“Inability to remember an important aspect of the traumatic event(s)” is, according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), a symptom of posttraumatic stress disorder (PTSD) in the cluster of “negative alterations in cognitions and mood” (American Psychiatric Association, 2013). This symptom description is based on the theory that persons with PTSD have difficulties intentionally recollecting memories of traumatic events and that their memories of trauma tend to be fragmented and poorly organized in comparison with persons without PTSD (Ehlers and Clark, 2000). According to this theory, because of peritraumatic dissociation, a highly traumatic event can result in inadequate encoding and integration of trauma into memory, resulting in a memory being difficult to access voluntarily (Brewin et al., 1996; Ehlers and Clark, 2000; van der Kolk and Fisler, 1995). However, this theory has been criticized, and contrasting views have argued that emotionally arousing experiences enhance encoding of memory and that experiences of unpleasant events therefore are remembered better than memories of a routine day (Cahill and McGaugh, 1995; McGaugh, 2013; Talarico et al., 2004). Enhanced encoding of a memory related to a traumatic event would result in the memory of trauma being more accurate and detailed than memories of less traumatic events.
Different measures have been used to study the quality and the accuracy of traumatic memories in persons with PTSD. There are three concepts frequently used when measuring memory accuracy: coherence of traumatic memories, recollection of trauma over time, and remembering details of trauma.
The concept of “coherence” has been widely used and can be explained as “the quality of being logical and consistent” or “the quality of forming a unified whole” (Coherence, 2019). Coherence and organization are used as synonyms across studies. Memory fragmentation, which is the notion that memories are not remembered as an integrated whole but come in pieces with important aspects missing (Rubin et al., 2004), is considered a part of incoherency. Coherence can be tested in different ways, which are further described in Results. The second concept used to investigate trauma memories is “recollection of trauma over time.” In studies using this concept, participants have been followed over time and have been asked the same questions at several time points to be able to study how memories of trauma develop over time and if there is a difference between persons with and without PTSD in this regard. The third concept studies the amount of details of a trauma remembered and compares this amount in persons with and without PTSD. This way of studying memory of trauma is in the present review called “remembering details of trauma.”
Traumatic events described in PTSD are commonly related to violence and crime such as war, murder, or sexual violence. A person's narrative of a traumatic event can therefore be an important tool for crime investigation, especially in cases with scarce evidence material. Furthermore, trauma narratives can be important evidence in asylum cases where the asylum seeker describes having experienced trauma such as torture in the home country (Herlihy and Turner, 2007; Herlihy et al., 2002). PTSD is a relatively common disorder with a lifetime prevalence of around 10% (Jorge, 2015; Kessler et al., 1995; Resnick et al., 1993), and point prevalence among refugees is as high as 30% (Steel et al., 2009). Considering this, and the fact that the accuracy of the traumatic memories in PTSD can be of importance for both authorities carrying out crime investigations and health professionals in the diagnostic process and in treatment, it is highly important whether trauma recollections in persons with PTSD are reliable and in accordance with the actual events.
Two literature reviews concerning the coherence and consistency of traumatic memories in persons with PTSD have previously been published. In a review from 2014, Brewin studied the quality of voluntary memory of traumatic events in acute stress disorder (ASD) and PTSD (Brewin, 2014). Nine articles were included. The review concluded that voluntarily recalled trauma memories in people with PTSD or ASD were disorganized and fragmented compared with healthy controls (HCs). However, in some of the included studies, the PTSD diagnosis was based solely on self-rated questionnaires (Berntsen et al., 2003; Halligan et al., 2003). Furthermore, in the included studies, there were both participants who had traumatic brain injury (TBI) (Jones et al., 2007) and children (Kenardy et al., 2007; Salmond et al., 2011). To have TBI or being a child is a factor that possibly alters trauma memory and could thereby confound the effect of PTSD on memory. In addition, as argued by Rubin et al. (2016), it is problematic to compare PTSD with ASD because of the fact that the diagnostic criteria for ASD, according to DSM-IV (used in the review of Brewin, 2014), include derealization, depersonalization, and amnesia, which can be interpreted as incoherency. Thus, having an incoherent trauma narrative is required in ASD in DSM-IV.
In the second review from 2005, van Giezen et al. included studies regarding consistency of memory from emotionally arousing events, such as sexual abuse, war zone exposure, or disasters (van Giezen et al., 2005). In total, 37 articles were reviewed. The inclusion criteria of the review were broad and included studies where participants did not have PTSD. In the articles studying the relationship between symptoms of PTSD and change in recollection of trauma over time, van Giezen et al. found PTSD symptoms to be significantly associated with inconsistencies in reports of traumatic memories over time. The more PTSD symptoms the participants reported, the more likely they were to amplify their memory of the traumatic event (van Giezen et al., 2005). For example, in one of the studies included in van Giezen's review (Southwick et al., 1997), a group of 59 veterans from the Operation Desert Storm completed a 19-item questionnaire about their combat experiences at 1 month and 2 years after returning from the Gulf War. The participants were asked questions about whether they had experienced certain traumatic events during their time in war, such as witnessing killings or sustaining injury. The results showed that persons who reported more PTSD symptoms 2 years after returning from war were more likely to have changed their answers to the questions about having experienced a traumatic event from “no” at 1 month to “yes” at 2 years compared with participants with less PTSD symptoms. However, similar to Brewin (2014), most studies included in van Giezen's review were based on self-rated PTSD symptoms and not clinically diagnosed PTSD.
Using only self-ratings implies a significant risk of including persons who do not fulfill the criteria for a PTSD diagnosis. Veterans assessed for PTSD have previously been found to overreport their psychopathology across a variety of diagnostical tests (Frueh et al., 2000). This could potentially lead to an overdiagnosing of PTSD where participants appear as having PTSD even if they do not have PTSD. Studies with an overdiagnosing of PTSD could result in an underestimation of the effect of PTSD on memory. In studies using self-ratings for diagnosing PTSD, the PTSD diagnosis is based solely on the report of symptoms. This makes it more vulnerable to overreporting than diagnosing done by a trained clinician using diagnostic interviews. The tendency of overreporting in veterans assessed for PTSD is not fully understood, but one theory is that veterans overreport their symptoms to receive financial compensation as a result of the PTSD diagnosis (Frueh et al., 2000). On the other hand, by using only self-ratings for diagnosing PTSD, there is a risk that participants fulfilling the criteria for PTSD instead have other psychiatric or neurological disorders because there is a certain overlap of symptoms between diagnoses. Thus, as self-rating of symptoms is a factor that may have influenced the results of the abovementioned reviews, this review will only include articles where the PTSD diagnosis of participants included is clinically verified.
Hence, the aim of the present review was to investigate the accuracy of traumatic memories in persons older than 15 years with a clinically verified diagnosis of PTSD, which, to the best of our knowledge, has never has been reviewed before.
The databases PubMed, PsycINFO, and Embase were systematically searched by the first author for publications on memory for trauma in persons with PTSD. A combination of MeSH/thesaurus terms and free text words was used. Because of the different features of the databases, the exact phrasing of the words differed slightly.
In PubMed, the following search was performed: (((“Stress Disorders, Post-Traumatic”[Mesh])) OR ((“posttraumatic stress disorder”[Text Word] OR “PTSD”[Text Word]))) AND (((((“memory”[Text Word]) OR “memory disorders”[Text Word])) OR ((((“Memory”[Mesh] OR “Memory disorders”[Mesh]))) OR ((memory[Title/Abstract] AND accuracy[Title/Abstract]))))). In PsycINFO, the following search was performed: MJ Stress Disorders, Post-Traumatic OR TX posttraumatic stress OR TX PTSD AND MJ Memory disorders OR MJ memory OR TX memory disorder*. In Embase, the following search was performed: (PTSD.mp. or posttraumatic stress disorder/) and (memory/or memory disorder/) (MAP-term).
The last search was performed the October 7, 2017. The search included every publication until the day of the last search. Furthermore, reference lists of studies included were searched manually by the first author, who also reviewed all of the titles and abstracts for relevance. Articles included were written in English. However, studies in other languages were assessed based on the English abstract, and no studies were excluded based on the language criterion. Only studies enrolling patients with a clinically verified diagnosis of PTSD, based on the World Health Organization International Classification of Diseases (ICD) (World Health Organization, 2004) or DSM (American Psychiatric Association, 2013) criteria, were included. All types of studies except reviews were included. Studies in which the trauma was related to a traumatic head injury or involved hospitalization in an intensive care unit department or anesthesia were excluded considering the influence these events could have on memory. In addition, publications studying children under the age of 15 were excluded. We did not exclude studies due to comorbidities as long as participants were diagnosed with PTSD.
The search strategy described above identified a total of 4911 studies (4902 from databases and 9 from reference lists), of which 58 were of possible relevance (Fig. 1). After reading through the 58 articles, 47 were deemed irrelevant and excluded because the PTSD diagnosis was not clinically verified, memories were not from trauma, or it was not clear which of the participants had PTSD and which ones did not. The last author was involved whenever there was any doubt regarding inclusion of articles.
Eleven articles were included in this review (Table 1). All studies except for one (Qin et al., 2003) used DSM-IV for diagnosing. The participants in Qin's study were diagnosed at a veteran center, but the diagnostic manual used is not further specified.
TABLE 1 -
Overview Over Included Articles
|Author and Year
||Type of Study
||Time Since Trauma
|Rubin et al. (2016)
||Trauma-exposed adults from the community n = 60 (PTSD n = 30, non-PTSD n = 30)
||• Beck Depression Inventory (BDI-II)
• Clinical Administrated PTSD Scale (CAPS)
• The Dissociative Experience Scale (DES)
• Hollingshead Index of Socioeconomic Status (Hollingshead)
• PTSD symptom severity (PCL)
• Global Coherence Measures
||No differences in coherence in narrative from trauma were found between the PTSD and non-PTSD-group.
|McKinnon et al. (2015)
||Trauma-exposed passengers from Flight AT236 n = 15 (PTSD n = 6, non-PTSD n = 7), non–trauma-exposed adults from the community n = 15
||• Beck Anxiety Inventory
• Impact of Event Scale–Revised (IES-R)
• NEO-Five Factor Inventory
• Diagnostic and Statistical Manual of Mental Disorders (SCID)
• Autobiographical interview (AI)
|For trauma exposed with PTSD mean = 43.04, trauma exposed non-PTSD mean = 53.26 mos
||Details recalled from a traumatic event for the trauma-exposed group or a highly negative event for the non–trauma-exposed group
||For the trauma-exposed group, there was no difference in accuracy (amount of verifiable details recalled from trauma) between persons with or without PTSD.
|Römisch et al. (2014)
||Trauma-exposed women with PTSD n = 14, non–trauma-exposed women from the community non-PTSD n = 14
||• Oral narrative interview of most distressing, most angering, and happiest event coded for evaluation, immersion and fragmentation
• Posttraumatic Diagnostic Scale (PDS)
• Sense of Coherence Scale (SOC)
|Distress memories mean = 78.64 mos in the control group and mean = 66.57 mos in the trauma group
||No differences in fragmentation of the most distressing event were found between the PTSD and the non-PTSD group.
||Trauma-exposed undergraduates n = 30 (PTSD n = 15, non-PTSD n = 15)
• The Centrality of Event Scale (CES)
• Written narrative
• Global coherency
||Memories from trauma were as coherent in the PTSD group as in the non-PTSD group.
|Rubin et al. (2011)
||Trauma-exposed adults from the community n = 117 (PTSD n = 75, non-PTSD n = 42)
• Alcohol Use Disorders Identification Test (AUDIT)
• Davidson Trauma Scale (DTS)
• Interpersonal Support Evaluation List (ISEL)
• NEO Personality Inventory (NEO)
• Positive and Negative Affect Schedule (PANAS)
• PTSD Check List (PCL)
• Traumatic Life Events Questionnaire (TLEQ)
||No difference in self-rated coherence for the most stressful event between the PTSD and non-PTSD group.
|David et al. (2010)
||Trauma-exposed population sample
Study 1: n = 53, Study 2: n = 27, Study 3: n = 36
|• The Peritraumatic Distress Inventory (PDI)
• The Peritraumatic Dissociation Experience Questionnaire (PDEQ)
• Mini International NeuropsychiatricInterview (MINI)
|Study 1: 5 days, 1 mo, and 6 mos after trauma.
Study 2: On average, 3 and 9 mos after trauma.
Study 3: 1 and 12 mos after trauma
|Consistency of retrospective report of peritraumatic emotional experience over time
||Consistency of memory over time was better in persons who did not develop PTSD or who remitted from PTSD over time, compared with those with PTSD who did not remit.
|Jelinek et al. (2009)
||Trauma-exposed adults from the community n = 81 (PTSD n = 26, non-PTSD n = 55)
||• Sociodemographic interview
• Hamilton Depression Rating Scale
• Narrative interview coded for disorganization
• Global rating of disorganization
• Narrative Memory Test (NMT)
• The Disorganization Scale of the Traumatic Memory Questionnaire
• The Test d2
• Mehrfachwahl-Wortschatz Intelligenztest
|PTSD group: mean = 20.93 mos
Non-PTSD: mean = 23.12 mos
||Trauma memories of persons with PTSD were more disorganized compared with memories in the non-PTSD group.
|Wessa et al. (2006)
||Trauma-exposed survivors from the air show disaster in Ramstein n = 31 (with PTSD n = 16, non-PTSD n = 15), non–trauma-exposed population n = 16
• The State Trait Anxiety-Inventory (STAI-T)
• Modified S1–S2-paradigm Wechsler Memory Scale-Revised (WMS-R)
• 500 Hz electroencephalogram
• 200 Hz electrocardiogram
||Declarative memory of trauma
||No difference in trauma-specific declarative memory in trauma-exposed persons with PTSD compared with trauma-exposed persons without PTSD.
|Rubin et al. (2004)
||Trauma exposed veterans with PTSD n = 50
• Davidson Trauma Scale (DTS)
• The Mississippi PTSD scale
||Correlation between PTSD symptoms and fragmentation
||No correlation between PTSD symptoms and incoherency of trauma memories.
|Yovell et al. (2003)
||Trauma-exposed patients from the emergency department n = 6 (PTSD n = 2, non-PTSD n = 4)
• Semistructured interview
|7, 30, and 120 days after trauma
||Recollection of trauma over time
||All participants had brief, stable, and persistent memory gaps from the day of trauma. Furthermore, participants with PTSD also developed longer, progressive, and unstable memory gaps.
|Qin et al. (2003)
||Trauma-exposed clients, staff, and visitors from a veterans hospital n = 25 (PTSD n = 12, non-PTSD n = 13)
||Questionnaires that were modifications of those developed by the 9/11 Memory Consortium.
||Approximately 1 and 10 mos after the attacks
||Consistency of autobiographical and event memory for the September 11, 2011, attacks over time
||No difference in autobiographical memory and event memory for factual details of the attacks between the PTSD and the non-PTSD group.
The studies had different methods of investigating the memory of trauma. The results are presented in accordance with the three concepts described in the Introduction: coherence (six studies), recollection of trauma over time (three studies), and remembering details of trauma (two studies). These three concepts were chosen post hoc (during the data collection) as they were used in several of the included articles to describe and compare the results across the articles.
Six studies investigating coherence were identified. Different methods for studying coherence were used in the articles and are described further below.
A cross-sectional study by Rubin et al. (2004) examined the correlation between PTSD symptoms and coherency and fragmentation of memories. The study population consisted of 50 veterans diagnosed with PTSD who completed the Autobiographical Memory Questionnaire (AMQ) for a preservice, a noncombat, a combat, and an intrusive memory. AMQ is a self-rated questionnaire consisting of questions about a memory that participants rate on a 7-point scale (1 = not at all and 7 = completely). The time since trauma at the time of study was unknown. Among other questions, participants were asked two questions related to coherency and fragmentation. The first question was “whether the memory was coherent or whether it came in isolated facts or observations.” Fragmentation was tested by asking “whether the memory came in pieces with parts missing.” Results showed no correlation between self-rated coherency or fragmentation of traumatic memories and severity of PTSD symptoms.
A case-control study by Rubin et al. (2011) also used the AMQ to test for coherency. The participants were 117 adults from a community sample, 75 with and 42 without PTSD, who were all asked to complete the AMQ for the seven most important, three most stressful, and three most positive autobiographical memories. Questions concerning the narratives were a) story (“It comes to me in words or in pictures as a coherent story”) and b) pieces (“My memory comes to me in pieces with missing bits”). All participants in the PTSD group had experienced a trauma that fulfilled the A-criteria according to DSM-IV compared with only 73% to 83% (depending on the test) in the control group. Time since trauma was unknown. For the most distressing memories, there was no significant difference between the PTSD and non-PTSD groups on question 1. However, concerning question 2, a significant higher number of participants in the PTSD group reported that the memory came to them in pieces (fragmentation). The authors of the article argue that the fragmentation found in the PTSD group may be related to other aspects such as alcohol abuse and lack of social support rather than PTSD severity, and the overall conclusion of the article is that there was no significant difference in coherency between the PTSD and the non-PTSD group.
Römisch et al. (2014) compared narratives of 14 women who have PTSD with 14 women not fulfilling the criteria for PTSD. The women with PTSD were recruited through helplines and clinics on the basis of a current PTSD diagnosis and had all experienced a trauma according to the A1 or A2 criterion of DSM-IV (American Psychiatric Association, 1995). The control group was recruited through word of mouth. All women were interviewed about their most distressing, most angering, and happiest event, and the narratives were transcribed and coded for fragmentation. The mean time since the distressing event was 79 months in the PTSD group and 67 months in the control group. Coding was done by dividing narratives into chunks and sorting each chunk into categories depending on the content. Categories representing incoherency were unfinished utterances, repetitions, and filled and unfilled pauses. The number of categories/chunks representing incoherency in the narrative was thereafter compared between participants with and without PTSD. The results showed no correlation between PTSD and incoherency of the narratives.
A similar study (Jelinek et al., 2009) compared trauma narratives of trauma survivors with (n = 26) and without (n = 55) PTSD. Participants were all victims of a single trauma, meeting the A-trauma criterion according to DSM-V. Mean time since trauma was 21 months in the PTSD group and 32 months in the non-PTSD group. Trauma narratives were recorded, transcribed, and scored by a rater blinded to diagnostic status. Similar to Römisch et al. (2014), ratings were done by dividing narratives into chunks and sorting these into four different categories: repetitions, disorganized thoughts, organized thoughts, and not coded. Calculations were made, and each narrative was then given a “total disorganizing score” based on the amount for each category. Furthermore, each narrative was given a “global disorganization score,” which is a method previously used by Halligan et al. (2003), where the rater grades the experience of disorganization on a scale from 0 (not at all disorganized) to 10 (extremely disorganized). In addition, participants were asked to rate their own sense of memory disorganization by completing the Disorganization Scale of the Traumatic Memory Questionnaire. For each of the rating scales, the authors aimed to investigate a) whether there was a significant group (PTSD versus non-PTSD) × event (traumatic event versus unpleasant event) interaction for memory disorganization and b) whether each type of memories (traumatic and unpleasant) was disorganized for each of the two groups (PTSD and non-PTSD). For the first aim, a significant interaction was found for the total disorganization score. For the second aim, a significant correlation between PTSD and disorganization of trauma memories was found only for the global disorganization score. Despite these mixed findings, the authors overall concluded that memory from trauma was more disorganized in persons with PTSD than in the non-PTSD group.
In another study by Rubin (2011), written narratives were used to study the coherence of memories. The participants were undergraduate students from Duke University who had all experienced a trauma. The study included 15 students with PTSD and 15 students without PTSD. All the participants had experienced a trauma that met the A-trauma criterion according to DSM-IV, but it was unknown how much time had passed since the trauma. Participants wrote down narratives about the trauma as well as the most important and the happiest event experienced. Both self-ratings and observer ratings were used as measures of coherence. The raters scored the written narratives for comprehension and disorganization, and the Narrative Coherence Coding Scheme (NACCS) was used to provide a standard measure of the narrative coherence. Comprehension was tested by questions that the raters scored on a 7-point scale. An example of a question of comprehension was “Do you understand what happened, that is, are the ideas that the author chose to describe clear?” Disorganization was initially tested in the same way that Jelinek et al. (2009) tested for global disorganization, by asking the raters about the experience of disorganization on a 10-point scale. However, it was found that this method was difficult for the raters to use, and the question was therefore changed to “How much of the text is disorganized—that is, how much of the writing does not add to the development of the narrative or the understanding of the narrator?” This question was rated on a 7-point scale. The three components measured in NACCS were coherence, temporal ordering, and theme. In NACCS, “coherence” describes whether the narrator provides enough information to place the event in time and space in relation to other events, whereas “temporal ordering” is a measure of the narrator's ability to provide enough information to put the individual parts of the particular event in order on a time line. As a third component, “theme” describes the narrator's ability to interpret and connect the event with other autobiographical experiences and the narrator's self-perception. Finally, the self-ratings consisted of the AMQ for each narrative. Of the six measures used, only the rater-graded disorganization measure showed a significant correlation between PTSD and disorganized memories. After comparing all the results, the authors concluded that there was no general tendency for participants with PTSD to produce less coherent narratives and that memories from a traumatic event were not more incoherent than memories from a positive or important event.
Rubin et al. (2016) aimed to explore the coherence of trauma memories in individuals with (n = 30) and without PTSD (n = 30). The participants were adults from the community and had all experienced a trauma meeting the A-trauma criterion in DSM-V. Time since trauma was unknown. The participants orally narrated their three most negative, stressful, or traumatic life events followed by their three most positive, and finally their three most important life events. Narratives were transcribed, and coherence was rated using several different methods including self-rating, observer rating, and computer scoring. The tools that were used were the AMQ, NACCS, and global coherence measures, as well as the Coh-Metrix and Linguistic Inquiry Word Count (LIWC). Coh-Metrix is a computer program, developed by one of the authors of the article (McNamara et al., 2010), that analyzes coherence and cohesion metrics in a written text. LIWC is also a computer program that analyzes written text and calculates the percentage of different types of words used. The results of a multivariate analysis showed no correlation between PTSD and incoherent memories of trauma.
Change in Recollection of Trauma Over Time
Three articles studied change in recollection of trauma over time.
One article (David et al., 2010) consisted of three different studies of trauma-affected individuals with and without PTSD. All participants met the A-trauma criterion from DSM-IV. The participants were asked about recollection of their cognitive, emotional, and autonomic reactions in connection with trauma, and the aim of the study was to assess the consistency of these retrospective reports of peritraumatic emotional reactions over time in persons with and without PTSD. In the first study, 53 persons were interviewed at 5 days, 1 month, and 6 months after trauma. The second study included 27 persons who were interviewed at 3 and 9 months after trauma, and the third study included 36 persons interviewed at 1 and 12 months after trauma. Furthermore, participants were assessed for PTSD at each interview. The results showed that people who developed PTSD after trauma were less consistent in their retrospective reports compared with those who did not develop PTSD or those who recovered from PTSD over time (did not fulfill the criteria for PTSD at follow-up).
Yovell et al. (2003) studied a group of six single trauma survivors for recollection of trauma at 7, 30, and 120 days after trauma. Participants were recruited from an emergency department but were excluded if they needed surgery, had a head trauma, had loss of consciousness, or had been intoxicated with alcohol/drugs. All participants had experienced a trauma that met the A-trauma criteria of DSM-IV. Two of the six trauma survivors developed PTSD over time, making it possible to investigate the effect PTSD has on memory of trauma. At each interview, participants were asked to recall the day of the traumatic event in detail. The interview included both an open-ended phase and a structured probing phase. In the open part of the interview, participants freely narrated the day of the event in detail, and in the probing part, details from the emergency department records and earlier interviews were used as cues to maximize recollection of the day. All participants in the study, regardless of diagnosis of PTSD, reported brief memory gaps surrounding the time of trauma. These memory gaps lasted from a few seconds to a few minutes and were not possible to recall in the cued phase of the interview. Furthermore, these memory gaps were persistent and were not recalled in any of the interviews. In the two subjects who developed PTSD, an additional and potentially reversible extension of the memory gaps was found. Differently from the memory gaps seen in all participants, memory gaps in participants who developed PTSD increased in length over time. In addition, these memory gaps were more unstable and could at times be remembered by hints and forced-choice questions.
In a case-control study (Qin et al., 2003), 12 persons with PTSD and 13 controls without PTSD completed questionnaires about the September 11, 2001, terrorist attacks in America approximately 1 and 10 months after the attacks. The participants were clients, staff, or visitors of the Veterans Administration Connecticut Healthcare System and had all experienced a trauma before the attacks. The persons in the PTSD group all reported that they had been in a life-threatening situation and had all been diagnosed with PTSD and treated in a health care facility for veterans. The author of the study did not specify whether participants had a first-hand experience of the terrorist attack, but from the description of the method, it does not seem to be the case. Among other things, participants were asked questions about how they first heard about the attacks (autographical memory), about factual details of the attacks (event memory), and background knowledge related to the attacks. Results showed no significant difference in autobiographical memory, event memory, or background knowledge about the attacks at follow-up between the groups.
Remembering Details of Trauma
Two studies looked at remembering details of trauma.
McKinnon et al. (2015) examined the recollection of a highly traumatic event. Participants were 15 passengers from Flight AT236, a transatlantic flight that nearly crashed at sea. Six passengers developed PTSD and seven did not, and for two passengers, the diagnosis status was unknown. The group of passengers was matched with 15 HCs who were not passengers of Flight AT236. For passengers with PTSD, the mean time since the trauma at the point of interview was 43 months, and for passengers without PTSD, the mean time was 53 months. Participants underwent the Autobiographical Interview, recalling three different events: the airline disaster (or a highly negative event for HCs), the September 11, 2001, terror attacks, and a nonemotional event. The passengers from Flight AT236 recalled a greater amount of internal (episodic) details of the trauma memory compared with the HCs, regardless of whether they have PTSD or not. In addition, passengers with PTSD generated a larger number of external (nonepisodic) details in their memories from the three different events, compared with the passengers without PTSD. Importantly, there was no significant difference in accuracy (number of verifiable details recalled from the air plane trauma) between the PTSD and the non-PTSD passengers.
Wessa et al. (2006) studied a group of people having experienced the air show disaster in Ramstein in 1988: 16 who were diagnosed with PTSD and 15 who had never had PTSD. The study also included 16 HCs who never experienced a trauma. The time between the trauma and the time of study interview was not mentioned in the article. Participants were asked questions about their specific trauma (the air show disaster) and questions about other disastrous and neutral, non–trauma-related events. Trauma-specific questions were, for example, “Which air force display team performed during the air show in Ramstein in 1988?” Results showed no difference between the PTSD and non-PTSD groups in the amount of correct answers to trauma-specific, other trauma-related, or neutral questions.
To the best of our knowledge, this is the first systematic review examining memory of trauma in people with clinically diagnosed PTSD.
Eleven articles were included in the review. Three of these studies found a deficiency in memory of trauma (David et al., 2010; Jelinek et al., 2009; Yovell et al., 2003)—one in the coherency of traumatic memories and two in change of recollection of trauma over time. Eight studies concluded that there was no difference in memory between participants with or without PTSD (McKinnon et al., 2015; Qin et al., 2003; Rubin, 2011; Rubin et al., 2004, 2011, 2016; Römisch et al., 2014; Wessa et al., 2006)
Among the six studies investigating the coherency/organization of traumatic events (Jelinek et al., 2009; Rubin, 2011; Rubin et al., 2004, 2011, 2016; Römisch et al., 2014), only one concluded that traumatic memories in people with PTSD were more incoherent than in people without PTSD (Jelinek et al., 2009). Some of the other studies found significant differences in traumatic memories according to some of the measures used, but concluded, after comparing different measures, that there was no overall difference (Rubin, 2011; Rubin et al., 2011, 2016). There seems to be only minor differences between the trauma types included in the studies. All studies except for Rubin et al. (2004), which only studied veterans, included participants with different types of trauma. The trauma were heterogeneous, and examples of these were combat, assault, accidents, or death a loved one. Only two of the studies (Jelinek et al., 2009; Römisch et al., 2014) state the time since trauma, which makes comparison across studies challenging. What seems to constitute the largest difference between the studies are the ratings used to test for coherency, which range from self-ratings to observer ratings and computer-based rating.
Jelinek et al. (2009) concluded, after calculating across all relevant measures, that trauma memories in persons with PTSD were more disorganized than in persons without PTSD. Global rating of disorganization, which is one of the ratings that shows significant results in the study by Jelinek et al., is a subjective measure, where a rater determines to what degree the narrative is considered disorganized (on a scale from 0 [not at all disorganized] to 10 [extremely disorganized]). Two other studies used similar ratings (Rubin, 2011; Rubin et al., 2016). Rubin et al. (2016) used the same method as that of Jelinek et al. and found no difference in disorganization between the PTSD and non-PTSD groups. In the other study by Rubin (2011), disorganization was initially tested with the above mentioned method, but because of difficulties for the raters in using this method, it was changed to the question: How much of the text is disorganized on a scale from 1 (none) to 7 (almost all)? Rubin (2011) found, similarly to Jelinek et al., a correlation between PTSD and disorganized memories for trauma according to this measure.
The global rating of disorganization has not been validated across studies. In the study by Jelinek et al. (2009), a satisfactory interrater reliability (intraclass correlation coefficients, >0.84) was found. However, it is possible that the reliability of this rating is low across studies, which could explain the difference in results. This could be supported by the fact that Rubin et al. (2016) found it difficult to use the rating they initially chose.
In summary, most of the available research could not identify a significant difference in coherence between people with or without PTSD. Four of the larger studies found small differences in either coherence or fragmentation on some of the measures included, but only one of the studies found a significant difference between participants with and without PTSD. Although there may well be interpersonal differences, the present review does not provide support for the theory that memory of trauma in general is less coherent than other memories.
Change in Recollection of Trauma Over Time
Of the three studies examining the change in recollection of trauma over time (David et al., 2010; Qin et al., 2003; Yovell et al., 2003), two studies found an association with PTSD (David et al., 2010; Yovell et al., 2003).
The included studies differ on some parameters. The amount of time passed since trauma was similar, with maximum 1 year between trauma and follow-up. Despite all studies being relatively small, there was a fairly large difference in number of participants across studies ranging from 6 participants (Yovell et al., 2003) to 116 participants (David et al., 2010). However, the article by David et al. consisted of three different study-populations, with the largest group consisting of 53 participants. The type of trauma also differed in the three studies. In the studies by Yovell et al. and David et al., the participants all had personal trauma experiences that differed in nature. In the study by Qin et al. (2003), however, the trauma investigated was the same for all participants (the 9/11 terrorist attack), but it was not the index trauma for the participants, who presumably did not experience the attack in person. A possible reason for the fact that Qin et al. did not find a link between change in recollection and PTSD could therefore be that it was the memory of 9/11 and not of the index trauma that was examined in this study.
Furthermore, the questions participants were asked regarding trauma varied. In study by Yovell et al., the participants were asked to describe the day of the trauma in detail, whereas David et al. only investigated the consistency of the memory of peritraumatic distress, the cognitive, emotional, and autonomic reactions experienced at the time of the trauma. The participants in the study by Qin et al. were asked about both autobiographical memory and memory of factual details about the trauma.
Yovell et al. found memory gaps in all the participants. Interestingly, in this study, the participants were asked what precisely happened during the day of the trauma, and short memory gaps were found in all the participants. It is possible that these short memory gaps are not found in other studies because other studies do not ask for as many details about the traumatic event. However, it is important to note that this study was very small with only six participants.
In summary, the three studies differed widely in size and method. Two studies found an association between PTSD and change in recollection over time, although the material overall is quite limited in size.
Remembering Details of Trauma
The two studies investigating the remembering of details of trauma (McKinnon et al., 2015; Wessa et al., 2006) both found that participants with PTSD remembered as many details about the trauma as participants without PTSD. Both studies were small and investigated the memory of a single trauma. The mean time since trauma was 43 months in the study by Wessa et al., whereas it was unknown in the study by McKinnon et al. However, in the study by McKinnon et al., the trauma happened in 1988, and the study was published in 2006, so presumably the time since trauma was considerably longer in this study compared with that of Wessa et al.
What is particularly interesting about these studies is that they examine the memory of factual details about trauma and find that there is no difference between people with or without PTSD. This is an important finding considering accuracy of traumatic memories. Assuming that there is a large difference in time since trauma in the two studies, it furthermore indicates that the memory of trauma stays intact over time in persons with PTSD. In both studies, it had, however, only been possible to recruit a small group of people. None of the articles inform about the total amount of persons affected by the event or specify any characteristics of the group that did not participate in the studies. Nonetheless, it can be assumed that the population studied is only a small fraction of the total amount of people who had experienced the trauma. It is therefore a possibility that persons with more PTSD symptoms and greater memory loss from trauma were unable to participate in the studies. In that case, the studies could be biased in a false-negative direction. Furthermore, because both studies only investigated a single trauma, it is not possible to transfer these results to a population with multiple traumas. Nonetheless, the present review found no difference in detail remembered about trauma between participants with and without PTSD.
Comparison With Findings From Previous Reviews
Similarly to the review from 2005 (van Giezen et al., 2005), the comparison of results of the reviewed articles suggested that persons with PTSD had a more unstable memory of trauma over time. However, in the studies reviewed by van Giezen et al., it was found that persons with more PTSD symptoms had more amplified memories over time. This is different to the findings of the present review, where two of the included studies found that memories of trauma were unstable, but no general tendency to amplify memories over time were found (David et al., 2010; Yovell et al., 2003). It has been argued that amplification of memories could be due to the general overreporting of psychopathology seen in persons assessed for PTSD as described in the Introduction (Frueh et al., 2000). Interestingly, this present review finds that the changes of memory over time are due not only to amplification but also to a more general inconsistency over time, suggesting that memories of trauma in persons with PTSD change over time and that it is not only due to overreporting. However, it is important to note that only three relatively small articles on recollection of trauma over time were included in the present review, which is less than that included by Van Giezen due to differences in inclusion criteria.
Opposite to Brewin's (2014) review, we found no consistent correlation between incoherency of trauma memory and PTSD. This difference in results could be because Brewin included studies where the PTSD diagnosis was based on self-ratings and thereby might have included participants with memory difficulties rooted in different pathologies. Unlike previous reviews, we did not include patients with ASD, which also might have affected the result. Furthermore, the two studies looking at remembering details of trauma, a concept that has not been studied in previous reviews, did not find any association between PTSD and altered memory of trauma.
Strengths and Limitations of the Included Studies
As discussed above, the studies included in the present review have a broad range of limitations. Especially, small sample sizes and lack of information about the time passed since the trauma may influence the conclusion that can be drawn based on the present data. Another limitation is that several of the studies investigated the effect of a single trauma (McKinnon et al., 2015; Wessa et al., 2006; Yovell et al., 2003). This will presumably make it difficult to generalize the results to patients with PTSD who have experienced multiple or prolonged trauma such as war or imprisonment. A wide range of methods were used to study the memories of trauma, which made it difficult to compare the studies, and only little information was provided about the validity of the included measures.
Strengths and Limitations of the Present Review
A systematic search of three of the major databases was undertaken, and references were examined to find additional studies. However, we did not try to identify unpublished studies or study protocols. Furthermore, only the first author performed the screening of articles, although, when in doubt, studies were conferred with the last author. Only studies in English were included. Finally, it is a limitation to this review that no systematic tool for quality appraisal of the included studies was used.
This systematic review aimed to examine the effects of PTSD on accuracy of traumatic memories. The only area in which we found a consistent difference between the PTSD and non-PTSD groups was change in recollection of trauma over time. Participants with PTSD changed their recollection of trauma over time to a higher extent than participants without PTSD. As for the other two concepts studied, coherence of trauma memory and details remembered of trauma, most studies found no difference between the PTSD and non-PTSD groups. Although included articles to some extent pointed in different directions, most studies concluded that memories of trauma are as accurate in people with PTSD as in people without PTSD. This is an important finding for people working with persons with PTSD, although results should be interpreted with caution due to small sample sizes and methodological inconsistencies. Interpersonal variance should also be acknowledged. Furthermore, it is questionable whether these results can be generalized to survivors of multiple or prolonged trauma, as repetition of, for example, torture or sexual assaults could make it difficult to tell events apart. Future research should therefore include the abovementioned groups, for example, by comparing survivors of single or multiple traumas. Attention is furthermore warranted as to how to measure the accuracy of traumatic memories. Despite obvious difficulties in checking for the truth of events that happened in the past, the use of standardized and well-validated measures would contribute to securing the quality of findings. Important for future research are prospective studies that include participants immediately after their trauma and follow them for a longer period to see how and to what extent their memory of trauma changes over time. Furthermore, future studies should ideally include HCs both with and without traumatic experiences.
The authors declare no conflict of interest.
American Psychiatric Association (1995) Diagnostic and statistical manual of mental disorders: DSM-IV: International version with ICD-10 Codes
(4th ed). Washington, DC: American Psychiatric Association.
American Psychiatric Association (2013) Citing the DSM V
. Diagnostic and statistical manual of mental disorders
(5th ed). Washington, DC: American Psychiatric Association. doi:10.1176/appi.books.9780890425596.744053.
Berntsen D, Willert M, Rubin DC (2003) Splintered memories or vivid landmarks? Qualities and organization of traumatic memories with and without PTSD. Appl Cogn Psychol
Brewin CR (2014) Episodic memory
, perceptual memory
, and their interaction: Foundations for a theory of posttraumatic stress disorder. Psychol Bull
Brewin CR, Dalgleish T, Joseph S (1996) A dual representation theory of posttraumatic stress disorder. Psychol Rev
Cahill L, McGaugh JL (1995) A novel demonstration of enhanced memory
associated with emotional arousal. Conscious Cogn
Coherence (2019) In Oxford Online Dictionary. Available at: https://www.lexico.com/en/definition/coherence
. Accessed January 15, 2019.
David AC, Akerib V, Gaston L, Brunet A (2010) Consistency of retrospective reports of peritraumatic responses and their relation to PTSD diagnostic status. J Trauma Stress
Ehlers A, Clark DM (2000) A cognitive model of posttraumatic stress disorder. Behav Res Ther
Frueh BC, Hamner MB, Cahill SP, Gold PB, Hamlin KL (2000) Apparent symptom overreporting in combat veterans evaluated for PTSD. Clin Psychol Rev
Halligan SL, Michael T, Clark DM, Ehlers A (2003) Posttraumatic stress disorder following assault: The role of cognitive processing, trauma memory
, and appraisals. J Consult Clin Psychol
Herlihy J, Scragg P, Turner S (2002) Discrepancies in autobiographical memories—Implications for the assessment of asylum seekers: Repeated interviews study. BMJ
Herlihy J, Turner SW (2007) Asylum claims and memory
of trauma: Sharing our knowledge. Br J Psychiatry
Jelinek L, Randjbar S, Seifert D, Kellner M, Moritz S (2009) The organization of autobiographical and nonautobiographical memory
in posttraumatic stress disorder (PTSD). J Abnorm Psychol
Jones C, Harvey AG, Brewin CR (2007) The organisation and content of trauma memories in survivors of road traffic accidents. Behav Res Ther
Jorge RE (2015) Posttraumatic stress disorder. Continuum (Minneap Minn)
Kenardy J, Smith A, Spence SH, Lilley PR, Newcombe P, Dob R, Robinson S (2007) Dissociation in children's trauma narratives: An exploratory investigation. J Anxiety Disord
Kessler RC, Sonnega A, Bromet E, Hughes M, Nelson CB (1995) Posttraumatic stress disorder in the National Comorbidity Survey. Arch Gen Psychiatry
McGaugh JL (2013) Making lasting memories: Remembering the significant. Proc Natl Acad Sci U S A
McKinnon MC, Palombo DJ, Nazarov A, Kumar N, Khuu W, Levine B (2015) Threat of death and autobiographical memory
: A study of passengers from Flight AT236. Clin Psychol Sci
McNamara DS, Louwerse MM, McCarthy PM, Graesser AC (2010) Coh-Metrix: Capturing linguistic features of cohesion. Discourse Process
Qin JJ, Mitchell KJ, Johnson MK, Krystal JH, Southwick SM, Rasmusson AM, Allen ES (2003) Reactions to and memories for the September 11, 2001 terrorist attacks in adults with posttraumatic stress disorder. Appl Cogn Psychol
Resnick HS, Kilpatrick DG, Dansky BS, Saunders BE, Best CL (1993) Prevalence of civilian trauma and posttraumatic stress disorder in a representative national sample of women. J Consult Clin Psychol
Römisch S, Leban E, Habermas T, Döll-Hentschker S (2014) Evaluation, immersion, and fragmentation in emotion narratives from traumatized and nontraumatized women. Psychol Trauma Theory Res Pract Policy
Rubin DC (2011) The coherence of memories for trauma: Evidence from posttraumatic stress disorder. Conscious Cogn
Rubin DC, Deffler SA, Ogle CM, Dowell NM, Graesser AC, Beckham JC (2016) Participant, rater, and computer measures of coherence in posttraumatic stress disorder. J Abnorm Psychol
Rubin DC, Dennis MF, Beckham JC (2011) Autobiographical memory
for stressful events: The role of autobiographical memory
in posttraumatic stress disorder. Conscious Cogn
Rubin DC, Feldman ME, Beckham JC (2004) Reliving, emotions, and fragmentation in the autobiographical memories of veterans diagnosed with PTSD. Appl Cogn Psychol
Salmond CH, Meiser-Stedman R, Glucksman E, Thompson P, Dalgleish T, Smith P (2011) The nature of trauma memories in acute stress disorder in children and adolescents. J Child Psychol Psychiatry Allied Discip
Southwick SM, Morgan CA 3rd, Nicolaou AL, Charney DS (1997) Consistency of memory
for combat-related traumatic events in veterans of operation desert storm. Am J Psychiatry
Steel Z, Chey T, Silove D, Marnane C, Bryant RA, Van Ommeren M (2009) Association of torture and other potentially traumatic events with mental health outcomes among populations exposed to mass conflict and displacement: A systematic review and meta-analysis. JAMA
Talarico JM, LaBar KS, Rubin DC (2004) Emotional intensity predicts autobiographical memory
experience. Mem Cognit
van der Kolk BA, Fisler R (1995) Dissociation and the fragmentary nature of traumatic memories: Overview and exploratory study. J Trauma Stress
van Giezen AE, Arensman E, Spinhoven P, Wolters G (2005) Consistency of memory
for emotionally arousing events: A review of prospective and experimental studies. Clin Psychol Rev
Wessa M, Jatzko A, Flor H (2006) Retrieval and emotional processing of traumatic memories in posttraumatic stress disorder: Peripheral and central correlates. Neuropsychologia
World Health Organization (2004) ICD-10 international statistical classification of diseases and related health problems
(Vol. 2, p 95). Geneva: World Health Organization.
Yovell Y, Bannett Y, Shalev AY (2003) Amnesia for traumatic events among recent survivors: A pilot study. CNS Spectr