IAT score and current (present/absent) or imminent SI (occurred/did not occur) were dichotomous measures. In participants receiving alcohol detoxification, the D/S-IAT significantly predicted only wishes to be dead (r = 0.33, p < 0.001).

CONCLUSION

The full-scale D/S-IAT score predicted the current intensity of wishes to be dead in both inpatient samples, and current and imminent SI in participants admitted for SI/SB. The dichotomized D/S-IAT score did not predict the simple occurrence of SI.

Reporting Ethical Procedures in Suicide Prevention Research: Current Status and Recommendations

OBJECTIVE

Ethical concerns frequently arise in suicide prevention research regarding participant safety and confidentiality. Despite a substantial literature on managing and navigating ethical concerns in suicide research, little attention has been paid to the reporting of ethical procedures. Furthermore, standard procedures for reporting ethical risk management procedures have not been developed.

METHOD

A review of the current literature was performed to examine the current state of reporting of ethical procedures within suicide research. Articles published in 2020 (N = 263) from three suicide-focused publications were screened and then coded (n = 131) to identify reporting of procedures for the ethical conduct of research and suicide risk management steps taken by the research teams.

RESULTS

The majority of articles reported ethical review or approval (84.7%) and reported the use of an informed consent process (77.9%). Only 28.2% included risk mitigation procedures. Of those 29.7% of those articles reported conducting risk evaluation, 66.7% reported resource dissemination, and 51.4% reported an intervention.

CONCLUSION

As empirical support for brief interventions accrues, suicide prevention researchers should consider establishing standards for the reporting of procedures to ensure the safety of participants with suicidal risk.HighlightsReporting suicide safety protocols helps ensure high ethical standards in research.Fewer than 1/3 of articles reviewed reported risk mitigation procedures in 2020.Standard procedures for reporting safety protocols in suicide research are needed.

Suicidality and mood: the impact of trends, seasons, day of the week, and time of day on explicit and implicit cognitions among an online community sample

Decades of research have established seasonality effects on completed and attempted suicides, with rates increasing in spring. Little advancements have been made to explain this phenomenon, with most studies focusing almost exclusively on the number of suicide attempts and deaths. Using more than six years of data collected among a US, UK, and Canadian online community sample (N > 10,000), we used newly developed Prophet forecasting and autoregressive-integrated moving average time-series models to examine the temporal dynamics of explicit and implicit self-harm cognitions. We created three groups (past suicide attempters; suicide ideation and/or non-suicidal self-injury; no previous self-harm, suicidal thoughts, or behaviors). We found a general increase of negative self-harm cognitions across the six years and seasonality effects for mood and desire to die, particularly among those who previously made a suicide attempt. Negative explicit self-harm cognitions peaked in winter (December), with implicit self-harm showing a lagged peak of two months (February). Moreover, daily negative self-harm cognitions consistently peaked around 4-5 am, with implicit cognitions again showing a lagged effect (1-hour). Limitations include the volunteer sample not being representative and the cross-sectional nature of the data being unable to separate between-subject and within-subject structural trends in the time series. Our findings show that negative explicit and implicit cognitions precede the rise in suicidal behaviors in spring. We proposed a conceptual model of seasonal suicide risk that may offer fertile ground for theoretical advancements, including implications for clinical risk assessment and public policies regarding the availability of health services.

The death-implicit association test and suicide attempts: a systematic review and meta-analysis of discriminative and prospective utility

Suicide risk assessment involves integrating patient disclosure of suicidal ideation and non-specific risk factors such as family history, past suicidal behaviour, and psychiatric symptoms. A death version of the implicit association test (D-IAT) has been developed to provide an objective measure of the degree to which the self is affiliated with life or death. However, this has inconsistently been associated with past and future suicidal behaviour. Here, we systematically review and quantitatively synthesize the literature examining the D-IAT and suicide attempts. We searched psychINFO, Medline, EMBASE, and the Cochrane Central Register of Controlled Trials (CENTRAL) from inception until 9 February 2021 to identify publications reporting D-IAT scores and suicide attempts (PROSPERO; CRD42020194394). Using random-effects models, we calculated standardized mean differences (SMD) and odds ratios (ORs) for retrospective suicide attempts. We then calculated ORs for future suicide attempts. ORs were dichotomized using a cutoff of zero representing equipoise between self-association with life and death. Eighteen studies met our inclusion criteria (n = 9551). The pooled SMD revealed higher D-IAT scores in individuals with a history of suicide attempt (SMD = 0.25, 95% CI 0.15 to 0.35); however, subgroup analyses demonstrated heterogeneity with acute care settings having lower effect sizes than community settings. Dichotomized D-IAT scores discriminated those with a history of suicide attempt from those without (OR 1.38 95% CI 1.01 to 1.89) and predicted suicide attempt over a six-month follow-up period (OR 2.99 95% CI 1.45 to 6.18; six studies, n = 781). The D-IAT may have a supplementary role in suicide risk assessment; however, determination of acute suicide risk and related clinical decisions should not be based solely on D-IAT performance.