3 - (IOP 32) Characterizing Latent Class Profiles of Young Adolescents Presenting to the Emergency Department

Background: A recent meta-analysis of SI and attempt among youth (younger than age 22) of 365 articles across 149 countries/territories identified only 10 studies that examined youth 13 years and younger. A significant gap exists in our understanding of youth in the transition period between preadolescence to adolescence – ages 12 and 13. The aim of the current investigation was to identify latent classes of suicidal thoughts and behaviors among youth ages 12 to 13, as well as to predict suicide attempt post-baseline.

Methods: We conducted a secondary analysis of data collected from the Emergency Department Screen for Teens at Risk for Suicide (ED-STARS), which aimed to improve risk detection of adolescents presenting to pediatric EDs. For this investigation, we included youth aged 12-13 from both ED-STARS Study 1 (N=2,094 enrolled between June 2015-July 2016) and Study 2 (N=1,233 enrolled between July 2017-October 2018). The following baseline variables were included in class enumeration: thoughts of death in past few weeks, SI in past week, lifetime history of suicidal behavior, non-suicidal self-injury (NSSI) in past year, aggression in past year, hopelessness, connectedness, and bullying or peer victimization. Due to low prevalence, we constructed the outcome variable as either youth- or parent-report of suicide attempt at either follow-up period (three or six months). We conducted initial data reduction in R (version 4.3.1), variable creation in SPSS, and latent class analyses (LCA) using the tidySEM package (0.2.7) in R.

Results: LCA suggested a four-class solution based on lowest Bayesian Information Criterion (32741.43), and the approximate correct model probability exceeded 0.99. Class 1 (“thoughts of death and SI”; 32.4% of sample) and Class 4 (“high SI”; 39.1%) consisted of low proportion of past suicidal behavior, NSSI, hopelessness, aggression, low connectedness, and bullying. However, the “thoughts of death and SI” class had the highest proportion of thoughts of deaths and the “high SI” class had the highest proportion of past-week SI. Class 2 (“moderate symptoms”; 12.8% of sample) had moderate proportion of all indicators. Class 3 (“numerous symptoms, minimal SI”; 12.9% of sample) had high proportion of nearly all variables except past-week SI. Figure 1 presents the variable probabilities by class. Using Bolck et al. (2004) auxiliary variable approach, class membership predicted any reported suicide attempt by six-month follow-up. Specifically, class 1 and class 4 had the highest probabilities of having a suicide attempt after baseline.

Discussion: Classes did not delineate clearly by severity. Rather, classes differed on presence of thoughts of death, SI, past behavior, and NSSI. Classes with the greatest probability of past-week SI had the highest probability of future suicide attempt, highlighting the importance of assessing SI directly among this age group. This study was strengthened by sampling from pediatric EDs and enriching the sample with youth presenting with psychiatric complaints, capturing a sample of higher-risk youth. Future research should examine the transition from preadolescence to adolescence with different enumeration variables, with particular attention to pubertal onset/development.