for WIZARD KIM: Review Paper—Topic Proposal & Reference Page

O R I G I N A L P A P E R

The role of stress sensitization in progression of posttraumatic distress following deployment

Geert E. Smid • Rolf J. Kleber • Arthur R. Rademaker •

Mirjam van Zuiden • Eric Vermetten

Received: 29 December 2012 / Accepted: 9 May 2013 / Published online: 29 May 2013

� Springer-Verlag Berlin Heidelberg 2013

Abstract

Purpose Military personnel exposed to combat are at risk

for experiencing post-traumatic distress that can progress

over time following deployment. We hypothesized that

progression of post-traumatic distress may be related to

enhanced susceptibility to post-deployment stressors. This

study aimed at examining the concept of stress sensitiza-

tion prospectively in a sample of Dutch military personnel

deployed in support of the conflicts in Afghanistan.

Method In a cohort of soldiers (N = 814), symptoms of

post-traumatic stress disorder (PTSD) were assessed before

deployment as well as 2, 7, 14, and 26 months (N = 433;

53 %) after their return. Data were analyzed using latent

growth modeling. Using multiple group analysis, we

examined whether high combat stress exposure during

deployment moderated the relation between post-deploy-

ment stressors and linear change in post-traumatic distress

after deployment.

Results A higher baseline level of post-traumatic distress

was associated with more early life stressors (standardized

regression coefficient = 0.30, p \ 0.001). In addition, a stronger increase in posttraumatic distress during deploy-

ment was associated with more deployment stressors

(standardized coefficient = 0.21, p \ 0.001). A steeper linear increase in posttraumatic distress post-deployment

(from 2 to 26 months) was predicted by more post-

deployment stressors (standardized coefficient = 0.29,

p \ 0.001) in high combat stress exposed soldiers, but not in a less combat stress exposed group. The group difference

in the predictive effect of post-deployment stressors on

progression of post-traumatic distress was significant

(v2(1) = 7.85, p = 0.005). Conclusions Progression of post-traumatic distress fol-

lowing combat exposure may be related to sensitization to

the effects of post-deployment stressors during the first

year following return from deployment.

Keywords Post-traumatic stress disorder � Delayed onset � Stress sensitization � Military deploymentPaper presented at The International Society for Traumatic Stress

Studies 28th Annual Meeting, Los Angeles, November 1–3, 2012.

Electronic supplementary material The online version of this article (doi:10.1007/s00127-013-0709-8) contains supplementary material, which is available to authorized users.

G. E. Smid (&) � R. J. Kleber Foundation Centrum’45/Arq, Nienoord 5,

1112 XE Diemen, The Netherlands

e-mail: [email protected]

R. J. Kleber

Department of Clinical and Health Psychology,

Utrecht University, Utrecht, The Netherlands

A. R. Rademaker � M. van Zuiden � E. Vermetten Research Centre-Military Mental Health,

Ministry of Defense, Utrecht, The Netherlands

M. van Zuiden

Department of Psychiatry, Academic Medical Center,

University of Amsterdam, Amsterdam, The Netherlands

M. van Zuiden

Laboratory of Neuroimmunology and Developmental Origins

of Disease, University Medical Center Utrecht, Utrecht,

The Netherlands

E. Vermetten

Rudolf Magnus Institute of Neurosciences, University Medical

Center Utrecht, Utrecht, The Netherlands

123

Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754

DOI 10.1007/s00127-013-0709-8

Introduction

Deployed military personnel are at risk of developing

symptoms of posttraumatic stress disorder (PTSD) fol-

lowing deployment. Among Dutch soldiers, prevalence of

PTSD 5 months after deployment to Iraq was estimated at

3–4 % [1]. Among U.S. soldiers, PTSD prevalence

3–4 months after deployment to Afghanistan was estimated

at 6 % using strict criteria [2], and PTSD prevalence with

serious functional impairment 3 and 12 months after

deployment to Iraq at 6–12 % [3]. Among UK soldiers,

PTSD prevalence after deployment to Iraq and Afghanistan

was estimated at 4 % [4].

During the past decades, numerous studies have docu-

mented progression of PTSD symptoms following return

from deployment. In soldiers and veterans initially

reporting subthreshold levels of post-traumatic distress,

progression of post-traumatic distress may lead to delayed-

onset PTSD. The estimated prevalence of delayed-onset

PTSD in military populations varies across prospective

studies, equaling 3 % with onset between 3 and 9 months

after deployment [5], 6.5 % with onset between 4 and

18 months [6], 7 % with onset between 1 and 7 months [7],

and 3.5 % with onset between 1 and 6 years [8]. In a

prospective study covering 20 years following the 1982

Lebanon war [9], veterans endorsing combat-related PTSD

with delayed onsets 3–20 years following combat exposure

were identified. In this study, delayed-onset PTSD was

endorsed by 23.8 % of veterans who did not have stress

reactions during the war and who did not meet PTSD cri-

teria 1 year after the war.

A large cross-sectional population study [10] evidenced

an increased likelihood of delayed onset of PTSD follow-

ing exposure to military combat as compared to other

potentially traumatic events. The association between

military combat and delayed-onset PTSD was subsequently

confirmed in a meta-analysis [11]. Consistent with these

findings, some prospective studies noted an overall trend of

symptom increase after combat exposure, i.e., an increase

of point prevalence of PTSD across subsequent assess-

ments. Following deployment to Iraq, a large sample

(N = 88,235) of US soldiers were screened immediately

after return and 3–6 months later [12]. Prevalence of

screening positive for PTSD increased across occasions

from 12 to 18 % in active soldiers, and from 13 to 25 % in

reserve component soldiers returning to civilian life. Pro-

spective studies of Gulf War veterans reported increases in

PTSD symptom levels over the course of 2 years following

return from deployment [13] as well as between 4 and

14 years following deployment [14]. Likewise, in UK

regular soldiers after deployment to Iraq and Afghanistan,

an increase in probable PTSD prevalence from 3.0 to 5.2 %

was reported between 1 and 6.5 years post-deployment [4].

Post-traumatic distress following deployment has been

shown to be related to several predictive factors. These

markers of risk can be categorized according to their

temporal origin in relation to the time of deployment. Thus,

pre-deployment (e.g., demographic), deployment-related,

and post-deployment risk markers can be outlined. Combat

exposure has been reported as the strongest deployment-

related predictor of subsequent posttraumatic distress [15].

Pre-deployment factors such as prior traumatic events [16]

and early life trauma [17] have also been shown to predict

posttraumatic distress. Recently, disciplinary offences

during military service, preceding any exposure to trauma,

were found to predict later delayed-onset PTSD [18].

Post-deployment stressors may additionally influence

the course of post-traumatic distress. In a sample of sol-

diers deployed to Iraq who were followed up between 2

and 6 months after deployment, post-deployment stressors

such as unemployment, broken relationships or illness of

significant others were independently associated with

PTSD symptom increases when controlling for baseline

(pre-deployment) symptoms and deployment stressors [19].

Also, 77 % of veterans endorsing delayed-onset PTSD

reported severe life stressors in the year preceding delayed

PTSD onset, against 32 % in veterans who reported no

PTSD [20]. Finally, in a prospective 20-year study in

veterans from the 1982 Lebanon war, post-war negative

life events were associated with delayed PTSD onset [21].

The effects of new stressors on posttraumatic distress

may operate in two different ways. First, the effect may be

simply additive, such that distress related to new stressors

adds to the posttraumatic distress. Second, an interactive

effect may occur if extreme traumatic exposure influences

the intensity with which survivors respond to subsequent

stressors. Indeed, exposure to extreme stressors may

enhance an individual’s reactivity to subsequent stressors, a

process that has been labeled sensitization to stress [22].

Sensitization refers to the situation in which an organism

responds more strongly to a variety of stimuli after expo-

sure to a potentially threatening or noxious stimulus. It

represents a form of non-associative learning that is likely

to cover a repertoire of mechanisms [23]. Sensitized

reactions may be both non-specific (e.g., depressed mood)

and specific to the stimulus that caused the sensitization

(e.g., trauma-related symptoms in PTSD). A role of stress

sensitization in delayed-onset PTSD has been suggested

[24]. A recent study among residents affected by a fire-

works disaster [25] found that residents whose home was

completely destroyed responded with greater distress to

stressful life events reported 18–20 months following the

disaster than residents whose home was less damaged,

using a stratified analysis. These results suggest that stress

sensitization effects are likely to manifest in groups char-

acterized by considerable stressor exposure.

1744 Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754

123

Stress sensitization can be described as a three–variable

relationship in which change in posttraumatic distress over

time constitutes the dependent variable, recent stressors

(e.g., post-deployment stressful life events) represent a

direct causal variable predicting change in distress, and

prior stressors (e.g., high combat stress exposure) represent

a temporally preceding interaction variable moderating the

direct effects of recent stressors on change in distress.

Simultaneous modeling of longitudinal change over time

based on repeated assessments as well as causal and

interaction effects can be achieved using structural equa-

tion modeling.

The aim of the current study was to examine whether

progression of post-traumatic distress after return from

military deployment could be explained with the stress

sensitization hypothesis. Specifically, our research ques-

tions were (1) to what extent are high combat stress

exposed (HCSE) soldiers more likely to manifest stress

sensitization, i.e., increased reactivity to post-deployment

stressors reported 1 year following deployment compared

with low combat stress exposed (LCSE) soldiers? (2) Is

stress sensitization likely to be involved in progression of

post-traumatic distress following deployment?

We hypothesized that pre-deployment levels of post-

traumatic distress and change in post-traumatic distress

during deployment would be predicted by early life

trauma and deployment stressors, respectively. We also

hypothesized that change in posttraumatic distress (slope)

over 2 years following deployment would be predicted

by life stressors reported 1 year following return from

deployment in HCSE soldiers, but not in LCSE soldiers.

Thus, we hypothesized a sensitization effect in the HCSE

group.

Methods

Participants and Procedures

This study is part of a prospective cohort study in the Dutch

Department of Defense. From 2006 to 2010, The Nether-

lands deployed approx. 20,000 soldiers to Afghanistan as

part of an International Security Assistance Force (ISAF).

Participants in this study volunteered to participate prior to

a 4-month deployment to Afghanistan between 2006 and

2008. Duties during deployment consisted of combat

patrols, clearing or searching buildings, participation in de-

mining operations, and transportation across enemy terri-

tory. Combat experiences included exposure to enemy fire,

participation in armed combat, seeing seriously injured

comrades and civilians, and witnessing the death of fellow

soldiers and civilians. The study comprised 5 assessments:

approximately 2 months prior to deployment (T1) and

approximately 2 (T2), 7 (T3), 14 (T4), and 26 months (T5)

following return from deployment. The first three assess-

ments took place at military bases, and the last two

assessments were mailed-in. At T1, N = 814 constituted

the initial sample. The number of participants at T2 was

N = 693 (85.1 % of the initial sample); at T3: N = 644

(79.1 %); at T4: N = 465 (57.1 %); and at T5: N = 433

(53.2 %). Of the 814 participants at T1, N = 345 (42.4 %)

participated in all 5 assessments; N = 146 (17.9 %) par-

ticipated in 4, N = 167 (20.5 %) in 3, N = 82 (10.1 %) in

2, and N = 74 (9.1 %) in 1 assessment. The study was

approved by the Institutional Review Board of the Uni-

versity Medical Center Utrecht, The Netherlands. Written

consent was obtained after a written and verbal description

of the study.

Measures

Socio-demographic characteristics

Participants were asked about number of previous

deployments, rank during deployment, age during deploy-

ment, gender, and education level. In addition, participants

indicated whether their function during deployment was

inside the military base or compound only or outside the

compound as well. Participants whose function during

deployment was outside the compound comprised the

HCSE group, and those having worked inside the com-

pound the LCSE group.

Early life trauma

Exposure to potential traumatic experiences before the age

of 18 was assessed at T1 using the Early Trauma Inventory

(ETI) self-report short form [26], Dutch version [27]. The

ETI is designed to assess exposure to potential traumatic

experiences before the age of 18 years (general trauma,

physical abuse, emotional abuse and sexual abuse) and

consists of 27 dichotomous items. The total score repre-

sents the number of experienced events.

Deployment stressors

At T2, exposure to potentially traumatic deployment

stressors was assessed with a 13-item Deployment Stress-

ors Checklist specifically developed for this study [28].

Items refer to specific events, for example, ‘‘Exposure to

enemy fire (yes/no)’’, ‘‘Being the target of enemy fire (yes/

no)’’, and ‘‘Being held at gunpoint (yes/no)’’ (the full text

of the checklist was published previously [28]). A total

score was obtained by summing affirmative responses.

Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754 1745

123

Post-deployment stressors

Exposure to stressful life events in the first year after

return from deployment was assessed at T4 using a

10-item yes/no checklist specifically developed for this

study (see the data supplement). The checklist comprised

items about divorce or broken relationship, accident or

assault to self or close other, severe illness to self or

close other, death of a close other, burglary or fire in

own home, financial problems, and being dismissed. A

total score was obtained representing the number of

endorsed stressors.

Self-report inventory for PTSD

PTSD symptom level over the past 4 weeks was assessed

with the 22-item Self-Report Inventory for PTSD (SRIP)

[29, 30]. A higher score indicates more PTSD symptoms

(range 22–88), i.e., higher levels of post-traumatic distress.

The SRIP has good concurrent validity with other PTSD

measures such as the Clinician Administered PTSD Scale

and Mississippi Scale for PTSD [30]. The SRIP does not

assess symptoms of PTSD with reference to a specific

event. Measurement invariance testing revealed that no

substantial differences in PTSD factor structure exist if

PTSD is assessed in trauma-exposed participants with vs.

without reference to a single traumatic event [31]. Partic-

ipants were assigned a probable PTSD diagnosis when their

score on the SRIP was C 38 [28]. This cutoff score cor-

responds to the mean plus two standard deviations, which

coincides with the 95th percentile of scores before

deployment within a population of 704 soldiers from the

Dutch Armed Forces [(mean: 26.91 SD = 5.34)]. The

validity of this cutoff score is supported by a study [29] that

compared SRIP scores with ratings from a diagnostic

interview within a community population of older adults,

showing adequate sensitivity and specificity.

Analyses

Analyses were performed using SPSS/Amos software

versions 20.0 and 17.0, respectively. For our structural

equation modeling (SEM) analyses, missing data due to

attrition were handled using the full information maximum

likelihood (FIML) procedure. It has been shown that under

ignorable missing data conditions, FIML estimates are

unbiased, and the bias in FIML parameter estimates is

relatively unaffected by the amount of missing data [32].

Descriptive analyses

Using Chi-square tests for categorical variables and t tests

for continuous variables, we evaluated differences between

the sample completing all assessments and the dropout

sample and between the HCSE and LCSE groups. Data

screening revealed moderate to severe non-normality in

PTSD symptom scores, early trauma, and number of pre-

vious deployments. This non-normality was expected

because low scores on these variables were the most

common, and higher scores increasingly rare. We, there-

fore, replicated our SEM analyses using Bayesian estima-

tion (see below).

Latent growth modeling (LGM)

Progression of post-traumatic distress may be flexibly

modeled using latent growth models. Interest of the present

analysis centered on changes during two distinct time

periods, specifically during deployment (from T1 to T2)

and following deployment (from T2 to T5). Piecewise

growth models can be used to subdivide a series of mea-

surements of PTSD into meaningful segments and to

summarize important aspects of change in each segment

[33]. We specified a piecewise growth model comprising

three factors: (1) pre-deployment (baseline) posttraumatic

stress symptom level (intercept); (2) linear change (slope)

in posttraumatic distress during deployment (pre-to-post);

and (3) linear change (slope) in posttraumatic distress after

deployment. The average posttraumatic stress symptom

level can be expressed using this model as the sum of the

pre-deployment level, the change during deployment, and

the change after deployment (slope * time after deploy-

ment). Therefore, the baseline level factor loaded on all

five indicators (T1–T5) with factor loadings set to 1 (see

Fig. 1). The slope during deployment factor loaded on the

four post-deployment indicators (T2–T5) with factor

loadings set to one to capture the change during the period

from pre- to post-deployment. The slope post-deployment

factor loaded on the indicators assessed at the 7–26-month

follow-up (T3 to T5). Factor loadings were set to 0.42, 1,

and 2, respectively, to capture the mean time in years since

the T2 assessment. Residual variances associated with the

T1 and T2 assessments were constrained to be equal to

enable the model to be identified. We evaluated model fit

using the discrepancy v2, comparative fit index (CFI), non- normed fit index (NNFI/TLI), root-mean-square error of

approximation (RMSEA), and Akaike information crite-

rion (AIC). Models that fit very well are indicated by CFIs

and NNFIs C 0.95 and RMSEAs B 0.06 [34].

Construction of a MIMIC model

To explore predictor effects on levels and course of post-

traumatic distress, we constructed a multiple indicators,

multiple causes (MIMIC) model [25]. MIMIC models are a

broad class of structural equation models where exogenous

1746 Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754

123

observed variables influence latent variables that in turn

have multiple indicators. We hypothesized that the early

trauma would predict pre-deployment distress; deployment

stressors would predict linear change during deployment;

and post-deployment stressors would predict linear change

post-deployment. To adjust for possible confounding

effects, we included gender, age, education, rank, and

number of previous deployments as covariates in the model

with paths to all factors.

Multiple group analysis

We applied multiple group analysis to test whether high

combat stress exposure moderated the relation between

post-deployment stressors and linear change in posttrau-

matic stress post-deployment. In multiple group analysis, a

model is estimated simultaneously across groups. Through

the specification of cross-group equality constraints, group

differences on any individual parameter or set of parame-

ters can be tested [35]. The fit of the model with parameters

constrained to be equal across groups is compared with that

of the unrestricted model with the Dv2 test [35]. Thus, we examined differences in paths and factor means between

the HCSE and LCSE groups by applying the MIMIC model

to these groups simultaneously and subsequently testing

models corresponding to increasingly restricted hypothe-

ses. Specifically, we subsequently applied cross-group

equality constraints corresponding to the following nil-

hypotheses: equality of early trauma effects on baseline

posttraumatic distress level; equality of deployment stres-

sor effects on slope during deployment; equality of post-

deployment stressor effects on slope post-deployment;

equality of adjusted baseline levels; equality of adjusted

slopes during deployment; and equality of adjusted slopes

post-deployment (see Table 1).

Bayesian replication

Following construction of our final models, we tested

robustness of the maximum likelihood (ML) estimation

method against normality violations in our data. Therefore,

we repeated all analyses using Bayesian estimation, which

is not based on normality assumptions [36]. For our

Bayesian implementation of SEM, we used the default

settings in Amos [36], including an uninformative flat prior

ranging from -3.4 * 10 38

to 3.4 * 10 38

. The Markov Chain

Monte Carlo (MCMC) sampling for the Bayesian estima-

tion was continued until subsequent runs were sufficiently

uncorrelated, i.e., when the value of the Gelman–Carlin–

Stern–Rubin convergence statistic was less than the con-

servative default value of 1.002. Because the Bayesian and

FIML estimates were essentially identical, we concluded

that ML estimation was robust against normality violations

in our data and chose to present FIML estimates. (The

Bayesian estimates are presented in Supplementary

Table 4.)

Results

Attrition and descriptive analyses

Study attrition was associated with the following baseline

characteristics: younger age (25.16 vs. 30.25, p \ 0.001), lower (soldier or corporal) rank (76.9 vs. 49.1 %,

p \ 0.001), lower education (51.2 vs. 28.6 %, p \ 0.001), and fewer previous deployments (0.67 vs. 1.07, p \ 0.001). In addition, study attrition was associated with working

outside the compound (79.2 % vs. 57.8 %, p \ 0.001), more deployment stressors (3.87 vs. 3.40, p = 0.030), and

higher PTSD symptoms at T3 (29.16 vs. 27.16, p \ 0.001)

27.08***

(26.66–27.49)

± 4.59

Baseline Level

1.12***

(0.68–1.56)

± 3.79

Slope During Deployment

PTSS T1

1

PTSS T2

1

1

PTSS T3

1

1

E1 1

E2 1

E3 1

-0.59***

(-0.87– -0.31)

± 2.35

Slope Post- Deployment

.42

PTSS T4

E4 1

1 1

1

-.05 -.25*

-.30***

PTSS T5

E5 1

2

1

1

.74 .81 .58 .55 .84

Fig. 1 Piecewise Growth Model. For the model:

v2(7) = 14.26, p = 0.047, CFI = 0.99, NNFI = 0.98,

RMSEA = 0.04, AIC = 40.26.

Factors (ovals) are shown with

unadjusted means (95 %

confidence intervals) ± SD,

factor loadings (arrows) with

fixed regression weights,

covariances (curved lines) with

correlations, and indicators

(squares) with explained

variance (R 2 ). PTSS

posttraumatic stress symptoms.

E1 to E5 represent residual error

variances. *p \ 0.05, **p \ 0.01, ***p \ 0.001

Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754 1747

123

as well as T4 (28.33 vs. 26.67, p = 0.036). Study partici-

pants who dropped out from one or more study assessments

did not differ from those who completed all assessments in

gender, early trauma, post-deployment stressors, and PTSD

symptoms at T1, T2, and T5. Missing data due to attrition

were handled using FIML, which is bias-free under

ignorable missing data conditions [32]. Because demo-

graphic variables (age, rank, education), number of previ-

ous deployments, and exposure variables associated with

study attrition were correlated, and analyses corrected for

demographic variables and number of previous deploy-

ments (see below) only showed negligible differences from

the unadjusted results, and because PTSD symptoms at

baseline did not predict study attrition, we concluded that

there were no indications of meaningful nonresponse bias.

Descriptive analyses are reported in Table 2. Of N = 455

participants providing sufficient data to establish a diagnosis

of probable deployment-related PTSD across at least four

assessments, n = 22 were excluded because they endorsed

PTSD before deployment. Participants who did not endorse

PTSD at any follow-up assessment (n = 368) constituted

85.0 % of the sample. Thirty-five participants (8.1 %)

endorsed probable PTSD at 2 months post-deployment.

Probable delayed-onset PTSD, defined as endorsing proba-

ble PTSD for the first time 7, 14, and/or 26 months after

deployment, was found in 30 participants (6.9 %).

Differences between high vs. low combat stress exposed

groups are also shown in Table 2. The HCSE group was

younger, more often male, lower educated and ranked than

the LCSE group. In addition, the HCSE group reported

fewer previous deployments, more deployment stressors,

and more post-deployment stressors than the LCSE group.

We, therefore, corrected for these variables in separate

analyses (see below). As expected, the HCSE group

reported higher PTSD symptoms at 1 and 6 months after

deployment, more probable PTSD at T2, and more prob-

able delayed-onset PTSD (see Table 2). The percentage of

participants meeting probable PTSD (including delayed-

onset PTSD) threshold levels in the HCSE group (19.0 %)

was over twice the percentage in the LCSE group (9.1 %).

Change in post-traumatic distress

during and after deployment

Figure 1 depicts the piecewise growth model for the whole

sample. As shown in Fig. 1, the mean increase in post-

traumatic distress during deployment was modest, equaling

1.12 points on the SRIP (95 % CI = 0.68–1.56, p \ 0.001). Following deployment, there was a significant linear

decrease in mean distress, equaling -0.59 points on the

SRIP each year (95 % CI = -0.87 to -0.31, p \ 0.001). The model fits the data well (see model fit indices in Fig. 1).

The three factors pre-deployment level, slope during

deployment, and slope post-deployment explained large

proportions of the variance in PTSD symptoms at all

assessments (55–85 %; see Fig. 1). Not withstanding, all

residual variances were significant (data not shown), indi-

cating that the longitudinal course of symptoms over time

departed from linearity in some individuals.

Stressor effects on post-traumatic distress

To examine effects of stressor exposure on pre-deployment

level of posttraumatic distress, slope during deployment, and

Table 1 Multiple group model selection

Nr. Ref. a

Cross-group equality constraints b,c

df v2 p Ddf Dv2 p CFI NNFI RMSEA AIC

1 – Unconstrained 38 67.03 0.003 0.97 0.95 0.03 167.03

2 1 Early trauma effects equal across groups 39 67.03 0.003 1 0.00 0.953 0.98 0.95 0.03 165.03

3 2 Deployment stressor effects equal

across groups

40 67.16 0.005 1 0.13 0.723 0.98 0.96 0.03 163.16

4 3 Post-deployment stressor effects equal across groups

41 75.01 0.001 1 7.85 0.005 0.97 0.95 0.03 169.01

5 3 Adjusted baseline level equal across

groups

41 67.62 0.006 1 0.46 0.496 0.98 0.96 0.03 161.62

6 5 Adjusted slope during deployment

equal across groups (Final)

42 68.48 0.006 1 0.86 0.353 0.98 0.96 0.03 160.48

7 6 Adjusted slope post-deployment equal across groups

43 72.54 0.003 1 4.06 0.044 0.97 0.96 0.03 162.54

Estimates from the final model are presented in Table 3 a

Reference model nr b

Models represent the reference model with equality constraint(s) added; the models shown in bold were rejected because of a significant

worsening of the model fit c

Groups: 1. Low combat stress exposed (N = 227), 2. High combat stress exposed (N = 520)

1748 Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754

123

slope post-deployment, we created a MIMIC model, i.e., the

piecewise growth model with added predictor variables. This

model fits the data well. A path diagram is shown in Fig. 2.

As expected, baseline level and slope during deployment of

post-traumatic distress were significantly associated with

early trauma (standardized regression weight = 0.30,

p \ 0.001) and exposure to deployment stressors (standard- ized regression weight = 0.21, p \ 0.001), respectively. For the sample as a whole, post-deployment stressors had no

significant effects on slope of PTSD symptoms post-

deployment (p = 0.114). Estimates from this model as well

as model fit indices are presented in Table 3.

Stress sensitization in high combat stress exposed

soldiers

To examine whether high combat stress exposure moder-

ated the relation between post-deployment stressors and

Table 2 Descriptive analyses

Total numbers vary due to missing responses a

Participants providing sufficient data to establish a diagnosis of probable deployment-related PTSD across at least four assessments, excluding

N = 22 who endorsed PTSD before deployment b

Differences between high vs. low combat stress exposed groups

* p \ 0.05 ** p \ 0.01 *** p \ 0.001

Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754 1749

123

linear change in post-traumatic distress post-deployment,

we applied the MIMIC model to the HCSE and LCSE

groups simultaneously and subsequently tested models

corresponding to increasingly restricted hypotheses. The

multiple group model selection is shown in Table 1. The

unconstrained model (i.e., paths and factors not set to be

equal across groups; model nr. 1 in Table 1) showed ade-

quate model fit. In model nr. 2, early trauma effects were

constrained to be equal across groups. The Dv2 test showed that the model fit did not worsen significantly, indicating

that early trauma effects on pre-deployment level were

equal across groups. Thus, this constraint was maintained

in subsequent models. The model constraining deployment

stressor effects on slope during deployment to be equal

across groups (nr. 3) again did not show worsening of

model fit, compared with the preceding model. However,

the model constraining post-deployment stressor effects on

slope post-deployment to be equal across groups (nr. 4)

showed a significant worsening of the model fit compared

with the preceding model. Therefore, model nr. 4 was

rejected and the constraint was not maintained in sub-

sequent models. The model constraining adjusted baseline

levels to be equal across groups (nr. 5) and the model

constraining adjusted slopes during deployment to be equal

across groups (nr. 6) again did not show a worsening of

model fit, compared with the respective preceding models.

The model constraining adjusted slopes post-deployment to

be equal across groups (nr. 7) showed a significant wors-

ening of the model fit compared with the preceding model.

Therefore, model nr. 7 was rejected, and model nr. 6

became the final most parsimonious model.

Estimates from the final multiple group model (model

nr. 6 in Table 1) are presented in Table 3. Early trauma

effects on baseline level and deployment stressor effects on

slope during deployment were equal across groups. Early

trauma explained 10 and 7 % of variance in the LCSE and

HCSE groups, respectively. Deployment stressors

explained 3 and 4 % of variance in slope during deploy-

ment in the LCSE and HCSE groups, respectively. Post-

deployment stressors effects on slope post-deployment

were not significant in the LCSE group. In contrast, in the

HCSE group, slope post-deployment was strongly pre-

dicted by post-deployment stressors (standardized regres-

sion weight = 0.29, p \ 0.001). Within the HCSE group, for each reported post-deployment stressor, the mean SRIP

score increased by 0.88 points (95 % CI = 0.37–1.38,

p \ 0.001) per year. Post-deployment stressors explained 4 % of variance in slope post-deployment in the LCSE

group, against 8 % in the HCSE group.

Factor means and confidence intervals are also reported

in Table 3. The factors representing baseline level of post-

traumatic distress (adjusted for early trauma) and slope

during deployment (adjusted for deployment stressors)

were equal across groups. The factor representing the slope

post-deployment (adjusted for post-deployment stressors)

differed between the LCSE and HCSE groups. Within the

HCSE group, the adjusted mean SRIP score decreased 1.22

points per year (95 % CI = -1.84 to -0.60, p \ 0.001). (Note that adjusted in this context means: when post-

deployment stressor effects equal zero.) This decrease was

steeper than in the LCSE group, where the adjusted mean

SRIP score decreased 0.44 points per year (95 % CI =

-0.77 to -0.06, p = 0.023). This difference appears to

represent regression of elevated PTSD scores to the mean,

reflecting an overall tendency towards recovery and resil-

ience. Thus, HCSE soldiers were likely to recover from

combat-related distress in the absence of post-deployment

stressors. In the presence of post-deployment stressors,

they were likely to experience persistence or progression of

post-traumatic distress during the 2 years following return

from deployment.

Analyses adjusted for demographic variables

and previous deployments

To adjust for possible confounding effects, we included

gender, age, education, rank, and number of previous

deployments as covariates in the MIMIC and multiple group

models with paths to all factors. Because the relevant results

of adjusted analyses only showed negligible differences from

Baseline Level

Slope During Deployment

Slope Post Deployment

Early Trauma

Deployment Stressors

Post- Deployment

Stressors

.31*** / .27*** .18*** / .21*** -.20 / .29***

.10 / .07 .03 / .04 .04 / .08

Fig. 2 Final Multiple Group Model. For the model: v2(42) = 68.48, p = 0.006, CFI = 0.98, NNFI = 0.96, RMSEA = 0.03, AIC =

160.48. Factors (ovals) are shown with explained variance (R 2 ), paths

with standardized regression weights for the LCSE/HCSE groups,

respectively. Curved lines represent covariances. Indicators PTSS T1 to

T5 (see Fig. 1) as well as residual error variances are not shown. See

Table 3 for factor intercepts and unstandardized regression weights

1750 Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754

123

the unadjusted results, we concluded there were no con-

founding effects of demographic variables and previous

deployments. The results of the adjusted analyses are pre-

sented in Supplementary Table 3. The multiple group model

selection is presented in Supplementary Table 2.

Discussion

Our results demonstrate that high combat stress exposed

(HCSE) soldiers responded more strongly to stressful life

events during the first year after deployment compared

with low combat stress exposed (LCSE) soldiers. Specifi-

cally, post-deployment stressors predicted persistence or

progression of post-traumatic distress in HCSE soldiers,

whereas no such predictive effects were found in LCSE

soldiers. Consistent with earlier findings, more early life

trauma was associated with higher baseline levels of post-

traumatic distress [17], and more exposure to deployment

stressors was associated with stronger increases in post-

traumatic distress during deployment [15].

To the best of our knowledge, our study is the first to

provide direct evidence that stress sensitization serves well

as a model explaining progression of post-traumatic dis-

tress. In addition, our findings provide prospective support

for the concept of the stress sensitization. At present, most

of the evidence in support of sensitization is based on

studies of trauma-exposed populations, showing elevated

Table 3 Model estimates

Parameter Mean/coeff. (95 % CI) Std. coeff. p

Full sample (N = 814)

Piecewise growth model a

Baseline level 27.08 (26.66 to 27.49) – 0.000

Slope during deployment 1.12 (0.68 to 1.56) – 0.000

Slope post-deployment -0.59 (-0.87 to -0.31) – 0.000

MIMIC model b

Early trauma ? baseline level 0.48 (0.36 to 0.60) 0.30 0.000

Deployment stressors ? slope during deployment 0.31 (0.16 to 0.45) 0.21 0.000

Post-deployment stressors ? slope post-deployment 0.27 (-0.06 to 0.60) 0.11 0.114

Baseline level 25.55 (24.99 to 26.11) – 0.000

Slope during deployment -0.12 (-0.85 to 0.61) – 0.745

Slope post-deployment -0.78 (-1.16 to -0.39) – 0.000

Multiple group model c

LCSE group (N = 227)

Early trauma ? baseline level 0.46 (0.34 to 0.59) 0.31 0.000

Deployment stressors ? slope during deployment 0.33 (0.18 to 0.48) 0.18 0.000

Post-deployment stressors ? slope post-deployment 20.19 (20.53 to 0.14) 20.20 0.114

Baseline level 25.60 (25.02 to 26.19) – 0.000

Slope during deployment -0.29 (-1.01 to 0.43) – 0.424

Slope post-deployment 20.41 (20.77 to 20.06) – 0.023

HCSE group (N = 520)

Early trauma ? Baseline level 0.46 (0.34 to 0.59) 0.27 0.000

Deployment stressors ? Slope during deployment 0.33 (0.18 to 0.48) 0.21 0.000

Post-deployment stressors ? Slope post-deployment 0.88 (0.37 to 1.38) 0.29 0.000

Baseline level 25.60 (25.02 to 26.19) – 0.000

Slope during deployment -0.29 (-1.01 to 0.43) – 0.424

Slope post-deployment 21.22 (21.84 to 20.60) – 0.000

Coeff coefficient, std standardized

Means indicate mean adjusted self-rating inventory for PTSD total scores a

For the model: v2(7) = 14.26, p = 0.047, CFI = 0.99, NNFI = 0.98, RMSEA = 0.04, AIC = 40.26. See path diagram in Fig. 1 b

For the model: v2(19) = 43.44, p = 0.001, CFI = 0.98, NNFI = 0.96, RMSEA = 0.04, AIC = 93.44. See path diagram in Fig. 2 c

For the model: v2(42) = 68.48, p = 0.006, CFI = 0.98, NNFI = 0.96, RMSEA = 0.03, AIC = 160.48. Parameters displayed in bold were not constrained to be equal across groups. See model selection in Table 1 (model nr. 6)

Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754 1751

123

risk of PTSD in individuals reporting prior trauma expo-

sure [37]. A recent prospective study highlighted the

strongly increased risk of PTSD following repeated sexual

victimization [38]. A prospective study in disaster survi-

vors found direct evidence for stress sensitization during

the first years following a disaster in survivors reporting

total home destruction due to the disaster [25]. In a pro-

spective study in young children, trauma-exposed children

with current life stressors had elevated internalizing and

externalizing problems compared with trauma-exposed

children without current stress and non-trauma-exposed

children with and without current stressors [39], consistent

with stress sensitization.

Stress sensitization may involve a repertoire of neuro-

biological mechanisms [40]. Changes in functioning of the

hypothalamus–pituitary–adrenal axis, autonomic nervous

system, as well as brain regions associated with threat

detection and fear expression (e.g., amygdala) may be

implicated in sensitized (behavioral) responses after stres-

sor exposure. The functioning of these systems and circuits

can be altered by prolonged, severe or traumatic stress [41].

In addition, the functioning of these systems is dysregu-

lated in individuals who developed PTSD [42]. Moreover,

several recent studies have shown that vulnerabilities in the

functioning of these systems, as assessed prior to or shortly

after the trauma exposure leading to PTSD, are associated

with subsequent development of PTSD [28, 43]. Future

research is needed to indicate whether identification of risk

of stress sensitization based on stressor exposure may be

complemented by neurobiological markers of stress

sensitization.

Stress sensitization also involves cognitive processes

used for meaning attribution that may be understood within

Conservation of Resources (COR) theory [25]. COR theory

states that people strive to retain, protect, and build

resources and that what constitutes a stressor to them is the

potential or actual loss of these resources [44]. According

to this theory, resource loss is disproportionally more

salient than resource gain. Therefore, those who already

lack resources are more vulnerable to resource loss [44].

Soldiers exposed to direct combat stressors may be at

increased risk of resource loss, for example, due to nega-

tive perceptions of the mission [6] or deployment-related

distress or impairment. Progression of post-traumatic dis-

tress may thus reflect increased vulnerability to further

resource loss in combat-exposed soldiers.

The clinical relevance of the differences in stress sen-

sitivity between the LCSE and HCSE groups may be

illustrated by the finding that the percentage of participants

meeting probable PTSD (including delayed-onset PTSD)

threshold levels in the HCSE group (20.1 %) was over

twice the percentage in the LCSE group (9.3 %). In addi-

tion, in our sample, HCSE soldiers reported more stressors

post-deployment than LCSE soldiers. Higher levels of

posttraumatic distress in the HCSE group may in turn

confer an increased risk of interpersonal stressors due to

conflicts or violence [10]. Notably, more post-deployment

stressors do not by themselves explain their stronger effect

(i.e., a stronger effect of each reported stressor) on slope

post-deployment in the HCSE group.

Strengths of the current study include the true prospec-

tive design, the inclusion of a pre-deployment assessment,

as well as the duration of follow-up covering 2 years fol-

lowing return from deployment. However, we also

acknowledge some methodological limitations to this

study. First, the distress assessments were restricted to

participant self-report. It has been suggested that military

personnel may be reluctant to endorse genuine distress

from fear of stigma in the military context [6]. Therefore,

estimates of distress based on these data may be conser-

vative. Notwithstanding, mean PTSD scores in our sample

were far below recommended cutoff scores for probable

PTSD, indicating generally mild distress. Second, stressor

exposure was assessed by self-report questionnaire. Use of

interview assessment of exposure to stressful life events

could be considered more valid. Given the large size of our

sample, questionnaire assessments were more feasible than

interviews. Third, differential attrition in our study repre-

sents another potential limitation. Given frequently high

rates of attrition in trauma research [45], our completion

rates are acceptable. Importantly, symptoms of posttrau-

matic distress at the initial assessment did not predict

attrition, and our adjusted analyses suggested that there

were no meaningful effects of differential attrition on our

results. Fourth, there were important differences between

the HCSE vs. LCSE groups in gender, age, education, rank,

and number of previous deployments. To prevent con-

founding, we corrected for these differences in our

analyses.

Our findings have several implications for practice. In

addition to levels of distress, levels of stressor exposure

during and following deployment play a central role in

explaining the trajectory of deployment-related distress.

We feel that these should, therefore, be routinely assessed

by clinicians dealing with soldiers as well as veterans. Our

results suggest that the absence of prominent signs and

symptoms of distress immediately following deployment

does not preclude distress in the long term that may in part

still be related to the deployment. Early interventions may

effectively be targeted at high risk groups based on combat

exposure [46]. When soldiers and veterans are seeking

mental health care at later stages following deployment,

progression of posttraumatic distress should seriously be

considered.

The duration of the sensitization effects may depend on

the intensity as well as the recency of combat stress

1752 Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754

123

exposure [25]; this would be an important avenue for fur-

ther research. Such research could provide arguments to

evaluate the length between deployments to reduce mental

health risks for combat-exposed soldiers. Foreseeable

stressors and resource losses, including unemployment and

physical impairments, may be an effective target for sec-

ondary prevention of psychological distress in military

personnel. Availability of practical assistance following

return from deployment tailored to individual concerns

related to the deployment and reintegration is, therefore,

essential. Health care providers, social workers, chaplains,

commanders, and peer supporters dealing with recently

combat-exposed soldiers should be aware that they may

show increased responsiveness to subsequent stressful life

events. They may signal possible signs of posttraumatic

distress, for example, mood instability [47], social with-

drawal, increased alcohol use, indifference, disciplinary

measures, and thrill-seeking behavior, and motivate indi-

viduals to seek appropriate care.

Acknowledgments Funded by a grant from the Dutch Ministry of Defense. The funders had no role in the design and conduct of the

study; collection, management, analysis, and interpretation of the

data; and preparation, review, or approval of the manuscript. Geert E.

Smid, MD, PhD had full access to all of the data in the study and

takes responsibility for the integrity of the data and the accuracy of

the data analysis. The authors thank Col. Kees IJzerman, MD, MPH

for his valuable comments and the commanders and troops for their

time and effort. The authors also thank Kim Kroezen, Anne Muilwijk,

Capt. Maurits Baatenburg de Jong, Capt. Jessie Smulders, Lt. Martijn

Derks, and Sgt. Loes van den Boomen for organizing the data

acquisition.

Conflict of interest None.

References

1. Engelhard IM, van den Hout MA, Weerts J, Arntz A, Hox JJCM,

McNally RJ (2007) Deployment-related stress and trauma in

Dutch soldiers returning from Iraq: prospective study. Br J Psy-

chiatry 191:140–145

2. Hoge CW, Castro CA, Messer SC, McGurk D, Cotting DI,

Koffman RL (2004) Combat duty in Iraq and Afghanistan, mental

health problems, and barriers to care. N Engl J Med 351:13–22

3. Thomas JL, Wilk JE, Riviere LA, McGurk D, Castro CA, Hoge

CW (2010) Prevalence of mental health problems and functional

impairment among active component and National Guard soldiers

3 and 12 months following combat in Iraq. Arch Gen Psychiatry

67:614–623

4. Fear NT, Jones M, Murphy D, Hull L, Iversen AC, Coker B,

Machell L, Sundin J, Woodhead C, Jones N, Greenberg N,

Landau S, Dandeker C, Rona RJ, Hotopf M, Wessely S (2010)

What are the consequences of deployment to Iraq and Afghani-

stan on the mental health of the UK armed forces? A cohort

study. Lancet 375:1783–1797

5. Dickstein BD, Suvak M, Litz BT, Amy BA (2010) Heterogeneity

in the course of posttraumatic stress disorder: trajectories of

symptomatology. J Trauma Stress 23:331–339

6. Gray MJ, Bolton EE, Litz BT (2004) A longitudinal analysis of

PTSD symptom course: delayed-onset PTSD in Somalia peace-

keepers. J Consult Clin Psychol 72:909–913

7. Grieger TA, Cozza SJ, Ursano RJ, Hoge C, Martinez PE, Engel

CC, Wain HJ (2006) Posttraumatic stress disorder and depression

in battle-injured soldiers. Am J Psychiatry 163:1777–1783

8. Goodwin L, Jones M, Rona RJ, Sundin J, Wessely S, Fear NT

(2012) Prevalence of delayed-onset posttraumatic stress disorder

in military personnel: is there evidence for this disorder?: results

of a prospective UK cohort study. J Nerv Ment Dis 200:429–437

9. Solomon Z, Mikulincer M (2006) Trajectories of PTSD: a

20-year longitudinal study. Am J Psychiatry 163:659–666

10. Prigerson HG, Maciejewski PK, Rosenheck RA (2001) Combat

trauma: trauma with highest risk of delayed onset and unresolved

post-traumatic stress disorder symptoms, unemployment, and

abuse among men. J Nerv Ment Dis 189:99–108

11. Smid GE, Mooren TTM, Van der Mast RC, Gersons BPR, Kleber

RJ (2009) Delayed post-traumatic stress disorder: systematic

review, meta-analysis, and metaregression analysis of prospective

studies. J Clin Psychiatry 70:1572–1582

12. Milliken CS, Auchterlonie JL, Hoge CW (2007) Longitudinal

assessment of mental health problems among active and reserve

component soldiers returning from the Iraq war. JAMA

298:2141–2148

13. Thompson KE, Vasterling JJ, Benotsch EG, Brailey K, Constans

J, Uddo M, Sutker PB (2004) Early symptom predictors of

chronic distress in Gulf War veterans. J Nerv Ment Dis

192:146–152

14. Li B, Mahan CM, Kang HK, Eisen SA, Engel CC (2011) Lon-

gitudinal health study of US 1991 Gulf War Veterans: changes in

health status at 10-year follow-up. Am J Epidemiol 174:761–768

15. Kang HK, Natelson BH, Mahan CM, Lee KY, Murphy FM

(2003) Post-traumatic stress disorder and chronic fatigue syn-

drome-like illness among Gulf war veterans: a population-based

survey of 30,000 veterans. Am J Epidemiol 157:141–148

16. Smith TC, Wingard DL, Ryan MAK, Kritz-Silverstein D, Slymen

DJ, Sallis JF, For the Millennium Cohort Study Team (2008)

Prior assault and posttraumatic stress disorder after combat

deployment. Epidemiology 19:505–512

17. Smith TC, Ryan MAK, Wingard DL, Slymen DJ, Sallis JF, Kritz-

Silverstein D (2008) New onset and persistent symptoms of post-

traumatic stress disorder self reported after deployment and

combat exposures: prospective population based US military

cohort study. BMJ 336:366–371

18. Brewin CR, Andrews B, Hejdenberg J, Stewart L (2012)

Objective predictors of delayed-onset post-traumatic stress dis-

order occurring after military discharge. Psychol Med

42:2119–2126

19. Vasterling JJ, Proctor SP, Friedman MJ, Hoge CW, Heeren T,

King LA, King DW (2010) PTSD symptom increases in Iraq-

deployed soldiers: comparison with non-deployed soldiers and

associations with baseline symptoms, deployment experiences,

and post-deployment stress. J Trauma Stress 23:41–51

20. Andrews B, Brewin CR, Stewart L, Philpott R, Hejdenberg J

(2009) Comparison of immediate-onset and delayed-onset post-

traumatic stress disorder in military veterans. J Abnorm Psychol

118:767–777

21. Horesh D, Solomon Z, Zerach G, Ein-Dor T (2011) Delayed-

onset PTSD among war veterans: the role of life events

throughout the life cycle. Soc Psychiatry Psychiatr Epidemiol

46:863–870

22. Post RM, Weiss SRB (1998) Sensitization and kindling phe-

nomena in mood, anxiety, and obsessive- compulsive disorders:

the role of serotonergic mechanisms in illness progression. Biol

Psychiatry 44:193–206

Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754 1753

123

23. Kandel ER, Schwartz JH (1982) Molecular biology of learning:

modulation of transmitter release. Science 218:433–443

24. McFarlane A (2010) The long-term costs of traumatic stress:

intertwined physical and psychological consequences. World

Psychiatry 9:3–10

25. Smid GE, Van der Velden PG, Lensvelt-Mulders GJLM,

Knipscheer JW, Gersons BPR, Kleber RJ (2012) Stress sensiti-

zation following a disaster: a prospective study. Psychol Med

42:1675–1686

26. Bremner JDM, Bolus RP, Mayer EAM (2007) Psychometric

properties of the early trauma inventory-self report. J Nerv Ment

Dis 195:211–218

27. Rademaker AR, Vermetten E, Geuze E, Muilwijk A, Kleber RJ

(2008) Self-reported early trauma as a predictor of adult per-

sonality: a study in a military sample. J Clin Psychol 64:863–875

28. van Zuiden M, Geuze E, Willemen HLDM, Vermetten E, Maas

M, Heijnen CJ, Kavelaars A (2011) Pre-existing high glucocor-

ticoid receptor number predicting development of post-traumatic

stress symptoms after military deployment. Am J Psychiatry

168:89–96

29. van Zelst WH, de Beurs E, Beekman ATF, Deeg DJH, Bramsen I,

van Dyck R (2003) Criterion validity of the self-rating inventory

for post-traumatic stress disorder (SRIP) in the community of

older adults. J Affect Disord 76:229–235

30. Hovens JE, Bramsen I, Van der Ploeg HM (2002) Self-rating

inventory for post-traumatic stress disorder: review of the psy-

chometric properties of a new brief Dutch screening instrument.

Percept Mot Skills 94:996–1008

31. Elhai JD, Engdahl RM, Palmieri PA, Naifeh JA, Schweinle A,

Jacobs GA (2009) Assessing posttraumatic stress disorder with or

without reference to a single, worst traumatic event: examining

differences in factor structure. Psychol Assess 21:629–634

32. Enders CK, Bandalos DL (2001) The relative performance of full

information maximum likelihood estimation for missing data in

structural equation models. Struct Equ Model 8:430–457

33. Duncan TE, Duncan SC, Strycker LA (2006) An Introduction to

Latent Variable Growth Curve Modeling: Concepts, Issues, and

Applications. Lawrence Erlbaum Associates, Mahwah (NJ)

34. Hu LT, Bentler PM (1999) Cutoff criteria for fit indexes in

covariance structure analysis: conventional criteria versus new

alternatives. Struct Equ Model 6:1–55

35. Kaplan DW (2008) Structural Equation Modeling: Foundations

and Extensions. Sage, Thousand Oaks (CA)

36. Arbuckle JL (2009) Amos 18.0 User’s Guide. Amos Develop-

ment Corporation, Crawfordville, Florida

37. Breslau N, Chilcoat HD, Kessler RC, Davis GC (1999) Previous

exposure to trauma and PTSD effects of subsequent trauma:

results from the Detroit area survey of trauma. Am J Psychiatry

156:902–907

38. Walsh K (2012) National prevalence of posttraumatic stress dis-

order among sexually revictimized adolescent, college, and adult

household-residing women. Arch Gen Psychiatry 69:935–942

39. Grasso DJ, Ford JD, Briggs-Gowan MJ (2012) Early life trauma

exposure and stress sensitivity in young children. J Pediatr Psy-

chol 38:94–103

40. Stam R (2007) PTSD and stress sensitisation: a tale of brain and

body: part 1: human studies. Neurosci Biobehav Rev 31:530–557

41. Van Wingen GA, Geuze E, Vermetten E, Fernández G (2011)

Consequences of combat stress on brain functioning. Mol Psy-

chiatry 16:583

42. Morris MC, Compas BE, Garber J (2012) Relations among post-

traumatic stress disorder, comorbid major depression, and HPA

function: a systematic review and meta-analysis. Clin Psychol

Rev 32:301–315

43. van Zuiden M, Geuze E, Willemen HLDM, Vermetten E, Maas

M, Amarouchi K, Kavelaars A, Heijnen CJ (2012) Glucocorticoid

receptor pathway components predict post-traumatic stress dis-

order symptom development: a prospective study. Biol Psychia-

try 71:309–316

44. Hobfoll SE (2001) The influence of culture, community, and the

nested-self in the stress process: advancing conservation of

resources theory. Appl Psychol 50:337–421

45. Scott CK, Sonis J, Creamer M, Dennis ML (2006) Maximizing

follow-up in longitudinal studies of traumatized populations.

J Trauma Stress 19:757–769

46. Adler AB, Bliese PD, McGurk D, Hoge CW, Castro CA (2009)

Battlemind debriefing and battlemind training as early interven-

tions with soldiers returning from Iraq: randomization by platoon.

J Consult Clin Psychol 77:928–940

47. Marwaha S, Parsons N, Broome M (2013) Mood instability,

mental illness and suicidal ideas: results from a household survey.

Soc Psychiatry Psychiatr Epidemiol (Online) 1–7

1754 Soc Psychiatry Psychiatr Epidemiol (2013) 48:1743–1754

123

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.