Suicide is the leading cause of trauma-related death in the US, and the second-leading cause of death overall in people aged 10–34 years (1). Yet, effective preventive measures have been elusive. Suicide is transdiagnostic (2, 3), crossing disorders of affect, behavior, substance abuse, and trauma (4–9), and frequently occurs in the absence of a conventional psychiatric diagnosis (10). Many, perhaps most, suicides are first attempts (11) that, even if previously planned, are completed in a window of ten minutes during a suicidal crisis (12). Suicidal crises are difficult to predict based on recent history (13). Current suicidal ideation (SI) fluctuates in a nonlinear manner during crises (14, 15), consistent with interacting short- and long-term behavioral mechanisms (16) and susceptibility to hyperarousal (5, 6). Practical prediction of risk must address interactions of long- and short-term behavior regulation that underlie suicidal behavior, and characteristics differentiating suicidal ideation from overt suicidal behavior (17, 18). As a proxy for suicide, we examined short- and long-term behavior regulation in people who survived a medically severe suicide attempt (MSSA), compared to people with similar demographics and symptoms who had not attempted suicide (NA) despite suicidal ideation.
MSSAs entail high risk for premature mortality beyond suicide. All-cause mortality in MSSA over five years was increased 15-fold in men and nine-fold in women, with a 5402-fold increase in suicide and 2480-fold increase in homicide mortality compared to the general population (19). Risk was highest during the first year after a MSSA (19, 20) and could persist for at least 14 years (20, 21). Post-hospital survival time after MSSA correlates negatively with trait-impulsivity (22), indicating that mechanisms underlying MSSAs are related to regulation of action (23) and susceptibility to hyperarousal (24). This is consistent with a two-stage model based on NIMH Research Diagnostic Criteria (RDoC), where persistent latent susceptibility to suicidal and related behavior (Suicidal Behavior Disorder, SBD) (25, 26) generates bouts of severe and fluctuating overt suicide risk (Suicide Crisis Syndrome, SCS) (27, 28). The unpredictable behavior in suicidal crises requires preventing SCS by identifying and treating SBD, susceptibility to suicidal crises. While most suicide attempts are survived, survived attempts have increased subsequent morbidity and premature mortality (19–22), including suicide (29, 30).
Figure 1 shows interrelating factors underlying suicidal behavior. The fluctuating, nonlinear nature of suicidal behavior is consistent with interacting mechanisms with different time courses (14). Long-term characteristics, potentially related to sensitization to addiction (31), trauma (32), or highly recurrent illness (33), could underlie latent risk for behavior dysregulation and suicidality (34–37). Trait-impulsivity is associated with susceptibility to sensitization in humans (38), consistent with its relationship to hyperarousal (24). In turn, sensitization increases state-related action-impulsivity, bypassing adaptive behavior regulation and facilitating aggressive and other suicide-related behavior (39, 40). MSSA might require transition from latent longer-term susceptibility (SBD) to fluctuating suicidality (17) with emergence of latent self-destructive characteristics through impaired action regulation and hyper-arousal (SCS) (34–41). This transition is compatible with an epigenetic mechanism with rapid activation of latent state-dependent characteristics (42).
To address these questions, we measured 1) characteristics associated with potential sensitization to trauma or substance use, hypothesizing that sensitization or a similar mechanism is necessary but not sufficient for suicidal behavior, producing a state of latent suicide risk; 2) trait-impulsivity, predisposing to hyperarousal and sensitization (24, 35), resulting in susceptibility to evoked or action-impulsivity, bypassing adaptive behavior by interfering with pre-attentional stimulus processing (35, 39) to facilitate self-destructive or maladaptive behavior (39); and 3) behavior dyscontrol, including aggression and psychiatric symptoms.
We recruited survivors of an MSSA and psychiatrically similar comparison subjects with history of suicidal ideation, but no history of suicide attempt (NA). Using Bayesian logistic regression and path analysis we investigated the hypothesis that suicide risk results from interacting long-term behavior regulation including sensitization to stress and/or addiction, and disruption of immediate behavior regulation facilitated by impulsivity. Suicidal ideation is prominent in many psychiatric disorders, and can predict eventual suicide if severe enough, but it is highly nonspecific and does not predict short-term risk (43). Therefore, we compared characteristics potentially related to suicide in terms of their relationship to suicidal ideation at its worst to those directly related to MSSA without correlation to SSI-W, as an initial step to distinguish relationships that were independent of suicidal ideation from those mediated by SSI-W.