10
British Journal of Clinical Psychology (2013), 52, 82–91 © 2012 The British Psychological Society www.wileyonlinelibrary.com The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5 Maj Hansen*, Mathias Lasgaard and Ask Elklit Institute of Psychology, National Centre for Psychotraumatology, University of Southern Denmark, Odense M, Denmark Objective. Acute stress disorder (ASD) was introduced into the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric Association, 1994) to identify posttraumatic stress reactions occurring within the first month after a trauma and thus help to identify victims at risk of developing posttraumatic stress disorder (PTSD). Since its introduction, research into ASD has focused on the prediction of PTSD, whereas only a few studies have investigated the latent structure of ASD. Results of the latter have been mixed. In light of the current proposal for the ASD diagnosis in the pending DSM-5, there is a profound need for empirical studies that investigate the latent structure of ASD prior to the DSM-5 being finalized. Design. Based on previous factor analytic research, the DSM-IV, and the proposed DSM-5 formulation of ASD, four different models of the latent structure of ASD were specified and estimated. Method. The analyses were based on a national study of bank robbery victims (N = 450) using the acute stress disorder scale. Results. The results of the confirmatory factor analyses showed that the DSM-IV model provided the best fit to the data. Thus, the present study suggests that the latent structure of ASD may best be characterized according to the four-factor DSM-IV model of ASD (i.e., dissociation, re-experiencing, avoidance, and arousal) following exposure to bank robbery. Conclusions. The results are pertinent in light of the pending DSM-5 and add to the debate about the conceptualization of ASD. Practitioner Points The present study supports the DSM-IV conceptualization of ASD and thus underlines the need of further research into the ASD structure before the launching of the DSM-5. *Correspondence should be addressed to Maj Hansen, Institute of Psychology, National Centre for Psychotraumatology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark (e-mail: [email protected]). DOI:10.1111/bjc.12002 82

The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

  • Upload
    ask

  • View
    212

  • Download
    1

Embed Size (px)

Citation preview

Page 1: The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

British Journal of Clinical Psychology (2013), 52, 82–91

© 2012 The British Psychological Society

www.wileyonlinelibrary.com

The latent factor structure of acute stress disorderfollowing bank robbery: Testing alternativemodelsin light of the pending DSM-5

Maj Hansen*, Mathias Lasgaard and Ask ElklitInstitute of Psychology, National Centre for Psychotraumatology, University ofSouthern Denmark, Odense M, Denmark

Objective. Acute stress disorder (ASD) was introduced into the fourth edition of the

Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric

Association, 1994) to identify posttraumatic stress reactions occurring within the first

month after a trauma and thus help to identify victims at risk of developing posttraumatic

stress disorder (PTSD). Since its introduction, research into ASD has focused on the

prediction of PTSD, whereas only a few studies have investigated the latent structure of

ASD. Results of the latter have been mixed. In light of the current proposal for the ASD

diagnosis in the pending DSM-5, there is a profound need for empirical studies that

investigate the latent structure of ASD prior to the DSM-5 being finalized.

Design. Based on previous factor analytic research, the DSM-IV, and the proposed

DSM-5 formulation of ASD, four different models of the latent structure of ASD were

specified and estimated.

Method. The analyses were based on a national study of bank robbery victims

(N = 450) using the acute stress disorder scale.

Results. The results of the confirmatory factor analyses showed that theDSM-IVmodel

provided the best fit to the data. Thus, the present study suggests that the latent structure

of ASD may best be characterized according to the four-factor DSM-IV model of ASD

(i.e., dissociation, re-experiencing, avoidance, and arousal) following exposure to bank

robbery.

Conclusions. The results are pertinent in light of the pending DSM-5 and add to the

debate about the conceptualization of ASD.

Practitioner Points

� The present study supports the DSM-IV conceptualization of ASD and thus underlines

the need of further research into the ASD structure before the launching of the DSM-5.

*Correspondence should be addressed to Maj Hansen, Institute of Psychology, National Centre for Psychotraumatology,University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark (e-mail: [email protected]).

DOI:10.1111/bjc.12002

82

Page 2: The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

� Clinical theory and practice may be affected in several ways if future research, such as

this study, fails to support the proposed structure of ASD in the DSM-5 and new

proposals of ASD in the DSM-5 are not set forward and tested.

� Treatment of acute posttraumatic symptoms will likely become less effective if itfocuses on an imprecise conceptualization of ASD.

� Although possible, it seems unlikely that the results of the present study may simply

reflect properties of the ASDS rather than the ASDdiagnosis, because theDSM-IVmodel

has been supported in prior studies using both a diagnostic interview and the ASDS.

� The results of the present study are based on bank robbery victims and thus should be

generalized to other trauma populations with caution.

The diagnosis of acute stress disorder (ASD) was introduced into the fourth edition ofDiagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric

Association, 1994) based on theoretical assumptions and very little empirical research

(Bryant & Harvey, 1997). In light of the current proposal for the ASD diagnosis in the

pending DSM-5 (Bryant, Friedman, Spiegel, Ursano, & Strain, 2011), there is a profound

need for empirical studies that investigate the latent structure of ASD before the finalizing

of the DSM-5. Hence, the present study tests different models of the latent structure of

ASD, including the diagnostic structure in the DSM-IV and the DSM-5.

Acute stress disorder was introduced into the DSM-IV with a twofold purpose (Bryant& Harvey, 1997). The first purpose was to recognize posttraumatic stress reactions

occurring within the first month after a trauma and the second purpose was to identify

victims at risk of developing posttraumatic stress disorder (PTSD). According to the DSM-

IV, an ASD diagnosis requires that the individual shows an intense emotional reaction to

experiencing, witnessing, or being confronted with a traumatic event (criterion A). The

individual must endorse at least three dissociation symptoms (criterion B) and one

re-experiencing symptom (criterion C). Furthermore, the individual must show symp-

toms of marked avoidance (criterion D), marked arousal (criterion E), significant clinicaldistress or functional impairment (criterion F), and the symptoms must be present

between 2 days and 1 month after the trauma (criterion G). Studies have found the

prevalence of ASD to be between 7 and 28% with a mean rate of 13% following different

forms of traumatic exposure (Bryant et al., 2011).

Even though the ASD diagnosis has been a part of the DSM-IV for over 15 years, few

studies have investigated the factor structure of ASD. On the contrary, much research has

focused on the ability of ASD to predict PTSD following different types of traumatic

exposure. However, according to Bryant’s (2011) recent systematic review, whichincludes 22 studies (19 of adults), ASD does not predict PTSD adequately. In particular,

many studies have reported a low to moderate sensitivity (i.e., the majority of the

participants who develop PTSD did not initially suffer from ASD). Thus, Bryant et al.

(2011) recommended that the diagnosis of ASD in the DSM-5 should only describe acute

stress reactions and not be regarded as a precursor of PTSD. Furthermore, due to research

suggesting that the emphasis on dissociation in DSM-IV is overly restrictive and not

recognizing the heterogeneity of traumatic stress reactions, Bryant et al. (2011)

concluded that ASDmay be better conceptualized as the severity of acute stress reactionswithout requiring the presence of specific clusters. Hence, the proposed DSM-5

conceptualization of ASD requires eight symptoms from across the full symptom profile

rather than requiring a number of symptoms from specific clusters. It is notable that no

new symptoms have been added to the proposal for ASD in the DSM-5 as compared to the

symptoms in the DSM-IV. Thus, the conceptualization of ASD in the DSM-5may be viewed

Acute stress disorder factor structure 83

Page 3: The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

as a single symptom cluster (or a one-factor model) of the DSM-IV four symptom clusters

(or four-factor model).

The proposal for ASD in the DSM-5 has been put forward, even though only two ASD

exploratory factor analyses (EFA; Bryant, Moulds, & Guthrie, 2000; Cardena, Koopman,Classen, Waelde, & Spiegel, 2000) and four ASD confirmatory factor analyses have been

conducted (CFA; Armour, Elklit, & Shevlin, 2011; Brooks et al., 2008; Edmondson, Mills,

& Park, 2010; Wang, Li, Shi, Zhang, & Shen, 2010). These studies have all tested the DSM-

IV structure in comparison to alternative structures. Using EFA and the acute stress

disorder scale (ASDS), Bryant et al.’s (2000) study supported two different models of the

ASD structure using two different trauma samples. In a sample of fire victims (N = 107)

Bryant et al. (2000) found support for a four-factor structure somewhat consistent with

theDSM-IV conceptualization ofASD, and support for a three-factor structure (acute stressreactions [all re-experiencing, avoidance, and arousal symptoms], dissociative symptoms,

and dissociative amnesia) in a sample of mixed trauma victims (N = 99, primarily motor

vehicle accident [MVA] victims). Cardena et al. (2000) also investigated theASD structure

in fire victims (N = 187) using EFA. However, they investigated ASD using the Stanford

Acute Stress Reaction Questionnaire and did not include avoidance symptoms in their

analysis. Cardena et al. (2000) found some support for theDSM-IV conceptualizationwith

dissociation, re-experiencing, and arousal being largely distinct factors. Brooks et al.

(2008) conducted the first CFA of the ASD structure using hospital admitted patientsexposed to mixed traumas (N = 587, mainly MVA, ASD prevalence = 7.5%) based on the

Acute Stress Disorder Interview (Bryant, Harvey, Dang, & Sackville, 1998). The results

supported the DSM-IV model of ASD. Wang et al. (2010) tested different models of the

latent structure of ASD in Chinese earthquake victims (N = 353, ASD preva-

lence = 15.3%) using the ASDS. The DSM-IV model provided the best fit to the data.

Edmondson et al. (2010) tested different models of the latent structure of ASD in

hurricane evacuees (N = 132, ASD prevalence = 77%) using the ASDS. None of the

models showed good fit to the data. However, a two-factor hierarchical model containinga second-order distress factor (re-experiencing, arousal, and avoidance) and one-first-

order factor (dissociation) achieved an acceptable model fit when allowing for eight

measurement error correlations.

To the best of our knowledge, only one study has tested the proposed DSM-5model of

ASD. Armour et al. (2011) tested different models of the latent structure of ASD including

the DSM-IV and the DSM-5 models in female rape victims, who all suffered from ASD

(N = 263) using the ASDS. The DSM-5 model of ASD was tested as a one-factor model of

the ASDS. Noteworthy, the study did not support the DSM-5 nor the DSM-IV model as allmodels failed to meet the standards for adequate model fit. Moreover, there were no

significant differences between the best fitted models. However, on the basis of

parsimony and the lowest Bayesian information criterion Armour et al. (2011) concluded

that a three-factor model (dissociation, avoidance, and re-experiencing/arousal) was

preferential. It is, however, notable that the Armour et al. (2011) study was based solely

on female rape victims with ASD, which may have influenced the results. Thus, there is a

great need to test theDSM-5model and alternativemodels of ASD following other forms of

traumatic exposure.In sum, the results of the CFA studies are split in two-two studies support the DSM-IV

conceptualization of ASD (Brooks et al., 2008;Wang et al., 2010), whereas the other two

studies supported a two-factor and a three-factor model of ASD (Armour et al., 2011;

Edmondson et al., 2010). The reason for the discrepancy in the results may be that the

structure of ASD varies between trauma types or to the use of different assessment tools

84 Maj Hansen et al.

Page 4: The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

etc. However, it is clear that the ASD structure needs to be tested further, especially, in

light of the release of the DSM-5. Indeed, for this reason it is crucial that different

conceptualizations of ASD are tested. Moreover, there is a need for studies based on large

samples, preferably representative samples, as well as studies based on other forms oftraumatic exposure than disaster, rape, and MVA. In particular, there is a lack of studies

investigating the ASD structure following nonsexual assault and a lack of studies which

test the current proposal of ASD in the DSM-5.

The present study, therefore, aims to test different models of the ASD structure using

the ASDS and CFA in a national study of bank robbery victims, that is, a form of nonsexual

assault, which is becoming increasingly more common in Denmark and most of the

western world (the Danish Bankers Association, 2011; Gilioli, Campanini, Fichera, Punzi,

& Gassitto, 2006). Due to the close resemblance between ASDS items and the ASDsymptoms listed in the DSM-IV and the DSM-5 proposal, the results of this study may shed

light on aspects of both the ASDS and the ASD construct in itself. Hence, the present study

is a contribution to the limited research on the ASD factor structure and adds to the debate

about how ASD should be conceptualized in the DSM-5. Using an alternative models

Table 1. Model specifications for the alternative models of acute stress disorder

ASDS items

Model 1 Model 2 Model 3

One-factor

model

(Proposed

DSM-5)

Two-factor

hierarchical

model (Edmondson

et al., 2010)

Three-factor

model

(Armour

et al., 2011)

Model 4

Four-factor

model

(DSM-IV)

ASDS1 (Numbness) ASD DIS1st order factor DIS DIS

ASDS2 (Dazed) ASD DIS1st order factor DIS DIS

ASDS3 (Derealization) ASD DIS1st order factor DIS DIS

ASDS4 (Depersonalization) ASD DIS1st order factor DIS DIS

ASDS5 (Amnesia) ASD DIS1st order factor DIS DIS

ASDS6 (Intrusive memories) ASD RE2nd order factor * RE/ARO RE

ASDS7 (Nightmares) ASD RE2tnd order factor* RE/ARO RE

ASDS8 (Flashbacks) ASD RE2nd order factor* RE/ARO RE

ASDS9 (Distress on

reminders)

ASD RE2nd order factor* RE/ARO RE

ASDS10 (Thought avoidance) ASD AVO2nd order factor* AVO AVO

ASDS11 (Conversation

avoidance)

ASD AVO2nd order factor* AVO AVO

ASDS12 (Reminders avoidance) ASD AVO2nd order factor* AVO AVO

ASDS13 (Emotional avoidance) ASD AVO2nd order factor* AVO AVO

ASDS14 (Difficulty sleeping) ASD ARO2nd order factor* RE/ARO ARO

ASDS15 (Irritability) ASD ARO2nd order factor* RE/ARO ARO

ASDS16 (Difficulty

concentrating)

ASD ARO2nd order factor* RE/ARO ARO

ASDS17 (Hypervigilance) ASD ARO2nd order factor* RE/ARO ARO

ASDS18 (Startle response) ASD ARO2nd order factor* RE/ARO ARO

ASDS19 (Physiological reactivity) ASD ARO2nd order factor* RE/ARO ARO

Note. ASD, acute stress disorder; ASDS, acute stress disorder scale; DIS, dissociation; RE, re-experience;

ARO, arousal; AVO, avoidance.

*Second-order factor, distress, by the three-first-order factors RE, AVO, and ARO.

Acute stress disorder factor structure 85

Page 5: The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

approach, the present study set out to test fourmodels,whichwere based on the results of

the previous CFA studies, and the conceptualizations of ASD in the DSM-IV and the

proposed DSM-5. The four tested models were the one-factor model proposed for

inclusion in the DSM-5 (Model 1) as specified by Armour et al. (2011), a replication ofEdmondson et al.’s (2010) two-factor hierarchical model (Model 2), a replication of

Armour et al.’s (2011) three-factor model (Model 3), and the DSM-IV’s four-factor model

(Model 4). The specific structure of the four models is shown in Table 1.

Method

The present study is part of a large national study of the psychological impact of bank

robberies committed inDenmark fromApril 2010 to April 2011. The studywas conducted

in collaboration between the Danish Bankers Association, all Danish Banks, and the

University Southern of Denmark. During this period, 626 bank employees were at work

during a robbery. A total of 450 employees filled out the questionnaire (response rate:

73%) aweek after the robbery (M = 9.89 days, SD = 6.30, range = 2–30).A total of 91.3%

(n = 411) of the participants were exposed to robberies, which involved the use of

weapon. The majority of the participants (80.7%, n = 364) were present in the roomduring the robbery, whereas 19.3% of the participants (n = 86) were in adjoining rooms.

As argued by Hansen and Elklit (2011), the stressor criterion A1 is met during a bank

robbery regardless of the presence of the participants, because a bank robbery involves

actual or threatened death or serious injury or a threat to the physical integrity of self or

others (i.e., colleagues). Ages ranged from 19 to 65 years (M = 42.2, SD = 12.54). A total

of 273 participants (60.7%) were women. The participants were informed of the purpose

of the questionnaire orally and in writing and instructed to fill out the questionnaire in

relation to the index robbery. Participation was voluntary and all necessary permissionsfor conducting questionnaire surveys, according to Danish Law were obtained.

The questionnaire assessed demographic variables and numerous pre-, peri-, and

posttraumatic variables, however, describing these are unnecessary for the aims and

objectives of this study. The ASD latent structure was assessed using the Danish version of

theASDS (Bryant et al., 2000). TheASDS is a 19-item self-reportmeasurewith four subscales

assessing the four separate DSM-IV ASD symptom clusters of dissociation, re-experiencing,

arousal, and avoidance. Answers are rated on a 5-point Likert-type scale (1 = not at all, to

5 = very much). Previous studies using the Danish ASDS have reported good reliabilitycoefficients of .80, .82, and .96 for the total score (Armour et al., 2011; Elklit&Christiansen,

2010; Hansen & Elklit, 2011). The Cronbach’s alpha values in the present study were all

satisfactory (dissociation = .85, re-experiencing = .79, avoidance = .73, arousal = .86,

total score = .93). Furthermore, Bryant et al. (2000) reported high internal consistency

(.96) for the ASDS total score, high 2–7-day test-retest reliability (.94) for the total scale, andgood convergent validity. The ASD symptom clusters were met if the participants reported

at least one re-experiencing symptom, one avoidance symptom, and one arousal symptom

in addition to at least three dissociative symptoms, all indicated by item scores >3 on theASDS. This procedure has been used previously (cf., Elklit & Christiansen, 2010).

Results

Four alternative CFA-models of ASD were specified and tested using LISREL 8.8 (Joreskog

& Sorbom, 2006). A covariance matrix and asymptomatic weight matrix were computed

86 Maj Hansen et al.

Page 6: The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

using PRELIS 2.8 (Joreskog & Sorbom, 2006). An asymptomatic weight matrix allows for

weaker assumptions regarding the distribution of the observed variables and results in

improved fit and test statistics (Curran, West, & Finch, 1996; Satorra, 1992).

The models were estimated using maximum likelihood, and as recommend by Hoyleand Panter (1995), the goodness of fit for each model was assessed with a range of fit

indices including the Satorra-Bentler scaled chi-square (S-Bv2), the incremental fit index

(IFI), and the comparative fit index (CFI). A nonsignificant S-Bv2 and values greater than

.95 for the IFI and CFI are considered to reflect acceptable model fit. In addition, the root

mean-square error of approximation (RMSEA) was reported. A RMSEA value up to .08 is

considered acceptable (Joreskog & Sorbom, 1993). Furthermore, we included the

standard root mean-square residual (SRMR) which has been shown to be sensitive to

modelmisspecification (Hu&Bentler, 1999). SRMR values <.08 are considered to indicateacceptable model fit. Moreover, the expected cross validation index (ECVI) was used for

the purpose of model comparison with the smallest value being indicative of the best

fitting model. No correlated errors were included in the models.

Prior to data analysis the data were screened for errors. The percentage of missing

values was very small (0–0.9%). Thus, the missing data were imputed using the

expectation maximization algorithm, which has been demonstrated as an effective

method of dealing with missing data (Bunting, Adamson, & Mulhall, 2002). The mean

ASDS total score was 34.86 (SD = 13.52, range: 19–86). The mean scores on the ASDSsubscales were 8.91 (SD = 4.14, range: 5–24) for dissociation, 7.69 (SD = 3.33, range: 4–20) for re-experiencing, 6.08 (SD = 2.74, 4–18) for avoidance, and 12.18 (SD = 5.32,

range: 6–30) for arousal. A total of 53 participants (11.8%) suffered from an estimated ASD

according to DSM-IV criteria. Data were screened for sex differences. Women scored

significantly higher on the ASDS (M = 37.72, SD = 14.57) compared to men (M = 30.39,

SD = 10.28), t(444.08) = 6.25, p < .001 and significantly more women suffered from an

estimated ASD than men, v2(1,450) = 9.40, p = .002. Furthermore, data were screened

for differences in the ASDS total score and estimated ASD prevalence in relation to theproximity to danger and weapon use during the robbery. The participants present during

the robbery scored significantly higher on the ASDS (M = 35.56, SD = 14.09) compared

to the participants in adjoining rooms (M = 31.78, SD = 10.33), t(168.61) = 2.84,

p = .005. Significantly more participants who were present during the robbery also

suffered from an estimated ASD compared to the participants in adjoining rooms,

v2(1,450) = 4.38, p = .036. There were no significant differences between the partici-

pants exposed to robberies with or without weapon use in relation to the estimated ASD

prevalence or ASDS total score.The fit indices of the different models are reported in Table 2. On the basis of the

criteria associated with the RMSEA, SRMR, IFI, and CFI, Model’s 2, 3, and 4were judged to

exhibit reasonable model fit. The RMSEA-values of Model 2, 3, and 4 (i.e., 6.8–7.6) do not

indicate good model fit, but are acceptable, indicating reasonable errors of approxima-

tion. The Santora-Bentler chi-squares were statistically significant. However, this should

not lead to rejection of the models, because caution is exercised when interpreting the

chi-square as ameasure ofmodel fit as the large sample size increases the power of the test

(Tanaka, 1987). Comparing Model 2, Model 3, and Model 4, the RMSEA, ECVI, and SRMRvalues were in favour of Model 4. Also, Model 4 showed a significantly better fit when

comparedwith the two othermodels (S-B v2Δ = 26–67; 3 dfmore, both ps < .005). Thus,

itwas judged that the four-factormodel represents an adequate description of the data and

is the best of the alternative models. The standardized factor loadings of Model 4 ranged

from .52 to .81 (M = 0.70) and were all statistically significant (p < .05). The four factors

Acute stress disorder factor structure 87

Page 7: The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

correlated very highly (r = .73–.96). All factor correlations were statistically significant

(p < .05).

Discussion

Consistent with the DSM-IV conceptualization of ASD and two previous CFA studies

(Brooks et al., 2008; Wang et al., 2010), this study suggests that the ASD, as measured by

the ASDS, is best described as a four-factor structure of dissociation, re-experiencing,

avoidance, and arousal. This suggests that immediately after the experience of bankrobbery the ASDS seems to capture the degree of distress that the participants are

experiencing as a distinctive set of four symptom clusters. Thus, this study could not

support the one-factor structure of ASD containing all ASD symptoms as proposed for the

DSM-5, the hierarchical two-factor model of ASD containing a second-order distress factor

(re-experiencing, arousal, and avoidance) and a dissociation factor found in Edmondson

et al.’s study (2010), or the three-factor model (dissociation, avoidance, and

re-experiencing/arousal) found in Armour et al.’s (2011) study. There are several

possible explanations for the mixed findings. First, the discrepancy in results may be dueto variation in ASD prevalence and ASD severity measured by the ASDS between the

studies and, thus, differences between studies using clinical and nonclinical samples. The

studies supporting the DSM-IV model including the present study reported a lower ASD

prevalence (7.5–15.3%) than themuch higher ASD prevalence (77–100%) reported in the

two remaining CFA studies. Likewise, ASD severity was lower in the studies supporting

the DSM-IV model. Of the three previous CFA studies using the ASDS only two studies

stated the mean scores on the ASDS. The mean ASDS total scores were lower in Wang

et al.’s (2010) study (M = 31.22, SD = 10.14) and in the present study (M = 38.86,SD = 13.86) compared to the much higher mean ASDS total score in Edmondson et al.

(2010) study (M = 61.61, SD = 19.38). A clinical sample is likely to reflect the ASD

structure in a different way than a mixed sample, because a clinical sample is more

symptomatic and, thus, would be expected to produce higher factor loadings than a

mixed sample. However, this was not the case with the only study using a full clinical

sample so far (Armour et al., 2011). In contrast to the standardized factor loadings found

in the DSM-IV model in the present study, several of the standardized factor loadings for

the three-factor model (dissociation, avoidance, and re-experiencing/arousal) found in

Table 2. Fit indices for the alternative models of acute stress disorder

Index Model 1 Model 2 Model 3 Model 4

S-Bv2 686.69 539.95 477.02 449.36

df (p) 152 (.00) 149 (.00) 149 (.00) 146 (.00)

RMSEA 90% CI .089 (0.082–0.095) .076 (0.070–0.083) .070 (0.063–0.077) .068 (0.061–0.075)ECVI 90% CI 1.70 (1.53–1.89) 1.39 (1.24–1.55) 1.25 (1.11–1.40) 1.20 (1.06–1.35)IFI .96 .97 .98 .98

CFI .96 .97 .98 .98

SRMR .064 .065 .058 .056

Note. S-Bv2, Satorra-Bentler scaled chi-square; RMSEA, rootmean-square error of approximation; ECVI,

expected cross validation index; IFI, incremental fit index,CFI, comparative fit index; SRMR, standard root

mean-square residual.

88 Maj Hansen et al.

Page 8: The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

Armour et al.’s (2011) study were only low to moderate. Furthermore, although the

specification of the three-factor model in Armour et al.’s (2011) study was based on high

correlations found between the re-experiencing and arousal symptom cluster, the three-

factor model still failed to meet established guidelines for adequate model fit. This couldindicate that the results of the Armour et al. (2011) studymay be sample specific. Second,

the discrepancy in the results may be due to methodological issues. Indeed, it is notable

that only the studies supporting the DSM-IV four-factor model of ASD, including the

present study, had a sample size and number of parameter estimates (N:q) in the range

around 10:1 (Kline, 2011). Hence, the models tested by Armour et al. (2011) and,

especially, Edmondson et al. (2010)may be too complex relative to the sample size of the

studies. Moreover, Edmondson et al. (2010) allowed for multiple measurement error

correlations to achieve a better model fit, making the results of the study sample-specific.In addition, the study byArmour et al.’s (2011) only included female participants. Third, it

is also possible that different ASD structures exist across different traumapopulations, this

is because the DSM-IV model is supported following exposure to mixed traumas,

earthquake, and bank robbery, but not following exposure to hurricane and rape.

Importantly, we did not find support for the one-dimensional structure of ASD

put forward for the DSM-5. This negative finding suggests that there is a need for

further research into the ASD structure and the acute phase of different traumas

before ASD is finally conceptualized in the DSM-5. Clinical theory and practice maybe affected in several ways if future research fails to support the proposed structure

of ASD in the DSM-5 and new proposals are not set forward and tested. In other

words, the diagnostic categorization of ASD must be clearer to lead to a better

understanding of the immediate impact of trauma. Evidently, treatment will likely

become less effective if it focuses on an imprecise conceptualization of ASD. Thus,

the proposal of ASD in the DSM-5 should accommodate empirical findings and

change diagnostic criteria accordingly. Furthermore, although the current proposal

of ASD in the DSM-5 allows for more heterogeneity of responses by requiring eightsymptoms rather than specific symptom clusters being present, we argue that the

proposal of ASD in the DSM-5 more seems to be a one-dimensional conceptualization

of the ASD diagnosis found in DSM-IV rather than a disorder including the wider

range of acute reactions proposed by Bryant et al. (2011). Thus, it remains possible

that ASD can be conceptualized better than done, so far, in the DSM-IV and the DSM-5 by

including a wider range of acute reactions. However, a new proposal of the ASD

structure in the DSM-5 incorporating this wider range of reactions needs to be put

forward to make testing possible.Currently, the majority of CFA research into the latent structure of ASD (including the

present study) suggests that the ASD structure is best conceptualized as a four-factor

model that corresponds to the DSM-IV model. Thus, changes to the conceptualization of

ASD may seem unnecessary and may have possibly negative clinical implications if the

conceptualization of the immediate impact of trauma is imprecise and treatment becomes

less focused. Due to the close resemblance between the ASDS items and the ASD

symptoms listed in the DSM-IV (and the DSM-5), the results of the present study also

provide empirical support for the structural validity of the ASDS and suggest that the ASDScan be used to screen for andmeasure ASD following acute traumatic exposure. However,

as highlighted by Brooks et al. (2008) it is important to remember that results should not

be interpreted as support of the utility of the ASD diagnosis, because it is not enough to

consider factor structures of proposed diagnoses. Concurrent and predictive functions of

the ASD also need to be considered.

Acute stress disorder factor structure 89

Page 9: The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

The present study has several limitations. First, the ASD diagnosis was based on a self-

report measure, rather than a clinical interview. Although possible, it seems unlikely that

the results may reflect properties of the ASDS rather than the ASD diagnosis, because the

ASDS items are specifically matched to the ASD symptoms listed in the DSM-IV and theproposed DSM-5. Furthermore, the DSM-IV model has been supported previously using

both a diagnostic interview and theASDS (Brooks et al., 2008;Wang et al., 2010). Second,

given the nonclinical nature of this sample, it remains a possibility that the study captured

an underlying factor of individual responses to acute trauma making it difficult to

comment on diagnostic issues. Third, even though the DSM-IVmodel has been supported

following other trauma types, the results of the present study are based on bank robbery

victims and should be generalized to other trauma populationswith caution. Although the

results showed significant differences inASDS total score and estimatedASDprevalence inrelation to proximity to danger, due to power issues we were not able to assess any

differences in the latent structure of ASD in relation to proximity to danger. However, the

latent structure of ASD should represent heterogeneous psychological responses to

traumatic events and thus all types of bank robbery victims should be included when

analysing the latent structure of ASD following bank robbery. Fourth, the sample in the

present study was not large enough to allow for the different models to be tested for sex

differences.

Despite its limitations, the results of the present study add to the current debate aboutthe latent structure of ASDprior to the finalizing of the pendingDSM-5, by testing different

models of the ASD structure in a national sample of nonsexual assault victims (i.e., bank

robbery victims). The present study supported the DSM-IV and not the proposed DSM-5

conceptualization of ASD, suggesting that although the DSM-IV conceptualization of ASD

may not be a goodpredictor of PTSD, it seems to describe acute stress responses following

bank robbery. Thus, the results underline the need for further research into the ASD

structure before launching the DSM-5. In particular, there is a need to specify and test

alternative models of ASD before the finalization of the DSM-5. Future research in this areashould expand on the current study by overcoming the reported limitations by basing the

ASD diagnosis on a clinical interview, and by including other trauma populations with

varying ASD prevalence rates.

References

American Psychiatric Association. (1994).Diagnostic and statistical manual of mental disorders.

(4th ed.) Washington, DC: Author.

Armour, C., Elklit, A., & Shevlin, M. (2011). The latent structure of acute stress disorder.

Psychological Trauma. doi:10.1037/a0024848

Brooks, R., Silove, D., Bryant, R., O’Donnell, M., Creamer, M., & McFarlane, A. (2008).

A confirmatory factor analysis of the acute stress disorder interview. Journal of Traumatic

Stress, 21, 352–355. doi:10.1002/jts.20333Bryant, R. (2011). Acute stress disorder as a predictor of posttraumatic stress disorder. Journal of

Clinical Psychiatry, 72, 233–239. doi:10.4088/JCP.09r05072bluBryant, R. A., Friedman, M. J., Spiegel, D., Ursano, R., & Strain, J. (2011). A review of acute stress

disorder in DSM-5. Depression and Anxiety, 28, 1–16. doi:10.1002/da.20737Bryant, R.,&Harvey, A. (1997). Acute stress disorder.Clinical PsychologyReview,17, 757–773. doi:

10.1016/S0272-7358(97)00052-4

Bryant, R. A., Harvey, A. G., Dang, S. T., & Sackville, T. (1998). Assessing acute stress disorder.

Psychological Assessment, 10, 215–220. doi:10.1037/1040-3590.10.3.215

90 Maj Hansen et al.

Page 10: The latent factor structure of acute stress disorder following bank robbery: Testing alternative models in light of the pending DSM-5

Bryant, R. A., Moulds, M. L., & Guthrie, R. M. (2000). Acute stress disorder scale. Psychological

Assessment, 12, 61–68. doi:10.1037//I040-3590.12.1.61Bunting, B. P., Adamson, G., &Mulhall, P. (2002). AMonte Carlo examination of MTMMmodel with

planned incomplete data structures. Structural Equation Modelling, 9, 369–389. doi:10.1207/S15328007SEM0903_4

Cardena, E., Koopman, C., Classen, C., Waelde, L. C., & Spiegel, D. (2000). Psychometric properties

of the Stanford Acute Stress Reaction Questionnaire (SASRQ). Journal of Traumatic Stress, 13,

719–734. doi:10.1023/A:1007822603186Curran, P. J., West, S. G., & Finch, J. F. (1996). The robustness of test statistics to nonnormality and

specification error in confirmatory factor analysis. Psychological Methods, 1, 16–29. doi:10.1037/1082-989X.1.1.16

the Danish Bankers Association. (2011). Robbery statistics. Retrieved from The Danish Bankers

Association via http://www.finansraadet.dk/tal–fakta/statistik-og-tal/roeveristatistik.aspxEdmondson, D.,Mills,M. A., & Park, C. L. (2010). Factor structure of the acute stress disorder scale in

a sample of Hurricane Katrina evacuees. Psychological Assessment, 22, 269–278. doi:10.1037/a0018506

Elklit, A., & Christiansen, D. (2010). ASD and PTSD in rape victims. Journal of Interpersonal

Violence, 25, 1470–1488. doi:10.1177/0886260509354587Gilioli, R., Campanini, P., Fichera, G. P., Punzi, S., & Gassitto, M. G. (2006). Emerging aspects of

psychosocial risks: Violence and harassment at work. La Medicina del Lavoro, 97, 160–164.Hansen, M., & Elklit, A. (2011). Predictors of acute stress disorder in response to bank robbery.

European Journal of Psychotraumatology, 2. doi:10.3402)ejpt.v2i0.5864

Hoyle, R. H., & Panter, A. T. (1995). Writing about structural equation models. In R. H. Hoyle (Ed.),

Structural equation modelling: Concepts, issues and applications (pp. 158–198). London:Sage.

Hu, L., & Bentler, P. M. (1999). Cut-off criteria for fit indexes in covariance structure analysis:

Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. doi:10.1080/10705519909540118

Joreskog, K., & Sorbom, D. (1993). Structural equation modeling with the SIMPLIS command

language. Chicago: Scientific Software.

Joreskog, K., & Sorbom, D. (2006). LISREL 8.8. Chicago: Scientific Software.

Kline, R. B. (2011). Principles and practice in structural equationmodelling. (3th ed.) New York:

Guildford Press.

Satorra, A. (1992). Asymptotic robust inferences in the analysis of mean and covariance structures.

Sociological Methodology, 22, 249–278.Tanaka, J. S. (1987). How big is big enough? Child Development, 58, 134–146. doi:10.1111/1467-

8624.ep7264172

Wang, L., Li, Z., Shi, Z., Zhang, Y., & Shen, J. (2010). Factor structure of acute stress disorder

symptoms in Chinese earthquake victims. Personality and Individual Differences, 48, 798–802. doi:10.1016/j.paid.2010.01.027

Received 12 January 2012; revised version received 29 May 2012

Acute stress disorder factor structure 91