Impact of behavioral inhibitory control and startle reactivity to uncertain threat on youth suicide risk

Abstract

Poor inhibitory control and exaggerated threat reactivity are two well-established risk factors for suicide. Theory suggests that these two factors may interact to influence suicide risk, although few studies have directly tested these relationships. In the present study, we examined the unique and interactive effects of inhibitory control (IC) and threat reactivity on self-reported suicide risk in a sample of 132 youth, ages 16–19. The stop signal task was used as a behavioral index of IC. Threat reactivity was captured using a modified version of the No-Predictable-Unpredictable threat paradigm that includes threat of predictable (P-) and unpredictable (U-) mild electrical shock. Startle eyeblink potentiation was measured throughout the task as an index of aversive responding. All participants completed a battery of well-validated self-report measures including current suicide risk. Hierarchical linear regression analyses controlling for age and sex revealed no main effects of IC or threat reactivity. However, there was a significant IC by reactivity to uncertain threat (U-threat) interaction. At lower levels of IC, greater startle reactivity to U-threat was associated with greater suicide risk. At higher levels of IC, there was no association between reactivity to U-threat and suicide risk. These results suggest that individual differences in IC and reactivity to U-threat interact to influence suicide cognitions, shedding light on potential subgroups of individuals who might be at elevated risk.

Introduction

Suicide constitutes a massive public health burden in the United States. Despite a global decrease in suicide rates since the year 2000, the suicide rate in the US continues to climb (Hedegaard, 2020; Naghavi, 2019). Likewise, suicide rates among young people in the US have steadily increased since 2007; as of 2021, suicide is now the second leading cause of death in people aged 10–24 years (Curtin & Garnett, 2023; Garnett, 2022). Alarmingly, suicide of youth ages 15–24 now accounts for 19.5% of total deaths by suicide in the US, up from 14.2% in 2007 (WISQARS Data Visualization (cdc.gov)). There is an urgent need to better understand who may be at increased risk of suicide, particularly among youth, to aid in the development of targeted and effective prevention and intervention strategies.

One risk factor for suicidal ideation and behavior is poor inhibitory control (IC). Behaviorally, IC is individual’s ability to choose to not take overt behavioral action in response to a cue stimulus. IC is a facet of impulsivity that reflects the tendency towards impulsive action (MacKillop et al., 2016). In the context of suicide, IC is involved in the inhibition of suicidal thoughts (Marzuk et al., 2005; Westheide et al., 2008) and behaviors (Sandoval et al., 2021) that are often elicited by strong negative emotions and the motivational drive for escape and relief (Baumeister, 1990). Indeed, prior work has demonstrated that poorer IC is associated with suicidal ideation (Keilp et al., 2013) and suicide attempts (Harfmann et al., 2019) across the lifespan (Connell et al., 2019). Situational escape and relief from negative emotions are the two most frequently endorsed motives for attempting suicide (Bryan et al., 2013; May et al., 2020; Tezanos et al., 2021). As a part of the larger executive functioning network, the IC system plays a role not only in inhibiting motor function (suicidal behavior), but in inhibition and regulation of suicidogenic thoughts and cognitions (Logan & Cowan, 1984). Notably, the relationship between IC and suicide risk may be especially salient in adolescence, which is a developmental period marked by emotional lability and under-developed impulse control (Casey et al., 2013).

Likewise, exaggerated threat reactivity has been identified as a risk factor for suicidal ideation and future suicide attempts in both adults and youth (Hazlett et al., 2016; Lieberman et al., 2020; Yang et al., 2023). Threat reactivity is a broad construct that encapsulates physiological and emotional responses to internal and external threats (Denefrio & Dennis-Tiwary, 2018). Individuals who are hypersensitive to threat exhibit increased physical, cognitive, and affective responses to perceived stressors. More specifically, research has shown that individuals who are sensitive to threat tend to overestimate the probability and consequences of stressors, exhibit sustained anticipatory arousal, and exhibit chronic negative affect (Grupe & Nitschke, 2013; Löw et al., 2008; Simmons et al., 2006). As such, these individuals are prone to frequent and intense negative emotions, which exacerbate suicide risk via negative reinforcement processes (Bryan et al., 2013; Millner et al., 2019). Youth may be particularly sensitive to these negative reinforcement processes in light of their decreased limbic-prefrontal functional connectivity and heightened emotional reactivity (Arain et al., 2013; Tzschoppe et al., 2014). Importantly, suicide risk has been linked to threat reactivity as measured using neural (Yang et al., 2023), psychophysiological (Ballard et al., 2014; Hazlett et al., 2016; Lieberman et al., 2020), and self-report assays (Venables et al., 2015), providing multimodal evidence for the relationship between threat sensitivity and suicide risk.

The aforementioned studies examined how poor IC and greater threat reactivity independently predict increased suicide risk. However, few studies have explored how these individual difference factors may interact. One notable exception is a study by Venables and colleagues (2015), which found evidence of an interaction between self-reported IC and threat sensitivity on suicide risk. Specifically, the authors found that in both outpatient clinical and male military cohorts, poor IC and greater threat reactivity both independently predicted suicide risk; however, these main effects were qualified by a significant two-way interaction. At high levels of IC, greater threat reactivity was associated with only a slight increase in suicide risk. However, at low levels of IC, greater threat reactivity was associated with a dramatic increase in suicide risk. The authors concluded that poor IC and high threat reactivity interact to increase dispositional risk for suicidal behavior via poor regulation of threat-induced self-injurious thoughts and behaviors (Venables et al., 2015).

The findings from Venables et al. 2015 provide valuable initial evidence regarding the synergistic impact of IC and threat sensitivity on suicidal thoughts and behaviors. Notably, this study utilized self-report to capture subjective perceptions of IC and threat sensitivity. This is important because self-report has certain limitations, including recall bias, systematic responding, and self-deception (Robins, 2007). Objective laboratory measures afford the opportunity to overcome these challenges by providing objective measurements of behavior and directly probing the target systems. One key example is the stop-signal task, which is a behavioral measure of latency of response inhibition (stop-signal reaction time; SSRT) and is used to assess IC. Similarly, a wide variety of behavioral threat tasks exist, all of which expose the participant to a mild aversive stimulus (e.g., a negative image (IAPS; Bradley & Lang, 2017)) or mild electric shock (Melzig et al., 2009) and collect psychophysiological measures of aversive responding such as startle eyeblink potentiation—a well-validated, cross-species reflex that reflects in-the-moment defensive reactivity (Lang, 1995; Lang et al., 1990).

As previously mentioned, the construct of threat reactivity is broad. Cross-species research shows that different forms of threat elicit different affective states (Barlow, 2000; Schmitz et al., 2011; Urien & Bauer, 2022). One important distinction is between threats that are predictable (P-threat) vs. unpredictable (U-threat). There is a growing body of literature suggesting that P- and U-threat can elicit qualitatively distinct aversive states. Specifically, predictable aversive stimuli have been shown to elicit a phasic response to an identifiable stimulus (often labeled fear), while unpredictable aversive stimuli elicit a general feeling of apprehension not tethered to a clearly identifiable stimulus (often labeled anxiety; Barlow, 2000; Davis, 1998). These responses have been shown to be pharmacologically distinct (Grillon et al., 2006, 2011) and mediated by overlapping, but separable, neural circuits (Alvarez et al., 2011; Davis, 2006).

This distinction between P- and U-threat is significant because converging research indicates that individual differences in response to U-threat relate to a variety of risk behaviors, including suicidal thoughts and behaviors. For instance, previous work from our lab has demonstrated that present and past suicidal ideation is associated with greater psychophysiological reactivity to U-threat but not P-threat (Lieberman et al., 2020). This suggests that risk for suicidal ideation may be associated specifically with ambiguous, anxiety-provoking threats rather than threat in general.

Therefore, the present study sought to expand upon prior research by examining the relationship between IC, U- and P-threat reactivity, and suicide risk in youth. More specifically, we sought to extend the findings by Venables et al. 2015 by focusing on behavioral markers of IC and threat sensitivity, parsing sensitivity to unpredictable versus predictable threat, and utilizing a sample 16–19 year olds. To measure IC, we used the well-validated stop signal task (Verbruggen et al., 2008). To capture threat reactivity, we administered a modified version of Grillon’s No-Predictable-Unpredictable threat paradigm (Schmitz & Grillon, 2012). Startle eyeblink potentiation was recorded during the NPU task as an index of aversive responding. Suicide risk was assessed using the Suicide Cognitions Scale – Revised (Bryan, May, et al., 2022), which taps into the schematic underpinnings of a suicidal belief system such as pervasive feelings hopelessness, perceived burdensomeness, thwarted belongingness, defeat, and unbearability (Bryan, May, et al., 2022; Rudd & Bryan, 2021). Unlike other suicide measures, the SCS-R successfully discriminates between suicide ideators and attempters (Bryan et al., 2020) and predicts future suicide attempts (Bryan et al., 2014). Based upon previous IC/suicide and U-threat/suicide literature, we hypothesized that IC, U-threat, and suicide risk would interact such that individuals with low IC and high U-threat reactivity are at increased risk for experiencing suicidal cognitions.

2. Methods

2.1. Participants

One hundred and thirty two participants between the ages of 16 and 19 years were recruited as part of a larger study examining the neurobiology of the association between trauma exposure, psychiatric symptoms, and alcohol use in young adults. Individuals were recruited via social media advertisements and flyers posted in the Columbus, Ohio community, including nearby high school and college campuses. Participants were enrolled into one of two groups: 1) lifetime history of interpersonal trauma exposure (i.e., physical assault, sexual assault, or immediate family violence) or 2) no lifetime history of interpersonal trauma exposure. Participants were required to have had minimal alcohol exposure at enrollment (i.e., self-reported lifetime consumption of >1 but <100 standard alcoholic drinks) but be at high risk for developing alcohol problems by virtue of self-reported affiliation with risky peers and having access to alcohol. Individuals with active suicidal intent who were in need of emergency services were excluded from the study for safety reasons; however, all other individuals with suicidal thoughts and behaviors were included. Exclusionary criteria included any major active medical or neurological illness, lifetime or current incidence of manic/psychotic symptoms, deafness, traumatic brain injury, current psychotropic medication use, lifetime history of alcohol or substance use disorder, and pregnancy. Individuals were instructed to abstain from drugs and alcohol at least 24 hours prior to the lab assessments, which was verified via screening of urine and breath alcohol content. All study procedures were approved by The Ohio State University Institutional Review Board. Participants provided written informed consent or assent with parental consent and were monetarily compensated for their time.

2.2. Suicide Risk and Other Clinical Measures

Participants first completed an initial screening session involving a battery of validated self-report measures. Suicide risk was determined using the Suicide Cognitions Scale – Revised (SCS-R; Bryan, May et al., 2022). The SCS-R is a 16-item self-report measure used to evaluate suicidogenic thoughts and beliefs including hopelessness, perceived burdensomeness, entrapment, and unbearability (Bryan, May et al., 2022). Numerous studies have demonstrated that SCS-R scores are superior to traditional measures of suicide risk in predicting suicidal behaviors (Bryan et al., 2014; Bryan, May et al., 2022; Rudd & Bryan, 2021). SCS-R scores also differentiate suicide attempters from non-attempters (Bryan, May et al., 2022, Bryan, Thomsen et al., 2022; Rudd & Bryan, 2021). Respondents rate their level of agreement with each statement on a 5-point Likert scale ranging from 0 (strongly disagree) to 4 (strongly agree). Item responses are summed to obtain a total SCS-R score ranging from 0 to 64, with higher scores reflecting increased vulnerability to suicidal behavior.

Additionally, the Beck Depression Inventory II (BDI-II; Beck et al., 2011) was collected as a self-report measure of depression. The BDI-II is a well-validated and widely used measure of depression in adolescents and adults and has been shown to reflect depression across age groups (Osman et al., 2004, 2008). The scale consists of 21 items graded on a 4-point Likert scale. Items are summed for a total score from 0 to 63, with higher scores reflecting more severe depressive symptoms.

Lifetime history of trauma exposure was assessed using the trauma screener from the UCLA PTSD Reaction Index (RI) for DSM-5 (Pynoos and Steinberg, 2015). The UCLA PTSD RI is a clinician-administered assessment designed to capture a wide range of traumatic and stressful life events in children and adolescents. The UCLA PTSD RI trauma history profile asks about lifetime exposure to 18 different traumatic and stressful events. Specifically, exposure to the following events was assessed: neglect/maltreatment, sexual abuse, physical abuse, emotional abuse, domestic violence, community violence, war/political violence, life-threatening medical illness, serious accident, school violence, disaster, terrorism, kidnapping, sexual assault/rape, interpersonal violence, bereavement, separation from caregiver, and impaired caregiver. All responses were made on a yes versus no checklist and a total count of trauma exposure was created by summing responses across all 18 events.

2.3. Stop-Signal Task and Behavioral Data Processing

Participants were presented with the computerized STOP-IT task, a Windows executable stop-signal program (Verbruggen et al., 2008). The task presents the participant with two visual go stimuli, a square and a circle, and one auditory stop-signal stimulus. Participants were instructed to respond to the visual stimuli by pressing the left key when they saw a square and the right key when they saw a circle (go trials). Participants were then informed that on occasion, they would hear a beep in their headphones; this beep (stop-signal) cued them to inhibit, or stop, their response to the presented shape (stop trials). It was made clear that participants were not to anticipate the stop-signal and should instead respond to the shape as quickly as possible, only halting their response when they heard the stop-signal. They were also informed that it would be harder to inhibit their response on some stop-trials than others, depending on the delay between the go stimulus and the stop-signal.

The stop-signal was presented on 25% of the trials, for a total of 56 stop trials out of 224 total trials. The experiment began with a practice block of 32 trials, wherein the stop-signal was presented after variable delay times (stop-signal delay; SSD). SSD was initially set at 250 ms and was continuously adjusted via a step-up step-down algorithm until the participant could successfully inhibit responding on 50% of trials. SSD was decreased by 50ms after an unsuccessful inhibition and increased by 50ms after a successful inhibition. The practice block was followed by 3 test blocks of 64 trials each. After each block, participants received feedback on their performance, including 1) number of incorrect responses on go trials (e.g., the participant responded but pressed the wrong button), 2) number of missed responses on go trials (e.g., the participant did not respond when they should have), 3) average reaction time on go trials, and 4) percentage of stop trials where they successfully inhibited their response. Data was compiled and analyzed using Verbruggen et al.’s (2008) ANALYZE-IT program, which was included in the download with their STOP-IT program. The program calculated the following values for each participant: 1) mean probability of participant’s responding on stop-signal trials; 2) mean stop-signal delay time (SSD); 3) mean stop-signal reaction time (SSRT); 4) mean reaction time of accidental responding on stop-signal trials; 5) mean reaction time on go trials; 6) mean percentage of correct responses on go trials; 6) mean percentage of missed responses on go trials; and 7) a z-score and the corresponding p-value to determine whether participants inhibit significantly more or less than 50% of the time. The SSRT (calculated as the mean reaction time on go (no signal) trials minus the mean SSD), or the latency of the inhibitory response, was then included as the behavioral measurement of inhibitory control.

2.4. NPU Task and Startle Data Processing

The NPU startle task and procedures have been extensively described by our group (Gorka, 2020; Gorka et al., 2016, 2017). Shock electrodes were first placed on participants’ left wrist, and a shock work-up procedure was completed to identify the level of shock intensity each participant described as “highly annoying but not painful” (i.e., 1–5 mA). Participants then completed a 2-min startle habituation task, wherein they were presented with 6 white-noise startle probes; this habituation phase serves to prevent major startle habituation responding during the first trials of the NPU task. The task itself was modeled after Grillon and colleagues’ NPU threat task and included three within-subject conditions: no shock (N), predictable shock (P), and unpredictable shock (U). Text at the bottom of the computer monitor informed participants of the current condition. Each condition lasted 145 s, during which a 4-s visual countdown (CD) was presented six times. The interstimulus intervals (ISIs; i.e., time between CDs) ranged from 15 s to 21 s, during which only the text describing the condition was on the screen. No shocks were delivered during the N condition. A shock was delivered every time the CD reached 1 during the P condition. Shocks were delivered at random during the U condition. For all conditions, startle probes were administered during both the CD (1–2 s following CD onset) and ISI (4–13 s following ISI onset). The time interval between a shock and the following startle probe was always greater than 10 s to ensure that the subsequent startle response was not significantly affected by an immediately preceding stimulus. Each condition was presented two times in a randomized order, counterbalanced. Participants received 24 total electric shocks (12 in P, 12 in U) and 60 total startle probes (20 in N, 20 in P, 20 in U).

Startle data were acquired using BioSemi Active Two system (BioSemi; Amsterdam, The Netherlands), and stimuli were administered using Presentation (Albany, CA). Electric shocks of participants’ calibration voltage lasted 400 ms; acoustic startle probes, which consisted of 103-dB bursts of white noise lasting 40 ms, were presented via headphones. Startle responses were recorded from two 4-mm Ag/AgCl electrodes placed over the orbicularis oculi muscle below the left eye. The ground electrode was located at the frontal pole (Fpz) of an electroencephalography cap that participants wore as part of the larger study. Data were collected using a bandpass filter of DC-500 Hz at a sampling rate of 2000 Hz.

Blinks were processed and scored according to published guidelines (Blumenthal et al., 2005): applied using a 28 Hz high-pass filter, rectified, and then smoothed using a 40 Hz low-pass filter. Peak amplitude was defined within 20–150 ms following the probe onset relative to baseline (i.e., average activity for the 50-ms preceding probe onset). Each peak was identified by software but examined by hand to ensure acceptability. Blinks were scored as nonresponses if activity during the post stimulus timeframe did not produce a peak that was visually differentiated from baseline. Blinks were scored as missing if the baseline period was contaminated with noise or movement artifact, or if a spontaneous or voluntary blink began before minimal onset latency. Blink magnitude values (i.e., condition averages include values of 0 for nonresponses) were used in all analyses. To balance the number of startle probes across all three conditions and match the conditions on visual stimuli, we only used the startle probes from the task CDs for each of our condition averages. This is consistent with many published papers from our lab (Gorka, 2020; Gorka et al., 2013; Kreutzer & Gorka, 2021; Manzler et al., 2022; Radoman et al., 2019). Using a repeated measures analysis of variance (ANOVA), we confirmed that the NPU paradigm elicited the intended task effects (effect of condition: F[2, 262]= 71.98, p<.001; η_p² = .36; N_CD M =100.4± 84.2; P_CD M =118.5± 87.5; U_CD M =139.7± 93.0) such that raw startle magnitude was significantly greater during U_CD (F[1, 131= 114.85, p<.001; η_p² = .47) and P_CD (F[1, 131= 33.41, p<.001; η_p² = .20) relative to N_CD. Startle magnitude was also greater during U_CD compared with P_CD (F[1, 131= 48.94, p<.001; η_p² = .27).

As discussed in Meyer et al. (2017), traditional subtraction-based difference measures are problematic for highly correlated variables. Therefore, to quantify the difference between threat and no-threat trials, we calculated a standardized residual score for U-threat (U-threat_resid) and P-threat (P-threat_resid) by saving the variance leftover (i.e., the amount of variability in a dependent variable [DV] that is not explained by an independent variable [IV]) in two simple linear regressions, where the N_CD (IV) was entered to separately predict the U_CD and P_CD (DVs). The U_CD residual score was used as the primary independent variable.

2.5. Data Analysis Plan

To test our hypothesis, we performed a hierarchical linear regression analysis with SCS-R scores as our outcome variable. Self-reported biological sex and age were included as covariates and entered in Block 1. Lifetime trauma exposure captured via the UCLA PTSD RI was also included as a covariate in Block 1 given that our sample was comprised of trauma-exposed youth and prior studies have demonstrated that trauma is robustly associated with suicide risk (Ásgeirsdóttir et al., 2018; Park et al., 2020). The main effects of IC and startle reactivity to U-threat were entered in Block 2. The IC by startle reactivity to U-threat interaction was entered in Block3.

Biological sex was dummy coded and all continuous variables were mean-centered. A significant two-way interaction was followed up using a standard simple slopes approach (Aiken et al., 1991). Specifically, the moderator was re-centered at 1 SD above the mean for “low IC” and 1 SD below the mean for “high IC.” Two new interaction terms were created and post-hoc additional follow-up linear regression models were run at high and low levels of IC. All analyses were conducted using SPSS v28 (IBM).

Post-hoc we conducted a series of sensitivity analyses to explore whether similar patterns of results were observed with startle reactivity to P-threat (as an independent variable) and current depression severity (as an outcome). Startle reactivity to P-threat reflects acute fear and is considered a distinct form of threat sensitivity (Davis et al., 2010). Meanwhile, depression and thoughts of suicide are closely related yet distinct outcomes (Tan & Wong, 2008; Ullman & Najdowski, 2009). Two additional hierarchical linear regression analyses were used to examine the impact of these variables. The final sample size was dictated by the aims of the larger project though a post-hoc sensitivity analysis was conducted using G*Power (Faul et al., 2009). With a sample size of 132, power 1-β = 0.80, and α = 0.05, the study was powered to detect at least a small-to-medium effect size of f²= .07.

3. Results

3.1. Sample Characteristics

The characteristics of the sample are presented in Table 1. The sample had a range of lifetime psychopathology including high prevalence rates for major depressive disorder. Regarding suicide risk, 12.1% of participants made at least one prior suicide attempt and 3.8% reported two or more prior attempts.

Table 1.

Participant Demographics and Characteristics

Demographics
Age (years)	18.1 (1.0)
Sex (% female)	68.2%
Ethnicity (% Hispanic)	10.6%
Race
White	62.9%
Black	12.1%
Asian	9.8%
American Indian or Alaskan Native	0.0%
Biracial, Other or Unknown	15.2%
Lifetime SCID Diagnoses
Major depressive disorder	53.8%
Generalized anxiety disorder	3.0%
Social anxiety disorder	18.9%
Panic disorder	0.8%
Specific phobia	2.3%
Post-traumatic stress disorder	21.2%
Alcohol use disorder	0.0%
Substance use disorder	0.0%
Suicide Risk and Clinical Variables
Past Suicide Attempt (% Yes)	12.1%
Suicide Cognitions Scale-Revised	10.4 (10.6)
BDI-II Total Score	12.2 (9.8)

Note. BDI-II = Beck Depression Inventory-II.

3.2. Unique and Interactive Effects of IC and Threat Sensitivity

The results of the hierarchical linear regression analysis are presented in Table 2. Younger age and higher lifetime exposure to trauma were uniquely associated with increased suicide risk. There was no significant main effect of IC (p = .14) or startle reactivity to U-threat (p = .34) on suicide risk. However, there was a significant 2-way interaction. Follow-up analyses revealed that at lower levels of IC, greater startle reactivity to U-threat was associated with greater suicide risk (β = .20, t = 2.33, p = .02). At higher levels of IC, there was no association between startle reactivity to U-threat and suicide risk (β = −.09, t = −0.76, p = .45). This interaction is illustrated in Figure 1.

Table 2.

Linear Regression Analyses Testing Unique and Interactive Effects of Inhibitory Control and Reactivity to Uncertain Threat on Suicide Risk

	β	t	p-value	Adj. R2	R2 Change	F Change	p-value Change
Block 1				.05	.08	3.45	.02
Age	−.19*	−2.23	.03
Sex	.02	.23	.82
Lifetime Trauma Exposure	.21*	2.42	.02
Block2				.06	.02	1.55	.22
Inhibitory Control	.13	1.50	.14
Startle reactivity to U-threat	.08	.96	.34
Block 3				.09	.03	4.20	.04
IC x Startle reactivity to U-threat	.19	2.01	.04

Note. U-threat = anticipation of unpredictable threat; IC = inhibitory control

Figure 1:

The effect of reactivity to U-threat on suicide risk at high and low levels of inhibitory control. “High” and “Low” inhibitory control refer to individuals scoring 1 SD below or 1 SD above the mean stop-signal reaction time, respectively. Gray shaded areas illustrate 95% confidence intervals.

3.3. Sensitivity Analyses

In the P-threat model, there was no main effect of IC (β = .12, t = 1.39, p = .17) or startle reactivity to P-threat (β = .10, t = 1.21, p = .23) on suicide risk. There was also no IC by P-threat reactivity interaction (β = .12, t = 1.42, p = .16). In the depression model, there was no main effect of IC (β = .17, t = 1.93, p = .06) or startle reactivity to U-threat (β = .01, t = 0.16, p = .87) on depression severity. There was also no IC by U-threat reactivity interaction (β = .12, t = 1.29, p = .20).

4. Discussion

The present study explored the unique and interactive effects of IC and reactivity to U-threat on youth suicide risk. Our work expanded upon existing studies by utilizing behavioral and psychophysiological measures of IC and U-threat reactivity and testing the relationship between these behavioral factors and suicide risk in a youth cohort. We found that there were no main effects of IC or reactivity to U-threat; however, there was a significant IC by U-threat reactivity interaction. At lower levels of IC, greater startle reactivity to U-threat was associated with greater suicide risk. At higher levels of IC, there was no association between reactivity to U-threat and suicide risk. Our sensitivity analyses indicated that this relationship is robustly related to U-threat, and there were no unique or interactive effects with P-threat. Additionally, we found that these predictors are more related to suicide risk than depression. This suggests that levels of IC uniquely influence the relationship between reactivity to U-threat and suicide, shedding light on potential subgroups of individuals who might be particularly at risk for suicide.

Our findings revealed that there were no main effects of IC or reactivity to U-threat on suicide risk. The absence of main effects conflicts with the findings of Venables et al. (2015), who found unique, independent effects of both poor IC and high threat reactivity on suicide risk using self-report scales. This discrepancy may point to inherent methodological differences between self-report and behavioral paradigms. Self-report scales typically capture one’s own perception of various traits and behaviors. Meanwhile, objective laboratory tasks elicit and record behavioral responses to relevant stimuli. In light of the present findings, it is possible that objective IC and threat reactivity do not exert robust independent effects on youth suicide risk, and that each measure, on its own, may have limited unique predictive validity. Another key difference between the study conducted by Venables et al. (2015) and the present study are the sample demographics. Venables et al. (2015) utilized two cohorts: one outpatient psychiatric cohort with a mean age of 26.7 years, and one Finnish soldier cohort with a mean age of 18 years. Our cohort, which was recruited from the greater Columbus, OH area, has a mean age of 18.1 years. Though the latter of Venables et al.’s cohorts is relatively similar to ours, the outpatient psychiatric cohort’s contribution to their sample may contribute to the differences between their findings and that of the present study. Finally, and perhaps most importantly, the specificity of the measurement differs greatly between the present study and that conducted by Venables et al. (2015). The threat and IC scales utilized by Venables et al. (2015) were intentionally broad to reflect multiple aspects of each construct. In contrast, our behavioral assessments were relatively narrow, particularly our measure of threat reactivity. The NPU task was developed to disentangle sustained (uncertain) from acute (predictable) threat. Prior work from our group has demonstrated that increased startle potentiation to U-threat is related to both current and lifetime suicidal ideation in a cohort of young adults (Lieberman et al., 2020). Thus, the present study builds upon these findings by identifying an interaction between U-threat reactivity and IC on suicide risk in youth.

Additionally, results revealed a significant IC by U-threat reactivity interaction. At lower levels of inhibitory control (only), greater startle reactivity to U-threat was associated with greater suicide risk. No such relationship was shown between P-threat and IC. Previous research has suggested that individuals with high reactivity to U-threat experience chronic anticipatory anxiety and greater subjective distress given the inherent uncertainty of daily life, and that this distress increases risk for suicide by means of avoidance-based escape (Bryan et al., 2013; Millner et al., 2019). In the context of poor IC, this risk may be more pronounced. Individuals who exhibit poor IC have difficulty exhibiting cognitive control over of a wide array of maladaptive motivated behaviors, including escape-motivated suicidal behavior brought on by acute distress (Devos et al., 2015; Harfmann et al., 2019; Lavagnino et al., 2016; Quach et al., 2020; Squeglia et al., 2014; Trotzke et al., 2020; Weafer & Fillmore, 2008; Westheide et al., 2008; Wu et al., 2013). As such, individuals who have high U-threat reactivity and poor IC may struggle to cognitively and behaviorally regulate periods of intense negative affect, amplifying the desire for escape via suicide. This aligns with the functional model of suicide (Allan et al., 2023; Bryan et al., 2013), which posits that negative reinforcement processes and the motivational drive to escape overwhelming distress are critically involved in the emergence and reoccurrence of suicidal cognitions and behaviors.

The present study elucidates the importance of providing threat-sensitive individuals with tools to manage their distress, specifically if they also struggle with inhibiting impulsive behavior. One approach to treating this subgroup is mindfulness training, which has been shown to improve inhibitory control in both a healthy adult sample (Pozuelos et al., 2019) and a youth sample of male juvenile offenders (Ron-Grajales et al., 2021). Additionally, inhibitory control training in the form of repeated behavioral task practice has been shown to increase IC, as evidenced by improved performance on behavioral IC tasks following training (Spierer et al., 2013) and changes in the activation of the inferior frontal gyrus and dorsolateral prefrontal cortex, brain regions thought to underlie top-down IC functioning (Berkman et al., 2014). Especially given the neuroplasticity of the age group, IC training is a promising avenue for helping offset suicide risk in threat-sensitive youth.

The present study has multiple strengths, including a youth sample and well-validated behavioral measures of IC and threat reactivity. That said, the study also has several limitations. First, the selected sample was recruited to have a wide range of trauma histories and be at-risk for the onset of alcohol problems. Although this cohort is known to be at elevated risk for suicide, there may be overlapping mechanisms implicated in risk for alcohol problems and suicide that impact the findings and limit the generalizability of the results to other samples. Related, all participants in this study were free from all psychoactive medications, including those prescribed for mental health conditions. Additionally, individuals with bipolar disorder and psychosis were excluded from the sample. This facilitates interpretation of the behavioral data, but may also have inadvertently excluded some of the individuals most vulnerable to suicide. Finally, we solely examined suicide cognitions in this study, which is a specific kind of suicide risk and does not include assessment of suicidal behavior, such as self-injury or suicide attempts. Therefore, the relationship between IC and U-threat reactivity might be specific to suicide cognitions and not other forms of suicide risk. Conversely, the relationship between IC and U-threat reactivity might be even more strongly related to suicide attempts and self-harm than suicide cognitions, given the role of IC in regulating impulsive behavior; however, the prevalence of lifetime suicidal behavior in the present sample was relatively low and thus, we did not have adequate statistical power to examine this outcome. Future research with adequate statistical power should consider testing the synergistic impact of poor IC and high U-threat on suicide attempts and behavior. Future studies should also consider how subjective, behavioral, and psychological measures of IC, threat sensitivity, and suicide risk may interact across dimensions.

In sum, the present findings expand upon previous research by identifying high U-threat reactivity and low IC as synergistic risk factors for youth suicide. These findings contribute to the wider body of literature aimed at identifying targeted treatments for suicide prevention.

Impact Statement.

The present study examined the unique and interactive effects of behavioral inhibitory control and threat reactivity on suicide risk in youth. Results reveled that greater reactivity to unpredictable threat was associated with increased suicide risk at low, but not high, levels of inhibitory control. These findings shed light on subgroups of youth who may be at elevated risk.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.