at SciVerse ScienceDirect

Behaviour Research and Therapy 49 (2011) 796e801

Contents lists available

Behaviour Research and Therapy

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

Shorter communication

Attentional bias in post-traumatic stress disorder diminishesafter symptom amelioration

Myriam El Khoury-Malhamea,b,*, Laura Lanteaumec, Eva Maria Beetzd, Jacques Roquese,Emmanuelle Reynauda,b, Jean-Claude Samuelianf, Olivier Blina,g, René Garciah, Stephanie Khalfaa,b

aNational Center for Scientific Research (CNRS), Mediterranean Institute for Cognitive Neuroscience (INCM, UMR 6193), CHU Timone Hospital, 13385 Marseille Cedex 5, FrancebDepartment of Neurosciences, Faculty of Life and Health Sciences, University of the Mediterranean, Aix Marseille 2, Marseille, FrancecMontreal Neurological Institute and Hospital, McGill University, Montreal, CanadadDepartment of Psychiatry and Medical Psychology, Princess Grace Hospital Center, Monaco, FranceeCenter for Treatment of Psychological Trauma, Montpellier, Francef Public Assistance for Marseille Hospitals (APHM) Psychiatry Center Pole, CHU Conception Hospital, 13005 Marseille, Franceg Public Assistance for Marseille Hospitals (APHM) Unit for Clinical Pharmacology and Therapeutic Evaluation (CIC-UPCET), CHU Timone Hospital, 13385 Marseille Cedex 5, FrancehNeurobiology and Psychotraumatology Laboratory (EA4321), University of Nice-Sophia Antipolis, UFR Sciences, 06108 Nice Cedex 2, France

a r t i c l e i n f o

Article history:Received 13 October 2010Received in revised form7 July 2011Accepted 26 August 2011

Keywords:Emotional stroopDetection of target (DOT)Post-traumatic stress disorder (PTSD)DisengagementEye movement desensitization andreprocessing (EMDR)

* Corresponding author. INCM, UMR CNRS 6193,Jean Moulin Boulevard, Bldg F, 1st Floor, 13385 Marse(0)4 91 38 46 60; fax: þ33 (0)4 91 47 21 40.

E-mail address: myriam.khoury@gmail.com (M. El

0005-7967/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.brat.2011.08.006

a b s t r a c t

Background: Avoidance and hypervigilance to reminders of a traumatic event are among the maincharacteristics of post-traumatic stress disorder (PTSD). Attentional bias toward aversive cues in PTSDhas been hypothesized as being part of the dysfunction causing etiology and maintenance of PTSD. Theaim of the present study was to investigate the cognitive strategy underlying attentional bias in PTSD andwhether normal cognitive processing is restored after a treatment suppressing core PTSD symptoms.Methods: Nineteen healthy controls were matched for age, sex and education to 19 PTSD patients. Weused the emotional stroop and detection of target tasks, before and after an average of 4.1 sessions of eyemovement desensitization and reprocessing (EMDR) therapy.Results:We found that on both tasks, patients were slower than controls in responding in the presence ofemotionally negative words compared to neutral ones. After symptoms removal, patients no longer hadattentional bias, and responded similarly to controls.Conclusion: These results support the existence of an attentional bias in PTSD patients due to a disen-gagement difficulty. There was also preliminary evidence that the disengagement was linked to PTSDsymptomatology. It should be further explored whether attentional bias and PTSD involve common brainmechanisms.

� 2011 Elsevier Ltd. All rights reserved.

Introduction

The post-traumatic stress disorder (PTSD) is an anxiety disorderthat occurs in the aftermath of a traumatic event. According tothe DSM-IV (American Psychiatric Association, 2004), it is a non-adaptive reaction to stress characterized by intrusive memories,avoidance, hypervigilance and social dysfunctions, persisting atleast one month after the trauma. From a cognitive point of view,one of the key factors of the emotional distress and maintenance ofanxiety disorder is the existence of non-adaptive attentional bias

CIC-UPCET, Timone Hospital,ille Cedex 5, France. Tel.: þ33

Khoury-Malhame).

All rights reserved.

toward information with aversive value (Mogg & Bradley, 1998).One way to investigate such bias is via interference tasks, such asthe emotional stroop (e-Stroop) and the detection of target (DOT),which involves a central task to be performed while ignoringemotional distracters (Mathews & MacLeod, 2005).

The most common paradigm indexing sensitivity to threat-related events is the e-Stroop (Williams, Mathews, & MacLeod,1996). In such tasks, anxious people tend to be generally slowerin responding to emotionally negative words than to neutral ones,implying the existence of selective attention to emotional cues(Fox, Russo, & Dutton, 2001). More specifically, PTSD patients areknown to have an attentional bias on e-Stroop tasks with traumacues and general words with negative valence (Hayes, Weber,Gallagher, & Drouillard, 2009; McNally, Otto, Yap, Pollack, &Hornig, 1999). For instance in motor accidents, PTSD patientswere slower in color-naming accident related words than neutral

M. El Khoury-Malhame et al. / Behaviour Research and Therapy 49 (2011) 796e801 797

ones (Beck, Freeman, Shipherd, Hamblen, & Lackner, 2001). The e-Stroop task thus provides clear evidence of the presence of atten-tional bias in PTSD, but gives no information on its underlyingaltered processing strategy.

The specific selectivity in strategic cognitive processing seems tobe best addressed by the DOT (Posner & Petersen, 1990). In this task,anxious patients have shown a selective disengagement bias fromtrauma-related material, with a difficulty detaching their attentionfrom threatening cues (Bar-Haimet al., 2007). Recent traumavictimsbehave alike, as they were found to view trauma-related pictureslonger than generally aversive pictures, unlike healthy controlparticipants (Elsesser, Sartory, & Tackenberg, 2005). Results for PTSDseem controversial. Some studies have shown that PTSD patientsorient away from threat. For instance, Vietnam veterans with PTSDwere found to ‘escape’ the presentation of combat scenes when theywere able to turn off the display (Blanchard, Kolb, Pallmeyer, &Gerardi, 1982). PTSD patients were also shown to name targetsfaster when in close proximity to mild threat words (Bryant &Subbiah, 1999). Recent studies, however, have shown that PTSDpatients orient toward the threat cues (Browning, Holmes, &Harmer,2010; Pineles, Shipherd, Mostoufi, Abramovitz, & Yovel, 2009).

Toomany stimulimodalities, intensities and durationswere usedto settle discrepant results as to whether PTSD have decreasedreaction time (RT) on emotional trials due to vigilance (facilitationby threat cues) or increased RT due to disengagement (interferenceof threat cues). Our first aim was to replicate PTSD attentional biason an e-Stroop task, and further refine the altered cognitive strategyon a DOT task, including generally negative and neutral words.

The PTSD model has been implemented by recent observationsof altered brain structures involved in attentional bias tasks such asthe amygdala (Bremner et al., 2004). Evidence is accumulating inPTSD that attentional bias and symptomatology are both positivelycorrelated with limbic functioning and negatively correlated withprefrontal activity (Cisler & Koster, 2010). These findings can beunderstood in the light of Eysenck theoretical model of attentionalbias toward threat in anxiety, stipulating that anxiety would favorbottom-up effect of amygdala on prefrontal cortex (PFC) andweaken top-down regulatory control which would manifest bydifficulty in disengaging attention from distracting threat cues(Eysenck, Derakshan, Santos, & Calvo, 2007).

At that stage, it remains unknown whether cerebral alterationsand attentional bias is initially present in PTSD patients beforetraumatic exposure or only occurs after it, alongside symptomdevelopment. Yet, the causal relationship between attentional biasand symptomatology seems bidirectional in anxiety disorders. Onone hand, attentional biases are reduced following interventionsusing for instance implementation intentions in social anxiety(Webb, Ononaiye, Sheeran, Reidy, & Lavda, 2010). On the otherhand, interventions designed to modify attentional biases areassociated with reductions in anxiety (Hakamata et al., 2010), innon-clinical as well as pathologically anxious populations (Koster,Verschuere, Crombez, & Van Damme, 2005). These interventionshave not been studied in PTSD; though studies show that initiallyattentional bias and symptom severity are correlated in PTSD(Pineles et al., 2009).

Our second aim was thus to monitor potential changes inattentional bias in PTSD using e-Stroop and DOT tasks when thesymptoms were relieved i.e. before and after successful treatment.The eye movement desensitization and reprocessing (EMDR)therapy was chosen as a validated treatment option for PTSD,rapidly and efficiently relieving PTSD symptoms (APA, 2004). So far,only one study had looked at the effect of psychotherapy (CognitiveBehavioral Therapy (CBT)), on PTSD attentional bias on a stroop task(Devieni, Blanchard, Hickling, & Buckley, 2004). It had failed toevidence initial e-Stroop interference effect in patients and thus

found no relation between changes in color-naming delays andtreatment response or modality.

Our study explored the attentional bias in PTSD on e-Stroop andDOT tasks, before and after symptom amelioration. This was doneto better define the impaired cognitive processing in PTSD andfurther examine its modulation with symptom amelioration. Wehypothesized that, similarly to anxious populations previouslystudied on these tasks; PTSD patients would have a difficulty dis-engaging their attention from aversive cues and would be slower inthe presence of emotionally negative stimuli. Based on aforemen-tioned premise, we also hypothesized that this bias would beinitially observed in PTSD but not after EMDR.

Methods

Subjects

A total of 23 adult outpatients were recruited among traumavictims at the medico-psychological crisis cell (CUMP) at thePsychiatry Pole of the Conception Hospital inMarseille, France. Theyall met the DSM-IV criteria for PTSD following a single traumaticevent (12 aggressions including hold-ups and rapes, 5 road acci-dents, 6 work related accidents) with no previous history of neuro-logic or psychiatric disorders. Four patients were excluded fromdataanalyses as they abandoned the study. Subsequent analysis included19 patients (7 males and 12 females, with mean age¼ 45�15 yearsand mean education¼ 7.6� 2.7 years after grade 7). Patients hadbeen exposed to their traumatic event for an average of 17.2months.Seven patients were on antidepressants and 7 took anxiolytics.

A total of 19 healthy adult controls (9 males and 10 females, withmean age¼ 38� 14 years and mean education¼ 9.0� 2 years aftergrade 7) with no history of neurologic or psychiatric disorders,were recruited via screening lists at the clinical investigation centreat the Timone Hospital (CIC-UPCET). Groups were matched for age,sex and education level. Patients and controls were also individu-ally matched. As such, each patient was matched with a controlhaving the same age (plus or minus 5 years) and education (plus orminus 2 years).

Psychological assessment

All participants were assessed by a psychiatrist, using thestructured Mini-Internal Neuropsychiatric Interview for DSM-IV(Lecrubier, Weiller, Hergueta, Bonora, & Lepine, 1998). This wasdone to check for the absence of psychiatric disorders prior to thetrauma and screen for PTSD and potential comorbid disorders.Participants responded to demographic questions and completedthe Beck Depression Inventory (BDI) (Cottraux, 1985) and the State-Trait Anxiety Inventory (STAI-Y) (Schweitzer & Paulhan, 1990).Patients also completed trauma-related scales: PTSD Check ListScale (PCL-S) (Ventureyra, Yao, & Cottraux, 2002) and ModifiedPTSD Symptoms Scale (MPSS) (Stephenson, Marchand, & LaVallee,1999). The validated French version was used for all the scales.

EMDR treatment

All PTSD patients underwent EMDR. According to the APAreports published in 2004, this eight-step standardized protocol isone of the validated treatments for PTSD. It is based on an infor-mation processing model (Shapiro & Maxfield, 2002). EMDR is aneffective rapid therapy with stable outcome demonstrated ina 35-month follow-up study (Högberg et al., 2008).

Patients were treated by one of 3 therapists, all trained by theFrench Institute of EMDR. There was no fixed number of sessions.Sessions were planned every 7e15 days according to patients and

M. El Khoury-Malhame et al. / Behaviour Research and Therapy 49 (2011) 796e801798

therapists availability. The treatment was considered successfuland complete when patients no longer reported distress whenthinking about their trauma. They were interviewed again by thepsychiatrist using the MINI. Patients required an average of 4.1treatment sessions (ranging from 1 to 9 sessions) lasting for anaverage of 2.5 month (0.5e4 months).

They were retested when they no longer met PTSD classificationaccording to DSM-IV criteria and had no more pathological scoreson PTSD scales.

Healthy controls were also tested twice, at time points matchingthe interval between patients testing sessions.

Cognitive tests

The cognitive tasks we used were the ones validated byLanteaume, Bartolomei, and Bastien-Toniazzo (2009). They wereadministered by the experimenters.

Emotional stroop taskThis task included 96 trials. Each trial consisted of a black fixa-

tion cross display of 500 ms at the center of a white screen, fol-lowed by a cue display in the middle of the screen until theresponse was given. The inter-trial interval was 500 ms. The cuedisplay consisted of either an emotionally negative word (such asaccident, raping,.), or a neutral one (such as sphere, housing,.).The type of word (emotional, neutral) and ink color (red, blue,green) were randomly counterbalanced across trials, with a newsequence for each participant. Participants were asked to fixate onthe black cross. After it disappeared, they had to identify the colorof the displayed word.

Detection of target taskThis task included 128 trials. Each trial consisted of a succession

of three steps: a black fixation cross display of 500 ms at the centerof a white screen, a cue display of 500 ms, and a dot display untilthe response was given. The inter-trial interval was 500 ms. Thecues consisted of either an emotionally negative word (such asaccident, raping,.) and a matched neutral word (such as sphere,housing,.), or a neutral word and another matched neutral word.Words were vertically opposite at equal distance from the center ofthe screen, 4 cm apart. The probe display consisted of a black circlethat appeared at the same spatial location as one of the two words.The emotional word position (lower, upper), the probe position(lower, upper), and the type of word pair (emotionaleneutral,neutraleneutral) were randomly counterbalanced across trialswith a new sequence for each participant. For the emotional pairs,trials were said to be congruent when the dot replaced theemotional word and incongruent when the dot replaced theneutral word (Fig. 1). Participants were asked to fixate on the black

Fig. 1. DOT task. Illustration of the succession of the three steps for the DOT task fora congruent and incongruent trial.

cross. They were told two words would appear immediately afterthe black cross and, that a dot target would appear after the words.They were asked to give the location of the target.

Procedure

The investigation was carried out in accordance to the latestversion of the Declaration of Helsinki. After receiving clear expla-nations of the procedure, participants provided informed consentin accordance with the guidelines set forth by the local ethicalcommittee CPP South Mediterranean 2. Before starting the exper-iment, all participants completed the BDI and STAI-Y and PTSDpatients completed additional PTSD scales.

Participants were comfortably seated at 60 cm viewing distancefrom a 1700 computer screen. In both tasks, participants were askedto answer as quickly as possible without sacrificing accuracy,by pressing corresponding color or location keys on an AZERTYkeyboard.

Data and statistical analyses

Trials in which participants gave the wrong word color (%errorrate for controls at session 1, 2, PTSD patients pre, post-EMDR is 1.3,1.2, 1.1 and 0.8% respectively), the wrong dot target location (0.6,0.8, 0.5 and 0.5% respectively) or in which reaction time (RT) wasabove or below 2 standard deviations of their mean RT wereconsidered error trials and removed from subsequent analysis,similarly to Lanteaume et al. (2009).

An e� Stroop index was calculated :

RTemotional words � RTneutral

When it differs from zero, it indicates the existence of an atten-tional bias; with a positive indexmeaning that attention is capturedby emotional words and a negative index meaning that emotionalwords are avoided.

Three indices were calculated for the DOT:

- Congruence: RTincongruent� RTcongruent. A positive index indi-cates a bias in threat detection, either on congruent or incon-gruent trials

- Disengagement: RTincongruent� RTneutral. A positive index indi-cates stronger attentional holding for negative cues; subjectsare slower to respond to neutral cues in presence of emotionalones.

- Vigilance: RTneutral� RTcongruent. A positive index indicatesenhanced attention capture for negative cues; subjects arefaster in responding to emotional cues in presence of neutralones (Salemink, Van Der Hout, & Kindt, 2007).

A two-way repeated measures ANOVA was used for the e-Stroop, disengagement and vigilance indices, with Group (2 levels:PTSD patients and Controls) as a between factor and Session(2 levels: for controls: session 1 and 2, and for patients: pre andpost-EMDR) as a within factor. Significant main effects at 0.05significance levels were followed by post-hoc tests using Bonfer-roni correction.

Results

Clinical data

Groups did not differ in terms of age, sex and education levels.Data for the various scales used was found to conform the normal

distribution according to the KolmogoroveSmirnov test results.

Table 1Characteristics of participants: mean (SD) for Beck Depression Inventory (BDI),State-Trait Anxiety Inventory (STAI-Y), and for the patients PTSD Check List Scale(PCL-S), Modified PTSD Symptoms Scale (MPSS), and Impact of Event Scale (IES).Significant p-value: *p< 0.05; **p< 0.001.

Control 1(n¼ 19)

Control 2(n¼ 19)

PTSDpre-EMDR(n¼ 19)

PTSDpost-EMDR(n¼ 19)

BDI depression 3.2 (2.6) 2.3 (2.6) 15 (8.1) 5.6 (3.9) F(1,36)¼ 22.61*STAI state

anxiety29.5 (8.1) 30.5 (8.4) 50.1 (11.3) 34.6 (8.7) F(1,36)¼ 20.47*

STAI traitanxiety

40.3 (7.9) 37 (8.9) 56.5 (10.5) 42.8 (7.6) F(1,36)¼ 14.43*

PCL-S e e 62.3 (11.4) 28.4 (6.6) t(1,18)¼ 11.28**MPSS e e 74.2 (22.7) 21.9 (10.8) t(1,18)¼ 9.93**IES e e 56.7 (13.4) 17.3 (10.0) t(1,18)¼ 9.89**

M. El Khoury-Malhame et al. / Behaviour Research and Therapy 49 (2011) 796e801 799

Table 1 shows the psychometric measures of controls andpatients. In accordance with clinical evaluations, PTSD patientsinitially scored higher than controls on BDI and STAI scales. Aftertreatment, both groups had comparable scores (Table 1).

Patients also scored higher than the cut-off for pathology onPTSD scales pre but not post-EMDR. They met the criteria for thefollowing major current comorbid diagnoses (pre/post-EMDR):major depression (n¼ 12/4), other anxiety disorders (n¼ 15/7) andhigh-medium suicidal risk (n¼ 7/0).

Emotional stroop

Statistical analysis revealed a significant Group� Session inter-action for the e-Stroop index (F(1,36)¼ 4.604, p< 0.05) (Fig. 2).Post-hoc tests showed that unlike controls who had a null e-Stroopindex, patients had a positive one (p< 0.05) with longer RT onemotional trials vs. neutral ones.

After treatment, we found no significant difference in e-Stroopindex between groups. There was no effect of test/retest as wefound no difference in controls RT at their 1st/2nd session. Sessionhad a significant effect only for the PTSD groupwith larger e-Stroopindex pre than post-EMDR (p< 0.05).

DOT

Only the PTSD group had a congruence index different from zero(t(1,18)¼ 2.32, p< 0.05).

Fig. 2. Stroop effect. Mean and error bars of the e-Stroop index. For the control group,digits 1 and 2 indicate testing sessions. *p< 0.05 according to Session�Groupinteraction.

Group� Session interaction was significant for the disengage-ment (F(1,36)¼ 4.94, p< 0.05) but not for the vigilance index(F(1,36)¼ 0.028, p> 0.05) (Fig. 3). Post-hoc tests showed thatpatients had larger disengagement index than controls (p< 0.05).Patients indeed had longer RT compared to controls on incongruentvs. neutral trials; they were slower to respond to a neutral word inthe presence of an emotional one compared to their response in thepresence of a neutral pair.

After treatment, there was no significant difference betweenPTSD and control groups on either index. There was also no effect oftest/retest as we found no difference in controls’ indices at their1st/2nd session. Session had a major effect for PTSD patients, withsignificantly larger disengagement index pre than post-EMDR(p< 0.05).

Adjustment for RT differences: analysis of covariance (ANCOVA)

To examine if comorbid anxiety, depression or use of medicationwould explain the effects found for PTSD on e-Stroop and disen-gagement index, they were entered separately as covariates in theGroup� Session ANOVA.

Statistical analysis revealed none of the three covariates weresignificantly related to any of patients’ attentional bias with thefollowing ANCOVA values respectively for anxiety, depression andmedication: for e-Stroop index (F(1,33)¼ 7.535, p< 0.05, F(1,34)¼4.25, p< 0.05 and F(1,34)¼ 5.29, p< 0.05), and disengagementindex (F(1,33)¼ 4.89, p< 0.05, F(1,33)¼ 7.15, p< 0.05 and F(1,34)¼5.69, p< 0.05).

Correlation analysis

To test for the correlation of bias modulation with changes inanxiety levels and in PTSD symptomatology, Pearson correlationanalyses were conducted between the difference in disengagementindex pre minus post-EMDR and the difference in STAI-A (stateanxiety) and MPSS (symptom severity) scores pre minus post-EMDR, at a 0.05 significance level, with Bonferroni corrections.Statistical analysis revealed significant correlation between

Fig. 3. DOT effect. Mean and error bars of congruence, disengagement and vigilanceindices. For the control group, digits 1 and 2 indicate testing sessions. *p< 0.05according to null hypothesis for the congruence index and according to theSession�Group interaction for the disengagement and vigilance indices.

M. El Khoury-Malhame et al. / Behaviour Research and Therapy 49 (2011) 796e801800

changes in disengagement index and changes in STAI-A scores withr¼ 0.49, n¼ 18, and p< 0.05 (Fig. 4). We also found a strongtendency for a significant correlation between changes in disen-gagement index and changes in MPSS scores.

Discussion

The present study illustrates a deficiency in PTSD’s cognitiveprocessing of emotional cues. It first and foremost replicates resultson the attentional bias in PTSD reflecting patients’ difficulty indisengaging their attention from aversive cues. The most importantresult is that the disengagement bias in PTSD diminishes aftersymptom amelioration post-EMDR.

We have shown that altered threat processing in PTSD is evi-denced on attentional bias tasks. Our results showed that patientswere slower than controls in indicating the color of emotionalwords on the e-Stroop and had difficulty orienting their attentionaway from negatively-valenced words on the DOT task. This resultsupports previous findings reviewed by Bush, Luu, and Postner(2000) on the existence of attentional bias in PTSD toward aver-sive words. It also points to the existence of a disengagementproblem in PTSD (Browning et al., 2010), rather than a vigilance one(Bryant et al., 1999). It is noteworthy that further investigations arestill needed to better define time course changes of attentional biasin PTSD in shorter (50, 100 and 200 ms) and longer (1250 ms)presentations, and its putative modulation by other somewhatmore ecologic stimuli; such as sounds and IAPS images or faces(Koster et al., 2005).

Taken together, our results indicate that emotional informationseems to reduce processing efficiency in PTSD. These findings ofpatients delayed RT in the presence of aversive cues go along thelines of what is generally held about patients with other anxietydisorders such as generalized anxiety disorder, social phobia, panicdisorder and non-clinical high trait anxiety subjects (Cisler &Koster, 2010).

The major finding here remains that the attentional bias in PTSDdiminishes after symptom removal by EMDR. We found thatsimilarly to controls, EMDR treated patients who were symptom-free had null e-Stroop and disengagement indices. To the best ofour knowledge, no studies have so far looked at the modulation of

Fig. 4. Correlation. Plot of correlation between difference in disengagement index anddifference in STAI-A (state anxiety) scores in PTSD patients. r¼ 0.493, n¼ 18 andp< 0.05.

attentional bias in PTSD before and after symptom removal.Restoration of processing bias has been found after CBT in patientswith chronic pain (Dehghani, Sharpe, & Nicholas, 2004) and alcoholdependence (Fadardi & Cox, 2009).

Moreover, changes in disengagement index after EMDR posi-tively correlated with changes in state anxiety, with larger disen-gagement bias found in highly anxious patients. This resultsuggests that such bias could be corrected by decreasing stateanxiety. It goes along recent findings that modification of attentionbias reduces anxious symptomatology (Browning et al., 2010; See,MacLeod, & Bridle, 2009). EMDR is thus one more approach tomodify attentional bias, alongside expressive writing for instance(Vedhara et al., 2010).

We additionally found a strong tendency for a correlationbetween attentional bias and PTSD symptom severity. The failure toreach significant levels might be due to the non-specificity of thewords used and/or to the heterogeneity of the PTSD population interms of trauma-type. A study by Sveen, Dyster-Aas, and Wille-brand (2009) had indeed evidenced a correlation between burn-specific e-Stroop bias in PTSD and symptom severity. We nowprovide preliminary evidence that attentional bias in PTSD wouldbe linked to symptomatology, and could be a marker of thepathology.

A first limitation of the study arises from our inability to retestthe 4 drop-outs, which were mostly out of reach or refused to beretested. Their inclusion, alternatively to the presence of a wait-listgroup of PTSD patients, would have provided more reliableevidence in correlating attentional bias to acquired symptom-atology rather than inherited factors or mere practice effect.

Another limitation is to assess for the effect of the delaybetween testing sessions. It is however not likely for mere passageof time to account for the restoration of cognitive processing ofaversive cues in our tasks, as patients were diagnosed with symp-toms lasting for an average of 17.2 months and showed no signs ofspontaneous recovery. It has been shown that treatment gain byEMDR is stable over 35 months (Högberg et al., 2008). Furtherstudies should address the current limitation to better assesstreatment benefit in attentional bias in follow-up studies.

A third limitation is that some of the patients evaluated were onstable medical regimen for antidepressants and/or anxiolytics (9 of19 patients) and had other comorbid anxiety and/or mood disor-ders. The meta-analysis by Bar-Haim et al. (2007) suggested thatco-occurrence of mood disorders with anxiety does not playa major role in the threat-related bias of anxious individuals. Also,attentional bias appears to occur in equal magnitudes in all anxietydisorders (Cisler & Koster, 2010). We can also argue that our find-ings are relevant to anxiety, which is associated with perturbationsof early attentional deployment (10e500 ms), rather than beingrelevant to depression, which is associated with later effects(500e1000 ms) (Browning et al., 2010).

None of the comorbid factors co-varied with our results;however, their alteration of cognitive processing cannot be totallyruled out. It would be useful in future studies, to further exploredrug and comorbidity interaction with attentional and cognitivebias in larger PTSD subpopulations with or without medication andwith or without comorbidities.

Inasmuch as the inherited and/or acquired facets of attentionalbias are yet undefined, it is important in clinical terms to betteroutline the altered strategy underlying attentional bias in PTSD, itsassociation to symptoms and its modulation by treatment; to betteroptimize therapeutic options. Our design addresses the correlationbetween attentional bias and anxious symptomatology in PTSD,before and after symptom removal. Our results do suggest biases arereduced following interventions that reduce anxiety. Our group haspreviously explored causal relationship between limbic functioning,

M. El Khoury-Malhame et al. / Behaviour Research and Therapy 49 (2011) 796e801 801

attention orientation and PTSD symptomatology. We had shownthat in symptomatic patients, amygdala over-activation causesdisengagement problems, which subsequently drives anxioussymptomatology (El Khoury-Malhame, Reynaud, & Khalfa, 2011).We have currently shown that the attentional bias diminishes aftersymptom removal by EMDR in PTSD. Further studies shouldmonitorwhether such therapeutic interventions relieving symptomsseverity and correcting attentional bias also re-establishes propercerebral functioning in PTSD.

To conclude, our study has shown that cognitive impairment inPTSD is substantiated via e-Stroop and DOT indices, as patientshave a disengagement bias toward emotionally negative words.More importantly, we have demonstrated for the first time that thispathologically impaired attentional bias diminishes after symptomamelioration by successful EMDR. These indices could constitutecognitive markers of PTSD. In order to verify that symptom removalalso restores normal functioning of the cerebral structures knownto be involved in PTSD and attentional bias, similar paradigmsshould be further explored at the brain level using fMRI.

Role of funding source

Funding for this research was provided by the French NationalCenter for Scientific Research CNRS for participants’ payment andcomputer purchase.

The CNRS had no further role in study design; the collection,analysis and interpretation of data; in writing the report nor in thedecision to submit the paper for publication.

Conflict of interest statementThe authors have no actual or potential conflict of interest

including financial or personal relationships that could influence orcould be perceived the work in this manuscript.

References

American Psychiatric Association (APA). (2004). Practice guideline for the treatmentof patients with acute stress disorder and posttraumatic stress disorder. Arlington,VA: American Psychiatric Association Practice Guidelines.

Bar-Haim, Y., Lamy, D., Pergamin, L., Bakermans-Kranenburg, & Van Ijzendoorn, M. H.(2007). Threat-related attentional bias in anxious and non-anxious individuals:a meta-analytic study. Psychological Bulletin, 133, 1e24.

Beck, J. G., Freeman, J. B., Shipherd, J. C., Hamblen, J. L., & Lackner, J. M. (2001).Specificity of stroop interference in patients with pain and PTSD. Journal ofAbnormal Psychology, 110(4), 536e543.

Blanchard, E. B., Kolb, L. C., Pallmeyer, T. P., & Gerardi, R. J. (1982). A psychophysi-ological study of PTSD in Vietnam veterans. Psychiatric Quaterly, 54(4),220e229.

Bremner, J. D., Vermetten, E., Vythilingam, M., Afzal, N., Schmahl, C., Elzinga, B., et al.(2004). Neural correlates of the classic color and emotional stroop in womenwith abuse-related posttraumatic stress disorder. Biological Psychiatry, 55(6),612e620.

Browning, M., Holmes, E., & Harmer, C. (2010). The modification of attentional biasto emotional information: a review of the techniques, mechanisms, and rele-vance to emotional disorders. Cognitive, Affective, & Behavioral Neuroscience,10(1), 8e20.

Bryant, D. J., & Subbiah, I. (1999). Subjective landmarks in perception and memoryfor spatial location. Canadian Journal of Experimental Psychology, 48(1), 119e139.

Bush, G., Luu, P., & Postner, M. I. (2000). Cognitive and emotional influences inanterior cingulate cortex. Trends in Cognitive Science, 4, 215e222.

Cisler, J. M., & Koster, E. H. (2010). Mechanisms of attentional biases toward treat inanxiety disorders: an integrative review. Clinical Psychology Review, 30,203e216.

Cottraux, J. (1985). Evaluation clinique et psychométrique des états dépressifs.Collection Scientifique Survector.

Dehghani, M., Sharpe, L., & Nicholas, M. K. (2004). Modification of attentional biasesin chronic pain patients: a preliminary study. European Journal of Pain, 8(6),585e594.

Devieni, T., Blanchard, E., Hickling, E., & Buckley, T. (2004). Effects of psychologicaltreatment on cognitive bias in motor vehicle accident-related post-traumaticstress disorder. Journal of Anxiety Disorders, 18, 211e231.

El Khoury-Malhame, M., Reynaud, E., & Khalfa, S. (2011). Amygdala activity corre-lates with attentional bias in PTSD. Neuropsychologia, 49(7), 1969e1973.

Elsesser, K., Sartory, G., & Tackenberg, A. (2005). Initial symptoms and reactions totrauma-related stimuli and the development of posttraumatic stress disorder.Depression and Anxiety, 21(2), 61e70.

Eysenck, M. W., Derakshan, N., Santos, R., & Calvo, M. G. (2007). Anxiety andcognitive performance: attentional control theory. Emotion, 7(2), 336e353,Review.

Fadardi, J. S., & Cox, W. M. (2009). Reversing the sequence: reducing alcoholconsumption by overcoming alcohol attentional bias. Drug and Alcohol Depen-dence, 101(3), 137e145.

Fox, E., Russo, R., & Dutton, K. (2001). Attentional bias for threat: evidence fordelayed disengagement from emotional faces. Cognition and Emotion, 16(3),355e379.

Hakamata, Y., Lissek, S., Bar-Haim, Y., Britton, J. C., Fox, N. A., Leibenluft, E., et al.(2010). Attention bias modification treatment: a meta-analysis toward theestablishment of novel treatment for anxiety. Biological Psychiatry, 68(11),982e990.

Hayes, P. C., Weber, C. L., Gallagher, M. B., & Drouillard, G. J. (2009). Attenuation ofapparent new-onset ocular tics with successful treatment of PTSD. CNS Spec-trum, 14(4), 214e220.

Högberg, G., Pagani, M., Sundin, O., Soares, J., Aberg-Wistedt, A., Tärnell, B., et al.(2008). Treatment of post-traumatic stress disorder with eye movementdesensitization and reprocessing: outcome is stable in 35-month follow-up.Psychiatry Research, 159(1e2), 101e108.

Koster, E. H., Verschuere, B., Crombez, G., & Van Damme, S. (2005). Time-course ofattention for threatening pictures in high and low trait anxiety. BehaviourResearch and Therapy, 43(8), 1087e1098.

Lanteaume, L., Bartolomei, F., & Bastien-Toniazzo, M. (2009). How do cognition,emotion, and epileptogenesis meet? A study of emotional cognitive bias intemporal lobe epilepsy. Epilepsy Behavior, 15(2), 218e224.

Lecrubier, Y., Weiller, E., Hergueta, P., Bonora, L., & Lepine, J. P. (1998). French versionM.I.N.I 5.0.0/DSM-IV. Paris, France: INSERM.

Mathews, A., & MacLeod, C. (2005). Cognitive vulnerability to emotional disorders.Annual Reviews of Clinical Psychology, 1, 167e195.

McNally, R. J., Otto, M. W., Yap, L., Pollack, M. H., & Hornig, C. D. (1999). Is panicdisorder linked to cognitive avoidance of threatening information? Journal ofAnxiety Disorders, 13(4), 335e348.

Mogg, K., & Bradley, B. P. (1998). A cognitive-motivational analysis of anxiety.Behaviour Research and Therapy, 36(9), 809e848.

Pineles, S. L., Shipherd, J. C., Mostoufi, S. M., Abramovitz, S. M., & Yovel, I. (2009).Attentional biases in PTSD: more evidence for interference. Behaviour Researchand Therapy, 47(12), 1050e1057.

Posner, M. I., & Petersen, S. E. (1990). The attention system of the human brain.Annual Review of Neuroscience, 13, 25e42.

Salemink, E., Van Der Hout, M., & Kindt, M. (2007). Selective attention and threat:quick orienting versus slow disengagement and two versions of the dot probetask. Behaviour Research and Therapy, 45, 607e615.

Schweitzer, M. B., & Paulhan. (1990). Manuel pour l’inventaire d’Anxiété Trait-Etat(forme Y). Laboratoire de Psychologie de la Santé, Université de Bordeaux II.

See, J., MacLeod, C., & Bridle, R. (2009). The reduction of anxiety vulnerabilitythrough the modification of attentional bias: a real-world study using a home-based cognitive bias modification procedure. Journal of Abnormal Psychology,118(1), 65e75.

Shapiro, F., & Maxfield, L. (2002). Eye movement desensitization and reprocessing(EMDR): information processing in the treatment of trauma. Journal of ClinicalPsychology, 58(8), 933e946.

Stephenson, R., Marchand, A., & LaVallee, M. C. (1999). Translation and transculturalvalidation of the modified PTSD symptom scale. In XXVIst International congressof psychology, Montreal.

Sveen, J., Dyster-Aas, J., & Willebrand, M. (2009). Attentional bias and symptoms ofpost-traumatic stress disorder one year after burn injury. The Journal of Nervousand Mental Disorder, 197(11), 850e855.

Vedhara, K., Brant, H., Adamopoulos, E., Byrne-Davis, L., Mackintosh, B., Hoppitt, L.,et al. (2010). A preliminary investigation into whether attentional bias influ-ences mood outcomes following emotional disclosure. International Journal ofBehavioral Medicine, 17(3), 195e206.

Ventureyra, V. A., Yao, N., & Cottraux, J. (2002). The validation of the post-traumaticchecklist scale (PCLS) in post traumatic stress disorder and non-clinical-subjects. Psychotherapy and Psychosomatics, 29(3 Pt 1), 232e238.

Webb, T. L., Ononaiye, M. S., Sheeran, P., Reidy, J. G., & Lavda, A. (2010). Usingimplementation intentions to overcome the effects of social anxiety on atten-tion and appraisals of performance. Personality and Social Psychology Bulletin,36(5), 612e627.

Williams, J. M., Mathews, A., & MacLeod, C. (1996). The emotional stroop task andpsychopathology. Psychology Bulletin, 120(1), 3e24.