Associations of active-duty mental health trajectories with post-military adjustment: Results from the STARRS Longitudinal Study

Abstract

Background:

Many servicemembers experience difficulties transitioning from military to civilian life. We examined whether changes in mental health observed during active duty were associated with indices of post-military adjustment.

Methods:

Survey data from the multi-wave Army STARRS Pre/Post Deployment Study (PPDS; conducted 2012–2014) were linked to follow-up data from wave 1 of the STARRS Longitudinal Study (STARRS-LS1; conducted 2016–2018). Empirical Bayes estimates of intercepts and slopes of posttraumatic stress, problematic anger, and depressive symptoms during the PPDS were extracted from mixed-effects growth models and evaluated as predictors of life stress among 1,080 participants who had separated or retired from the Army at STARRS-LS1; and of job satisfaction among 586 veterans who were employed at STARRS-LS1.

Results:

Higher average levels and larger increases in posttraumatic stress, anger, and depression over the deployment period were each associated with increased stress and (in the case of anger and depression) reduced job satisfaction. Posttraumatic stress and anger slopes were associated with overall stress (b=5.60, p<.01 and b=15.64, p=.04, respectively) and relationship stress (b=5.50, p=.01 and b=22.86, p=.01, respectively) beyond the average levels of those symptoms. Limitations: Some transition-related difficulties may have resolved before outcome assessment; some measures were not previously validated.

Conclusions:

Larger increases in posttraumatic stress and anger over a deployment period were associated with increased stress after leaving the Army, even after controlling for average symptom levels during the same period. Monitoring changes in mental health during active duty may help identify personnel who need additional support to facilitate the military-to-civilian transition.

Introduction

The first few years after leaving the military are a critical transition time. During this period, veterans often confront stressors that may include identity strain, relationship and family problems, under- or unemployment, and financial pressure (Kintzle and Castro, 2018; McAllister et al., 2015; Mobbs and Bonanno, 2018; Thomas, 2022). These common challenges help explain why over 40% of U.S. veterans who served in Iraq and Afghanistan report difficulties in adjusting to civilian life (Morin, 2011). For high-risk veterans, transition stress may also contribute to acute outcomes such as suicidal behavior and homelessness (Brenner and Barnes, 2012; Chu et al., 2022; Tsai and Rosenheck, 2015).

With approximately 200,000 service members leaving the U.S. Armed Forces each year (U.S. Department of Veterans Affairs, 2020), it is important to develop strategies for identifying and engaging those at high risk for transition challenges. Specialized care geared towards veterans’ transition needs is available, but usage is limited. Indeed, while approximately 60% of veterans are eligible for healthcare provided by the U.S. Department of Veterans Affairs, most either do not use the benefit, or do so for only a portion of their healthcare needs (Farmer et al., 2016). This presents limited opportunities for detecting transition stress that could lead to more severe problems—and highlights the value of being able to identify at-risk personnel before they leave service. This would facilitate tailoring of transition services (e.g., those available through the U.S. Department of Defense Transition Assistance Program or inTransition mental health service) based on a service member’s risk level, allowing for more support to be provided to at-risk members while they remain “accessible” (i.e., on active duty).

In support of this overarching objective, we investigate whether characteristics measured during active duty provide information regarding a soldier’s risk of post-military adjustment problems (e.g., higher life stress or lower job satisfaction). Mental health symptoms during a critical active-duty period comprise the focal predictor variables, as data suggest that mental health problems are key risk factors for poor post-military adjustment (Adler et al., 2022; Karney and Crown, 2007; Morissette et al., 2021; Zivin et al., 2011). Importantly, we extend the existing evidence by evaluating a novel analytic approach with the potential to improve risk stratification. This method leverages repeated measures of symptoms to evaluate whether changes in mental health that soldiers experience while on active duty (e.g., increases in symptoms over a deployment period, which could signal a “deteriorating” trajectory/slope) explain variance in post-military outcomes, beyond that explained by average levels of symptoms during the same timeframe.

The dual consideration of average levels and trajectories of mental health symptoms represents an opportunity to identify at-risk soldiers who may not otherwise be recognized. To illustrate, a dynamic model that incorporates trajectories may yield different predictions for two soldiers who each report mild depressive symptoms overall while on active duty. The first case would be a soldier who consistently reports mild symptoms across measurement occasions, while the second would be a soldier who reports no depressive symptoms at time 1, mild symptoms at time 2, and moderate symptoms at time 3. The latter trajectory appears to suggest an elevated risk of poor outcomes despite a relatively low average depression score across occasions. While plausible, empirical investigation is needed to determine whether trajectories offer unique information regarding risk of later problems.

The purpose of this research was to evaluate this approach using longitudinal survey data from the Pre/Post Deployment Study (PPDS) of the Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS; Kessler et al., 2013; Ursano et al., 2014). The PPDS assessed soldiers on multiple occasions during a period involving preparation for deployment, deployment to a combat zone, and post-deployment reintegration. In the current study, individual differences in mental health were captured by estimating two parameters over the PPDS: average levels of symptoms and changes in symptoms. These parameters were then linked to outcomes assessed approximately four years later in wave 1 of the STARRS Longitudinal Study (STARRS-LS1; Stanley et al., 2022).

Our analysis specifically tested whether average levels and changes in posttraumatic stress, problematic anger, and depression during the PPDS were prospectively associated with stress and job satisfaction after leaving the Army (measured in the STARRS-LS1 survey). Given possible impacts of mental health problems on interpersonal functioning (Hammen, 2005; Taft et al., 2011), we also derived subscales of the stress and job satisfaction measures that captured relational aspects of those domains. We hypothesized that both the average level and the amount of change in posttraumatic stress, problematic anger, and depression during the PPDS would be associated with STARRS-LS1 outcomes. Specifically, we anticipated that both higher average scores and larger increases in each symptom domain would be associated with higher overall stress and relationship stress, and lower overall job satisfaction and work relationship satisfaction at STARRS-LS1. We also predicted that trajectories would explain incremental variance in post-military outcomes beyond that explained by average levels of mental health symptoms.

Methods

Overview/Participants

The PPDS component of Army STARRS is a multi-wave panel survey of three Brigade Combat Teams that deployed to Afghanistan in 2012. Baseline surveys occurred 1–2 months before deployment, and follow-ups used here were administered approximately 3 and 9 months after return from deployment. The STARRS Longitudinal Study is a follow-up study of Army STARRS participants consisting of periodic surveys starting in 2016 and continuing until 2025. This study uses wave 1 (STARRS-LS1) data to examine post-military outcomes of PPDS participants. STARRS-LS1 data were collected September 2016-April 2018 resulting in a mean interval of 46 months (range=36–58) between the last wave of the PPDS and STARRS-LS1. Written informed consent was obtained for survey participation and linkage of responses to Army/Department of Defense records. Procedures were approved by the Institutional Review Boards of the collaborating institutions. Further details regarding the PPDS and STARRS-LS are available elsewhere (Kessler et al., 2013; Stanley et al., 2022).

Of the 9949 soldiers present for duty in the Brigade Combat Teams, 8776 consented to participate in the PPDS baseline survey and link their survey responses and Army/Department of Defense records. The current sample was restricted to those who had also: deployed to Afghanistan with their team (n=7741); participated in a follow-up survey at 3 and/or 9 months post-deployment (n=6747); been recruited for and participated in the STARRS-LS1 survey [n=2627; see Stanley et al. (2022) for details]; provided complete information for study measures (n=2269); and indicated that their military status at STARRS-LS1 was separated or retired (n=1080). On average, veterans comprising the sample had been out of the Army for 2.72 years (SD=1.21). Job satisfaction analyses were conducted in the subsample of 586 veterans who reported being employed at the time of their STARRS-LS1 survey. The remaining veterans were looking for work (n=100), pursuing education/training (n=122), retired (n=44), unemployed for other reasons (e.g., disability, homemaker; n=64); or missing employment information (n=164).

Measures

Posttraumatic stress.

PPDS baseline assessment of past-30-day posttraumatic stress was based on six items from the Posttraumatic Stress Disorder Checklist-Civilian version (Weathers et al., 1993). These items were also available at 3- and 9-months post-deployment. Participants rated intrusive recollections, physical reactions to trauma cues, avoidance of internal cues (e.g., memories), situational avoidance, trouble concentrating, and feeling jumpy/easily startled on a 5-point scale ranging from “never” (coded “1”) to “6 or more times a week” (coded “5”). Ratings were averaged to create a total posttraumatic stress score at each assessment point (Cronbach’s alpha=.91-.96).

Depression.

PPDS assessment of past-30-day depressive symptoms was based on items adapted from the Composite International Diagnostic Interview Screening Scales (Kessler and Ustun, 2004). Respondents rated four items assessing low mood, pessimism about the future, worthlessness, and anhedonia on a 5-point scale ranging from “none of the time” (coded “1”) to “all or almost all the time” (coded “5”). These items were administered in surveys conducted at pre-deployment baseline and at 3- and 9-months post-deployment. Respondents’ ratings were averaged to create a total depression score (Cronbach’s alpha=.91-.94).

Problematic anger.

The PPDS measure of problematic anger is described elsewhere (Campbell-Sills et al., 2021). Respondents rated how often they felt irritated, annoyed, or grouchy; so angry that they might explode; a lot angrier than most people would be in the same situation; and that their anger was uncontrollable on a 5-point scale from “none of the time” (coded “1”) to “all or almost all of the time” (coded “5”). These ratings were averaged to create a problematic anger score (Cronbach’s alpha=.88-.92), which was available at pre-deployment baseline and at 3- and 9-months post-deployment.

Life stress.

The STARRS-LS1 survey evaluated current stress related to finances, career, health, close personal relationships, relationships with family/friends, health of loved ones, other problems experienced by loved ones, problems getting along with others, and “your life overall”. Stress in each domain was rated on an 11-point scale from “no stress” (coded “0”) to “very severe stress” (coded “10”). An overall stress score was derived by averaging the ratings of each item (Cronbach’s alpha=.92). We also derived a subscale capturing relationship stress, which included the items that queried stress related to close personal relationships, relationships with family/friends, and problems getting along with others; Cronbach’s α=.88).

Job Satisfaction.

For those who reported being employed at STARRS-LS1, the survey evaluated perceptions of their job security, salary and benefits, opportunity for advancement, enjoyment of work, work conditions (pace, control, stress), relationships with coworkers, and relationships with supervisors [rated on a 5-point scale from “poor” (coded “1”) to “excellent” (coded “5”)]. Ratings were averaged to create an overall job satisfaction score (Cronbach’s α=.84). To evaluate work relationship satisfaction, we also created a subscale comprised of two items assessing relationships with coworkers and supervisors (Cronbach’s α=.86).

Controls.

Socio-demographic characteristics were assessed in the baseline PPDS survey and included sex, age, race and ethnicity, education level, and marital status. These variables were used for describing the sample and as controls in regression models. Race and ethnicity categories were derived based on participants’ responses to two survey items. Participants who gave an affirmative response to the item “Are you Spanish/Hispanic/Latino?” were classified as Hispanic or Latino. All other participants were categorized based on their response to the item “What is your race?” Response options were American Indian or Alaskan Native, Asian, Black or African American, Native Hawaiian or Other Pacific Islander, White, or “Other”. All race and ethnicity categories were used in descriptive statistics, and in regression models a dichotomous variable was used (White vs all other categories).

Data Analysis

Calculation of posttraumatic stress, problematic anger, and depression trajectories in the PPDS was based on mixed-effect growth models using the nlme package (Pinheiro et al., 2022) in R (R Core Team, 2022). Mixed-effect growth models estimate trajectories over time (Singer and Willett, 2003) and can be used to extract empirical Bayes estimates of the intercept (average) and slope for each individual. We followed a four-step model building procedure (Bliese and Ployhart, 2002) detailed in Supplementary Methods, Supplementary Results, and Supplementary Table 1. Given that slopes were estimated based on three PPDS assessments (pre-deployment and 3- and 9-months post-deployment), we examined only linear trajectories. Empirical Bayes estimates of the intercept and slope for posttraumatic stress, problematic anger, and depression during the PPDS were calculated for each participant and linked to the STARRS-LS1 outcomes (overall stress, relationship stress, overall job satisfaction, and work relationship satisfaction) using ordinary least squares regression models.

Results

The sample of 1,080 veterans was 93.1% male (n=1006) and 6.9% (n=74) female, with a mean age of 26.3 (SD=6.3) at baseline. The race and ethnicity distribution within this sample was: 1.1% (n=12) American Indian or Alaskan Native, 3.4% (n=37) Asian, 8.2% (n=89) Black or African American, 16.5% (n=178) Hispanic or Latino, 1.1% (n=12) Native Hawaiian or Other Pacific Islander, 68.3% (n=738) White, and 1.3% (n=14) “Other.” In the models, a dichotomous race and ethnicity variable was used as a control [68.3% (n=738) White; 31.7% (n=342) other race and ethnicity]. The education level of the majority (73.1%; n=789) was high school diploma or GED/equivalent, 25.1% (n=271) had an undergraduate college degree, and 1.9% (n=20) had a graduate degree. About half of the sample was married (51.2%; n =553), with 39.3% (n=424) never married and 9.5% (n=103) either divorced, separated, or widowed.

Correlations among the PPDS symptom measures and the STARRS-LS1 stress and job satisfaction measures are shown in Supplementary Table 2. The first hypothesis tests evaluated, in separate models controlling for socio-demographic characteristics, the effects of average levels (intercepts) of mental health symptoms over the deployment period and slopes of mental health symptoms over the deployment period. The models of overall life stress and job satisfaction at STARRS-LS1 are shown in Tables 1 and 2, respectively; with the models of relationship stress and work relationship satisfaction shown in Supplementary Tables 3 and 4, respectively. The results show that, generally, both the intercepts and the slopes of posttraumatic stress, problematic anger, and depressive symptoms over the PPDS displayed significant associations with all four STARRS-LS1 outcomes. Exceptions were that neither the intercepts nor the slopes of posttraumatic stress symptoms were associated with overall job satisfaction or work relationship satisfaction. In all other cases, higher average levels and larger increases in posttraumatic stress, problematic anger, and depressive symptoms over the PPDS were each positively associated with overall life stress (Table 1) and relationship stress (Supplementary Table 3), and negatively associated with overall job satisfaction (Table 2) and work relationship satisfaction (Supplementary Table 4).

Table 1.

Associations of mean values and trajectories of posttraumatic stress, problematic anger, and depression over the Pre/Post Deployment Study with overall life stress at wave 1 of the STARRS Longitudinal Study

	Model 1a		Model 1b		Model 2a		Model 2b		Model 3a		Model 3b
	Est	SE	Est	SE	Est	SE	Est	SE	Est	SE	Est	SE
Intercept	3.72**	0.42	3.49**	0.43	3.51**	0.41	3.10**	0.42	3.74**	0.41	3.02**	0.42
Posttraumatic Stress Intercept	1.38**	0.18
Posttraumatic Stress Slope			10.45**	1.59
Problematic Anger Intercept					1.49**	0.14
Problematic Anger Slope							42.51**	4.19
Depression Intercept									2.02**	0.18
Depression Slope											45.30*	4.25
Observations	1080		1080		1080		1080		1080		1080
R2	0.08		0.07		0.13		0.12		0.13		0.13
Adjust R2	0.08		0.06		0.12		0.11		0.12		0.12
Residual Std. Error	2.12		2.14		2.07		2.09		2.07		2.08
F Statistic	12.17**		10.08**		19.73**		17.75**		19.97**		19.17**

Note. Estimates are from ordinary least squares models that controlled for socio-demographic characteristics (sex, age, race and ethnicity, education, and marital status).

^**p<0.01

^*p<0.05

^†p<0.10

Table 2.

Associations of mean values and trajectories of posttraumatic stress, problematic anger, and depression over the Pre/Post Deployment Study with overall job satisfaction at wave 1 of the STARRS Longitudinal Study

	Model 1a		Model 1b		Model 2a		Model 2b		Model 3a		Model 3b
	Est	SE	Est	SE	Est	SE	Est	SE	Est	SE	Est	SE
Intercept	3.18**	0.24	3.23**	0.24	3.23**	0.24	3.29**	0.24	3.14**	0.24	3.27**	0.24
Posttraumatic Stress Intercept	−0.19†	0.10
Posttraumatic Stress Slope			−1.30	0.85
Problematic Anger Intercept					−0.27**	0.08
Problematic Anger Slope							−8.04**	2.31
Depression Intercept									−0.34**	0.10
Depression Slope											−8.00**	2.36
Observations	586		586		586		586		586		586
R2	0.01		0.01		0.03		0.03		0.03		0.03
Adjust R2	0.00		0.00		0.02		0.01		0.01		0.01
Residual Std. Error	0.83		0.83		0.82		0.82		0.82		0.82
F Statistic	1.05		0.91		2.22*		2.15*		1.96*		2.07*

Note. Estimates are from ordinary least squares models that controlled for socio-demographic characteristics (sex, age, race and ethnicity, education, and marital status).

^**p<0.01

^*p<0.05

^†p<0.10

The next set of models tested the hypothesis that the slopes of posttraumatic stress, problematic anger, and depressive symptoms would exhibit associations with the STARRS-LS1 outcomes beyond the effects of the corresponding intercepts (i.e., average symptom levels). Table 3 shows that the posttraumatic stress slope displayed unique associations with overall stress and relationship stress, but not with overall job satisfaction or work relationship satisfaction. Beyond the effect of a soldier’s average level of posttraumatic stress over the deployment period, a steeper increase in posttraumatic stress was associated with higher overall stress and relationship stress after leaving the Army. The same pattern of results was observed for problematic anger (Table 4); namely, larger increases in anger during the PPDS were uniquely associated with higher overall life stress and relationship stress after leaving the Army. While we intended to also test this hypothesis using the PPDS depression measure, we found the model coefficients were unstable due to excessive collinearity between depression intercepts and slopes (r=.97) and thus did not retain the models with both entered as predictors.

Table 3.

Independent associations of mean values vs trajectories of posttraumatic stress over the Pre/Post Deployment Study with post-military stress and job satisfaction outcomes

	Model 1		Model 2		Model 3		Model 4
	Overall stress		Relationship stress		Overall job satisfaction		Work relationship satisfaction
	Est	SE	Est	SE	Est	SE	Est	SE
Intercept	3.57**	0.42	3.26**	0.51	3.21**	0.24	3.39**	0.29
Posttraumatic Stress Intercept	1.04**	0.21	1.12**	0.25	−0.15	0.13	−0.04	0.16
Posttraumatic Stress Slope	5.60**	1.85	5.50*	2.21	−0.56	1.06	−2.04	1.29
Observations	1080		1080		586		586
R2	0.09		0.07		0.01		0.02
Adjust R2	0.08		0.07		0.00		0.00
Residual Std. Error	2.12		2.54		0.83		1.00
F Statistic	11.92**		9.51**		0.96		1.45

Note. Estimates are from ordinary least squares models that controlled for socio-demographic characteristics (sex, age, race and ethnicity, education, and marital status).

^**p<0.01

^*p<0.05

^†p<0.10

Table 4.

Independent associations of mean values vs trajectories of problematic anger over the Pre/Post Deployment Study with post-military stress and job satisfaction outcomes

	Model 1		Model 2		Model 3		Model 4
	Overall stress		Relationship stress		Overall job satisfaction		Work relationship satisfaction
	Est	SE	Est	SE	Est	SE	Est	SE
Intercept	3.33**	0.42	2.89**	0.50	3.26**	0.24	3.44**	0.29
Problematic Anger Intercept	1.06**	0.25	1.18**	0.29	−0.16	0.14	−0.10	0.17
Problematic Anger Slope	15.64*	7.53	22.86**	8.94	−3.93	4.18	−7.63	5.06
Observations	1080		1080		586		586
R2	0.13		0.13		0.03		0.04
Adjust R2	0.12		0.13		0.02		0.02
Residual Std. Error	2.07		2.45		.82		1.00
F Statistic	18.07**		18.43**		2.07*		2.47**

Note. Estimates are from ordinary least squares models that controlled for socio-demographic characteristics (sex, age, race and ethnicity, education, and marital status).

^**p<0.01

^*p<0.05

^†p<0.10

Discussion

Our study contributes to a growing literature that seeks to determine whether information collected during active military duty can be used to identify personnel who need higher-intensity services to facilitate the transition to civilian life (Adler et al., 2022; Koh et al., 2022; Stanley et al., 2022). The study makes several contributions. First, we showed the potential utility of incorporating simple trajectory variables into models of post-military outcomes. Slopes of posttraumatic stress and problematic anger across a deployment and reintegration period explained incremental variance (beyond that explained by average symptom levels) in post-military stress outcomes, with greater increases in posttraumatic stress and problematic anger during the PPDS associated with higher overall stress and relationship stress at follow-up. Posttraumatic stress and problematic anger slopes did not explain incremental variance in civilian job satisfaction measures perhaps in part due to smaller sample sizes reducing power to detect effects.

In practice, monitoring symptom trajectories may provide military organizations with an additional tool for identifying at-risk individuals while they remain on active duty. For instance, service members are required to participate in annual Periodic Health Assessments that evaluate general health, posttraumatic stress, and depression. Our study represents a proof-of-concept within a sample assessed at predetermined times before and after a critical exposure (i.e., combat deployment). Findings using this study design suggest that quantifying trajectories of symptoms that are already routinely evaluated may provide unique information about whether the soldier is at increased risk of post-military adjustment difficulties such as elevated life stress. It is not clear whether symptom trajectories would demonstrate comparable associations with post-military outcomes when assessed under more routine circumstances versus over a deployment cycle; however, work in non-military settings suggests that even under routine conditions (e.g., annual engagement surveys in firms) trajectories convey valuable information (Chen et al., 2011). Additional work is needed to ensure that extracting symptom trajectories from administrative records would be a cost-effective endeavor (i.e., to justify modifying existing systems to calculate trajectories and channel the relevant information appropriately).

The finding that higher average scores and steeper increases in posttraumatic stress symptoms during the PPDS were prospectively associated with higher stress ratings several years later builds on evidence from retrospective studies indicating that servicemembers with posttraumatic stress disorder report more life stressors than those without posttraumatic stress disorder in the years after their combat exposure (Benyamini and Solomon, 2005). Further, the significant associations of both average levels and slopes of posttraumatic stress with later relationship stress converge with other results suggesting possible negative impacts of posttraumatic stress on intimate relationships and social functioning (Creech et al., 2019; Riggs et al., 1998; Taft et al., 2011). Moreover, this is now the second study to document prospective associations between problematic anger during active duty and post-military problems. A recent Millennium Cohort Study report indicated that problematic anger was associated with a wide range of subsequent behavioral health, relational, and economic problems (Adler et al., 2022); providing additional evidence that monitoring problematic anger during military service may help identify personnel at increased risk of post-military difficulties.

More generally, our study highlights potential long-term impacts of combat service on veterans. The approximate time lag between the last post-deployment survey and the collection of stress and job satisfaction data was four years. It is reasonable to expect that numerous life events occurred in the interim, yet mental health symptom trajectories from the deployment period remained significant predictors of certain distal outcomes. The prospect that deployment-related symptom trajectories might contribute to a series of events that continue to have ramifications years later underscores the importance of efforts to monitor the mental health of soldiers during and after deployments.

This study has some limitations. First, self-report measures are susceptible to response bias. Second, the symptom scales were abbreviated versions of validated measures, and did not assess all the criteria for posttraumatic stress disorder or major depressive disorder. However, they represent brief measures that may be more likely to be collected at scale in military settings. Third, the stress and job satisfaction measures had not been previously validated, although they had high face validity and internal consistency. Finally, the veterans who comprised the sample had been out of the Army for two-and-a-half years on average. Some may have experienced transition problems (e.g., elevated stress, job dissatisfaction) that had resolved by the time of the STARRS-LS1 survey. Future studies should attempt to characterize associations between active-duty mental health trajectories and indices of adjustment during earlier phases of adaptation to civilian life (e.g., the first year of transition).

In conclusion, mental health symptom trajectories across a deployment and reintegration cycle explained variance in post-military life stress outcomes, beyond that explained by average levels of mental health symptoms during the same active-duty period. Information regarding how service members’ mental health changes over key active-duty timeframes may be useful for identifying those at greater risk of severe transition stress—which may precede more serious and costly outcomes (e.g., affective disorders, suicidal behavior). Further research is warranted to replicate the current findings, and to investigate whether similar associations are observed in other military contexts (e.g., among members of other service branches, or during active-duty periods that do not include deployment) or in non-military settings [e.g., among college students, whose patterns of alcohol/substance use may relate to outcomes during their transition to full-time work (Arria et al., 2013)]. Efforts to improve servicemember engagement with existing transition programs (United States Government Accountability Office, 2022) and to tailor these resources to address the needs of at-risk personnel are also needed.

Highlights.

• This study investigates whether changes in mental health observed during active military duty are associated with post-military adjustment problems.

• Higher average levels and larger increases in posttraumatic stress, anger, and depression over a deployment period were associated with increased stress after leaving the Army.

• Higher average levels and larger increases in anger and depression over the deployment period were also associated with lower civilian job satisfaction scores.

• Larger increases in posttraumatic stress and anger over the deployment period were associated with more overall stress and relationship stress after leaving the Army, even after controlling for average levels of those symptoms.

• Monitoring changes in mental health during active duty may help identify soldiers who could benefit from extra support to facilitate the military-to-civilian transition.

Abbreviations used in this article:

Army STARRS

Army Study to Assess Risk and Resilience in Servicemembers

PPDS

Pre/Post Deployment Study

STARRS-LS1

wave 1 of the Study to Assess Risk and Resilience in Servicemembers Longitudinal Study