lable at ScienceDirect

Journal of Psychiatric Research 59 (2014) 60e67

Contents lists avai

Journal of Psychiatric Research

journal homepage: www.elsevier .com/locate/psychires

DSM-5 posttraumatic stress disorder: Factor structure and rates ofdiagnosis

Emily L. Gentes a, b, Paul A. Dennis a, c, d, Nathan A. Kimbrel a, b, c, Michelle B. Rissling a, b,Jean C. Beckham a, b, c, VA Mid-Atlantic MIRECC Workgroup, Patrick S. Calhoun a, b, c, d, *

a Durham Veterans Affairs Medical Center, Durham, NC, USAb The VA Mid-Atlantic Mental Illness Research, Education, and Clinical Center, Durham, NC, USAc Department of Psychiatry and Behavioral Sciences, Duke University Medical Center, Durham, NC, USAd VA Center for Health Services Research in Primary Care, Durham, NC, USA

a r t i c l e i n f o

Article history:Received 3 June 2014Received in revised form19 August 2014Accepted 21 August 2014

Keywords:DSM-5Posttraumatic stress disorderLatent factor structureCFA

* Corresponding author. VA Mid-Atlantic MIRECC,508 Fulton Street, Durham, NC 27705, USA. Tel.: þ1 9

E-mail addresses: emily.gentes@va.gov (E.L. Gent(P.S. Calhoun).

http://dx.doi.org/10.1016/j.jpsychires.2014.08.0140022-3956/Published by Elsevier Ltd.

a b s t r a c t

Posttraumatic stress disorder (PTSD) is a significant problem among Iraq/Afghanistan-era veterans. Todate, however, there has been only limited research on how the recent changes in DSM-5 influence theprevalence and factor structure of PTSD. To address this key issue, the present research used a modifiedversion of a gold-standard clinical interview to assess PTSD among a large sample of Iraq/Afghanistan-eraveterans (N ¼ 414). Thirty-seven percent of the sample met DSM-5 criteria for PTSD compared to a rate of38% when DSM-IV diagnostic criteria were used. Differences in rates of diagnosis between DSM-IV andDSM-5 were primarily attributable to changes to Criterion A and the separation of the “avoidance” and“numbing” symptoms into separate clusters. Confirmatory factor analysis (CFA) was used to compare thefit of the previous 3-factor DSM-IV model of PTSD to the 4-factor model specified in DSM-5, a 4-factor“dysphoria” model, and a 5-factor model. CFA demonstrated that the 5-factor model (re-experiencing,active avoidance, emotional numbing, dysphoric arousal, anxious arousal) provided the best overall fit tothe data, although substantial support was also found for the 4-factor DSM-5 model. Low factor loadingswere noted for two of the symptoms in the DSM-5 model (psychogenic amnesia and reckless/self-destructive behavior), raising questions regarding the adequacy of fit between these symptoms andthe other core features of PTSD. Overall, findings suggest the DSM-5 model of PTSD is an improvementover the previous DSM-IV model of PTSD, but still may not represent the true underlying factor structureof PTSD.

Published by Elsevier Ltd.

The publication of the fifth edition of the Diagnostic and Statis-tical Manual of Mental Disorders (DSM-5; American PsychiatricAssociation, 2013) introduced several significant changes to thediagnosis of posttraumatic stress disorder (PTSD; for a completedescription, see Calhoun et al., 2012; Friedman, 2013; Miller et al.,2013). The present study examined changes in the factor struc-ture and rates of PTSD diagnosis between DSM-IV and DSM-5.

The first major change introduced by DSM-5 involves PTSDCriterion A, which defines the set of events that constitute a trau-matic exposure. Specifically, the definition of a traumatic event wasclarified to include “exposure to actual or threatened death, serious

Durham VA Medical Center,19 286 0411.es), Patrick.calhoun2@va.gov

injury, or sexual violence” through directly experiencing, witness-ing, or learning about an event happening to a close familymemberor friend. Notably, though, actual or threatened death of a loved oneis now considered traumatic only if the event was either violent oraccidental. In addition, witnessing traumatic events through elec-tronic media, television, video games, or pictures is now explicitlyexcluded from Criterion An unless these experiences occurred aspart of one's professional responsibilities (e.g., as a first responder).Finally, Criterion A2, which required that the individual's responseinclude fear, helplessness, or horror, was removed from the DSM-5PTSD criteria because of evidence that many people deny experi-encing these intense emotional reactions and because the presenceor absence of these reactions has not been shown to predict risk forPTSD (Friedman, 2013).

DSM-5 now includes four symptom clusters rather than thethree symptom clusters included in DSM-IV. In addition, the totalnumber of possible PTSD symptoms has increased from 17 in DSM-

E.L. Gentes et al. / Journal of Psychiatric Research 59 (2014) 60e67 61

IV to 20 in DSM-5. DSM-5 Criterion B symptoms remain largely thesame as those in the DSM-IV reexperiencing cluster. However, DSM-5 Criterion C is now comprised of only two items reflecting effortfulavoidance, whereas Criterion D contains symptoms of negativealterations in cognitions and mood. Of the seven symptoms nowincluded in Criterion D, two are new symptoms that were added toDSM-5 to capture distorted blame of self or others about the causeor consequence of the traumatic event and pervasive negativeemotional state. Additionally, the DSM-IV symptom of sense offoreshortened future has been modified to reflect persistent andexaggerated negative expectations.

Finally, DSM-5 Criterion E is comprised of six symptomsreflecting alterations in arousal and reactivity associated with thetraumatic event (previously DSM-IV hyperarousal symptoms). Cri-terion E contains one new symptom, which captures reckless orself-destructive behavior. Additionally, the symptom of irritabilityand outbursts of anger has been modified to focus exclusively onirritable or aggressive behavior. A diagnosis of PTSD in DSM-5 re-quires one symptom each from Clusters B and C and two symptomseach from Clusters D and E, all of which must begin followingexposure to a qualifying Criterion A event.

Many of these changes e particularly the reorganization ofsymptoms into four clusters - were intended to bring the PTSDdiagnostic criteria into closer agreement with research on the un-derlying factor structure of the disorder (Friedman, 2013). Researchconducted to date has found little support for the three-factormodel presented in DSM-IV. Rather, the research has supportedtwo four-factor models (King et al., 1998; Simms et al., 2002) and,more recently, a five-factor model (Elhai et al., 2011). The four-factor “emotional numbing” model (King et al., 1998) separatessymptoms of active avoidance (e.g., avoiding reminders of trauma)from emotional numbing (e.g., restricted range of affect), retainingthe reexperiencing and hyperarousal factors as they are presentedin DSM-IV. The four-factor “dysphoria” model (Simms et al., 2002)similarly retains the reexperiencing factor, along with an activeavoidance factor. However, it includes a broadened version of theemotional numbing factor, in which emotional numbing symptomsare combined with symptoms of sleep disturbance, irritability, anddifficulty concentrating to form a factor reflecting general distressor dysphoria. In this model, the hyperarousal factor contains onlytwo items (hypervigilance and exaggerated startle). Finally, thefive-factor model (Elhai et al., 2011) consists of reexperiencing,active avoidance, and emotional numbing factors, but separatessymptoms of sleep disturbance, irritability, and concentration dif-ficulties into a separate dysphoric arousal factor. The two remaininghyperarousal symptoms comprise an anxious arousal factor.

Although the changes to the PTSD diagnosis are largely consis-tent with findings from extant literature on the factor structure ofDSM-IV PTSD symptoms, the inclusion of new symptom criteria inDSM-5 will result in changes to estimates of the latent structure ofthe disorder. To date, however, there have only been a handful ofstudies of the factor structure of DSM-5 PTSD symptomatology.These have examined the factor structure of DSM-5 symptomsamong college students (Elhai et al., 2012), US adults (Biehn et al.,2013; Contractor et al., 2014; Miller et al., 2013), Chinese adults(Liu et al., 2014), and veterans (Miller et al., 2013). While thesestudies found evidence of adequate fit for the proposed DSM-5factor structure, there is a clear need for additional research on thefactor structure of DSM-5 PTSD symptoms. In particular, additionalstudies that use more traditional methods of assessment (e.g.,structured clinical interviews) with traumatized samples areneeded.

There is also a significant need for additional research on howthe new DSM-5 criteria will affect rates of diagnosis. For example,some individuals who did not previously meet Criterion A2 (fear,

helplessness, or horror) may now qualify for a PTSD diagnosis,whereas other individuals may no longer meet criteria for PTSD dueto lack of sufficient symptoms across each of the revised fourclusters. To date, studies based onDSM-IV criteria have shown a oneto two percent decrease in PTSD prevalence when requiring bothactive avoidance and numbing symptoms, as is required for diag-nosis in DSM-5 (Contractor et al., 2014; Elhai et al., 2009; Forbeset al., 2011). However, a recent self-report internet survey con-ducted in non-clinical college students showed a slight increase inthe observed prevalence rate associated with DSM-5 criteria (Elhaiet al., 2012). Only one study to date has examined prevalencechanges using a clinical sample and an interview-based measure,which is considered the gold standard in the field (Calhoun et al.,2012). Results from this study showed that seven percent of thosewho met DSM-IV Criterion A reported events that would no longerqualify under the new DSM-5 stressor criterion, while only oneindividual who did not meet DSM-IV Criterion A met the new DSM-5 criterion. Results further showed that changes in prevalence mayrange from a five percent decrease to an eight percent increase,depending on the base-rate of DSM-IV PTSD diagnosis in thesample.

1. Study objectives

The objective of the present study was to examine the impact ofchanges to the PTSD diagnostic criteria on both factor structure andrates of diagnosis among a large sample of Iraq/Afghanistan-eraveterans. As interview-based methods are considered to be themost valid approach to diagnostic assessment of psychiatric diag-nosis and PTSD (Jablensky, 2002; McDonald and Calhoun, 2010;Miller, 2005), an adapted version of the Structured Clinical Inter-view for DSM-IV (First et al., 1998) was used to assess PTSD in thecurrent study.

2. Method

2.1. Participants

The sample was comprised of 414 volunteers enrolled in theDepartment of Veterans Affairs (VA) Mid-Atlantic Mental IllnessResearch, Education and Clinical Center multisite study of post-9/11U.S. military veterans. Eligibility criteria included military serviceon or after September 11, 2001 and fluency in English. The currentsample was limited to those with complete data for the PTSDmodule of the SCID and the additional interview prompts assessingthe new DSM-5 items. Of the 443 individuals who completedinterview prompts assessing newDSM-5 items, 29 individuals wereeliminated from analyses due to missing SCID data. Potential par-ticipants were recruited via fliers, clinic referrals, and letters ofinvitation. Written consent was obtained from all participants, whowere compensated up to $175 for study enrollment. Study pro-cedures were approved by the Durham, Salisbury, Richmond andHampton VA Medical Centers. Participant characteristics are listedin Table 1.

2.2. Measures

The electronic version of the SCID was used to assess DSM-IVAxis I diagnoses, including PTSD. To ensure that complete data onPTSD symptoms would be available for all participants in the study,the PTSD module of the SCID was administered in full to all par-ticipants (i.e., without skip out rules). The SCID has been found to beboth clinically sensitive and reliable (Keane and Barlow, 2002), withgood to excellent interrater reliability for current disorders andmoderate test-retest reliability for lifetime disorders (Rogers,

Table 1Sample characteristics.

Range Mean (SD) Freq (%)

Age 21e65 36.47 (9.60)Female 82 (20%)EthnicityHispanic or Latino 30 (7%)Not Hispanic or Latino 382 (93%)

RaceCaucasian 212 (53%)African-American 191 (46%)American Indian 6 (1%)Asian 3 (1%)

Years of school 10e24 13.38 (3.79)Number of deployments 0e8 1.55 (1.11)Married/cohabitating 187 (45%)Employed 211 (51%)Years since trauma 2e50 10.70 (7.95)Combat Exposure Scale 0e40 12.58 (10.42)Beck Depression Inventory 0e57 15.81 (12.58)Davidson Trauma Scale 0e136 45.70 (39.38)

E.L. Gentes et al. / Journal of Psychiatric Research 59 (2014) 60e6762

2001). SCID interviewers received initial and ongoing training andsupervision by psychologists and other experienced interviewers.The interviewers demonstrated excellent mean interrater reli-ability for any Axis I diagnosis (Fleiss's kappa¼ .94), and specificallyfor current PTSD (Fleiss's kappa ¼ 1.0; Fleiss and Cohen, 1973). Toassess newly revised DSM-5 symptoms of PTSD that did not overlapwith DSM-IV SCID items, four additional interview prompts wereconstructed and administered. These items included measurementof DSM-5 Criteria D2 (formally DSM-IV C7), D3, D4, and E2 and areprovided in Calhoun et al., 2012.

To address the changes to Criterion A in DSM-5, two doctorallevel psychologists experienced in the assessment of PTSD ratedwhether or not each individual's index trauma met the new defi-nition of a qualifying traumatic event based on DSM-5 Criterion A.Observed agreement for DSM-5 Criterion A was 92%. In addition,the raters reached consensus on the remaining 8% of events.

Several well-validated and widely-used self-report measureswere used to assess combat exposure, depression and PTSDsymptom severity, global symptom severity, and suicidal ideation.Specifically, the Combat Exposure Scale (CES; Keane et al., 1989)was used to assess combat exposure, the Beck DepressionInventory-II (BDI-II; Beck et al., 1996) was used to assess depressivesymptomatology, the Davidson Trauma Scale (DTS; Davidson et al.,1997) was used to assess PTSD symptom severity; the total score forthe Symptom Checklist-90 (SCL-90; Derogatis, 1994) was used toassess global symptom severity, and the Beck Scale for SuicideIdeation (BSS; Beck et al., 1979) was used to evaluate the presenceand severity of suicidal thoughts.

2.3. Data analysis plan

Three sets of analyses were conducted. First, descriptive ana-lyses were performed on trauma exposure and PTSD symptomendorsement rates documented during the SCID interview. Preva-lence rates of PTSD according to DSM-IV and DSM-5 criteria werecompared and sensitivity and specificity analyses were conductedusing the DSM-IV rates as the reference standard. Second, confir-matory factor analysis (CFA) was conducted on the dichotomousPTSD symptom data from the SCID. Given categorical data, CFAwasperformed using robust (mean- and variance-adjusted) weightedleast squares method, available throughMplus 7. Four models werecompared: the DSM-IV three-factor model, the DSM-5 four-factormodel, Simms and colleagues' “dysphoria” model (Simms et al.,2002), and Elhai and colleagues' five-factor model (Elhai et al.,

2011). Each of the four models was constructed using 16 of the 17SCID items (dropping item C7 and replacing this item with DSM-5item D2, which reflects persistent and exaggerated negative ex-pectations) plus the 4 questions constructed to tap new DSM-5criteria (Calhoun et al., 2012). Model fit was evaluated using stan-dard fit criteria (Hoyle, 1995; Hu and Bentler, 1999): root meansquare error of approximation (RMSEA)� .05, comparative fit index(CFI) � .90, TuckereLewis index (TLI) � .95, and weighted rootmean square residual (WRMR) � 1.00. Finally, item-response the-ory (IRT) analysis was conducted to evaluate the utility of each itemfor determining symptom severity. The basic concept of IRT is tolink performance on a given test to the underlying trait it wasdesigned to measure. As such, an assumption of IRT is that the itemdata are unidimensional. To meet this assumption, separate IRTanalyses were performed for each symptom cluster. Item-characteristic curves (ICC) were generated depicting the stan-dardized latent symptom cluster score on the x-axis and the cor-responding probability of item endorsement on the y-axis. For thepurposes of the present analysis, two relevant pieces of informationmay be gleaned from each curve (Crocker and Algina, 1986). Thefirst, item “difficulty,” corresponds to the x-value of the inflectionpoint of a given curve and reflects item-endorsement rate. Thefurther to the right the inflection point, the lower the endorsementrate. The second piece of information, item “discrimination,” cor-responds to the slope of the curve at the inflection point. Thesteeper the slope, the better an item serves as an indicator of thelatent trait. IRT was conducted in Mplus 7 using robust maximumlikelihood estimation with a logit link.

3. Results

3.1. Trauma exposure and symptom endorsement

A categorization of each individual's index trauma is listed inTable 2. All participants described an event that met DSM-IV Cri-terion A1. Of these, 87% (n ¼ 359) also met DSM-IV Criterion A2.Thus, 87% of all participants endorsed a DSM-IV Criterion A trau-matic event. Indexed traumas were also rated for whether theywould qualify as a DSM-5 Criterion A event. Ninety-sevenpercent oftraumas (n ¼ 401) qualified.

Ultimately, 38% of the sample (n ¼ 157) met DSM-IV criteria forPTSD. Thirty-seven percent (n ¼ 153) met DSM-5 criteria for PTSD.Agreement between the two sets of diagnostic criteria was high(Cohen's kappa ¼ .88, sensitivity of DSM-5 criteria to detect DSM-IVdiagnosis ¼ .91, specificity ¼ .96). Nevertheless, 14 of the partici-pants who met DSM-IV criteria (9%) did not meet DSM-5 criteria. Ofthese, two did not meet Criterion A, two did not meet Criterion C,and 10 did not meet Criterion D. Compared to those who met bothDSM-IV and DSM-5 criteria, these 14 participants reported less se-vere PTSD symptoms on the DTS and less severe depressionsymptoms on the BDI-II, both t(155) > 1.99, both p < .048. Of par-ticipants who met DSM-5 criteria for PTSD, 10 (7%) did not meetDSM-IV criteria. Of these, six denied experiencing fear, helpless-ness, or horror (i.e., DSM-IV Criterion A2); hence, their failure tomeet DSM-IV criteria for PTSD. Another four did not meet DSM-IVCriterion C. There were no significant differences in severity ofPTSD or depression symptoms between these 10 participants andthose who met both DSM-IV and DSM-5 criteria, both t(151) < 0.95,both p > .341.

Intercorrelations between PTSD status and clinical outcomes asa function of concordance between DSM-IV and DSM-5 (i.e., BDI,BSS, SCL-90 GSI, and CES) are listed in Table 3. DSM-5 PTSD corre-lated more highly with the BDI than DSM-IV PTSD; however, thisdifference is quite small and likely does not constitute a clinicallymeaningful difference. Twenty-two percent of participants (n¼ 92)

Table 2Trauma exposure reported for the SCID Interview and PTSD prevalence.

Index trauma description n (%) Met DSM-IV Met DSM-IV Met DSM-IV Met DSM-5 Met DSM-5

Criterion A1 Criterion A2 PTSD Criteria Criterion A PTSD Criteria

Childhood physical/sexual abuse 23 (6%) 23 (100%) 20 (87%) 10 (43%) 23 (100%) 9 (39%)Witnessed/experienced violence as child 15 (3%) 15 (100%) 14 (93%) 2 (13%) 15 (100%) 2 (13%)Adult domestic violence 6 (1%) 6 (100%) 6 (100%) 6 (100%) 6 (100%) 6 (100%)Adult physical/sexual abuse 14 (3%) 14 (100%) 14 (100) 8 (57%) 14 (100%) 8 (57%)Witnessed/experienced violence as adult 24 (6%) 24 (100%) 20 (83%) 8 (33%) 24 (100%) 7 (29%)Serious accident (e.g., motor vehicle, fire) 34 (8%) 34 (100%) 29 (85%) 4 (12%) 33 (97%) 5 (15%)Natural disaster 8 (2%) 8 (100%) 7 (88%) 1 (13%) 8 (100%) 1 (13%)Exposure to combat/war 226 (55%) 226 (100%) 197 (87%) 114 (50%) 225 (100%) 113 (50%)Death of family member/close friend 45 (11%) 45 (100%) 36 (80%) 7 (16%) 38 (84%) 5 (11%)Other trauma 19 (5%) 19 (100%) 16 (84%) 3 (16%) 15 (79%) 3 (16%)

Table 5

E.L. Gentes et al. / Journal of Psychiatric Research 59 (2014) 60e67 63

met SCID criteria for current major depressive disorder (MDD).Rates of comorbid depressionwere statistically equivalent with theDSM-5 PTSD (48% 73/153) as compared to DSM-IV (46% 72/157).Other diagnoses in the present sample included bipolar disorder(n ¼ 8), dysthymic disorder (n ¼ 17), depression NOS (n ¼ 3),psychotic disorder (n ¼ 2), alcohol abuse or dependence (n ¼ 33),cannabis abuse (n ¼ 6), other drug abuse or dependence (n ¼ 1),panic disorder (n ¼ 3), social phobia (n ¼ 11), specific phobia(n ¼ 9), obsessive-compulsive disorder (n ¼ 4), generalized anxietydisorder (n ¼ 8), anxiety disorder NOS (n ¼ 10), somatization dis-order (n ¼ 1), and pain disorder (n ¼ 1).

Endorsement rates of PTSD symptoms and item correlationswith DSM-5 symptom clusters are depicted in Table 4. All itemsloaded most heavily on their presumptive symptom cluster; how-ever, items D1 (inability to recall important aspects of the trauma)

Table 3Correlations between PTSD diagnostic criteria and clinical symptom measures.

Mean (SD) Intercorrelations

DSM-IV DSM-5 Test of difference

BDI 15.81 (12.57) .54** .56** Steiger's z ¼ 2.47, p < .01BSS .62 (2.24) .14** .16** Steiger's z ¼ 1.89, p ¼ .06SCL-90 GSI .98 (.83) .54** .55** Steiger's z ¼ 1.14, p ¼ .25CES 12.58 (10.42) .39** .39** Steiger's z ¼ 0.48, p ¼ .63

**p < .01.

Table 4DSM-5 symptom endorsement and itemecluster correlations.

DSM-5symptoms

Endorsedn (%)

Item-cluster severity correlations (r) Scale a

if itemremoved

B (a ¼ .83) C (a ¼ .74) D (a ¼ .83) E (a ¼ .81)

B-1 173 (42%) .85 .51 .66 .61 .77B-2 155 (37%) .77 .48 .55 .60 .81B-3 43 (10%) .55 .34 .40 .37 .85B-4 196 (47%) .86 .63 .64 .59 .77B-5 172 (42%) .81 .63 .60 .58 .78C-1 211 (51%) .61 .89 .59 .58 e

C-2 184 (44%) .60 .89 .59 .55 e

D-1 53 (13%) .31 .26 .39 .23 .85D-2 126 (30%) .54 .46 .73 .53 .80D-3 74 (18%) .34 .30 .59 .32 .83D-4 158 (38%) .57 .51 .77 .56 .80D-5 161 (39%) .63 .55 .79 .66 .79D-6 172 (42%) .66 .61 .83 .67 .78D-7 127 (31%) .55 .48 .75 .55 .80E-1 189 (46%) .56 .47 .63 .77 .76E-2 17 (4%) .08 .12 .22 .23 .84E-3 226 (55%) .57 .54 .55 .80 .76E-4 186 (45%) .52 .46 .45 .76 .77E-5 172 (42%) .58 .52 .60 .77 .76E-6 209 (50%) .61 .47 .59 .81 .75

and E2 (reckless or self-destructive behavior), demonstrated weakitemecluster correlations. In addition, B3 (flashbacks) was onlymoderately correlated with its cluster. The remaining symptomsloaded strongly on their corresponding clusters (rs � .73).

3.2. CFA

Table 5 provides the symptom-scale assignments for each of themodels that was examined. The model fit statistics for the four CFAmodels are listed in Table 6. Of the four models, the three-factorDSM-IV model demonstrated the weakest fit according to chi-square test of difference, p < .001 with the Elhai model demon-strating the strongest, p < .001. The fit of the DSM-5 and Simms

Table 6Confirmatory factor analysis (CFA) fit statistics.

Model Х 2 (df) RMSEA CFI TLI WRMR

DSM-IV (3 factors) 287.45 (167) .042 .991 .990 0.958DSM-5 (4 factors) 260.20 (164) .038 .993 .992 0.897Simms (4 factors) 258.21 (164) .037 .993 .992 0.895Elhai (5 factors) 199.23 (160) .034 .994 .993 0.777

Note. RMSEA ¼ root mean square error of approximation; CFI ¼ comparative fitindex; TLI ¼ TuckereLewis Index; WRMR ¼ weighted root mean square residual.

Symptom-scale assignments across PTSD models.

DSM-5 symptoms Models

DSM-IV DSM-5 Simms Elhai

(B-1) intrusive recollections R R R R(B-2) distressing dreams R R R R(B-3) dissociative reactions (e.g., flashbacks) R R R R(B-4) psychological distress at exposure to cues R R R R(B-5) physiological reactivity on exposure to cues R R R R(C-1) avoidance of internal reminders A/N A A A(C-2) avoidance of external reminders A/N A A A(D-1) inability to recall important

aspects of traumaA/N N D N

(D-2) negative expectations aboutself/others/world

A/N N D N

(D-3) distorted blame of self or others A/N N D N(D-4) pervasive negative emotional state A/N N D N(D-5) diminished interest in activities A/N N D N(D-6) detachment or estrangement A/N N D N(D-7) inability to experience positive emotions A/N N D N(E-1) irritable or aggressive behavior H H D DA(E-2) reckless or self-destructive behavior H H D DA(E-3) hypervigilance H H H AA(E-4) exaggerated startle response H H H AA(E-5) problems with concentration H H D DA(E-6) difficulty falling or staying asleep H H D DA

Note. R, reexperiencing; A ¼ avoidance; N ¼ numbing; H ¼ hyperarousal;D ¼ dysphoria; DA ¼ dysphoric arousal; AA ¼ anxious arousal.

Fig. 1. Confirmatory factor analysis of PTSD symptomatology. All estimates are standardized.

E.L. Gentes et al. / Journal of Psychiatric Research 59 (2014) 60e6764

model fell between these two. The DSM-5 and Elhai models aredepicted in Fig. 1.

In both the DSM-5 and Elhai models, items D1 (dissociativeamnesia) and E2 (reckless or self-destructive behavior) demon-strated the weakest loadings on their respective factors. Item D1demonstrated a .51 loading on the negative alterations in the DSM-5 model and a .59 loading in the Elhai model. E2 loaded .41 on thehyperarousal factor in the DSM-5 model and .43 on the anxiousarousal factor in the Elhai model. Otherwise, the remaining itemsloaded strongly on their respective factors (i.e., .75 or greater).

3.3. IRT

The results of the IRT analyses (Fig. 2) largely corroborated theCFA findings. Within the B cluster, item B3 (flashbacks) demon-strated the greatest difficulty (i.e., the lowest prevalence rate), item

B1 (intrusive recollections), the lowest discrimination. By contrast,B4 (distress at cue exposure) demonstrated the greatest discrimi-nation.Within the C cluster, C1 (avoidance of internal reminders) andC2 (avoidance of external reminders) yielded similar difficulties, yetitem C2 was far more discriminant. Within the D cluster, items D1(dissociative amnesia) and, to a lesser extent, D3 (pervasive negativeemotions) were relatively difficult and showed low discrimination.Of the E items, E2 (reckless or self-destructive behavior) had thegreatest difficulty and the lowest discrimination. In comparison tothe remaining items, E1 (irritable or aggressive behavior), alsodemonstrated relatively low discrimination.

4. Discussion

The objective of the present study was to examine the impact ofchanges to the PTSD diagnostic criteria on both factor structure and

Fig. 2. Item characteristic curves for items corresponding to the B, C, D, and E criteria. In each figure, the x-axis represents the standardized symptom cluster score, with a mean of0 and a standard deviation of 1. The y-axis represents the probability of item endorsement.

E.L. Gentes et al. / Journal of Psychiatric Research 59 (2014) 60e67 65

rates of diagnosis utilizing data obtained from a large sample of U.S.veterans using a gold-standard structured clinical interview. One ofthe major changes to the PTSD diagnostic criteria in DSM-5was themodification of Criterion A1 and the removal of Criterion A2. Pre-vious studies of PTSD among combat veterans have indicated thatmany deny experiencing fear, helplessness, or horror (DSM-IV Cri-terion A2) during combat, instead noting they experienced anger orthat “their training took over” (Roemer et al., 1998). Thus, thischange may have increased the probability of combat veteransreceiving a diagnosis of PTSD. Current results indicated that allparticipants in the present study reported an event that met DSM-IV Criterion A1, with 87% meeting both DSM-IV A1 and A2. Incontrast, 97% of participants met DSM-5 Criterion A. Not surpris-ingly the most common reason that participants in the presentstudy met DSM-5 but not DSM-IV Criterion A was that they had notendorsed fear, helplessness, or horror (DSM-IV Criterion A2).However, several participants who had met DSM-IV Criterion A nolonger met under DSM-5, primarily because their index eventinvolved actual or threatened death of a loved one that was notviolent or accidental (e.g., death from chronic illness). Overall, notmeeting Criterion A was one of the most common reasons thatparticipants were excluded from a PTSD diagnosis under both DSM-IV and DSM-5.

In addition to the changes to Criterion A, several modificationsto the PTSD symptoms and symptom clusters in DSM-5 affectedrates of diagnosis. In the current study, where the DSM-IV base ratewas 38%, we observed a 1% decrease in prevalence under DSM-5.These differences in diagnostic rates between DSM-IV and DSM-5are consistent with those found in previous research (Calhoun et al.,2012; Contractor et al., 2014; Elhai et al., 2009; Forbes et al., 2011).The most common reasons that people whomet DSM-IV criteria for

PTSD no longer met under DSM-5 criteria were tightening of thetrauma criterion, lack of sufficient symptoms of active avoidance(Cluster C) and/or alterations in cognitions and mood (Cluster D).Major depressive disorder was highly comorbid with both DSM-IVand DSM-5 PTSD, but the comorbidity ratewas slightly higher usingtheDSM-5 definition. Associations of PTSD status with othermentalhealth outcomes (BDI, SCL-90) were also highly similar for bothDSM-IV and DSM-5-based PTSD, despite more depressive content insymptoms for DSM-5 PTSD.

Overall, the CFA results provide support for the DSM-5 modeland suggest that it is a significant improvement over the previousDSM-IV model of PTSD. While this finding is consistent with pre-vious studies of DSM-5 PTSD symptom structure (Biehn et al.,2013; Contractor et al., 2014; Elhai et al., 2011; Liu et al., 2014;Miller et al., 2013), the present research makes a unique contri-bution by examining the factor structure of DSM-5 PTSD symp-toms that were assessed via a structured clinical interview. Inaddition, the present study builds upon previous research byexamining the fit of Elhai and colleagues' 5-factor model (Elhaiet al., 2011) within the DSM-5 set of symptoms. Notably, the 5-factor model provided the best overall fit to the data, which sug-gests that further alterations to the DSM criteria may still benecessary to truly reflect the underlying factor structure of PTSD.

In general, the factor loadings for the DSM-5 model were rela-tively high, although two notable exceptions to this finding werethe factor loadings for the psychogenic amnesia and reckless/self-destructive behavior symptoms. Results from the IRT analysiscorroborated this finding, showing that these items also demon-strated high difficulty and low discrimination. The poor perfor-mance of these two items is consistent with prior research usinginternet-based studies (Miller et al., 2013). In addition,

E.L. Gentes et al. / Journal of Psychiatric Research 59 (2014) 60e6766

psychogenic amnesia consistently demonstrated low factor load-ings in studies of DSM-IV PTSD symptoms (King et al., 1998; Simmset al., 2002). However, there is some evidence to suggest that thisitem may be more likely to be endorsed by individuals with moresevere symptomatology (Miller et al., 2013) and severe dissociativesymptoms (Wolf et al., 2012). In sum, while it is clear that addi-tional research on the newDSM-5 PTSD symptom set is still needed,the findings from the present study and the only other study to useIRT to examine DSM-5 PTSD symptom performance (Miller et al.,2013) suggest that both the psychogenic amnesia and reckless/self-destructive behavior symptoms perform poorly, raising ques-tions regarding the adequacy of fit between these symptoms andthe other core features of PTSD.

4.1. Limitations and future directions

Several limitations should be noted when interpreting thesefindings. First, while a clinical interview was used to assess PTSDsymptoms, the DSM-5 symptoms were assessed using a modifiedversion of the original DSM-IV based SCID interview. Thus, thedegree to which the current findings will generalize to those ob-tained using other measures of DSM-5 PTSD symptoms is unclear.Additionally, the amount of time between trauma exposure anddiagnostic assessment was variable, with many of the assessmentsin the current study conducted well after the time of the traumaticexposure. Finally, Axis II disorders were not assessed in the presentsample. An additional limitation concerns the sample composition.Specifically, because the samplewas derived from an ongoing studyof post-9/11 military veterans, the degree to which the findingsfrom the current study will relate to other trauma samples (e.g.,rape survivors, domestic violence victims) is unclear.

4.2. Summary and conclusion

In summary, using DSM-5 criteria resulted in a slightly loweroverall rate of PTSD in the current study compared with the rate ob-tained using DSM-IV criteria (38% vs 37%). Still, nine percent of vet-erans who met DSM-IV PTSD did not meet DSM-5 PTSD. Changes indiagnostic statuswere largelyattributable todifferences inCriterionAdefinitionandthe separationof “avoidance”and “numbing” symptomclusters. We also used CFA to evaluate four different models of PTSDfactor structure using the new DSM-5 PTSD symptom set. We foundthat the DSM-5 4-factor model was a significant improvement overthe DSM-IV 3-factor model; however, we also found that a 5-factormodel provided the best overall fit to the data. Thus, while the DSM-5 model of PTSD appears to be a significant improvement over theDSM-IV model, it still may not represent the true underlying factorstructure of PTSD. Overall, this study supports our understanding ofPTSD as a disorder involving dysphoric mood and cognitions, alongwith symptoms of reexperiencing, active avoidance, and hyper-arousal, and suggests that the disorder may be even better concep-tualizedby separatingdysphoric andanxious arousal factors. Changesin our understanding of the disorder may have implications forintervention. In particular, the present study highlights the need forPTSD treatments to effectively target symptoms of active avoidanceand dysphoria. However, it is notable that DSM-5criteria identifiedlargely the same cohort of patients as DSM-IV, supporting thecontinued use of interventions previously found to be effective.

Contributors

We confirm that the manuscript has been read and approved byall named authors and that there are no other persons who satisfiedthe criteria for authorship but are not listed. We further confirm

that the order of authors listed in the manuscript has beenapproved by all of us.

We confirm that we have given due consideration to the pro-tection of intellectual property associated with this work and thatthere are no impediments to publication, including the timing ofpublication, with respect to intellectual property. In so doing weconfirm that we have followed the regulations of our institutionsconcerning intellectual property.

We understand that the Corresponding Author is the sole con-tact for the Editorial process (including Editorial Manager anddirect communications with the office). He/she is responsible forcommunicating with the other authors about progress, sub-missions of revisions and final approval of proofs. We confirm thatwe have provided a current, correct email address which is acces-sible by the Corresponding Author and which has been configuredto accept email from Journal of Psychiatric Research.

Conflict of interest statement

Wewish to confirm that there are no known conflicts of interestassociated with this publication and there has been no significantfinancial support for this work that could have influenced itsoutcome.

Acknowledgments

Preparation of this manuscript was supported, in part, by re-sources and facilities at the VA Mid-Atlantic MIRECC, Durham VAMedical Center; by the Department of Veterans Affairs Office ofAcademic Affiliations Advanced Fellowship Program in MentalIllness Research and Treatment; and by the Clinical ScienceResearch and Development Service of the VA Office of Research andDevelopment. Dr. Kimbrel was supported by a Career DevelopmentAward-2 (1IK2CX000525-01A1) from the Clinical Science Researchand Development Service of the VA Office of Research and Devel-opment. The views expressed in this article are those of the authorsand do not necessarily represent the views of the Department ofVeterans Affairs or any of the other institutions with which theauthors are affiliated.

The VA Mid-Atlantic MIRECC workgroup for this publicationincludes John A. Fairbank, Harold Kudler, Christine E. Marx, Scott D.Moore, Rajendra A. Morey, Mira Brancu, Jennifer J. Runnals, KristyStraits-Troster, Larry A. Tupler, Elizabeth Van Voorhees, H. RyanWagner, and Richard D. Weiner from the Durham VA MedicalCenter, Durham, NC; Marinell Miller-Mumford from the HamptonVA Medical Center, Hampton, Virginia; Scott D. McDonald andTreven Pickett from the Hunter Holmes McGuire VA Medical Cen-ter, Richmond, Virginia; and Robin Hurley, Katherine H. Taber, andRuth E. Yoash-Gantz from the W.G. Hefner VA Medical Center,Salisbury, NC. We thank those who kindly volunteered to partici-pate in this study.

References

American Psychiatric Association. Diagnostic and statistical manual of mental dis-orders. 5th ed. Arlington, VA: American Psychiatric Association; 2013.

Beck AT, Brown G, Steer RA. Beck depression inventory II manual. San Antonio, TX:The Psychological Corporation; 1996.

Beck AT, Kovacs M, Weissman A. Assessment of suicidal intention: the scale forsuicide ideation. J Consult Clin Psychol 1979;47:343e52.

Biehn TL, Elhai JD, Seligman LD, Tamburrino M, Armour C, Forbes D. Underlyingdimensions of DSM-5 posttraumatic stress disorder and major depressive dis-order symptoms. Psychol Inj Law 2013;6:290e8.

Calhoun PS, Hertzberg JS, Kirby AC, Dennis MF, Hair LP, Dedert EA, et al. The effect ofdraft DSM-V criteria on posttraumatic stress disorder prevalence. DepressAnxiety 2012;29:1032e42.

Contractor AA, Durham TA, Brennan JA, Armour C, Wutrick HR, Frueh BC, et al.DSM-5 PTSD's symptom dimensions and relations with major depression's

E.L. Gentes et al. / Journal of Psychiatric Research 59 (2014) 60e67 67

symptom dimensions in a primary care sample. Psychiatry Res 2014;215:146e53.

Crocker LM, Algina J. Introduction to classical and modern test theory. New York:Holt, Rinehart, and Winston; 1986.

Davidson JR, Book SW, Colket JT, Tupler LA, Roth S, David D, et al. Assessment of a newself-rating scale for post-traumatic stress disorder. Psychol Med 1997;27:153e60.

Derogatis LR. SCL-90-R: symptom checklist-90-R: administration, scoring, andprocedures manual. Minneapolis, MN: National Computer System, Inc; 1994.

Elhai JD, Biehn TL, Armour C, Klopper JJ, Frueh BC, Palmieri PA. Evidence for aunique PTSD construct represented by PTSD's D1-D3 symptoms. J AnxietyDisord 2011;25:340e5.

Elhai JD, Ford JD, Ruggiero KJ, Frueh BC. Diagnostic alterations for posttraumaticstress disorder: examining data from the National Comorbidity Survey Repli-cation and National Survey of Adolescents. Psychol Med 2009;39:1957e66.

Elhai JD, Miller ME, Ford JD, Biehn TL, Palmieri PA, Frueh BC. Posttraumatic stressdisorder in DSM-5: estimates of prevalence and symptom structure in anonclinical sample of college students. J Anxiety Disord 2012;26:58e64.

First MB, Spitzer RL, Gibbon M, Williams JBW. Structured clinical interview forDSM-IV axis I disorders, research version, patient edition (SCID-I/P). New York:Biometrics Research, New York State Psychiatric Institute; 1998.

Fleiss JL, Cohen J. The equivalence of weighted kappa and the intraclass correlationcoefficient as measures of reliability. Educ Psychol Meas 1973;33:613e9.

Forbes D, Lockwood E, Elhai JD, Creamer M, O'Donnell M, Bryant R, et al. An ex-amination of the structure of posttraumatic stress disorder in relation to theanxiety and depressive disorders. J Affect Disord 2011;132:165e72.

Friedman M. Finalizing PTSD in DSM-5: Getting here from there and where to gonext. J Trauma Stress 2013;26:548e56.

Hoyle RH. Structural equation modeling. Thousand Oaks, CA: Sage Publications,Inc.; 1995.

Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis:convention criteria versus new alternatives. Struct Equ Model 1999;6:1e55.

Jablensky A. Research methods in psychiatric epidemiology: an overview. Aust N. Z JPsychiatry 2002;36:297e310.

Keane TM, Barlow D. Posttraumatic stress disorder. In: Barlow D, editor. Anxiety andits disorders: the nature and treatment of anxiety and panic. New York: ThePress; 2002. p. 418e53.

Keane T, Fairbank J, Caddell J, Zimering R, Taylor K, Mora C. Clinical evaluation of ameasure to assess combat exposure. Psychol Assess 1989;1:53e5.

King DW, Leskin GA, King LA, Weathers FW. Confirmatory factor analysis of theClinician-Administered PTSD Scale: evidence of the dimensionality of post-traumatic stress disorder. Psychol Assess 1998;10:90e6.

Liu P, Wang L, Cao C, Wang R, Zhang J, Zhang B, et al. The underlying dimensions ofDSM-5 posttraumatic stress disorder symptoms in an epidemiological sampleof Chinese earthquake survivors. J Anxiety Disord 2014;28:345e51.

McDonald SD, Calhoun PS. The diagnostic accuracy of the PTSD checklist: a criticalreview. Clin Psychol Rev 2010;30:117e74.

Miller AB.General principles of screening. In: AhrensW,Pigeot I, editors.Handbook ofepidemiology. Berlin: Berlin Springer Science & Business Media; 2005.p. 1268e300.

Miller MW, Wolf EJ, Kilpatrick D, Resnick H, Marx BP, Holowka DW, et al. Theprevalence and latent structure of proposed DSM-5 posttraumatic stress dis-order symptoms in U.S. national and veteran samples. Psychol Trauma TheoryRes Pract Policy 2013;5:501e12.

Roemer L, Orsillo SM, Borkovec TD, Litz BT. Emotional response at the time of apotentially traumatizing event and PTSD symptomatology: a preliminaryretrospective analysis of the DSM-IV criterion A2. J Behav Ther Exp Psychiatry1998;29:123e30.

Rogers R. Structured clinical interview for DSM-IV disorders (SCID) and other Axis Iinterviews. In: Rogers R, editor. Handbook of diagnostic and structured inter-viewing. New York: The Guilford Press; 2001. p. 103e48.

Simms LJ, Watson D, Doebbeling BN. Confirmatory factor analyses of posttraumaticstress symptoms in deployed and nondeployed veterans of the Gulf War.J Abnorm Psychol 2002;111:637e746.

Wolf EJ, Miller MW, Reardon AF, Ryabchenko KA, Castillo D, Freund R. A latent classanalysis of dissociation and posttraumatic stress disorder: evidence for adissociative subtype. JAMA Psychiatry 2012;69:698e705.