Identifying when an individual is at imminent (i.e., within minutes, hours, or days) risk of suicide is a critical endeavor in suicide risk assessment and prevention. Thus, the validation of warning signs for suicidal intent and behavior, and an understanding of the timeframes in which these clinically actionable, is a national priority. Although an expert panel in 2003 produced a list of ten suicide-specific warning signs that has since been disseminated broadly across suicide prevention websites, public health/educational campaigns, and clinical settings, previous research has failed to sufficiently or prospectively validate these warning signs. Research has been limited, among other methodological constraints, by a lack of psychometrically-sound instruments to assess suicide warning signs repeatedly over time.
This theoretically-driven and clinically-relevant study will address critical gaps in our assessment and understanding of publicized warning signs for suicide through a multifaceted approach. First, an initial item pool will be generated to assess each warning sign that will be iteratively refined based on quantitative and qualitative feedback from content area experts (suicide researchers, clinicians, and those with lived experience) with diverse perspectives across four waves (N = 40). Second, the resultant item set will be repeatedly administered to a sample of 160 severely suicidal adults recruited from both community and clinical settings, who will participate in an ecological momentary assessment (EMA) study with 6 prompts per day across 21 consecutive days. Participants will also report on suicidal intent and behavior at one-month follow-up.
Rigorous evaluation of the resultant measure’s psychometric properties (within- and between-person factor structure, reliability, and validity) will be conducted using (1) multilevel confirmatory factor analysis to test the performance of each individual warning sign’s items, allowing for estimation of within- and between-person factor loadings and reliability coefficients; (2) multilevel exploratory factor analysis to examine the potential higher-order structure of these warning signs across levels; (3) multilevel structural equation modeling to establish initial evidence of validity; and (4) multilevel, linear, and logistic regression models to examine concurrent and prospective relationships between each warning sign and suicide-related outcomes (i.e., suicidal intent at the current and subsequent EMA-assessed time-point, suicidal intent and behaviors at one-month follow-up).
The product developed because of this proposal, a set of psychometrically-sound, brief measures assessing warning signs for suicide, has implications for researchers, clinicians, and the public. Researchers focused on understanding near-term suicide risk can select reliable and valid brief measures for use in intensive longitudinal designs, which are currently lacking. Clinicians can collaborate with their patients using the list of warning signs to incorporate personally relevant warning signs into treatment. Finally, a validated and updated list of warning signs can be disseminated as part of public education campaigns and in community-facing settings (e.g., schools, churches). Following the validation of assessments of warning signs, the research team plans to conduct future studies comprehensively (1) investigating the short-term clinical utility of warning signs in predicting suicide attempts and deaths and (2) understanding variability in the relevancy and utility of various warning signs across populations and settings.
