R

Eps

MEa

b

a

ARRA

KPPCPVTHMTE

dbITfpeM

nPaeteA

U

0h

Journal of Anxiety Disorders 28 (2014) 108–114

Contents lists available at ScienceDirect

Journal of Anxiety Disorders

esearch paper

valuating treatment of posttraumatic stress disorder with cognitiverocessing therapy and prolonged exposure therapy in a VHApecialty clinic

atthew D. Jeffreysa,∗, Courtney Reinfelda, Prakash V. Nairb, Hector A. Garciaa,mma Mata-Galana, Timothy O. Rentza

South Texas Veterans Healthcare System, 5788 Eckhert Road, San Antonio, TX 78240, United StatesUniversity of Texas Health Science Center, Department of Epidemiology and Biostatistics, 7703 Floyd Curl Drive, San Antonio, TX 78229, United States

r t i c l e i n f o

rticle history:eceived 19 October 2012eceived in revised form 5 April 2013ccepted 15 April 2013

eywords:TSDsychotherapyognitive processing therapy

a b s t r a c t

This retrospective chart review evaluates the effectiveness of manualized cognitive processing therapy(CPT) protocols (individual CPT, CPT group only, and CPT group and individual combined) and manualizedprolonged exposure (PE) therapy on veterans’ posttraumatic stress disorder (PTSD) symptoms in oneVeterans Health Administration (VHA) specialty clinic. A total of 517 charts were reviewed, and analysesincluded 178 charts for CPT and 85 charts for PE. Results demonstrated CPT and PE to significantly reducePTSD Checklist (PCL) scores. However, PE was significantly more effective than CPT after controllingfor variables of age, service era, and ethnicity. Additional findings included different outcomes amongCPT formats, decreased treatment dropouts for older veterans, and no significant differences in outcome

rolonged exposure therapyeteransreatmentealth servicesental health

ranslational research

between Hispanic and White veterans. Study limitations and future research directions are discussed.© 2013 Published by Elsevier Ltd.

vidence based treatment

Posttraumatic stress disorder (PTSD) is highly prevalent andisabling affecting veterans of all service eras. One study foundetween 15% and 17% of veterans returning from Afghanistan and

raq at 1-year follow-up met screening criteria for PTSD (Hoge,erhakopian, Castro, Messer, & Engel, 2007). Prevalence estimatesor Vietnam veterans include a 9.1% current and a 19% lifetimerevalence (Dohrenwend et al., 2007). A rate of 10.1% has beenstimated for veterans of Operation Desert Storm (Kang, Natelson,ahan, Lee, & Murphy, 2003).Because of this high prevalence of PTSD in combat veterans, a

eed was identified to deliver evidence-based psychotherapies forTSD in Veterans Health Administration (VHA) settings immedi-tely following the start of the wars in Afghanistan and Iraq (Rosent al., 2004). VHA began a national initiative in 2006 to formally

rain clinicians in cognitive processing therapy (CPT) and prolongedxposure (PE) therapy (Karlin et al., 2010; Ruzek & Rosen, 2009).lthough there are increasing numbers of clinicians trained in CPT

∗ Corresponding author at: UTHSCSA, 5788 Eckhert Road, San Antonio, TX 78240,nited States. Tel.: +1 210 699 2145; fax: +1 210 699 2257.

E-mail addresses: Matthew.Jeffreys@va.gov, m.jeffreys59@att.net (M.D. Jeffreys).

887-6185/$ – see front matter. © 2013 Published by Elsevier Ltd.ttp://dx.doi.org/10.1016/j.janxdis.2013.04.010

and PE, there has been limited systematic evaluation of the effec-tiveness of these treatments in VHA settings (Alvarez et al., 2011;Chard, Schumm, Owens, & Cottingham, 2010; Morland, Hynes,Mackintosh, Resick, & Chard, 2011; Rauch et al., 2009; Schnurr et al.,2007; Tuerk et al., 2011; Yoder et al., 2012).

There is strong support for the efficacy and tolerability ofcognitive-behavioral therapies for PTSD treatment (Bisson &Andrew, 2007; Bradley, Greene, Russ, Dutra, & Westen, 2005;Mendes, Mello, Ventura, Passarela Cde, & Mari Jde, 2008; Ponniah &Hollon, 2009). Both CPT and PE are efficacious treatments for PTSDrelated to non-combat and combat traumas alike (Foa, Zoellner,Feeny, Hembree, & Alvarez-Conrad, 2002; Hembree, Foa, et al.,2003; Monson et al., 2006; Resick, Nishith, Weaver, Astin, & Feuer,2002; Schnurr et al., 2007). This finding is reflected in the revisedVA/Department of Defense (DoD) Clinical Practice Guideline whichstrongly recommends CPT and PE for PTSD treatment (VA/DoD,2010). Although data are limited, trials directly comparing CPT andPE have demonstrated similar efficacy between the two treatments

(Nishith, Resick, & Griffin, 2002; Resick et al., 2002). One long-termstudy has demonstrated lasting improvement in PTSD symptomsat 10-year follow-up for both treatments (Resick, Williams, Suvak,Monson, & Gradus, 2011). Further, similar outcomes have been

nxiety

fw

itdmseodCnior2C

rmFpdMO

b

123

4

5

c

1

1

csbrwelswtJr

1

svto

M.D. Jeffreys et al. / Journal of A

ound between variations of CPT conducted with and without aritten trauma account (CPT-C) (Resick et al., 2008).

Translation of CPT and PE into clinical practice has raised somenteresting implementation questions regarding dropout rates andreatment response by service era and ethnicity. Two studiesemonstrated Vietnam veterans less likely to drop out of treat-ent, but there were differing results for treatment outcomes by

ervice era (Chard, Schumm, Owens, & Cottingham, 2010; Yodert al., 2012). Chard and colleagues found greater improvement inutcome for Operation Enduring Freedom/Operation Iraqi Free-om/Operation New Dawn (OEF/OIF/OND) veterans treated withPT as compared to Vietnam veterans. Yoder and colleagues foundo difference in outcome between the two service eras for PE. Stud-

es of CPT and PE in veterans have not demonstrated differences inutcome by ethnicity, though Hispanic veterans have been under-epresented (Chard et al., 2010; Monson et al., 2006; Tuerk et al.,011). This is the first study the authors are aware of comparingPT and PE in veterans.

The study identified 528 records for review of veterans whoeceived CPT or PE in one VHA specialty clinic. In particular, treat-ent outcome and drop outs for CPT and PE were examined.

eatures unique to this program included a large number of His-anic veterans and veterans of different service eras receivingiffering treatments based upon patient and therapist preferences.ost Vietnam veterans in this program received CPT and mostEF/OIF/OND veterans received PE.

The following study hypotheses were evaluated for this articleased upon the literature reviewed above:

) Both CPT and PE show equal benefit for PTSD symptoms.) Differing CPT formats are equally effective.) Dropout rates are higher for the OEF/OIF/OND veterans as com-

pared to veterans of other eras.) OEF/OIF/OND veterans show greater reduction in their PTSD

symptoms than veterans of other eras.) Hispanic and White veterans respond similarly to treatment.

The implications and limitations of the study findings are dis-ussed for clinical practice and future research.

. Methodology

.1. Study overview

The study was a retrospective chart review of one specialty PTSDlinic beginning January 1, 2006 through January 21, 2011. Thetudy was approved through the study site’s institutional reviewoard and research and development committee. Charts wereeviewed for veterans treated in the clinic who were diagnosedith military-related PTSD. A data collection sheet was followed to

nsure consistent recording from each chart. As an exempt study,imited data were collected and did not include treating therapist,pecific dates of treatment, prior treatment, concurrent treatmentith medications, specific trauma types, service connected sta-

us, or comorbidities. Outcome data on cases in progress as ofanuary 21, 2011 were not collected due to exceeding the timeange allowed for the chart review.

.2. Clinician training

Therapists were licensed clinical psychologists, licensed clinical

ocial workers, and pre- and post-doctoral trainees. CPT was pro-ided initially by two psychologists with VHA training in CPT beforehe official rollout who subsequently were trained through the roll-ut. Initial training in CPT consisted of a two day workshop followed

Disorders 28 (2014) 108–114 109

by phone consultation. All pre- and post-doctoral trainees co-facilitated CPT groups with licensed clinicians and received weeklysupervision. Training, supervision, and fidelity monitoring for thetrainees was provided by the two psychologists. Because the cur-rent data were collected before and during the national rollout, twoversions of the CPT manual were used (Chard, 2006; Resick, Monson& Chard, 2007). CPT group only therapy was provided according toits treatment manual (Chard, Resick, Monson, & Kattar, 2009).

PE was provided by one psychologist initially with non-VHAtraining who subsequently completed the rollout training, andby one social worker trained through the rollout. VHA trainingconsisted of a four day workshop with subsequent supervision andsuccessful completion of two cases using the PE treatment manual(Foa, Hembree, & Rothbaum, 2007). Therapists who joined the teamwere subsequently trained in CPT and PE through the VHA rollout.

1.3. Treatment sample

A total of 528 medical records were identified for reviewfrom a clinical database and appointment logs for calendar years2006–2010. There were 396 records with PTSD Checklist (PCL;Weathers, Huska, & Keane, 1991) values available one month pre-treatment (268 for CPT and 128 for PE). A total sample of 263veteran charts identified as completing treatment and having PCLvalues available one month pre-treatment and one month post-treatment were included in the data analysis for CPT and PE (178for CPT and 85 for PE). Dropout from treatment was definedas completing less than 2/3 of the recommended appointments.Demographics for veterans with pre- and post-treatment dataavailable are shown in Table 1.

1.4. Interventions

CPT is a twelve-session, manualized treatment protocol whichfocuses on modifying cognitions surrounding five main traumathemes including safety, trust, power and control, esteem and inti-macy (Resick et al., 2008). CPT with the written trauma accountwas offered in group only, individual only, or combined group andindividual formats. CPT without the written trauma account wasprovided also and was identified as CPT-C in this study. Sessionswere 60 min for individual therapy and 90 min for group therapyas directed by CPT treatment protocols. Patients were expected tocomplete practice assignments, but no measures of adherence wereavailable for this study.

PE is delivered in an individual format for approximately 10–15weekly 90 min sessions and consisted of the following core compo-nents: psycho education, breathing retraining, in vivo-exposures,and imaginal exposures as described in the literature (Hembree,Rauch, et al., 2003). Patients were expected to complete in vivo andimaginal exposure exercises, breathing exercises, and listen to thesession recording as homework to benefit from the therapy, but aswith CPT no measures of adherence were available for this study.

1.5. Measures

The Clinician Administered PTSD Scale (CAPS; Blake et al., 2000)was used diagnostically in records reviewed starting January 1,2006 and was replaced later with the Mini Neuropsychiatric Inter-view (MINI) for PTSD during the latter half of 2010 (Lecrubier et al.,1997). Both the CAPS and the MINI are accepted structured diag-nostic interviews for the assessment of PTSD. The clinicians utilizedonly the PTSD diagnostic category of the MINI as part of the assess-

ment protocol.

Progress in treatment was monitored through the PCL com-pleted by the patient and recorded by the clinician at differingtreatment intervals to measure PTSD symptom improvement. The

110 M.D. Jeffreys et al. / Journal of Anxiety Disorders 28 (2014) 108–114

Table 1Demographic tables for subjects with non-missing baseline PCL values at both 1 month prior to Treatment and 1 month post Treatment.

Covariate Treatment Total P-value

CPT PE

Age N 178 85 263 <0.001a

Missing 0 0 0Mean (SD) 57.1 (9.38) 38.2 (13.26) 51.0 (13.92)Median 60 33 57Min, Max 26, 77 23, 73 23, 77

Gender, n (%) Male 177 (99.4) 80 (94.1) 257 0.007b

Female 1 (0.6) 5 (5.9) 6Missing 0 (0) 0 (0) 0Total 178 85 263

Ethnicity, n (%) Hispanic 101 (56.7) 46 (54.1) 147 0.34c

White 59 (33.1) 26 (30.6) 85African American 15 (8.4) 9 (10.6) 24Asian 0 (0) 2 (2.4) 2Other 2 (1.1) 2 (2.4) 4Unknown 1 (0.6) 0 (0) 1Missing 0 (0) 0 (0) 0Total 178 85 263

PbJ

1

WwPtscsw(bpdcaamewaasspeoPut

2

2

l

a t-Test (t = 13.27, df = 261).b Pearson Chi-square test (chi-square = 7.31, df = 1).c Fisher’s exact test.

CL has demonstrated a strong correlation to the CAPS and is muchriefer to complete as a measure of treatment progress (Blanchard,ones-Alexander, Buckley, & Forneris, 1996).

.6. Statistical interpretation

All analyses were performed using Stata/SE version 11.2 forindows. Demographic variables and the number of appointmentsere summarized for all subjects and by treatment group (CPT or

E). The variables were also compared between groups using t-ests, Chi-square tests, or Fisher’s Exact Tests as appropriate. PCLcores within 1 month prior to treatment (pre-PCL scores) wereompared with those within 1 month post-treatment (post-PCLcores) for both CPT and PE. Significant differences in PCL scoresere obtained from a simple linear regression of PCL scores on time

pre-PCL vs. post-PCL at 1 month). The regression was performed onootstrap samples from subjects with complete PCL data at 1 monthrior to treatment (referred to as ‘baseline’) since the PCL scoresid not have a normal distribution. A 95% bias corrected bootstraponfidence interval for the pre-post-PCL difference in both CPTnd PE was constructed. The post-PCL scores were modeled withge, gender, ethnicity, CPT type (for patients receiving CPT treat-ent), treatment year, and service era as covariates. Models where

ach covariate was examined individually were also fit. All modelsere adjusted for baseline PCL values after subtracting the mean

nd were performed on patients having non-missing PCL scorest baseline. Furthermore, the linear models were run on bootstrapamples. Linear models were fit for patients who received CPT andeparately for those who received PE. A fully adjusted model waserformed with treatment included as a covariate. Logistic mod-ls were used to model drop-out status (dropped out/did not droput) using the same independent predictors used in the models ofCL score and constructed in a similar manner. The logistic modelssed the unrestricted dataset. Effect size was computed for eachreatment group using Cohen’s d statistic.

. Results

.1. Overall outcomes for CPT and PE

For veterans receiving CPT with non-missing PCL values at base-ine (n = 268), the mean pre-PCL score was 64.49 with standard

deviation of 10.54 while the mean post-PCL score (n = 178) was53.12 with standard deviation of 13.72. For PE, the mean pre-PCLscore (n = 128) was 56.67 with standard deviation of 11.35 whilethe mean post- PCL score (n = 85) was 32.76 with standard devia-tion of 11.44. The 95% bias corrected bootstrap confidence intervalfor the pre-post-PCL difference in CPT was (−13.2, −9.4) while the95% bias corrected bootstrap confidence interval for the pre-post-PCL difference in PE was (−26.4, −21). Both comparisons indicateda significant decrease in PCL scores. Pairwise correlations betweenthe pre-PCL scores and the magnitude of the drop in PCL scoresafter treatment in OEF/OIF/OND veterans produced a value of 0.41indicating a moderate positive correlation. This implies that thereis a possibility that higher pre-PCL scores leads to a greater dropin PCL scores after treatment in OEF/OIF/OND veterans though theresult did not reach statistical significance.

2.2. Demographic variables and outcomes

Initially, demographic variables were compared between thetwo treatments using t-tests, Chi-square tests or Fisher’s ExactTests as appropriate for subjects with non-missing pre- and post-PCL values. Mean age was significantly different between the twotreatments, but gender and ethnicity did not vary significantly bytreatment (see Table 1). In the overall general linear model, wheretreatment is a predictor along with other covariates such as age,gender, ethnicity and OEF/OIF/OND status (n = 256, total df = 255),both being a OEF/OIF/OND veteran (p = 0.049) and receiving PEtreatment (p < 0.001) led to a significantly greater decrease in post-PCL scores. In the overall model, with treatment year includedas a covariate (n = 248, df = 247), being an OEF/OIF/OND veteran(p = 0.047) and receiving PE treatment (p < 0.001) contributed sig-nificantly to a greater decrease in post-PCL scores. In the overalllinear model with Vietnam status and treatment as a covariate(n = 256, df = 255), receiving PE treatment (p < 0.001) was signif-icant. In the overall model with Desert Storm status (n = 256,df = 255), receiving PE treatment (p < 0.001) led to a significantlygreater decline in post-PCL scores.

Age category was a significant predictor of dropout from

therapy. In the overall logistic model with treatment as a predictorand covariates including age, gender, ethnicity and OEF/OIF/ONDstatus (n = 500, df = 6), having an age between 30 and 50 (p = 0.02)and having an age over 50 (p < 0.001) were significant compared

M.D. Jeffreys et al. / Journal of Anxiety Disorders 28 (2014) 108–114 111

Table 2Comparison of differing CPT formats.

Covariate PCL Scores P-valuea

Pre 1 month Post 1 month

CPT Group andIndividual

N 201 150 <0.001b

Missing 0 51Mean (SD) 65.15 (10.454) 54.11 (13.482)Median 66 55Min, Max 32, 85 21, 85

CPT Group only N 40 20 0.002c

Missing 0 20Mean (SD) 63.13 (9.557) 52.45 (11.427)Median 65 52Min, Max 46, 84 29, 72

CPT Individual only N 20 7 0.018d

Missing 0 13Mean (SD) 61.45 (12.717) 32.43 (8.867)Median 64 33Min, Max 30, 78 17, 43

a The CPT Group and Individual had 150 complete pairs of Pre-Post scores. The CPT Group only had 20 complete pairs. The CPT Individual only had 7 complete pairs.

tth53o

fgCftuae

2

fmhCoapcc

TMn

b Paired t-test (t = 11.38, df = 149).c Wilcoxon signed rank test (z = 3.05).d Wilcoxon signed rank test (z = 2.37).

o having an age less than 30. In the overall logistic model withreatment year included and service era (n = 498, df = 10), againaving an age between 30 and 50 (p = 0.03) and having an age over0 (p < 0.001) were significant compared to having an age less than0 with odds ratios of 0.54 and 0.23, respectively, indicating thatlder veterans were less likely to drop out.

The small number of females (19 in CPT and 15 in PE in theull dataset) produced unstable results and, therefore, the effect ofender was not considered. There were 41 African Americans inPT treatment and 24 African Americans in PE treatment for theull dataset. Although the number of African Americans was small,heir numbers were not too small that their ethnic status producednstable results. Therefore, African Americans were included in thenalysis and showed significantly more improvement than otherthnicities in PE only.

.3. CPT outcomes and dropouts

Table 2 summarizes the pre-post-PCL scores for differing CPTormats. All CPT treatment types, with the exception of CPT-C treat-

ent, led to significantly reduced scores after treatment. It should,owever, be noted there was only one post-PCL score available forPT-C, so it was not included in the analysis. It must be pointedut that only pre-post pairs with complete data were used in the

nalysis. For the CPT combined group and individual treatment, theaired t-test used 150 samples; for CPT group only treatment, 20omplete pairs were used, and for CPT individual only treatment, 7omplete pairs were used.

able 3odel of post 1 month PCL for CPT patients using age category, ethnicity, CPT type and O

on-missing baseline PCL values at 1 month prior to treatment) (n = 174, total df = 173).

Covariate Coefficient Bootstrap standard error

30 ≤ age ≤ 50a 9.72 5.581

Age > 50a 7.03 6.383

Hispanicb 2.02 1.9

African-Americanb −1.52 3.339

CPT group onlyc 1.76 2.5

CPT individual onlyc −13.68 3.564

OEF/OIF/ONDd −2.76 4.195

a Age < 30 is the referent.b Caucasian is the referent.c CPT group and individual is the referent.d Non OEF/OIF/OND service era is the referent.

In the CPT treatment, based on the individual linear model(n = 177, df = 176), the CPT format veterans received had a signif-icant association with the magnitude of the drop in PCL scoresas given by the individual model. Those who received CPT indi-vidual only treatment had a significantly greater drop in PCLscores than those who received combined group and individ-ual treatment (p < 0.001). In the fully adjusted model for the CPTgroup, post-PCL scores were modeled on age, ethnicity, CPT type,and OEF/OIF/OND status. In this model, CPT type was signifi-cant even after adjusting for the other covariates and where CPTindividual only treatment had a p value of less than 0.001 (seeTable 3). However, this result should be considered with cautiongiven the smaller numbers in group only and individual ther-apy.

Veterans from the OEF/OIF/OND service era were significantlycorrelated with a drop in post-PCL scores. Based on the individuallinear model (n = 178, df = 177), OEF/OIF/OND veterans had a 4.87greater drop in PCL scores as compared to veterans of other serviceeras for CPT.

The fully adjusted model was rerun with year of treatmentincluded as a covariate. For CPT treatment, the covariates includedage category, gender, ethnicity, CPT type, treatment year andOEF/OIF/OND status. In this model (n = 174, df = 173), having CPTindividual only treatment (p < 0.001) as well as treatment year 2009

(p = 0.04; and compared to 2006) were significant. Similar fullyadjusted models with Vietnam and Desert Storm veteran statussubstituted for OEF/OIF/OND status were run and the same rela-tionship was found.

EF/OIF/OND (yes/no) as predictors and adjusted for baseline PCL (for subjects with

z P-value (95% confidence interval)

1.74 0.08 (−1.21, 20.66)1.1 0.27 (−5.48, 19.54)1.06 0.29 (−1.71, 5.74)

−0.46 0.65 (−8.07, 5.02)0.71 0.48 (−3.14, 6.66)

−3.84 <0.001 (−20.67, −6.7)−0.66 0.51 (−10.98, 5.46)

112 M.D. Jeffreys et al. / Journal of Anxiety Disorders 28 (2014) 108–114

Table 4Model of post 1 month PCL for PE patients using age category, ethnicity and OEF/OIF/OND (yes/no) as predictors and adjusted for baseline PCL (for subjects with non-missingbaseline PCL values at 1 month prior to treatment) (n = 81, total df = 80).

Covariate Coefficient Bootstrap standard error z P-value (95% confidence interval)

30 ≤ age ≤ 50a −2.49 2.167 −1.15 0.25 (−6.73, 1.76)Age > 50a −4.7 5.862 −0.8 0.42 (−16.19, 6.79)Hispanicb 0.92 2.831 0.32 0.75 (−4.63, 6.47)African−Americanb −5.45 2.693 −2.02 0.04 (−10.72, −0.17)OEF/OIF/ONDc −9.92 7.046 −1.41 0.16 (−23.73, 3.89)

uasuaovtw3CaAmC

gt(wswts

2

APwge

ciTAI(sWswAmiym

(d

a Age < 30 is the referent.b Caucasian is the referent.c Non OEF/OIF/OND service era is the referent.

The odds of a patient dropping out were modeled in individ-al logistic models using individual predictors and also in a fullydjusted logistic model as had been done while modeling post-PCLcores with the regression models. For this, the entire dataset wassed. In the individual model (n = 326, df = 1), continuous age hadn odds ratio of 0.95 and was significant (p < 0.001), implying thatlder veterans were less likely to drop out. Similarly, in the indi-idual model with age category substituted for age (n = 326, df = 2),hose having an age greater than 50 (p < 0.001; odds ratio 0.19)ere much less likely to drop out than those who were less than

0 years of age. Looking at CPT type (n = 316, df = 2), those receivingPT group only treatment were twice as likely to drop out (p = 0.03)s those receiving combined CPT individual and group treatment.lso, those receiving CPT individual only treatment were 2.6 timesore likely to drop out (p = 0.01) than those receiving combined

PT individual and group treatment.The individual model for OEF/OIF/OND veterans in the CPT

roup (n = 316, df = 1) shows that OEF/OIF/OND veterans were threeimes more likely to drop out than members of other service erasp < 0.001). In the fully adjusted logistic model for the CPT patientsith covariates age category, ethnicity, CPT type and OEF/OIF/OND

tatus (n = 307, df = 7), only having an age greater than 50 (p = 0.02)as significant compared to having an age less than 30. This implies

hat after adjusting for all covariates, age rather than OEF/OIF/ONDtatus predicted the odds of dropping out.

.4. PE outcomes and dropouts

For the PE subjects, in the individual model (n = 81, df = 80), beingfrican American caused a significantly greater decrease in post-CL scores (by as much as 7.25 points; p = 0.01) as compared to thoseho were White. Although OEF/OIF/OND veterans had a 7.8 point

reater decrease in post-PCL scores than those from other serviceras, the p value was not significant (p = 0.06, n = 85, df = 84).

In the fully adjusted model for the PE subgroup with multipleovariates such as age category, ethnicity, and OEF/OIF/OND statusncluded, only being African American was significant (see Table 4).his result should be viewed with caution, because only 9 Africanmericans were used in the model after adjusting for all covariates.

n the model with Vietnam veteran status in place of OEF/OIF/ONDn = 81, df = 80), African American race (p = 0.02) contributed to aignificant decrease in PCL scores (decline of 6.3) compared to

hites. Considering the fully adjusted model with the Desert Stormtatus variable (n = 81, df = 80), being African American (p = 0.02)as the only significant predictor, but the small number of Africanmericans in the Vietnam and Desert Storm models may produceisleading results. In the additional model with year of treatment

ncluded, the covariates included age category, ethnicity, treatmentear, and OEF/OIF/OND status. No covariates were significant in this

odel (n = 81, df = 80).Considering the logistic models, in the individual logistic model

n = 191, df = 1), increasing age is again significantly associated withecrease in likelihood of dropping out (p = 0.01). Similarly with age

category used instead of age in the individual model (n = 191, df = 2),having an age greater than 50 (p = 0.004; odds ratio = 0.23) was asso-ciated with much less likelihood of dropping out than having anage less than 30. In the fully adjusted model for PE (n = 184, df = 5),only age category was significant in the model after adjusting forother covariates with those having an age above 50 (p = 0.01; oddsratio = 0.10) much less likely to drop out than those under 30.In the fully adjusted PE model with year of treatment included(n = 184, df = 9), again those having an age above 50 (p = 0.01; oddsratio = 0.11) were much less likely to drop out than those under 30.This implies that after adjusting for age, OEF/OIF/OND status andother covariates, age rather than OEF/OIF/OND status predicted theodds of dropping out.

2.5. Effect sizes and treatment completion

Effect size was computed for each treatment type using Cohen’sd statistic. For CPT, the effect size was 0.96 and for PE the effectsize was 2.0, both indicating a large effect. Dropout rates werealso computed. For CPT the dropout rate was 32.2% and for PE itwas 44%. The mean number of treatment sessions for those whohad non missing pre-PCL scores was 15.5 for CPT and 5.6 for PE.Using a Wilcoxon rank sum test there was a significant differencein the number of sessions between treatments (p < 0.001, z = 12.36,n = 396). Additionally, the mean number of sessions for dropoutsfrom the whole dataset was 7.4 for CPT and 3.8 for PE. The ranksum test confirmed that the mean number of sessions was signifi-cantly different between the two groups (p < 0.001, z = 5.41, n = 189)among dropouts.

3. Discussion

3.1. Differing outcomes for CPT and PE

Both CPT and PE significantly reduced PCL scores, but PE reducedscores significantly more than CPT. The hypothesis that CPT and PEwould have equal benefit for PTSD symptoms in this clinic was notconfirmed by the study. The effect size for CPT and PE were 0.96and 2.01, respectively. This is not consistent with direct comparisonof CPT and PE in a study of female sexual assault survivors whichdemonstrated no significant difference in outcomes between thetwo treatments (Resick et al., 2002). However, the effect sizes forCPT and PE in this study are consistent with other studies. Monsonand colleagues (2006) found an effect size of 1.0 across outcomesconsistent with the effect size of 0.96 found in this clinic. Two stud-ies of PE with veterans in clinical settings found effect sizes of 2.07and 2.73, which also are consistent with the effect size of 2.01 inthis study (Tuerk et al., 2011; Yoder et al., 2012).

There are a number of possible explanations for these findings.

Treatment adherence, including assignment completion, may havediffered between CPT and PE, though data were not available forthis study. Selection bias may have affected the results as assign-ment to CPT or PE was based upon therapist and patient preference.

nxiety

Ablmoc

latefi

PrrVtacfvAcd

3

rblm

onOvo(aw

rswtA

3

riAridct

hcitt

M.D. Jeffreys et al. / Journal of A

djunctive therapies, such as pharmacotherapy, may have differedetween the veterans receiving CPT and PE, and no data were col-

ected concerning pharmacotherapy. Thus, the potential effects ofedication on treatment outcome cannot be ruled out, although

ther controlled trials of CPT and PE have allowed medicationsoncurrently (Monson et al., 2006; Schnurr et al., 2007).

Differing CPT formats used in this clinic did not yield equiva-ent results. CPT individual only had significantly better outcomes compared to CPT group and CPT combined group and individualreatment. One possibility is that individual therapy may providenhanced oversight and support for treatment adherence, but thesendings require further research.

Another factor which could obscure the effectiveness of CPT vs.E is the difference in dropout rates with PE having the higher dropate of 44.4% for the full data set as compared to CPT with a dropoutate of 32.2%. Also, differing CPT formats had differing dropout rates.eterans were 2.6 times more likely to drop out of CPT individual

herapy and 2 times more likely to drop out of CPT group only ther-py as compared to the combined group and individual format. Inomparison, examples of dropout rates in the literature include 34%or PE (Tuerk et al., 2011) and dropout rates of 35% for OEF/OIF/ONDeterans and 26% for Vietnam veterans for CPT (Chard et al., 2010).

partial explanation for differences in dropout rates for this studyould be the conservative definition of dropout used, which wasefined as completion of at least 2/3 of the recommended sessions.

.2. Outcomes for other hypotheses

The hypothesis that OEF/OIF/OND veterans have higher dropoutates than Vietnam veterans was accounted for by age rather thany service era. For both CPT and PE veterans older than 50 had much

ess likelihood of dropping out. A possible explanations could beore competing life responsibilities for the younger veterans.CPT outcomes differed across service eras after adjusting for

ther covariates with only Desert Storm veterans showing a sig-ificant decrease in PCL scores as compared to other service eras.EF/OIF/OND veterans did not have better outcomes than Vietnameterans. This differed from one study which demonstrated betterutcomes in OEF/OIF/OND veterans compared to Vietnam veteransChard et al., 2010). PE outcomes differed between OEF/OIF/ONDnd Vietnam veterans but did not reach statistical significance,hich is consistent with the literature (Yoder et al., 2012).

Hispanic and White veterans did not differ significantly in theiresponses to CPT or PE. Surprisingly, African American veterans hadignificantly better outcomes for PE than other ethnicities. Thereas no difference for African Americans in CPT. The explanation for

his difference is unknown, but given the small number of Africanmericans sampled, this could have skewed the results.

.3. Study implications and limitations

The study findings have important implications for futureesearch. Providing individual psychotherapy is more resourcentensive, but might also lead to better treatment outcomes.ssuming CPT and PE are equally efficacious, further translationalesearch regarding their implementation into the clinical settings needed. Some reasons for differing outcomes could includeeviation from treatment fidelity by both patients and therapists,oncurrent medications, therapist proficiency, and patient andherapist selection bias.

There were a number of important limitations to this study. Per-aps the most salient limitation is that this was a retrospective

hart review rather than a prospective randomized controlled trialntroducing the potential for selection biases to have an impact onreatment outcome. PCL data are not available for 254 of the 517otal charts reviewed, which may have skewed the results and led

Disorders 28 (2014) 108–114 113

to incorrect conclusions. Data on pharmacotherapy were not col-lected and between group differences in pharmacotherapy cannotbe ruled out as a confound.

Currently, we are unaware of any direct comparisons betweenCPT and PE with veterans in the literature. The main strengths of thestudy are the number of charts reviewed and conditions reflectingactual clinical practice. Future research directions suggested by thisstudy include randomized controlled trials in real world clinicalsettings in differing geographical areas. Controlling for treatmentfidelity, homework adherence, patient comprehension, adjunctivetreatments, and consistently recording treatment outcomes wouldhave great clinical and theoretical impact on PTSD treatment.

Acknowledgements

The authors would like to thank Dr. John Cornell in the Depart-ment of Epidemiology and Biostatistics and the Practice BasedResearch Network (PBRN) at the University of Texas Health Sci-ence Center San Antonio for support in statistical analysis for thispaper.

References

Alvarez, J., McLean, C., Harris, A. H., Rosen, C. S., Ruzek, J. I., & Kimerling, R. (2011).The comparative effectiveness of cognitive processing therapy for male vet-

erans treated in a VHA posttraumatic stress disorder residential rehabilitationprogram. Journal of Consulting and Clinical Psychology, 79, 590–599.

Bisson, J., & Andrew, M. (2007). Psychological treatment of post-traumatic stressdisorder (PTSD). Cochrane Database of Systematic Reviews, 18, CD003388.

Blake, D., Weathers, F., Nagy, L., Kaloupek, D., Klauminzer, G., Charney, D., et al.(2000). Clinician-administered PTSD scale (CAPS). West Haven, CT: National Cen-ter for Posttraumatic Stress Disorder Behavioral Science Division.

Blanchard, E. B., Jones-Alexander, J., Buckley, T. C., & Forneris, C. A. (1996). Psycho-metric properties of the PTSD Checklist (PCL). Behaviour Research and Therapy,34, 669–673.

Bradley, R., Greene, J., Russ, E., Dutra, L., & Westen, D. (2005). A multidimensionalmeta-analysis of psychotherapy for PTSD. American Journal of Psychiatry, 162,214–227.

Chard, K. (2006). Cognitive processing therapy 13 session residential treatmentmanual. Unpublished manuscript.

Chard, K. M., Resick, P. A., Monson, C. M., & Kattar, K. A. (2009). Cognitiveprocessing therapy therapist group manual: Veteran/military version. Washington,DC: Department of Veterans Affairs.

Chard, K. M., Schumm, J. A., Owens, G. P., & Cottingham, S. M. (2010). A comparisonof OEF and OIF veterans and Vietnam veterans receiving cognitive processingtherapy. Journal of Traumatic Stress, 23, 25–32.

Dohrenwend, B. P., Turner, J. B., Turse, N. A., Adams, B. G., Koenen, K. C., & Marshall,R. (2007). Continuing controversy over the psychological risks of Vietnam forU.S. Veterans. Journal of Traumatic Stress, 20, 449–465.

Foa, E. B., Hembree, E. A., & Rothbaum, B. O. (2007). Prolonged exposure therapy forPTSD: Emotional reprocessing of traumatic experiences. New York: Oxford Univer-sity Press.

Foa, E. B., Zoellner, L. A., Feeny, N. C., Hembree, E. A., & Alvarez-Conrad, J. (2002).Does imaginal exposure exacerbate PTSD symptoms? Journal of Consulting andClinical Psychology, 70, 1022–1028.

Hembree, E. A., Foa, E. B., Dorfan, N. M., Street, G. P., Kowalski, J., & Tu, X. (2003).Do patients drop out prematurely from exposure therapy for PTSD? Journal ofTraumatic Stress, 16, 555–562.

Hembree, E. A., Rauch, S. A. M., & Foa, E. B. (2003). Beyond the Manual: The insider’sguide to prolonged exposure therapy for ptsd. Cognitive and Behavioral Practice,10, 22–30.

Hoge, C. W., Terhakopian, A., Castro, C. A., Messer, S. C., & Engel, C. C. (2007). Associa-tion of posttraumatic stress disorder with somatic symptoms, health care visits,and absenteeism among Iraq war veterans. American Journal of Psychiatry, 164,150–153.

Kang, H. K., Natelson, B. H., Mahan, C. M., Lee, K. Y., & Murphy, F. M. (2003). Post-traumatic stress disorder and chronic fatigue syndrome-like illness among GulfWar veterans: A population-based survey of 30,000 veterans. American Journalof Epidemiology, 157, 141L 148.

Karlin, B. E., Ruzek, J. I., Chard, K. M., Eftekhari, A., Monson, C. M., Hembree, E. A.,et al. (2010). Dissemination of evidence-based psychological treatments forposttraumatic stress disorder in the Veterans Health Administration. Journal ofTraumatic Stress, 23, 663–673.

Lecrubier, Y., Sheehan, D. V., Weiller, E., Amorim, P., Bonova, I., Sheehan, K. H.,

et al. (1997). The Mini International Neuropsychiatric Interview (MINI). A shortdiagnostic structured interview: Reliability and validity according to the CIDI.European Psychiatry, 12, 224–231.

Mendes, D. D., Mello, M. F., Ventura, P., Passarela Cde, M., & Mari Jde, J. (2008).A systematic review on the effectiveness of cognitive behavioral therapy for

1 nxiety

M

M

N

P

R

R

R

R

14 M.D. Jeffreys et al. / Journal of A

posttraumatic stress disorder. International Journal of Psychiatry in Medicine, 38,241–259.

onson, C. M., Schnurr, P. P., Resick, P. A., Friedman, M. J., Young-Xu, Y., & Stevens,S. P. (2006). Cognitive processing therapy for veterans with military-relatedposttraumatic stress disorder. Journal of Consulting and Clinical Psychology, 74,898–907.

orland, L. A., Hynes, A. K., Mackintosh, M. A., Resick, P. A., & Chard, K. M. (2011).Group cognitive processing therapy delivered to veterans via telehealth: A pilotcohort. Journal of Traumatic Stress, 24, 465–469.

ishith, P., Resick, P. A., & Griffin, M. G. (2002). Pattern of change in prolongedexposure and cognitive-processing therapy for female rape victims with post-traumatic stress disorder. Journal of Consulting and Clinical Psychology, 70,880–886.

onniah, K., & Hollon, S. D. (2009). Empirically supported psychological treatmentsfor adult acute stress disorder and posttraumatic stress disorder: A review.Depression and Anxiety, 26, 1086–1109.

auch, S. A., Defever, E., Favorite, T., Duroe, A., Garrity, C., Martis, B., et al. (2009).Prolonged exposure for PTSD in a Veterans Health Administration PTSD clinic.Journal of Traumatic Stress, 22, 60–64.

esick, P. A., Galovski, T. E., Uhlmansiek, M. O., Scher, C. D., Clum, G. A., &Young-Xu, Y. (2008). A randomized clinical trial to dismantle components ofcognitive processing therapy for posttraumatic stress disorder in female vic-tims of interpersonal violence. Journal of Consulting and Clinical Psychology, 76,

243–258.

esick, P. A., Monson, C. M., & Chard, K. M. (2007). Cognitive processing therapy:Veteran/military version. Washington. DC: Department of Veterans Affairs.

esick, P. A., Nishith, P., Weaver, T. L., Astin, M. C., & Feuer, C. A. (2002). A com-parison of cognitive-processing therapy with prolonged exposure and a waiting

Disorders 28 (2014) 108–114

condition for the treatment of chronic posttraumatic stress disorder in femalerape victims. Journal of Consulting and Clinical Psychology, 70, 867–879.

Resick, P. A., Williams, L. F., Suvak, M. K., Monson, C. M., & Gradus, J. L. (2011).Long-term outcomes of cognitive-behavioral treatments for posttraumatic

stress disorder among female rape survivors. Journal of Consulting and ClinicalPsychology, 80, 201–210.

Rosen, C. S., Chow, H. C., Finney, J. F., Greenbaum, M. A., Moos, R. H., Sheikh, J. I., et al.(2004). VA practice patterns and practice guidelines for treating posttraumaticstress disorder. Journal of Traumatic Stress, 17, 213–222.

Ruzek, J. I., & Rosen, R. C. (2009). Disseminating evidence-based treatments forPTSD in organizational settings: A high priority focus area. Behaviour Researchand Therapy, 47, 980–989.

Schnurr, P. P., Friedman, M. J., Engel, C. C., Foa, E. B., Shea, M. T., Chow, B. K.,et al. (2007). Cognitive behavioral therapy for posttraumatic stress disorderin women: A randomized controlled trial. Journal of the American Medical Asso-ciation, 297, 820–830.

Tuerk, P. W., Yoder, M., Grubaugh, A., Myrick, H., Hamner, M., & Acierno, R. (2011).Prolonged exposure therapy for combat-related posttraumatic stress disor-

der: An examination of treatment effectiveness for veterans of the wars inAfghanistan and Iraq. Journal of Anxiety Disorders, 25, 397–403.

U.S. Department of Veterans Affairs & Department of Defense. (2010). VA/DoD clinicalpractice guideline for management of post-traumatic stress (2nd ed.). Washington.

Yoder, M., Tuerk, P. W., Price, M., Grubaugh, A. L., Strachan, M., Myrick, H., et al.

(2012). Prolonged exposure therapy for combat-related posttraumatic stressdisorder: Comparing outcomes for veterans of different wars. Psychological Ser-vices, 9, 16–25.

Weathers, F. W., Huska, J., & Keane, T. (1991). The PTSD checklist-military version.Boston, MA: National Center for PTSD.