J. Behav. Ther. & Exp. Psychiat. 46 (2015) 19e26

Contents lists available at ScienceDirect

Journal of Behavior Therapy and Experimental Psychiatry journal homepage: www.elsevier.com/locate/jbtep

How does immediate recall of a stressful event affect psychological response to it? Catherine B. Gittins, Helen M. Paterson*, Louise Sharpe** School of Psychology, The University of Sydney, NSW, 2006, Australia

a r t i c l e i n f o

a b s t r a c t

Article history: Received 15 October 2012 Received in revised form 18 July 2014 Accepted 21 July 2014 Available online 14 August 2014

Background and objectives: In forensic settings, individuals who experience a traumatic event are often encouraged to recall it soon afterwards to preserve their memory for it. Some theories of the development of post-traumatic stress disorder (PTSD) suggest that this may increase psychopathology. The primary aim of the study was to examine the effect of immediate recall of a trauma analogue video on psychopathology. Method: Eighty-five undergraduate students were randomised to view a video of a car accident, described as either a real event (high stress) or training event (low stress). They then completed either the Self-Administered Interview (SAI©, Gabbert, Hope, & Fisher, 2009) or a filler task. All participants returned one week later to provide an account of the event. Results: As predicted, participants in the SAI early recall task condition remembered the video content better one week after seeing the video, shown both by their greater recall of correct details and greater rejection of misinformation. However, completing the SAI resulted in higher anxiety immediately afterwards, and more severe PTSD-like symptoms one week later, compared to control condition. PTSD intrusion-like symptoms also predicted more accurate recall, while avoidance predicted poorer memory. Limitations: While the trauma analogue video used in this study has been previously used, and did effectively trigger post-traumatic-like symptoms, it is unclear how well these results generalise to actual trauma situations. Conclusions: These results suggest the relationship between PTSD symptoms and memory might be more complex than previously recognised, with intrusive phenomena possibly promoting memory and avoidance symptoms compromising memory. Crown Copyright © 2014 Published by Elsevier Ltd. All rights reserved.

Keywords: Trauma Memory Post-traumatic stress disorder

1. Introduction Traumatic response to an event and memory for that event are closely intertwined. The way people respond psychologically to stressful events is influenced by how they remember them (Brewin, 2011; McNally, 2003). In turn, memory for an event is impacted by psychological response it (Christianson, 1992). However, how they interact remains unclear. In some circumstances, details of a traumatic event need to be recalled soon after the event for forensic purposes. One tool that has been developed to record eyewitness memory immediately after a crime is the Self-Administered Interview

* Corresponding author. Tel.: þ61 2 9036 9403, fax: þ61 2 9036 5223. ** Corresponding author. Tel.: þ61 2 9351 4558. E-mail addresses: [email protected] (H.M. Paterson), louise. [email protected] (L. Sharpe). http://dx.doi.org/10.1016/j.jbtep.2014.07.006 0005-7916/Crown Copyright © 2014 Published by Elsevier Ltd. All rights reserved.

(SAI©; Gabbert, Hope, & Fisher, 2009). The SAI is a paper-and-pen booklet that instructs witnesses to answer general questions about the event in as much detail as they can. Like the Cognitive Interview (Fisher & Geiselman, 1992) witnesses are instructed to imagine the location of the event, what they saw, thought and felt at the time of the event, and to report everything they can remember. This is aimed to increase clarity of details recalled through a process known as context reinstatement (Hope, Gabbert, & Fisher, 2011). The SAI is regularly used in the UK, Norway, and the Netherlands, and as such, was used in this research to increase applicability of the findings. The SAI was designed to be given to witnesses as soon after a crime as possible to reduce delay between encoding and recall, thus optimising memory. Furthermore, this immediate recall is intended to protect individuals from incorporating incorrect postevent information into their memories (Gabbert, Hope, Fisher, & Jamieson, 2012). The SAI encourages individuals to retain as many details about the event as possible by offering an early recall

20

C.B. Gittins et al. / J. Behav. Ther. & Exp. Psychiat. 46 (2015) 19e26

opportunity. Research suggests that it improves recall of the event one week later, and is protective against misinformation (Gabbert et al., 2009, 2012). However, previous research has been conducted using low stress, non-violent mock-crime videos as memory stimuli. No comparison of its effect on events of differing levels of stress has been examined. Whilst not all crimes are traumatic in nature, some clearly are. How recalling a stressful event shortly after experiencing it interacts with the processes by which traumatic responses develop remains unclear. Cognitive theories of PTSD have not specifically addressed the impact of actively promoting better recall of the event on psychological outcomes. However, they do discuss the impact of processing a traumatic event. As Brewin, Dalgleish, and Joseph (1996) point out, most cognitive theories agree that integration of the memory with pre-existing cognitive structures is integral to psychological recovery. Writing about the details of a recently experienced stressful event may encourage this. Ehlers and Clark's (2000) model of PTSD asserts that organisation of trauma representations within the context of other autobiographical memories is crucial for successful processing. If this does not occur, trauma memories will be continually and unintentionally retrieved, resulting in the intrusion symptoms. Foa's model of PTSD (Foa & Kozak, 1986; Foa, Steketee, & Rothbaum, 1989) also suggests that processing supports psychological recovery. Accordingly, exposure to fear memories activates the trauma schema, allowing new information to be incorporated into the memory of the event, thus reducing anxiety. By encouraging individuals to recall the details of the event, an early initial recall opportunity could help individuals organise their memories and reduce anxiety. However, at a very early stage, some types of processing may not be beneficial. Meta-analyses of psychological debriefing have found that debriefed individuals are no less likely to develop PTSD and, in some cases, are more likely to develop symptoms (e.g. Rose, Bisson, & Wessely, 2003). Paterson, Whittle, and Kemp (2014) examined debriefing in which individuals reconstructed the details of the event. They found that, after viewing a stressful video, debriefed participants reported significantly more intrusion symptoms than non-debriefed participants. Some theories of PTSD make predictions about the impact of recalling a traumatic event in the period immediately following the event. Brewin et al.'s (1996) theory suggests that this period is particularly anxiety provoking due to both emotional reactions and conditioned responses to threat stimuli. In this initial phase, rather than engaging extensively with trauma representations, they claim individuals need to gradate their exposure to trauma memories and gradually process them. Otherwise, the anxiety may be so great that they avoid processing altogether. Similarly, Horowitz's (1986) theory of PTSD suggests that individuals may experience a phase, immediately after the traumatic event, during which they cannot begin processing. This phase occurs because the discrepancy between the new information and established schemas creates extreme anxiety. Any attempt at processing during this period leads to increased avoidance symptoms. Thus, both theories (Brewin et al., 1996; Horowitz, 1986) suggest that encouraging extensive processing immediately after the trauma may increase anxiety and avoidance symptoms. The effect of stress on memory consolidation is another area that remains unclear due to mixed findings. In his review, Christianson (1992) concluded that stressful events were well retained, sometimes better retained, than non-stressful events. Recent research has similarly found that higher levels of stress during encoding are associated with better recall (e.g. Paz-Alonso, Goodman, & Ibabe, 2013; Wiemers, Sauvage, Schoofs, HamacherDang, & Wolf, 2013). However, a meta-analysis by Deffenbacher,

Bornstein, Penrod, and McGorty (2004) suggested that stressful events are less well remembered than non-stressful events. A recent study has supported this contention, finding worse memory in individuals who had anxiety induced during encoding (Attwood, Penton-Voak, Burton, & Munafo, 2013). These contradictory findings are consistent with the disagreement within the PTSD literature. Some researchers assert that, although some details may be distinctly recalled, when asked to deliberately recall the event, individuals with PTSD generally display an inaccurate and incomplete memory (Brewin, 2011; Halligan, Michael, Clark, & Ehlers, 2003). However, other theorists assert that trauma memories are enhanced compared to non-trauma memories. McNally (2003) contends that the inherent intensity of the trauma experience heightens recall. The empirical evidence for these assertions is mixed, with some studies indicating that traumatic events are less well remembered than non-traumatic events (Byrne, Hyman, & Scott, 2001; Tromp, Koss, Figueredo, & Tharan, 1995) and others suggesting the opposite (Berntsen, 2001). As such, whether individuals with PTSD have heightened or decreased recall of the trauma experience is unclear. This research aimed to answer two questions. First, how does recall soon after a stressful event affect psychological response to it? As argued above, we hypothesised that participants asked to recall information immediately following exposure to a traumatic video would have higher anxiety than control participants and that this would lead to increased avoidance over time. Second, are high stress events differentially recalled than low stress events? Due to the mixed nature of the previous research (described above), we did not hypothesise whether stress would lead to better or worse memory. 2. Method 2.1. Design The study was a 2  2 design conducted in two sessions. In session 1, participants watched a video depicting the aftermath of a car accident. To manipulate stress, in the high stress condition the video was described as being a real event (Real video). In the low stress condition, the video was described as being an event staged for training purposes (Training video). Participants were cluster randomised in small groups into these conditions. After viewing the video on individual computers using headsets, participants were re-randomised into writing conditions. One group completed the SAI, writing about the events of the video (SAI). The other group completed a booklet of filler questions (Control). All participants completed measures of psychological responses and memory for the event (Fig. 1). 2.2. Participants Eighty-five (52 female) undergraduate students participated in the study for course credit. The mean age was 19.56 years (SD ¼ 3.93). Five participants did not return for session 2, however, their data from the session 1 was included. Participants were excluded if they, or a close friend or family member, had been involved in a serious car accident. They were advised not to sign up if they had a history of anxiety disorder. 2.3. Materials 2.3.1. Video The visual content of the videos was identical. The video (10 min) presented live footage of the aftermath of a car accident. Three victims are lying next to a road beside an upturned car. They

C.B. Gittins et al. / J. Behav. Ther. & Exp. Psychiat. 46 (2015) 19e26

21

85 Participants

SESSION ONE Baseline measure of psychological wellbeing: DASS21

Randomisation into video stress conditions

Real video n = 43

Training video n = 42

Test stress: STAI

Randomisation into writing conditions

SAI n = 22

Filler n = 21

SAI n = 21

Filler n = 21

Test stress: STAI, IES-R

One week delay

SESSION TWO n = 22

n = 20

n = 20

n = 18

Exposure to post event information (‘eyewitness statement’)

Test stress: STAI, IES-R, DASS21

Test memory: free-recall

Manipulation check questionnaire Fig. 1. Outline of study design. DASS21 Depression Anxiety Stress Scales, STAI State-Trait Anxiety Inventory, IES-R Impact of Event Scale-Revised.

have sustained serious injuries and display distress (e.g. groaning, crying). Paramedics, firefighters and a doctor attend to the victims. A fourth victim (deceased) is later shown with a mutilated face. This video has previously been used as a trauma analogue (Devilly & Annab, 2008; Devilly, Varker, Hansen, & Gist, 2007). Two voice-overs were added to differentiate high and low stress. In the Real video (high stress), the victims are described as people injured in a car accident, whereas, in the Training video (low stress) characters are described as actors simulating an event.

the original text to make it appropriate for the laboratory, for example referring to ‘victims’ rather than ‘offenders’. 2.3.2.2. Filler booklet. Control participants were given a booklet of equal length to the SAI. The questions bore no relation to the video, asking participants to describe their activities of the previous day and week and their plans for the rest of the day, without referring to their emotional response. Similar questions have previously been used as neutral filler tasks in other studies (e.g. Pennebaker, Kiecolt-Glaser, & Glaser, 1988).

2.3.2. Writing tasks 2.3.2.1. SAI. Participants were given the 16-page SAI booklet (Hope et al., 2011). It asks witnesses to visualise the event and write down everything they can recall about it. Finally, they are asked to respond to more specific questions. Minor changes were made to

2.3.3. Psychological response measures 2.3.3.1. State-Trait Anxiety Inventory-state (STAI-state). The STAIstate (Spielberger, 1984) was used to assess participant anxiety. The 20-item scale requires individuals to indicate how they

22

C.B. Gittins et al. / J. Behav. Ther. & Exp. Psychiat. 46 (2015) 19e26

currently feel on a Likert-scale ranging from 1 (Not at all) to 4 (Very much so). The STAI has high convergent validity (Gros, Antony, Simms, & McCabe, 2007) and Cronbach's alphas of .93e.94 were found in this sample. 2.3.3.2. Impact of Event Scale-Revised (IES-R). The IES-R (Weiss & Marmar, 1997) was used to assess participants' PTSDelike symptoms. The 22-item scale asks individuals to what degree they have experienced symptoms over the past week, using Likert-scales ranging from 0 to 4. The measure contains three subscales (Intrusion, Avoidance and Hyper-Arousal). It has good discriminant (Beck et al., 2008) and convergent (Erickson & Steiner, 2000) validity. For the first administration the instructions were changed to indicate symptoms since viewing the film and three questions about sleep were removed. Cronbach's alphas were between .82 and .88. 2.3.3.3. Depression Anxiety Stress Scales-21 (DASS21). The DASS-21 (Lovibond & Lovibond, 1995) was used to assess participants' anxiety, stress and depression. The 21-question measure asks individuals to rate to what extent statements have applied to them over the past week. Responses are indicated on Likert-scales ranging from 0 to 4. The DASS-21 has been well validated in a non-clinical sample (Henry & Crawford, 2005). Cronbach's alphas of .86e.89 were found. 2.3.4. Post-Event Information (PEI) Participants were given 10 items of correct and 10 items of incorrect information (misinformation) about the video before the memory test (Appendix A). The misinformation entailed changes of existing information and addition of nonexistent details. The PEI was presented in the form of a hand-written statement, described as an eyewitness statement. Twelve participants reported that they had recognised that the eyewitness's statement was misleading. The analyses were re-run without these participants, however the pattern of results remained unchanged, so they were reinstated. 2.3.5. Memory measure Participants were asked to complete a questionnaire that asked participants to type responses to broad questions about the video, which were designed to reflect cross-examination style questions (Appendix B). They included a general question about what they remembered and specific questions about the appearance of individuals, dialogue between individuals, the setting and the information in the narration. Both responses to the recall questionnaire and the SAI were deidentified and coders were blind to conditions. Details were divided into three categories: 1) no post-event information (no-PEI; information recalled by participants but not mentioned in the eyewitness statement; either accurate or incorrect [i.e. details that were not present in the video, for example the doctor was wearing glasses]), 2) incorrect PEI (misinformation from the ‘eyewitness statement’; either accepted, resisted or incorrect [i.e. details reported regarding the misinformation item but not reflecting the specific incorrect misinformation]) and correct PEI (correct PEI from the ‘eyewitness statement’; scored as correct PEI). Each complete detail reported was given one point. Partial answers (for example if the participant stated that the event occurred in the afternoon rather than specifying the time) were given a score of .5. While this scoring system is arguably in contrast to the approach of Koriat and colleagues (Goldsmith, Koriat, & Pansky, 2005; Goldsmith, Koriat, & Weinberg-Eliezer, 2002), it is consistent with previous research (Monds, Paterson, Kemp, & Bryant, 2013) and allows partial credit to be given for answers where the participant neglects details in order to preserve accuracy.

For the open-recall questionnaire, inter-rater reliability was calculated by comparing the scores of the main coder with those of another, independent coder for ten questionnaires. Cohen's Kappa indicated substantial inter-rater reliability (k ¼ .68, p < .001). For the SAI, two coders independently coded the same 10 SAIs, and the remaining SAIs were coded by one of the coders. Cohen's Kappa again indicated substantial inter-rater reliability (k ¼ .73, p < .001). These assessments were conducted on the basis of recommendations by Landis and Koch (1977). 2.3.6. Manipulation check measure Participants were asked three questions that tested whether they believed that the video depicted a real event or a training event. On a Likert-scale from 1 (Not at all) to 6 (Very) participants rated how realistic, disturbing and typical of emergency services they considered the footage. 2.4. Procedure Participants were tested in small groups but did not interact during the study. Prior to meeting the experimenter, participants were cluster randomised to conditions. Participants were seated individually and, after giving consent, completed the DASS21. The experimenter reminded participants that upcoming video may be distressing. Next, individuals in the Real video condition were told they would view a video of the aftermath of a real car accident and shown the Real video. Participants in the Training condition were told that they would view a video of an event staged for training purposes. They were told that ‘victims’ were actors and that nobody was actually injured. They then viewed the Training video. Prior to viewing, all participants were told that they would later be asked questions about the video and instructed to pay close attention. Once the videos finished, all participants completed the STAI and demographic questions. Participants then completed the writing task. In the SAI condition participants were given the SAI and asked to write responses. They were asked to answer it as if they were actually at the scene of the accident. All participants were told the booklet might take around 30 min. Participants who were unable to finish (23 in SAI, 27 in control) were stopped by the experimenter at that time. An ANOVA revealed that there was no significant difference in the amount of accurate or incorrect information reported by those who finished compared to those who did not. Following the writing task, all participants completed the STAI and IES-R. Seven days later the participants returned. They were given a copy of the eyewitness statement and told that it was written by another participant. Participants were instructed to read it to remind themselves of the video. All participants then completed the STAI, IES-R, DASS21, memory test and manipulation check questionnaire and debriefed. In both sessions, participants who finished early waited until their fellow participants had finished. The protocol was approved by the University of Sydney Human Research Ethics Committee. 2.5. Planned analysis A power analysis using G*Power3.1 revealed that 42 participants per group would be needed, given an effect size of .8, with a level of significance at .05 (Faul, Erdfelder, Buchner, & Lang, 2009). Since the primary analyses were 2  2 ANOVAs, this required 84 participants in total. Preliminary analyses revealed that all measures were normally distributed except DASS21. However, as DASS21 was not the primary outcome, and evidence suggests that ANOVA is robust enough to overcome violations of normality (Schmider,

C.B. Gittins et al. / J. Behav. Ther. & Exp. Psychiat. 46 (2015) 19e26

23

Ziegler, Danay, Beyer, & Buhner, 2010), the data was not normalised. Further preliminary analysis was conducted using ANOVAs to assess for covariates by comparing baseline characteristics across groups. To examine the hypothesised effects of the SAI and stress conditions on psychological responses, two 2(stress: higher, lower) 2(writing: SAI, control) ANOVAs were conducted on STAI scores pre- and post-writing task, as was a mixed design 2(stress) 2(writing) (2)(time: pre-, post-writing task) ANOVA. IES-R scores (overall and separate factors) post-writing task were also analysed using a 2(stress) 2(writing) ANOVA. STAI, DASS21 (overall and factors) and IES-R (overall and factors) scores after a one-week delay were analysed using 2(stress) 2(writing) ANOVAs. To examine the hypothesised effects of the SAI and stress condition on memory for the event, a second series of ANOVAs was conducted. A 2(stress) 2(writing) ANOVA was conducted for each category of detail: total accurate (no PEI), incorrect, misinformation accepted, misinformation resisted and correct PEI reported. 3. Results Preliminary analysis indicated a significant interaction between stress and writing conditions for DASS-Stress, F(1,81) ¼ 9.71, p ¼ .003. Therefore, DASS-Stress was included as a covariate in all analyses. However, when analyses were run without this covariate, a similar pattern of results emerged. No other significant differences were observed between groups. 3.1. Video stress manipulation Real video participants rated the video as significantly more realistic (M ¼ 5.33, F(1,75) ¼ 17.20, p < .001; Hedges's g ¼ .92, 95% CI ¼ .66, 1.18), more disturbing (M ¼ 4.26, F(1,75) ¼ 13.66, p < .001; Hedges's g ¼ .82, 95% CI ¼ .5, 1.13) and less typical of the emergency services (M ¼ 4.30, F(1,75) ¼ 5.35, p ¼ .023; Hedges's g ¼ .50, 95% CI ¼ .73, .27) than training video participants (Ms ¼ 4.16, 3.03, 4.85, respectively). Real video participants also had significantly higher STAI-State scores immediately after the video than training video participants, F(1,80) ¼ 21.30, p < .001, Hedges's g ¼ .96, 95% CI ¼ 1.29, 3.20 (See Table 1). 3.2. Psychological responses Immediately following the writing task, a 2  2 ANOVA analysis revealed a main effect for stress, with Real participants showing higher STAI-State scores than Training participants (F(1,80) ¼ 11.37, p ¼ .001; Hedges's g ¼ .60, 95% CI ¼ 1.53, 2.73, see Table 1). Furthermore, SAI participants had significantly higher STAI-State

Fig. 2. Effect of writing condition on STAI-State scores pre- and post-writing task. STAI State-Trait Anxiety Inventory, SAI Self-Administered Interview.

scores immediately after the writing task than control participants (F(1,80) ¼ 22.31, p < .001; Hedges's g ¼ .87, 95% CI ¼ 1.16, 2.91). No significant interaction was found (F(1,80) ¼ 1.15, p ¼ .286). A mixed design 2  2  (2) ANOVA comparing stress condition, writing condition and time of testing found that the reduction in STAI-State scores from pre- to post-writing task was significantly more marked for control participants than for SAI participants (F(1,80) ¼ 13.28, p < .001) (Fig. 2). There was no significant effect of stress or 3-way interaction. A 2  2 ANOVA demonstrated that SAI participants had significantly higher IES-R scores (M ¼ 22.02, SD ¼ 10.72) than controls (M ¼ 15.79, SD ¼ 11.04) immediately following the writing task (F(1,80) ¼ 7.42, p ¼ .008; Hedges's g ¼ .57, 95% CI ¼ 1.75, 2.88). There were no significant effects for stress nor significant interactions. Separate analyses of IES-R factors revealed that SAI participants had significantly higher IES-Intrusion scores (M ¼ 8.40, SD ¼ 4.95) than controls (M ¼ 5.00, SD ¼ 4.22, F(1,80) ¼ 12.62, p < .001; Hedges's g ¼ .73, 95% CI ¼ .25, 1.71). Additionally, real video participants had significantly higher IES-Intrusion scores (M ¼ 7.86, SD ¼ 4.84) than training video participants (M ¼ 5.55, SD ¼ 4.70, F(1,80) ¼ 5.82, p ¼ .018; Hedges's g ¼ .48, 95% CI ¼ .54, 1.49). No interaction was found for IES-Intrusion scores, nor were main effects or interactions found for IES-Avoidance and IESHyperarousal scores.

Table 1 Mean STAI-State scores, immediately after the video but before the writing task, and immediately after the writing task. Writing condition

Stress condition Pre-writing task Real

SAI Control Total

Post-writing task Training

Total

Real

Training

Total

M (SD)

M (SD)

M (SD)

M (SD)

M (SD)

M (SD)

57.50 (8.47) 53.81 (12.25) 55.70 (10.53)a

44.10 (8.78) 46.90 (12.15) 45.50 (10.57)a

50.95 (10.89)c 50.36 (12.55)c

48.27 (10.21) 35.38 (7.78) 41.98 (11.11)a

37.81 (9.00) 34.10 (8.31) 35.95 (8.76)a

43.16 (10.90)b,c 34.74 (7.97)b,c

STAI: State-Trait Anxiety Inventory, SAI: Self-Administered Interview. a Stress conditions significantly different p < .001. b writing conditions significantly different p < .001. c interaction significant p < .001.

24

C.B. Gittins et al. / J. Behav. Ther. & Exp. Psychiat. 46 (2015) 19e26

One week after the video, 2  2 ANOVAs showed no significant between-group differences on DASS-21, STAI-State or IES-R scores. When IES-R factors were analysed separately, Real participants had significantly higher IES-Intrusion scores (M ¼ 5.45, SD ¼ 4.26) than Training participants (M ¼ 3.37, SD ¼ 3.63, F(1,75) ¼ 5.51, p ¼ .022; Hedges's g ¼ .52, 95% CI ¼ .32, 1.36). Moreover, SAI participants had significantly higher IES-Avoidance scores (M ¼ 8.07, SD ¼ 5.43) than Control participants (M ¼ 5.71 SD ¼ 4.82, F(1,75) ¼ 4.36, p ¼ .040; Hedges's g ¼ .46, 95% CI ¼ .64, 1.55). There were no other main effects and no significant interactions for IES-Intrusion, -Avoidance and -Hyperarousal scores. 3.3. Memory for the event A 2  2 ANOVA revealed that SAI participants recalled significantly more accurate (no PEI) details than controls (F(1,75) ¼ 4.91, p ¼ .030; Hedges's g ¼ .50, 95% CI ¼ 2.92, 3.90) (See Table 2). They also resisted significantly more misinformation than controls (F(1,75) ¼ 8.46, p ¼ .005; Hedges's g ¼ .65, 95% CI ¼ .45, .85). There were no significant effects of stress condition nor interactions (ps > .068) and no significant between-group differences for incorrect, misinformation accepted or correct PEI reported (ps > .101). There were no significant differences between groups for overall accuracy (total correct details divided by total correct and incorrect details; ps > .830). Change in memory over time was examined for participants in the SAI condition using a (2) (SAI, open-recall) 2 (stress condition) ANOVA. Participants who completed the SAI reported significantly more of the misinformation items in the recall questionnaire than they had in the SAI (F(1.39) ¼ 13.51, p < .001; Hedges's g ¼ 1.40, 95% CI ¼ 1.62, 1.19). In the recall questionnaire, SAI participants recalled more incorrect items than they had in the SAI (F(1.39) ¼ 4.73, p ¼ .036; Hedges's g ¼ .97, 95% CI ¼ 1.56, .38). All other main effects and interactions were non-significant (all ps > .05). 3.4. Effect of PTSD symptoms on memory The effect of IES-Intrusion and IES-Avoidance scores (immediately following SAI) on SAI scores was analysed using the hierarchical regression. Neither variance in total correct recall nor total incorrect scores were significantly predicted in either steps of the models. However, in a model in which Step 1 contained stress condition and DASS-Stress score and Step 2 contained STAI scores (immediately following SAI), Step 2 accounted for significantly more variation in SAI total accurate scores (R2 change ¼ .1, F change (1,39) ¼ 5.03, p ¼ .031). STAI scores were significantly predictive of accurate recall (b ¼ .56, t(39) ¼ 2.24, p ¼ .031). This model did not significantly predict SAI total incorrect scores. The effect of IES-Intrusion and IES-Avoidance scores (at followup) on open-recall scores was analysed using hierarchical regression. Step 1 of the model included writing condition, stress

condition and DASS-Stress scores. Step 2 contained IES-Intrusion and IES-Avoidance scores. Step 1 of the model predicting correct recall (no PEI) was not significant (R2 ¼ .08, F(3,76) ¼ 2.06, p ¼ .113). However, Step 2 accounted for significantly more variance in scores (R2 change ¼ .23, F change (2,74) ¼ 12.09, p < .001). Both IES-Intrusion and IES-Avoidance scores independently predicted correct recall (no PEI). IES-Intrusion scores predicted increased recall, (b ¼ 2.07, t(74) ¼ 4.69, p < .001), while IES-Avoidance scores predicted decreased recall (b ¼ 1.04, t(74) ¼ 3.09, p ¼ .003). Step 1 of the model predicting incorrect scores was also not significant, (R2 ¼ .06, F(3,76) ¼ 1.72, p ¼ .170). However, when IESIntrusion and IES-Avoidance scores were added to the model, significantly more variation was accounted for (R2 change ¼ .08, F change (2,74) ¼ 3.52, p ¼ .035), with IES-Intrusion scores being the only independent predictor (b ¼ .31, t(74) ¼ 2.57, p ¼ .012). PTSD symptoms did not significantly predict misinformation accepted, misinformation resisted or correct PEI reported scores. 4. Discussion We hypothesised that immediate written recall of the stressful video would have a deleterious effect on stress and avoidance symptoms. This hypothesis was supported. While differences between groups in state anxiety were not evident a week later, participants who had an early recall opportunity displayed significantly more anxiety than control participants immediately after the writing task. One week later, despite no difference in PTSD-like symptoms overall, the SAI group displayed significantly higher avoidance-like symptoms than control participants. Participants in the Real video condition experienced significantly higher anxiety and, although ratings were relatively high for both groups, rated the video as significantly more realistic and disturbing than those in the Training video condition. However, the Real group also rated the video as significantly less typical of emergency services than the Training group. This may have been because they felt that the accident shown was more disturbing than was typical for these types of incidents. However, self-reported stress did not differentially affect memory. Real and Training condition participants did not display significantly different levels of recall. An early recall opportunity did help maintain memory for the event, preserving the quality of recall as reported in previous research (Gabbert et al., 2009, 2012). One week after viewing the video, SAI participants recalled a greater number of accurate details and were more resistant to misinformation than control participants. To further investigate the relationship between PTSD and memory, we examined the influence of PTSD-like symptoms on memory for the event. Intrusion-like symptoms were found to predict greater accurate video recall. However, they also predicted

Table 2 Mean number of items recalled. Writing condition

Stress condition Correct items (no PEI)

SAI Control Total

Misinformation resisted

Real

Training

Total

Real

Training

M (SD)

M (SD)

M (SD)

M (SD)

M (SD)

M (SD)

43.71 (15.95) 37.65 (15.36) 40.82 (15.78)

49.40 (17.73) 39.31 (12.94) 44.62 (17.26)

46.42 (17.73)a 38.43 (14.10)a

1.50 (.91) .85 (.93) 1.19 (.97)

1.85 (1.04) 1.28 (.83) 1.58 (.98)

1.67 (.98)a 1.05 (.90)a

PEI: Post-Event Information, SAI: Self-Administered Interview. a Writing conditions significantly different p < .05.

Total

C.B. Gittins et al. / J. Behav. Ther. & Exp. Psychiat. 46 (2015) 19e26

greater incorrect responses. Avoidance-like symptoms predicted reduced accurate recall. 4.1. Effect of immediate recall of a stressful event on psychological outcomes These findings suggest that completing a written recall task about a stressful event shortly following the event increases certain types of anxiety symptoms (specifically avoidance), although other types of anxiety do not appear to be affected. These effects are not necessarily specific to the SAI, and may result from any form of early recall. It is unclear whether or not these findings are specifically due to elements which are unique to the SAI, such as the context reinstatement. The results suggest that processing immediately after the traumatic event is not ideal. It supports Brewin et al. (1996) and Horowitz's (1986) theories, which suggest that, during this initial phase, anxiety is so great that extensive processing increases stress and thereby promotes avoidance. Thus, this study suggests that, despite the benefits of processing of trauma stimuli (Harvey, Bryant, & Tarrier, 2003), there is a period immediately following trauma during which extensive processing is not helpful and potentially detrimental. 4.2. Impact of memory for the event on psychological response The relationship between trauma and memory remains contentious, with some arguing that people with PTSD have reduced recall for their traumatic experience (e.g. Brewin, 2011) and others arguing they have enhanced recall (e.g. McNally, 2003). This study suggests a more complicated relationship between PTSD and memory for the event. In this sample, intrusionlike symptoms predicted greater accurate recall, consistent with Monds et al. (2013). Intrusion-like symptoms also predicted more incorrect responses. Thus, frequent recollection of an event through intrusion symptoms appears to both increase accurate memories and also increase likelihood of developing false memories. On the other hand, avoidance-like symptoms were associated with decreased memory. While it is difficult to extrapolate a causal relationship between trauma symptoms and memory for the event, this research suggests that certain PTSD symptoms may support recall of the traumatic event while others may interfere with it. It has previously been suggested that intrusion and avoidance symptom clusters may form two subtypes of PTSD. Brewin et al.'s (1996) theory suggests that continual focus on intrusive symptoms results in ‘chronic emotional processing’, while repeated avoidance of trauma memories results in ‘premature inhibition of processing’. The findings of the present study indicate that these subtypes may have different relationships to memory for the event. Chronic processing may result in enhanced recall, while inhibited processing may result in reduced recall. 4.3. Limitations, strengths and future directions This study has limitations, particularly the use of a traumaanalogous video. It is possible that real-life traumatic events are qualitatively different to analogous videos. However, other research which uses the same video stimulus and manipulates stress levels (Devilly & Varker, 2008) suggests there is not a difference, demonstrating variability in anxiety responses which reflects results found in field research (Mayou, Ehlers, & Hobbs, 2000). Although there were differences between the real and training groups on stress, we only measured self-reported stress and whether the manipulation affected other aspects of the stress response (e.g. cortisol response, heart rate) is unknown.

25

There were also limitations to the ecological validity of the implementation of the SAI. It was implemented within 10 min of the video, which would be rarely possible in real life, and some participants were unable complete all the items due to time constraints. In the field, witnesses are given unlimited time to complete the SAI. These issues should be considered when interpreting the results. These results are preliminary and future research should further investigate the psychological effects of recall of a stressful event. Given the increasingly widespread usage of the SAI, field testing should be conducted to measure the effects of the instrument, both short-term and long-term, on the psychological responses of witnesses to real-life crimes. 4.4. Conclusion This study suggests that writing about a stressful event soon after it may lead to more anxiety than completion of a non-recall task. Participants who completed the SAI were significantly more anxious immediately after completing the SAI and had significantly higher PTSD-like avoidance symptoms one week later than control participants. Thus, while completion of the SAI appears to significantly benefit memory for the event, it also appears to have a psychologically detrimental effect, at least in the short-term. Practises that encourage individuals who have experienced a traumatic event to cognitively engage with it soon afterwards should be cautious about potentially negative psychological effects. Acknowledgements This research was supported by two Australian Research Council Linkage Grants (LP110100220 & LP0989719). The authors would like to thank Celena Mckay and Li-Chin Wang for assistance with coding. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jbtep.2014.07.006. References Attwood, A. S., Penton-Voak, I. S., Burton, A. M., & Munafo, M. R. (2013). Acute anxiety impairs accuracy in identifying photographed faces. Psychological Science, 24, 1591e1594. Beck, J., Grant, D. M., Read, J. P., Clapp, J. D., Coffey, S. F., Miller, L. M., et al. (2008). The impact of event scale-revised: psychometric properties in a sample of motor vehicle accident survivors. Journal of Anxiety Disorders, 22, 187e198. Berntsen, D. (2001). Involuntary memories of emotional events: do memories of traumas and extremely happy events differ? Applied Cognitive Psychology, 15(7), S135eS158. Brewin, C. R. (2011). The nature and significance of memory disturbance in posttraumatic stress disorder. In S. Nolen-Hoeksema, T. D. Cannon, & T. Widiger (Eds.), Annual review of clinical psychology (Vol. 7, pp. 203e227). Brewin, C. R., Dalgleish, T., & Joseph, S. (1996). A dual representation theory of posttraumatic stress disorder. Psychological Review, 103, 670e686. Byrne, C. A., Hyman, I. E., & Scott, K. L. (2001). Comparisons of memories for traumatic events and other experiences. Applied Cognitive Psychology, 15, S119eS133. Christianson, S.-A. (1992). Emotional stress and eyewitness memory: a critical review. Psychological Bulletin, 112, 284e309. Deffenbacher, K. A., Bornstein, B. H., Penrod, S. D., & McGorty, E. (2004). A metaanalytic review of the effects of high stress on eyewitness memory. Law and Human Behavior, 28, 687e706. Devilly, G. J., & Annab, R. (2008). A randomised controlled trial of group debriefing. Journal of Behavior Therapy and Experimental Psychiatry, 39, 42e56. Devilly, G. J., & Varker, T. (2008). The effect of stressor severity on outcome following group debriefing. Behaviour Research and Therapy, 46, 130e136. Devilly, G. J., Varker, T., Hansen, K., & Gist, R. (2007). An analogue study of the effects of psychological debriefing on eyewitness memory. Behaviour Research and Therapy, 45(6), 1245e1254.

26

C.B. Gittins et al. / J. Behav. Ther. & Exp. Psychiat. 46 (2015) 19e26

Ehlers, A., & Clark, D. M. (2000). A cognitive model of posttraumatic stress disorder. Behaviour Research and Therapy, 38, 319e345. Erickson, S. J., & Steiner, H. (2000). Trauma spectrum adaptation: somatic symptoms in long-term pediatric cancer survivors. Psychosomatics: Journal of Consultation Liaison Psychiatry, 41, 339e346. Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses. Behavior Research Methods, 41, 1149e1160. Fisher, R. P., & Geiselman, R. E. (1992). Memory-enhancing techniques for investigative interviewing: The cognitive interview. Springfield, IL: Charles C. Thomas. Foa, E. B., & Kozak, M. J. (1986). Emotional processing of fear - exposure to corrective information. Psychological Bulletin, 99, 20e35. Foa, E. B., Steketee, G., & Rothbaum, B. O. (1989). Behavioral cognitive conceptualizations of Post-Traumatic stress disorder. Behavior Therapy, 20, 155e176. Gabbert, F., Hope, L., & Fisher, R. P. (2009). Protecting eyewitness evidence: examining the efficacy of a self-administered interview tool. Law and Human Behavior, 33, 298e307. Gabbert, F., Hope, L., Fisher, R. P., & Jamieson, K. (2012). Protecting against misleading post-event information with a self-administered interview. Applied Cognitive Psychology, 26, 568e575. Goldsmith, M., Koriat, A., & Pansky, A. (2005). Strategic regulation of grain size in memory reporting over time. Journal of Memory and Language, 52, 505e525. Goldsmith, M., Koriat, A., & Weinberg-Eliezer, A. (2002). Strategic regulation of grain size memory reporting. Journal of Experimental Psychology: General, 131, 73. Gros, D. F., Antony, M. M., Simms, L. J., & McCabe, R. E. (2007). Psychometric properties of the state-trait inventory for cognitive and somatic anxiety (STICSA): comparison to the state-trait anxiety inventory (STAI). Psychological Assessment, 19, 369e381. Halligan, S. L., Michael, T., Clark, D. M., & Ehlers, A. (2003). Posttraumatic stress disorder following assault: the role of cognitive processing, trauma memory, and appraisals. Journal of Consulting and Clinical Psychology, 71, 419e431. Harvey, A. G., Bryant, R. A., & Tarrier, N. (2003). Cognitive behaviour therapy for posttraumatic stress disorder. Clinical Psychology Review, 23, 501e522. Henry, J. D., & Crawford, J. R. (2005). The short-form version of the depression anxiety stress scales (DASS-21): construct validity and normative data in a large non-clinical sample. The British Journal of Clinical Psychology, 44, 227e239. Hope, L., Gabbert, F., & Fisher, R. P. (2011). From laboratory to the street: capturing witness memory using the self-administered Interview. Legal and Criminological Psychology, 16, 211e226.

Horowitz, M. J. (1986). Stress response syndromes (2nd ed.). Northvale, NJ: Jason Aronson. Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 159e174. Lovibond, S. H., & Lovibond, P. F. (1995). Manual for the depression anxiety stress scales (2nd ed.). Sydney: Psychology Foundation. Mayou, R. A., Ehlers, A., & Hobbs, M. (2000). Psychological debriefing for road traffic accident victims: three-year follow-up of a randomised controlled trial. British Journal Psychiatry, 176, 589e593. McNally, R. J. (2003). Remembering trauma. Cambridge, Mass: Belknap Press of Harvard University Press. Monds, L., Paterson, H., Kemp, R., & Bryant, R. (2013). Do distress responses to a traumatic film predict susceptibility to the misinformation effect? Journal of Trauma and Disassociation, 14, 562e575. Paterson, H. M., Whittle, K., & Kemp, R. I. (2014). Detrimental effects of postincident debriefing on memory and psychological response. Journal of Police and Criminal Psychology. http://dx.doi.org/10.1007/s11896-014-9141-6. Paz-Alonso, P. M., Goodman, G. S., & Ibabe, I. (2013). Adult eyewitness memory and compliance: effects of post-event misinformation on memory for a negative event. Behavioral Sciences & the Law, 31, 541e558. Pennebaker, J. W., Kiecolt-Glaser, J. K., & Glaser, R. (1988). Disclosure of traumas and immune function: health implications for psychotherapy. Journal of Consulting and Clinical Psychology, 56, 239e245. Rose, S., Bisson, J., & Wessely, S. (2003). A systematic review of single-session psychological interventions (‘debriefing’) following trauma. Psychotherapy and Psychosomatics, 72, 176e184. Schmider, E., Ziegler, M., Danay, E., Beyer, L., & Buhner, M. (2010). Is it really robust? Reinvestigating the robustness of ANOVA against violations of the Normal Distribution Assumption. Methodology European Journal of Research Methods for the Behavioral and Social Sciences, 6, 147e151. Spielberger, C. D. (1984). State-trait anxiety inventory: A comprehensive bibliography. Palo Alto, CA: Consulting Psychologists Press. Tromp, S., Koss, M. P., Figueredo, A. J., & Tharan, M. (1995). Are rape memories different - a comparison of rape, other unpleasant, and pleasant memories among employed women. Journal of Traumatic Stress, 8, 607e627. Weiss, D. S., & Marmar, C. R. (1997). The impact of event ScaleeRevised. In J. P. Wilson, & T. M. Keane (Eds.), Assessing psychological trauma and PTSD (pp. 399e411). New York, NY: Guilford Press; US. Wiemers, U. S., Sauvage, M. M., Schoofs, D., Hamacher-Dang, T. C., & Wolf, O. T. (2013). What we remember from a stressful episode. Psychoneuroendocrinology, 38, 2268e2277.