Exploring Uncoded Events in Suicide Attempt Classification Scales: A Comparison and Narrative Review of Two Interview Measures of Suicide Related Behavior (Board#14)

Abstract : Research

Aims: The measurement of suicidal behavior is critical for the field to draw conclusions about when youth are at risk for suicidal behaviors and how to promote recovery following suicidal behaviors. This study focused on two of the most commonly used interview measures to capture suicidal behavior, the Columbia Suicide Severity Rating Scale (C-SSRS; Posner et al., 2011) and the Suicide Attempt and Self Injury Interview (SASII; Linehan et al., 2006). Both interviews capture a broad range of suicidal behaviors from preparatory behaviors to suicide attempts; however, the coding of suicide attempts differs with the C-SSRS defining a suicide attempt as “a potentially self-injurious act committed with at least some wish to die as a result of the act” whereas the SASII defines it as “any overt, acute, nonfatal self-injurious act where both act and bodily harm or death are clearly intended, have occurred, and results in actual or expected tissue damage, illness, or, if no intervention from others, risk of death or serious injury.” Differences in clinical trial data were identified in operational definitions and resultant classifications of suicide attempts between the C-SSRS and the SASII. Given that finding, we aim to conduct a qualitative review of those attempts which were not coded or unable to be coded by the SASII that were coded as suicide attempts by the C-SSRS.

Methods: Data for the current study was drawn from a multi-site clinical trial Advancing Suicide Intervention Strategies for Teens During High-Risk Periods (ASSIST; R01 MH123442), which evaluates brief interventions to reduce suicide risk among adolescents following discharge from acute care. At their baseline study visit, participants were interviewed using the C-SSRS measure. Our team re-coded the interview data related to suicide attempts gathered in the C-SSRS according to the conventions outlined in the SASII instructions. Our coding team consisted of six researchers, ensuring reliability by having each participant’s data independently coded by at least two team members. Regular coding meetings were held to maintain consistency and resolve any discrepancies. Next steps to be completed include qualitative exploration of narrative explanations around uncoded categories (‘unable to code’ and ‘not coded’) to understand the impact of coding distinctions on outcomes.



Results: The final sample consisted of adolescent participants between ages 11-17 (n=113, Mage=14), who mostly identified as female (56%) and White (81.4%). Out of the 113 adolescents in the sample, 78 had at least one suicide attempt in the C-SSRS data. For the current study, we coded the first, most recent, and most lethal suicide attempts for individuals with multiple attempts. Descriptive analyses revealed discrepancies in suicide attempt classifications between the C-SSRS and the SASII. For cases classified as suicide attempts by the C-SSRS, only 69.2% were similarly classified by the SASII. The SASII demonstrated greater variability in its classifications of suicidal behavior after re-coding: 4.5% of cases were identified as aborted attempts, 2.0% as interrupted attempts, 9.6% of cases that could not be coded due to limited information, and 14.7% of the SASII classifications were ‘not coded’ as they did not meet the criteria to be considered suicide attempts. Those events which could not be coded or were ‘not coded’ will be examined qualitatively prior to final presentation of these data. Preliminary themes for differences in SASII coding include ‘youth refused to answer’ for ‘unable to code’ events and ‘no physiological effect’ under ‘not coded’ events.

Conclusions: Study results indicate that the SASII has more stringent requirements around defining a suicide attempt (e.g., the act results in physiological harm), compared to C-SSRS where any action that meets intent criteria is coded. Thus, nearly 15% of suicide attempts coded by the C-SSRS were not coded as a suicide attempt by the SASII definition and almost an additional 10% were unable to be coded due to limited information. Given the differences in the observed classification in suicide attempts between the C-SSRS and the SASII, there may be different conclusions drawn by researchers in the context of a clinical trial based on the instrument. The C-SSRS might be a more comprehensive tool to identify suicidal thoughts and behaviors of teens compared to the SASII measurement. Qualitatively examining narrative descriptions will likely provide more detailed information about differences between events that are coded by the C-SSRS but not the SASII which may inform future work around whether risk profiles and future suicidal behavior differ based on coding and classification by different measures.

Learning Objective: At the conclusion of this presentation, the participant should be able to discuss differences in two commonly used interview measures of suicidal behavior.