Association Between Responsibility for the Death of Others and Postdeployment Mental Health and Functioning in US Soldiers

This cohort study examines the risk of posttraumatic stress disorder, suicidality, depression, and functional impairment in US Army service members who were exposed to combat experiences.


Introduction
The association between deployment experiences and postdeployment mental health outcomes in military personnel has been extensively studied. [1][2][3] Combat exposure is associated with postdeployment posttraumatic stress disorder (PTSD), depression, and suicidal thoughts and behaviors (STBs). 4-9 Consistent with fear-based conceptualizations of trauma response, 10 most studies examining combat have focused on the impact of threats to one's life (eg, taking enemy fire) or witnessing harm done to others. [11][12][13][14][15][16] However, evidence has shown that other specific combat experiences may confer greater risk. 6, [17][18][19] In studies of combat-deployed military personnel, many soldiers report being responsible for death and violence, but little attention has been paid to the long-term outcomes of such a traumatic event. 2,[20][21][22] The current diagnosis of PTSD does not explicitly identify participation in harming or killing others as meeting criterion A of the DSM-5 (Diagnostic and Statistical Manual of Mental Disorders [Fifth Edition]) PTSD diagnosis. 20,22,23 However, despite soldiers' preparation and training to use violence, being responsible for someone else's death may adversely affect their mental health, in which case screenings and interventions could be developed to mitigate these outcomes. 20,21,24,25 Most evidence of mental health sequelae is found in studies of veterans. Across multiple war eras (Vietnam, Gulf, and Iraq/Afghanistan), being responsible for the death of others has been associated with a PTSD diagnosis and the most severe PTSD symptoms. 21,[26][27][28][29] Furthermore, this association persists after adjusting for other combat exposures. Being responsible for the death of others during combat has also been associated with STBs, demonstrating large effect sizes compared with other types of combat experiences. 17,18,28,30 However, whether these findings extend to active-duty soldiers in the post-September 11, 2001 era, [31][32][33] a group whose suicide rates have increased substantially over the past decade, remains relatively unexplored. 34,35 In the current study, we examined the association between responsibility for the death of others in combat and postdeployment mental health outcomes (eg, PTSD, major depressive episode [MDE], STBs, and functional impairment) among active-duty US Army personnel. Because delayed onset of mental health problems may occur after deployment, 36 this study assessed outcomes at both 2 to 3 months and 8 to 9 months postdeployment.

Methods
Data for this cohort study were obtained from the Pre/Post Deployment Study of the Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS), 37,38 a prospective, multiwave panel survey of 3 US Army brigade combat teams (BCTs) that deployed to Afghanistan in 2012 for approximately 10 months on average. Baseline evaluation occurred 1 to 2 months predeployment.
Follow-up assessments were conducted at approximately 2 to 3 weeks, 2 to 3 months, and 8 to 9 months postdeployment. Surveys were conducted at the BCTs' home posts, except the 8-to 9-month postdeployment survey, which was conducted online or by telephone. Respondents

Responsibility for the Death of Others in Combat
Survey items were taken from the combat stress scale and included "have direct responsibility for the death of" an enemy combatant, a noncombatant, or an ally or US personnel. A binary (Ն1 time or never) composite variable was created to indicate any endorsement of responsibility for the death of others in combat. For sensitivity analyses, binary variables were created per type of responsibility (ie, enemy combatant, noncombatant, or ally) and ordinal variables were created by frequency (never, 1 time, or Ն2 times) for any responsibility and by type of responsibility. Variables were created for both predeployment (or previous deployment) and postdeployment (or index deployment) time points.

Suicidality
Composite variables were created to assess lifetime (yes or no) and past-30-day (yes or no) STBs, combining the endorsement of suicidal ideation, nonsuicidal self-injury, and suicide attempt. The STB items were evaluated using a self-report version of the Columbia-Suicide Severity Rating Scale. 46

Statistical Analysis
Missing data were treated as missing (observations deleted), except for missing combat stress and responsibility for death items, which were imputed as 0. Few participants (22 of 4645 [0.5%]) were missing combat stress and responsibility for death item-level scores, and no participant was missing all responsibility for death or combat stress items. Combined analysis weights were applied in all analyses; these included propensity-based adjustment for baseline attrition (because of incomplete surveys or inability to link to administrative records), poststratification to map the eligible baseline sample to known demographic and service characteristics of soldiers in the 3 BCTs that deployed to Afghanistan after the predeployment interview dates, and propensity-based attrition adjustment to account for the loss of respondents owing to incomplete data in 1 or more follow-up survey waves.
Statistical analyses were conducted using R software, version 3.6.2 (R Foundation for Statistical Computing). 47 Primary analyses were a series of multivariable logistic regressions that were performed separately for the outcomes at 2 to 3 months and at 8 to 9 months postdeployment. Four outcomes were examined: past 30-day PTSD, MDE, STBs, and functional impairment. Models were adjusted for potential risk factors, including age, sex, race and ethnicity (Black or African American, Hispanic, non-Hispanic White, or other [Asian, American Indian or Alaska Native, Native Hawaiian or Other Pacific Islander, or other], which was self-reported in the survey), marital status, BCT, predeployment lifetime internalizing and externalizing disorders, and combat stress severity (minus responsibility for death items) during the index deployment. Hypothesis tests were 2-sided with a priori P < .05 significance level. Data analysis was performed from December 12, 2020, to April 23, 2021.
A series of sensitivity analyses were also conducted that repeated the multivariable logistic regressions: (1) using only responsibility for the death of an enemy combatant, first as a binary variable (yes or no) and then as an ordinal variable (0 = 0 times, 1 = 1 time, or 2 = Ն2 times); (2) using all 3 types of responsibility (enemy combatant, noncombatant, or ally) variables (binary); and (3) including a binary endorsement of responsibility for death during a previous deployment, the index deployment, and their interaction, to assess the potential cumulative effect within the subsample of soldiers who had previously deployed. In addition, given that the study was underpowered to test for specific STBs in regression analyses, we performed Fisher exact tests to examine the bivariate associations between suicidal ideation, nonsuicidal self-injury, and suicide attempt at 2 to 3 vs 8 to 9 months postdeployment and each binary type of responsibility (enemy combatant, noncombatant, ally, or any responsibility). Sensitivity analysis findings are shown in eTables 2 to 4 in the Supplement.
The mean (SD) combat stress score during the index deployment was 2.54 (1.70). Correlation strength among combat stress items ranged from small to moderate, with no evidence of

Sensitivity Analyses
Repeating the adjusted regressions by entering into the models only the responsibility for the death of an enemy combatant (binary) showed that being responsible remained significantly associated with PTSD at 8 to 9 months postdeployment (OR, 1.44; 95 CI%, 1.10-1.90; P = .009) ( Table 4).
However, the association with STBs at 8 to 9 months postdeployment was attenuated (OR, In adjusted regressions that simultaneously included in the models all 3 types of responsibility

Discussion
In a sample of post-September 11, 2001 active-duty soldiers who were deployed to Afghanistan, those who reported responsibility for the death of others were 42% (OR, 1.42) more likely to have PTSD and were 55% (OR, 1.55) more likely to have STBs at 8 to 9 months postdeployment. These associations were not significant at 2 to 3 months postdeployment, suggesting the possibility of a critical window for intervention. Being responsible for the death of others was not associated with MDE or functional impairment at any time after deployment.
These findings replicate and build on the evidence in veterans from multiple war eras that demonstrated an association between being responsible for another's death during combat and subsequent PTSD and STBs. 17,18,21,[26][27][28][29] We found that these associations are detectable while soldiers are still on active duty, which suggests that an opportunity exists for screening and intervention, potentially disrupting the often chronic, treatment-refractory trajectories in veterans with military-related PTSD. 48 In addition, these findings identify the psychological sequelae of being responsible for another's death as a potential intervention target for PTSD psychotherapy at 8 to 9    months postdeployment. The pattern remained the same for PTSD when examining specifically the responsibility for an enemy combatant's death, although the association with STB was attenuated.

JAMA Network Open | Psychiatry
These findings contrast with those of a study of 400 soldiers that found that unjust war events (but not killing enemy combatants) were associated with adverse mental health outcomes. 31 However, operationalization differences may at least partially explain the discrepancies. Specifically, in that study, "unjust war events" included being responsible for noncombatant deaths (as well as witnessing brutality toward noncombatants, ill or injured women or children, and Geneva Convention violations), and "killing enemy combatants" included an item that can be categorized as a fighting variable ("shooting/directing fire at enemy"). 49 These alternative ways of categorizing combat-related events, therefore, do not address directly the question assessed here. Although multivariable logistic regression models did not show associations between being responsible for the deaths of noncombatants or allies and PTSD or STBs, the study was underpowered to detect such an association, and those results should be interpreted cautiously.
Shifting away from a primarily fear-based trauma framework, we believe that the current findings shed light on the association between active participation in death in combat and mental health. Moving forward, research should examine the predictive value of expanding the conceptualizations of criterion A of the DSM-5 PTSD diagnosis to include such events. Responsibility for the death of others during combat may confer unique cognitive or affective mechanisms (eg, shame and guilt) that contribute to a more complex pattern of PTSD. 24 Similarly, suicidality at 8 to 9 months after being responsible for another's death may reflect a costly attempt at making amends or the consequence of separating from a unit that provided meaning-making from the experience. 50 Future investigation into soldiers' perceptions of moral transgressions and moral injury 51,52 will help clarify the basis for the associations found in the current study, thus identifying actionable intervention targets. [53][54][55] A broader perspective of PTSD mechanisms beyond fear will be important in identifying possible subtypes of the disorder, additional neurobiological mechanisms, and precision treatment needs.
These alternative conceptualizations may be a possible explanation for the low success rates of first-line psychotherapy approaches for military-related PTSD in active-duty soldiers, with only 31% respectively. These treatment options may benefit soldiers with postdeployment PTSD who reported responsibility for the death of others, but additional research is needed. It is unknown whether current first-line psychotherapy is useful for PTSD that is associated with being responsible for another's death. It is critical to ascertain whether therapeutically addressing such topics while soldiers remain in service, with the possibility of future combat deployment, is beneficial or iatrogenic. Until then, a possible primary prevention strategy may be for units to prepare soldiers before deployment by discussing the possibility and impact of being responsible for another's death in combat (including fratricide and noncombatants).

Limitations
This study has several limitations. First, symptom measures and combat experiences were assessed with self-reported questionnaires, 59 which are subject to response biases, including social desirability. Responsibility for the death of others was as reported by the soldier not as ascertained or adjudicated by any military or other source. Respondents may have endorsed being responsible across a range of conditions (eg, active killing, giving orders, legal responsibility, and feeling responsible), and future studies should identify whether these differences have meaningful implications for the associations with mental health symptoms. Second, these findings in a sample of US Army personnel who were deployed to Afghanistan in 2012 may not be generalizable to other periods or conflicts. The sample was restricted to soldiers with data at all 4 survey waves, which may have altered the results, although weights were used to mitigate the impacts of nonparticipation and attrition. Third, because of low endorsements, the study was underpowered to test for differential associations of ally and noncombatant deaths with postdeployment mental health outcomes.
Similarly, the study was underpowered to examine specific STB outcomes (eg, suicidal ideation and suicide attempts), which demonstrate distinct associations with risk factors. 60 Future research should focus on the association between these specific combat experiences and suicide attempt and death by suicide, separately. Fourth, future studies should examine common pathways and additive risk models of PTSD and STBs.

Conclusions
This cohort study found that being responsible for the death of others during combat is associated with PTSD and STB at 8 to 9 months, but not 2 to 3 months, postdeployment. These findings shed light on complex war traumas and their sequelae. Identifying soldiers who report this type of responsibility on their return from deployment and delivering interventions early may mitigate subsequent PTSD and STBs.