PTSD and Suicidality in a Longitudinal Nationally Representative Sample of Youth

Background and Purpose: Studies have reported associations between PTSD and suicidality (ideation and attempts) in adolescents. To date, there has been no prospective, longitudinal study with a nationally representative U.S. sample examining the relationship between suicidality and PTSD diagnosis. Additionally, risk and protective factors for suicidality among young people are not well understood. This study examines the factors associated with suicide ideation and attempts over time. It specifically investigates the relationship between PTSD diagnosis and suicidal thoughts or attempts.

Methods: The data source is Add Health, a longitudinal, and nationally representative survey of adolescents in the U.S. Students in grades 7-12 were interviewed in schools in Wave I and re-interviewed in homes in Waves II, III and IV. The survey asked about suicidal thoughts and attempts (with or without serious injury) in the year before the interview. This study included all who reported ever having a PTSD diagnosis at Wave 4 (n = 450) and a random sample control group of 450 without a PTSD diagnosis. Time (years since baseline), PTSD diagnosis, and suicidal friends or family were measured across all four waves, and the other predictors were measured from baseline only. Each predictor and their PTSD diagnosis interaction were examined for suicidal ideation, suicide attempts, and suicide attempts needing medical care, over time, using longitudinal logistic regression.

Results: Significant predictors for suicidal ideation were PTSD diagnosis (OR = 3.08, p < .001), time (OR = 0.95, p < .001), adult support (OR = 0.74, p < .001), suicidal friends or family (OR = 3.10, p < .001), and depression (OR = 1.06, p < .001). Gender, adult support, and depression had significant interactions with PTSD. PTSD showed stronger risk for males than females (p < 0.01). PTSD diagnosis decreased the risk of depression (p < 0.001) and the protective effect of adult support (p < 0.001).

Significant predictors for suicidal attempts were PTSD diagnosis (OR = 3.53, p < .001), time (OR = 0.92, p < .001), suicidal friends or family (OR = 4.10, p < .001), hopelessness (OR = 1.57, p < .05), and depression (OR = 1.06, p < .001). Adult support and parental support showed significant interactions with PTSD diagnosis, where PTSD diagnosis decreased the protective effect of adult support (p < .001) and parental support (p < .01).

Significant predictors for suicide attempt with injury were PTSD diagnosis (OR = 2.66, p < .001), time (OR = 0.98, p < .05), and suicidal friends or family (OR = 3.53, p < .001).

Conclusions and Implications: Risk and protective factors contributed differentially based on morbidity. Overall, PTSD diagnosis was a significant risk factor and having suicidal friends or family was the most robust predictor of suicidality over time. The interactive effects showed that for both suicide ideation and attempts without injury, having a PTSD diagnosis significantly decreased the protective effect of adult support over time. These findings suggest elements to include in suicide preventions.