Chelsea Dyan Gober Dykan, M.A. https://orcid.org/0000-0001-9954-7572 [email protected], Yoav Levinstein, M.A., Lucian Tetse-Laur, M.D., M.B.A., Ariel Ben-Yehuda, M.D., Jacob Rotschield, M.D., M.H.A., Daniel S. Pine, M.D., Paul D. Bliese, Ph.D., and Yair Bar-Haim, Ph.D.Authors Info & Affiliations

Publication: American Journal of Psychiatry

• Abstract

• METHODS

• RESULTS

• DISCUSSION

• Footnotes

• Supplementary Material

• REFERENCES

• Information & Authors

• Metrics & Citations

• View Options

• References

• Figures

• Tables

• Media

• Share

Abstract

Objective:

Evidence suggests that attentional threat avoidance is associated with increased risk for posttraumatic stress disorder (PTSD). This study evaluated the efficacy of two attention bias modification (ABM) protocols designed to enhance attention toward threats as a primary prevention of PTSD.

Methods:

The efficacy of the two ABM protocols was assessed using a three-arm randomized controlled trial in 501 male combat-bound soldiers. One protocol used response-time (RT)–based ABM to train attention toward threat over neutral stimuli (dot-probe task); the other used an eye-tracking-based ABM employing instrumental reward to enhance sustained attention to threat over neutral stimuli. Each intervention was compared to a sham RT-based task (dot-probe) presenting only neutral stimuli. Participants underwent four sessions of active or sham training. Threat-related attention was measured before and after training. Self-reported symptoms of PTSD (primary outcome) and of depression and anxiety (secondary outcomes), were assessed at baseline and postcombat 1 year later.

Results:

RT-based ABM delivered prior to combat exposure was associated with lower symptom severity and lower prevalence of probable PTSD postcombat relative to sham training (number needed to treat=22.7). A significant association was noted between training-induced threat attention and postcombat PTSD symptom severity in the RT-based ABM group. Eye-tracking-based ABM was not effective as a primary prevention protocol for PTSD symptoms.

Conclusions:

Consistent with a previous randomized controlled trial, RT-based ABM reduced risk for PTSD relative to sham ABM when implemented prior to combat exposure. These findings support the integration of RT-based ABM into resilience-building programs in military settings.

Primary prevention of posttraumatic stress disorder (PTSD) aims to reduce symptom risk by intervening before trauma exposure. Unlike with strategies targeting individuals already exposed to trauma and displaying symptoms (1, 2), less agreement exists on effective methods for PTSD prevention (3, 4). This study evaluated the efficacy of two attention bias modification (ABM) protocols as primary prevention for PTSD in combat-bound soldiers.

ABM shows promise in targeting attentional threat avoidance and reducing PTSD risk in combat-bound soldiers. Although attentional threat avoidance may provide short-term relief, it has been linked to greater long-term PTSD, depression, and anxiety severity (5–7). Integrating evidence linking threat avoidance to PTSD risk (8–10), preventive ABM trains soldiers to direct attention toward threat-related stimuli. A randomized controlled trial (RCT) suggested that ABM delivery immediately before combat disrupts the common association between combat exposure and PTSD symptoms (11). In another RCT, ABM delivered before combat was found to reduce PTSD risk 4 months postcombat relative to no training (12).

To manipulate attention toward threat, response-time (RT)-based protocols have used the dot-probe task with targets presented with high probability at the location of threat over neutral stimuli. The assumption is that participants implicitly learn the contingency between threat and target location and use it to enhance performance. A recently developed eye-tracking-based ABM method, called gaze-contingent music reward training (GC-MRT), uses feedback and operant conditioning principles. GC-MRT combines gaze-tracking with music rewards to train gaze behavior. Evidence supports GC-MRT’s efficacy in reducing threat-related attention biases and anxiety symptoms (13, 14), along with strong reliability in measurement (15). However, GC-MRT has not yet been applied to PTSD treatment or prevention.

In this study, we had three goals. First, we sought to replicate and extend the RT-based ABM findings from Wald et al. (12), which showed that four ABM sessions reduced PTSD risk in combat-deploying soldiers. Whereas Wald et al. used a treatment-as-usual control, we applied a tighter sham training control. Additionally, we used face rather than word stimuli in training. Based on the findings of the Wald et al. study, we expected that soldiers receiving RT-based ABM would show lower postcombat PTSD severity and prevalence compared to those receiving sham ABM. Second, we expected that soldiers receiving GC-MRT would exhibit lower PTSD severity and prevalence postcombat relative to sham ABM. Third, we compared the efficacy of RT-based ABM and GC-MRT in reducing PTSD risk. As no prior RCTs have tested GC-MRT for PTSD, this contrast was exploratory.

METHODS

Design

This three-arm RCT tested RT-based ABM and GC-MRT against sham control as primary prevention of PTSD in combat-bound soldiers. Participants underwent four sessions of active or sham interventions during a 3-month period of advanced combat training in 2021, before any combat exposure. Participants’ first combat deployment was in January 2022. Threat-related attention was measured pre- and postintervention. Self-reported symptoms of PTSD (primary outcome) and depression and anxiety (secondary outcomes) were assessed at predeployment baseline and 1-year postdeployment in January 2023. Combat experiences during deployment were documented.

Participants

Figure 1 depicts the flow of participants through the study. Participants were drawn from a cohort of male Israel Defense Force (IDF) recruits to a tier 1 infantry maneuver brigade (N=501; mean age at enrollment, 19.49 years [SD=0.85, range=18–25]). To be included in the study, soldiers had to be fluent in Hebrew. Of 570 potential participants (the full cohort), 3.0% declined to participate, 3.5% were not fluent in Hebrew, 5.1% were not available during baseline measurement, and 0.5% were removed due to technical issues. Participants were randomized into three groups: RT-based ABM (N=164), GC-MRT (N=167), and sham RT-based control (N=170). There were no differences between groups at baseline in demographic characteristics or self-reported symptoms (Table 1). The study was approved by the IDF Ethics Committee, the Tel Aviv University Institutional Review Board, and the Office of Human Research Oversight of the United States Army Medical Research and Development Command.

TABLE 1. Characteristics of participants in a study of attention bias modification for PTSD prevention in combat-deploying soldiers, by training groupa

Characteristic	RT-based ABM (N=164)	GC-MRT (N=167)	Sham control (N=170)	Mean	SD	Mean	SD	Mean	SD
Age (years)	19.4	0.8	19.5	0.8	19.6	0.9
Duration of education (years)	12.0	0.6	12.0	0.8	12.0	0.5
Premilitary traumatic exposure	1.0	1.1	1.0	1.2	0.9	1.0
Attention bias	0.6	18.9	−0.4	18.2	−2.0	16.2
Dwell time (%)	51.3	6.1	51.1	6.8	50.5	6.0
Combat exposure	3.3	2.5	3.4	2.8	2.8	2.4
PTSD (PCL-5)
Baseline	11.3	12.9	11.2	12.8	10.1	13.3
Postcombat	4.4	5.8	4.9	7.4	6.1	10.1
Depression (PHQ-9)
Baseline	5.9	5.3	5.1	4.7	5.3	5.1
Postcombat	1.9	3.1	2.2	3.2	2.5	4.0
Anxiety (GAD-7)
Baseline	3.2	4.3	2.8	3.7	3.1	4.4
Postcombat	1.4	2.7	1.3	2.7	1.6	3.2

a

GAD-7, 7-item Generalized Anxiety Disorder scale; GC-MRT, gaze-contingent music reward training; PHQ-9, Patient Health Questionnaire–9; PCL-5, PTSD Checklist for DSM-5; RT-based ABM, response-time-based attention bias modification.

a Baseline measurement was during basic training, before combat exposure. RT-based ABM, response-time-based attention bias modification; GC-MRT, gaze-contingent music reward training.

FIGURE 1. Participant flow through a study of attention bias modification for PTSD prevention in combat-deploying soldiersa

Attention Bias Measurements

Patterns of attention allocation to threat were measured in all participants before training or sham sessions began and after they concluded.

RT-based assessment.

A faces-based dot-probe task (Figure 2A) (16) consisting of 160 trials with full counterbalance of angry face location, probe location, probe type, and actors was used. Actors were 10 males and 10 females (Karolinska Directed Emotional Faces database) (17). In each trial, a fixation cross was presented (500 ms), followed by a pair of angry-neutral faces of the same actor (500 ms), followed by an arrowhead pointing either left or right at the location of one of the faces (until response). Participants were instructed to respond as quickly as possible without compromising accuracy by pressing the matching arrow on the keyboard. Threat-related attention bias was calculated by subtracting the mean RT to arrowheads appearing at the angry face location from the mean RT to arrowheads appearing at the neutral face location. Positive attention bias values indicate attention bias toward threat, and negative values indicate attention bias away from threat.

FIGURE 2. Examples of experimental trials of the response-time-based and eye-tracking-based assessment tasks

Eye-tracking-based assessment.

The free viewing task (15, 18) was used to measure threat-related gaze behavior with a high-speed eye tracker (Eyelink Portable Duo, SR Research, Ottawa). Calibration was followed by 30 trials, each presenting a matrix of 16 faces (Figure 2B). Pictures of faces were of actors from the same standardized database as was used in the RT-based assessment. Each trial began with a fixation cross shown until a fixation of 1,000 ms was recorded, thus verifying that a trial began only when a participant’s gaze was fixated at the center of the matrix. Each matrix was presented for 6,000 ms, followed by an intertrial interval of 2,000 ms. The matrices upheld the following parameters: each actor appeared only once in a matrix; each matrix contained eight male and eight female faces; half of the faces were angry and half neutral; and the four inner faces were two angry and two neutral. Participants were instructed to look freely at each matrix in any way they chose. Threat-related attention bias (dwell time percent) was calculated as the total dwell time on threat stimuli divided by the total dwell time on threat and neutral stimuli.

Interventions

Participants in all groups underwent four sessions of their respective intervention within the 3-month period of military basic training before any combat exposure.

RT-based ABM.

RT-based ABM (dot-probe task) followed the TAU-NIMH Initiative protocol (16). It applied the same display characteristics as those used for the RT-based attention assessment task described above, with two modifications. First, the faces used were from a different standardized faces database (19). Second, target probes appeared at the threat faces location 100% of the time. Participants were instructed to “Indicate the arrowhead’s direction as quickly as possible without compromising accuracy by pressing the matching arrow on the keyboard.” With repetitive threat-congruent trials, participants were expected to gradually learn the predictive value of threat faces in relation to probe location and thus shift their attention accordingly to improve task performance. Each of the four sessions consisted of 160 trials and lasted approximately 7 minutes.

Gaze-contingent music reward training (GC-MRT).

GC-MRT was a modified version of the gaze-tracking free viewing assessment task described above, designed to divert participants’ attention toward threat faces. The face stimuli were taken from the same database as those in the RT-based ABM task. At the beginning of each training session, participants selected a 12-minute music track to which they wished to listen. An extensive menu reflecting the most popular musicians according to published rating charts with a wide diversity of music styles was offered. Each session began with eye-tracking calibration followed by 30 face matrices adhering to the same parameters as the matrices described for the eye-tracking-based assessment task. Unlike in the measurement task, in GC-MRT, each matrix was shown for 24 seconds with no intertrial intervals. Participants were instructed to “Look freely at each matrix of faces in any way you like.” Participants heard their selected music play only when fixating on one of the threat faces in a matrix. When the participant was fixating on a neutral face, the music stopped. This instrumental reward setup induces attentional preference for threat over neutral stimuli through music reinforcement and has been shown to be efficacious in modifying threat-related gaze patterns in various psychopathologies (13, 14, 20). Each session lasted approximately 15 minutes.

Sham control.

The control condition consisted of the same number of trials and the same settings as the RT-based ABM condition but presented neutral faces in all trials, with the probe equally appearing at the location of the top or bottom face. This condition provides a similar experience to active ABM without exposure to threat stimuli or picture-probe contingency of target location. Participants were instructed to “Indicate the arrowhead’s direction as quickly as possible without compromising accuracy by pressing the matching arrow on the keyboard.” Each session lasted approximately 7 minutes.

Randomization

Participants were randomly allocated in a 1:1:1 ratio to RT-based ABM, GC-MRT, or sham control. Randomization occurred within each of the 15 platoons of the cohort. The allocation list was created using a computer-generated random number sequence prior to recruitment into the study. Enrolled participants were assigned to intervention based on their pregenerated allocation.

Symptom Measures

The primary outcome was self-reported PTSD symptoms, measured using the PTSD Checklist for DSM-5 (PCL-5) (21). Total severity score was used as a continuous index of symptom severity. Probable PTSD was determined based on a PCL-5 cutoff score ≥28 (21). Cronbach’s alphas for the sample were 0.93 and 0.91 for the baseline and postcombat assessments, respectively. Secondary outcomes were self-reported depression, measured using the Patient Health Questionnaire–9 (PHQ-9) (22); Cronbach’s alphas in the sample were 0.87 for both assessment points. Anxiety was measured with the 7-item Generalized Anxiety Disorder scale (GAD-7) (23); Cronbach’s alphas for the sample were 0.89 and 0.88 for the baseline and postcombat assessments, respectively.

Trauma Exposure Measures

Premilitary traumatic experiences were assessed using an 8-item yes/no questionnaire (10). This probed whether participants were present and/or injured in a terrorist attack, were in an area attacked by missiles or artillery or injured by such an attack, were present and/or injured in a motor vehicle accident, or experienced sexual or physical assault. The total score reflects a count of the endorsed events. The Combat Experiences Scale (10, 24) was used to track combat exposure. This 20-item questionnaire describes different combat events that soldiers may encounter during combat deployment. Items were yes/no questions concerning each combat event. A total score was derived as a count of the combat events endorsed.

Statistical Analysis

Power analyses were conducted using G*Power, version 3.1.9.7 (25). We estimated the necessary sample size for an analysis of covariance (ANCOVA) F-test with two groups, two repeated measurements, and one covariate to allow detection of a significant small effect size (Cohen’s f=0.15) (26) at 0.80 power and an alpha of 0.05 to require 175 participants per group. Therefore, all available participants from an enlistment cycle to a prespecified IDF infantry brigade were invited to participate, resulting in the enrollment of 164, 167, and 170 participants for the three groups, respectively.

Univariate analysis of variance was used to test group differences in baseline symptom and trauma exposure measures, and adherence to the intervention. Intervention completion and measurement attrition were examined using chi-square tests.

Intervention effects on symptoms for each of the specified hypotheses were evaluated using generalized estimating equations (GEE) (27) according to the prespecified hypotheses: 1) RT-based ABM versus sham control; 2) GC-MRT versus sham control; and 3) RT-based ABM versus GC-MRT. Separate GEE models were constructed for each pairwise comparison, corresponding to each of the three hypotheses. GEE applies an intention-to-treat approach, accommodating missing data and correlations among repeated measures by estimating marginal means, including data from all randomized participants based on the missing-at-random principle. To represent within-subject dependencies, an unstructured correlation matrix was used. For each of the continuous symptom outcomes, a Wald chi-square test was used to evaluate the time-by-group interaction term representing differential effects among study intervention conditions over two time points (baseline and postcombat). A linear model was used, including main effects for group and time, group-by-time interaction, and combat exposure as a covariate. Significant interactions were decomposed by comparing the groups at baseline and postcombat. To evaluate the robustness of significant pairwise findings, sensitivity analyses were conducted using multiple imputation along with ANCOVA models as a complementary analytic approach to provide an alternative perspective on group differences. This approach accounts for covariates (baseline symptoms and combat exposure) and provides a way to evaluate the potential impact of missing data by imputing 20 datasets simulating a design where no dropout occurred (28, 29). The imputed values were based on an expanded dataset to help evaluate whether differential dropout biased results. Pooled results from the 20 imputed datasets were compared to the complete-case results.

Group differences in the percentage of soldiers with probable postcombat PTSD were examined using likelihood ratio chi-square tests to account for the low expected frequency of probable PTSD cases anticipated in this preventive context. Odds ratios of probable PTSD and number needed to treat (NNT), indicating how many soldiers would need to receive the intervention for one to not develop probable PTSD, were calculated for the active interventions.

Partial correlations (with baseline PCL-5 as a covariate) were used to examine the associations between pre- and postintervention attention bias change (RT attention bias and dwell time percent) on postcombat PTSD symptom severity (PCL-5). No corrections were made for multiple comparisons.

RESULTS

Dropout

Seventy-three percent of the participants completed the postcombat assessment, and there was no significant difference between groups in attrition (χ2=0.50, df=2, p=0.78). Study completers and participants who dropped out did not differ significantly in baseline attention measures, symptoms, or premilitary trauma (Table S1).

Trauma Exposure

The groups did not differ significantly in premilitary traumatic experiences (F=0.38, df=2, 495, p=0.683). Participants reported a mean of 3.15 (SD=2.6) types of combat exposure, with no significant difference between the groups (F=2.29, df=2, 361, p=0.103).

Intervention Adherence

Participants completed an average of 3.6 (SD=0.80) of four training sessions. Participants in the RT-based ABM, GC-MRT, and sham control groups attended an average of 3.59 (SD=0.79), 3.54 (SD=0.90), and 3.69 (SD=0.72) sessions, respectively, with no significant differences between groups (F=1.52, df=2, 498, p=0.22). Training completion did not differ significantly between groups (χ2=2.23, df=2, p=0.30).

Hypothesis 1: Preventive RT-Based ABM Is Associated With Reduced Risk for PTSD Symptoms Relative to Sham Control

The GEE analysis indicated a group-by-time interaction for PTSD symptom severity (Wald χ2=3.15, df=1, p=0.038). Follow-up contrasts indicated that the groups did not differ significantly at baseline (Table S2), and a lower postcombat PTSD severity was shown for RT-based ABM relative to sham control (mean difference=2.11, p=0.048). ABM was associated with significantly lower PTSD severity compared to sham control in both the complete-case model (β=−2.07, SE=0.97, p=0.03) and the imputed model (β=−1.80, SE=0.86, p=0.04). Probable PTSD rates were significantly lower in the RT-based ABM group (0.9%) relative to the sham control group (5.3%) (likelihood ratio χ2=3.90, df=1, p=0.048; odds ratio=6.06; NNT=22.7) (Figure 3).

a GC-MRT, gaze-contingent music reward training; NNT, number needed to treat; RT-based ABM, response-time-based attention bias modification.

FIGURE 3. Percentage of participants with probable PTSD postcombat by training groupa

GEE analyses indicated a significant group-by-time interaction for depression severity (Wald χ2=3.86, df=1, p=0.05) and a nonsignificant interaction for anxiety severity (Wald χ2=0.008, df=1, p=0.93). Follow-up contrasts on depression scores indicated that the groups did not differ significantly at baseline or at postcombat assessment (Table S2).

Hypothesis 2: Preventive GC-MRT Is Associated With Reduced Risk for PTSD Symptoms Relative to Sham Control

The GEE analysis indicated a nonsignificant group-by-time interaction for PTSD symptom severity (Wald χ2=1.37, df=1, p=0.243). Follow-up contrasts indicated that the groups did not differ significantly on symptom severity at baseline or at postcombat assessment (Table S2). There was no significant difference between GC-MRT (2.7%) and sham control (5.3%) in probable PTSD prevalence (likelihood ratio χ2=1.00, df=1, p=0.32; GC-MRT odds ratio=2.02; NNT=38.5).

GEE analyses indicated nonsignificant group-by-time interactions for both depression (Wald χ2=0.02, df=1, p=0.90) and anxiety (Wald χ2=0.18, df=1, p=0.669). Follow-up contrasts indicated that the groups did not differ significantly at baseline or at postcombat assessment (Table S2).

Is RT-Based ABM Associated With Reduced Risk for PTSD Symptoms Relative to GC-MRT?

The GEE analysis indicated a nonsignificant group-by-time interaction for PTSD symptom severity (Wald χ2=0.336, df=1, p=0.562). Follow-up contrasts revealed that the groups did not differ significantly at baseline or at the postcombat assessment (Table S2). There was no significant difference between RT-based (0.9%) and GC-MRT (2.7%) in the rates of probable PTSD (likelihood ratio χ2=1.03, df=1, p=0.31).

GEE analysis indicated a significant group-by-time interaction for depression (Wald χ2=4.402, df=1, p=0.036) and a nonsignificant interaction for anxiety (Wald χ2=0.298, df=1, p=0.585). Follow-up contrasts indicated that the groups did not differ significantly at baseline or at the postcombat assessment (Table S2).

Associations Between Attention Bias Change and Postcombat PTSD Severity

Pretraining to posttraining change in attention bias measured by the eye-tracking-based dwell time percent index was negatively correlated with postcombat PTSD symptom severity in the group receiving RT-based ABM (r=−0.275, df=71, p=0.018). No such association was noted in the GC-MRT group (r=0.137, df=72, p=0.244) (Figure 4). Pre- to-posttraining change in attention bias measured by the RT-based dot-probe task was not correlated with postcombat PTSD severity in either of the active training groups (ABM: r=−0.05, df=71, p=0.67; GC-MRT: r=−0.115, df=72; p values >0.33).

a GC-MRT, gaze-contingent music reward training; PCL-5, PTSD Checklist for DSM-5; RT-based ABM, response-time-based attention bias modification.

FIGURE 4. Scatterplot of pre- to posttraining change in percent dwell time on threat (DT%) and postcombat PTSD severity in the response-time-based attention bias modification (ABM) groupa

DISCUSSION

This study evaluated the preventive potential of mechanized cognitive training interventions for PTSD in combat-bound soldiers. Three key findings arose. First, prior findings from Wald et al. (12) were extended; both studies support the efficacy of RT-based ABM as a prevention for PTSD delivered prior to combat exposure. Second, a significant association was observed between training-induced changes in attention bias and PTSD symptom severity in the RT-based ABM group, consistent with target engagement by RT-based ABM. Third, levels of PTSD symptoms did not differ significantly posttraining between the sham control and GC-MRT groups. Together, these findings carry implications for the design of effective PTSD preventions and the mechanism through which they reduce risk.

Attending to threats is advantageous in combat, where vigilance facilitates survival (7, 30). In extreme situations such as combat, and possibly in less extreme dangerous scenarios, attending to threat may support adaptation. RT-based ABM may reduce risk for PTSD by instantiating such adaptive forms of vigilance. Notably, the approach in this study differs from the one applied in ABM for anxiety disorders, where attention is trained away from threats. Findings in the present trial can be compared with the expected effect sizes in other RCTs for PTSD primary prevention (31), which often yield small or null effects due to program inefficacy or low-risk populations. Participants in our sample reported low symptom severity, with many reporting negligible symptoms. Nevertheless, the results demonstrated statistically significant effects, replicating and extending findings by Wald et al. (12), who used a similar design and population. With additional independent verification, ABM could be applied to at-risk populations and improve public health outcomes.

RT-based ABM reduced postcombat PTSD severity and probable PTSD prevalence compared to sham control. These findings, consistent with prior RCTs (11, 12), support the potential of RT-based ABM as primary prevention for PTSD in high-risk populations. Probable PTSD rates were considerably lower in the group receiving RT-based ABM (0.9%) compared to the group receiving the sham control intervention (5.3%), along with number needed to treat, as in the Wald et al. study (12), establishing that training ~20 soldiers could prevent postcombat PTSD in one soldier. In addition, the findings in the present study emerged while using a more credible control than the control used in the Wald et al. study. Further, unlike in the Wald et al. study, we used face stimuli instead of words, increasing potential dissemination to individuals with reading difficulties or with different linguistic backgrounds. While our findings are insufficiently strong to justify broad clinical application, the low-touch, low-cost nature of RT-based ABM renders it highly applicable in military contexts, pending additional independent verification.

Although ABM has been shown to reduce symptoms in depressed (32) and anxious (33, 34) patients, its efficacy has never been directly examined as a preventive protocol for these disorders. Based on research indicating excessive attention toward negative stimuli in depression (35) and anxiety (36, 37), treatment-oriented ABM trains attention away from negative stimuli. In contrast, based on studies showing elevated threat avoidance in PTSD (8–10), preventive ABM trains attention toward threatening stimuli. Future research could more fully characterize the nature of these pathways and the associated distinct impact of ABM on depression and anxiety in preventive contexts.

The negative correlation between training-induced changes in attention bias and postcombat PTSD severity in RT-based ABM suggests that altering threat-related attentional processes partially mediates the clinical effect. This supports theories linking attentional threat avoidance in dangerous contexts to PTSD vulnerability (7, 8, 38–40). However, the absence of a similar association in GC-MRT and the lack of significant correlations with RT-based measures highlight the complexity of cognitive target engagement mechanisms. Detecting such associations may depend on the reliability of attention assessment and the specific characteristics of cognitive tasks. Future studies could refine and standardize threat-related attention measures to optimize training protocols (15).

The lack of significant differences in PTSD severity between the GC-MRT and the sham control groups raises questions about its suitability as a preventive intervention in military settings. The absence of meaningful postcombat differences limits its clinical relevance. GC-MRT’s limited efficacy in this context may stem from targeting less relevant cognitive mechanisms or challenges in delivering this more complex intervention in military environments. Additionally, although RT-based ABM shows greater promise than GC-MRT, the lack of significant differences between the two approaches in clinical outcomes suggests that both protocols may warrant further refinement and testing. Unlike previous applications of GC-MRT, which focused on training patients with anxiety disorders to engage neutral over threat stimuli (13, 14), the GC-MRT variant applied here trained healthy participants to maintain attentional engagement with threat over neutral stimuli. Whereas substantial training effects were noted in previous GC-MRT studies, here, only minimal attentional change in the intended direction occurred. This discrepancy from past findings suggests the need for additional research designed to improve the training in PTSD prevention-oriented GC-MRT. For example, instead of 100% reinforcement of desired responses, learning through partial reinforcement may be tested.

The results should be interpreted with study limitations in mind. First, unlike Wald et al. (12), who applied clinical interviews, in this study we relied on self-reports. Although the PCL-5 shows high concordance with structured clinical interviews, clinician-rated assessments offer more nuanced insights into clinical status. Second, a notable number of participants did not complete postcombat assessments (27%). This limitation may relate to the difficult circumstances of following soldiers after combat; concerns with this may be partially addressed by three factors: 1) there was no significant difference on any of the baseline variables between completers and participants who dropped out; 2) dropout was equally distributed between groups; and 3) multiple imputation with an expanded dataset simulating a situation where no dropout occurred indicated no biasing in the results. Third, the lack of long-term follow-up restricts conclusions about the sustainability of protective effects over time and multiple deployments. Future studies could include longitudinal assessments to evaluate the durability of this intervention and how it might be affected by multiple deployments, a common occurrence in many armies. Fourth, because the soldier participants interacted throughout the study, there is a possibility that the intervention assignment of one participant indirectly influenced the outcomes of another. Such potential interference should be considered when interpreting the results. Fifth, the overall decline in group mean symptoms over time may reflect regression to the mean in symptom reporting, or alternatively, general resilience among combat soldiers. Another possibility is selective dropout of those who became more symptomatic after combat. Sixth, although the study demonstrated a significant relation between threat attention bias changes and PTSD outcomes in the RT-based ABM group, the correlational nature of this finding precludes causal inference. Additional experimental controls or mediation analyses requiring much larger samples are needed to confirm whether modifying threat-related attentional processes directly drives symptom reduction. Finally, the study was conducted on male IDF infantry recruits, limiting the extent to which the findings can be generalized to other combat roles (e.g., artillery, armor, air force, navy), female soldiers, or soldiers from different nationalities and cultural backgrounds. Expanding this research to diverse groups and contexts could enhance the applicability of primary prevention for PTSD, highlighting areas for future research.

The findings from this study have several practical and theoretical implications. From a clinical perspective, the demonstrated replicable efficacy of RT-based ABM highlights its potential for widespread implementation in military training programs as a preventive measure against PTSD in combat-bound troops. The modest effect sizes and the number needed to treat values indicate that although potentially beneficial, RT-based ABM should probably be integrated as part of a comprehensive resilience-building framework rather than a stand-alone solution. Future research, preferably in other militaries and by other researchers, could focus on refining cognitive training protocols, elucidating mechanisms of action, and exploring individual differences to maximize the clinical utility of these interventions.

Footnotes

ClinicalTrials.gov identifier: NCT05294848.

Supplementary Material

REFERENCES

1.

Bryant RA: Post‐traumatic stress disorder: a state‐of‐the‐art review of evidence and challenges. World Psychiatry 2019; 18:259–269

2.

Freedman SA: Psychological interventions to prevent PTSD. Psychiatr Ann 2019; 49:314–319

3.

Skeffington PM, Rees CS, Kane R: The primary prevention of PTSD: a systematic review. J Trauma Dissociation 2013; 14:404–422

4.

Sun Z, Song J, Chen J, et al: Preventing and mitigating post-traumatic stress: a scoping review of resilience interventions for military personnel in pre deployment. Psychol Res Behav Manag 2024; 17:2377–2389

5.

Constans JI, McCloskey MS, Vasterling JJ, et al: Suppression of attentional bias in PTSD. J Abnorm Psychol 2004; 113:315–323

6.

Lupien SJ, McEwen BS, Gunnar MR, et al: Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nat Rev Neurosci 2009; 10:434–445

7.

Shechner T, Bar-Haim Y: Threat monitoring and attention-bias modification in anxiety and stress-related disorders. Curr Dir Psychol Sci 2016; 25:431–437

8.

Beevers CG, Lee H-J, Wells TT, et al: Association of predeployment gaze bias for emotion stimuli with later symptoms of PTSD and depression in soldiers deployed in Iraq. Am J Psychiatry 2011; 168:735–741

9.

Sipos ML, Bar-Haim Y, Abend R, et al: Postdeployment threat-related attention bias interacts with combat exposure to account for PTSD and anxiety symptoms in soldiers. Depress Anxiety 2014; 31:124–129

10.

Wald I, Degnan KA, Gorodetsky E, et al: Attention to threats and combat-related posttraumatic stress symptoms: prospective associations and moderation by the serotonin transporter gene. JAMA psychiatry 2013; 70:401–408

11.

Wald I, Bitton S, Levi O, et al: Acute delivery of attention bias modification training (ABMT) moderates the association between combat exposure and posttraumatic symptoms: a feasibility study. Biol Psychol 2017; 122:93–97

12.

Wald I, Fruchter E, Ginat K, et al: Selective prevention of combat-related post-traumatic stress disorder using attention bias modification training: a randomized controlled trial. Psychol Med 2016; 46:2627–2636

13.

Arad G, Azriel O, Pine DS, et al: Attention bias modification treatment versus a selective serotonin reuptake inhibitor or waiting list control for social anxiety disorder: a randomized clinical trial. Am J Psychiatry 2023; 180:357–366

14.

Lazarov A, Pine DS, Bar-Haim Y: Gaze-contingent music reward therapy for social anxiety disorder: a randomized controlled trial. Am J Psychiatry 2017; 174:649–656

15.

Shamai-Leshem D, Abend R, Arad G, et al: The free-viewing matrix task: a reliable measure of attention allocation in psychopathology. J Anxiety Disord 2023; 100:102789

16.

Badura-Brack AS, Naim R, Ryan TJ, et al: Effect of attention training on attention bias variability and PTSD symptoms: randomized controlled trials in Israeli and US combat veterans. Am J Psychiatry 2015; 172:1233–1241

17.

Lundqvist D, Flykt A, Öhman A: Karolinska Directed Emotional Faces. Stockholm, Karolinska Institutet, Department of Clinical Neuroscience, Psychology Section, 1998

18.

Lazarov A, Abend R, Bar-Haim Y: Social anxiety is related to increased dwell time on socially threatening faces. J Affect Disord 2016; 193:282–288

19.

Tottenham N, Tanaka JW, Leon AC, et al: The NimStim set of facial expressions: judgments from untrained research participants. Psychiatry Res 2009; 168:242–249

20.

Shamai‐Leshem D, Lazarov A, Pine DS, et al: A randomized controlled trial of gaze‐contingent music reward therapy for major depressive disorder. Depress Anxiety 2021; 38:134–145

21.

Blevins CA, Weathers FW, Davis MT, et al: The Posttraumatic Stress Disorder Checklist for DSM‐5 (PCL‐5): development and initial psychometric evaluation. J Trauma Stress 2015; 28:489–498

22.

Kroenke K, Spitzer RL, Williams JB: The PHQ‐9: validity of a brief depression severity measure. J Gen Intern Med 2001; 16:606–613

23.

Spitzer RL, Kroenke K, Williams JBW, et al: A brief measure for assessing generalized anxiety disorder: the GAD-7. Arch Intern Med 2006; 166:1092–1097

24.

Hoge CW, Castro CA, Messer SC, et al: Combat duty in Iraq and Afghanistan, mental health problems, and barriers to care. N Engl J Med 2004; 351:13–22

25.

Faul F, Erdfelder E, Lang A-G, et al: G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007; 39:175–191

26.

Cohen J: Statistical Power Analysis for the Behavioral Sciences, 2nd ed. Hillsdale, NJ, Erlbaum, 1988

27.

Zeger SL, Liang KY, Albert PS: Models for longitudinal data: a generalized estimating equation approach. Biometrics 1988; 44:1049–1060

28.

Lee KJ, Simpson JA: Introduction to multiple imputation for dealing with missing data. Respirology 2014; 19:162–167

29.

Hayati Rezvan P, Lee KJ, Simpson JA: The rise of multiple imputation: a review of the reporting and implementation of the method in medical research. BMC Med Res Methodol 2015; 15:30

30.

LeDoux JE: Emotion circuits in the brain. Annu Rev Neurosci 2000; 23:155–184

31.

Adler AB, Bliese PD, McGurk D, et al: Battlemind debriefing and Battlemind training as early interventions with soldiers returning from Iraq: randomization by platoon. J Consult Clin Psychol 2009; 77:928–940

32.

Xia H-S, Li Y-X, Zhang Q-Y, et al: Attention bias modification for depression: a systematic review and meta-analysis. Front Psychiatry 2023; 14:1098610

33.

Price RB, Wallace M, Kuckertz JM, et al: Pooled patient-level meta-analysis of children and adults completing a computer-based anxiety intervention targeting attentional bias. Clin Psychol Rev 2016; 50:37–49

34.

Jones EB, Sharpe L: Cognitive bias modification: a review of meta-analyses. J Affect Disord 2017; 223:175–183

35.

Peckham AD, McHugh RK, Otto MW: A meta-analysis of the magnitude of biased attention in depression. Depress Anxiety 2010; 27:1135–1142

36.

Bar-Haim Y, Lamy D, Pergamin L, et al: Threat-related attentional bias in anxious and nonanxious individuals: a meta-analytic study. Psychol Bull 2007; 133:1–24

37.

Armstrong T, Olatunji BO: Eye tracking of attention in the affective disorders: a meta-analytic review and synthesis. Clin Psychol Rev 2012; 32:704–723

38.

Bar-Haim Y, Holoshitz Y, Eldar S, et al: Life-threatening danger and suppression of attention bias to threat. Am J Psychiatry 2010; 167:694–698

39.

Wald I, Shechner T, Bitton S, et al: Attention bias away from threat during life threatening danger predicts PTSD symptoms at one‐year follow‐up. Depress Anxiety 2011; 28:406–411

40.

Wald I, Lubin G, Holoshitz Y, et al: Battlefield-like stress following simulated combat and suppression of attention bias to threat. Psychol Med 2011; 41:699–707