Posttraumatic Amnesia and Recall of a Traumatic Event Following Traumatic Brain Injury

There has been a growing debate in the neuropsychiatric literature on the question of whether patients who sustain a traumatic brain injury (TBI) can develop symptoms of acute stress disorder (ASD) and posttraumatic stress disorder (PTSD). The controversy revolves around the issue of posttraumatic amnesia and the ability to reexperience the traumatic event. ASD and PTSD are two of the few diagnoses within DSM-IV¹ in which etiological assumptions are made; that is, a particular stressor is implicated as the causative event. If, however, because of a period of posttraumatic amnesia, patients cannot remember how they were injured or events surrounding the injury, their ability to reexperience the stressor as it occurred must remain questionable.

Despite this seeming paradox, numerous case reports have documented PTSD following mild to severe traumatic brain injury.^2–9 Furthermore, a series of studies limited to patients with mild TBI have consistently documented the presence of ASD and PTSD in a subset of patients. Thus, Bryant and Harvey¹⁰ noted that in the acute phase of recovery following a motor vehicle accident, symptoms such as fear and intrusive recollections of the accident were present in subjects with and without a head injury, albeit more commonly in the latter. Their subsequent studies reported that 14% of subjects with a mild TBI following a road traffic accident developed ASD and a further 4% to 5% were diagnosed with subsyndromal ASD; depressive, dissociative, and avoidant symptoms were predictive of ASD development and severity.^11,12 A follow-up of their patients 6 months later revealed that 24% had gone on to develop PTSD, although not all patients with ASD followed this course, suggesting that although ASD was a risk factor for subsequent PTSD, other modifying factors also played a role.¹³ Bryant and Harvey¹⁴ also compared accident survivors with and without mild TBI at 1 and 6 months post injury and found comparable rates of ASD and PTSD, respectively. Significantly, although traumatic memories of the accident were less common in the mild TBI group at the early assessment, this was no longer the case 6 months later, underscoring their point that mild TBI may not affect the longer-term PTSD profile.

Data disputing an association between traumatic brain injury and PTSD come from three sources. Warden et al.¹⁵ examined 47 patients with moderately severe TBI and found that although a minority of subjects endorsed the avoidance and autonomic arousal symptoms of PTSD, none reported any reexperiencing. This result was partially replicated by Sbordone and Liter,¹⁶ who compared two groups of subjects, those with PTSD and those with mild TBI. Their methodology differed from other studies' in that a standardized assessment of PTSD was not used. Rather, patients were asked to provide a detailed chronological history of events surrounding their traumatic event. The PTSD group provided emotionally charged recollections of the traumatic event, including nightmares, flashbacks, and intrusive imagery, whereas such symptomatology was absent from the TBI group. Furthermore, the patients with mild TBI did not display any emotional arousal when asked to describe what they remembered of their traumatic event, a finding at odds with Warden et al. Finally, a study of victims of road traffic accidents reached similar conclusions after finding a dearth of PTSD symptoms in a subgroup of patients who had briefly lost consciousness.¹⁷

Although there are adherents of the view that any period of PTA effectively rules out the likelihood of PTSD symptoms, the number of studies attesting to the contrary suggests that further research along these lines is called for. To date, all studies have limited patient selection to traumatic brain injuries of similar severity, i.e., either mild or moderate to severe, with the emphasis on the former given their greater numbers. No study has yet to investigate the prevalence of PTSD symptoms, in particular the reexperiencing phenomena, in a group of TBI patients stratified according to their degree of posttraumatic amnesia. Such an approach offers the opportunity of investigating the extent to which PTA influences symptom development and is the focus of the present report.

METHODS

A consecutive sample of 300 patients attending a traumatic brain injury clinic were enrolled in the study. As part of clinical care at our tertiary trauma center, all patients who have sustained a traumatic brain injury are given a TBI clinic appointment on discharge. At each visit, patients are seen by a neuropsychiatrist and a standardized assessment is completed. As a routine procedure prior to the assessment, patients are asked if their case notes may be used for research purposes, and if they agree, a consent form is signed attesting to this fact. The sample thus represents a consecutive series of referrals to our clinic. Written informed consent was also obtained after the procedures in this study had been explained to the subjects. The study received the approval of the Hospital Research Ethics Committee.

The following information was obtained from each patient in a set order:

First, a clinical coordinator interviewed the patients to record demographic data. It was at this point that the duration of posttraumatic amnesia was determined retrospectively, using the criteria of Russell and Smith.¹⁸ The trauma department's case notes were then scrutinized for details pertaining to the injury, such as the mechanism of injury (motor vehicle, pedestrian, industrial, assault) and indices of injury severity, namely Glasgow Coma Scale (GCS) scores and abnormalities on CT brain scan. For consistency, the GCS score recorded on arrival in the emergency department was chosen.

Thereafter, patients were examined by a neuropsychiatrist and completed two self-report questionnaires. The first was the 15-item Impact of Event Scale¹⁹ (IES), which measures symptoms of PTSD divided into two subscales: 7 questions exploring reexperiencing phenomena (also termed the intrusion subscale) and 8 questions dealing with avoidant behavior in relation to the traumatic event. The questions documenting intrusive symptoms includes the following: 1) I thought about it (the accident) when I didn't mean to; 2) I had dreams about it; 3) pictures about it popped into my mind; 4) other things kept making me think about it; 5) any reminders brought back feeling about it; 6) I had strong waves of feeling about it; 7) I had trouble falling asleep or staying sleep, because of pictures or thoughts about it that came into my mind. Subjects were instructed to choose one of four responses to each question: not at all; rarely; sometimes; often.

The second questionnaire was the 28-item General Health Questionnaire (GHQ), which was used as an overall index of psychological distress.²⁰ The GHQ, which has been used previously in traumatic brain injury research,²¹ has 4 responses to each question, and each was scored 0-0-1-1. A few patients could not complete the IES and GHQ because of difficulties understanding English. This is reflected by differences in the degrees of freedom that accompany the statistical analyses. The order of the assessment, with rating scales completed last, was such that the clinic coordinator who recorded details of PTA from patients did not know their IES and GHQ scores at the time.

For the purpose of statistical analysis, subjects were divided by their duration of posttraumatic amnesia into 4 groups according to the criteria of Russell and Smith¹⁸: <1 hour; 1–24 hours; 24 hours–1 week; >1 week.

The four PTA groups were compared across indices of traumatic brain injury (GCS, CT abnormalities) in addition to symptoms of PTSD (i.e., IES scores) and psychological distress (GHQ scores) using either one-way analysis of variance with post hoc Tukey comparisons or, in the case of ordinal data, chi-square analyses.

RESULTS

Of the 300 clinic attenders enrolled in the study, 18 did not sign consent, thereby reducing our sample to 282. This sample was divided according to PTA as follows: 147 subjects with a PTA <1 hour, 70 with a PTA of 1–24 hours, 40 with a PTA of 24 hours to 1 week, and 25 with a PTA >1 week. Subjects were assessed on average within 2 months of injury: mean=52.9±31.9 days (means reported with standard deviations throughout).

Demographic Data

The mean age of the sample was 35.2±14.9 years; 66% were male. Motor vehicle accidents accounted for 66% of all injuries. A breakdown of the demographic data according to duration of PTA is provided in Table 1, which also includes the manner in which subjects sustained their injuries. There were no statistically significant differences between the four PTA groups with respect to age, gender, marital status, and the mechanism of sustaining their injury, but a one-way analysis of variance demonstrated a difference in the time between sustaining the accident and having the assessment. Post hoc Tukey analysis revealed it was the patients with PTA >1 week who were seen after more time had elapsed since the accident. The large standard deviation, however, pointed toward a considerable degree of overlap with the other three groups.

Traumatic Brain Injury Data

As PTA lengthened, other indices of traumatic brain injury severity also worsened. Thus, Glasgow Coma Scale scores deteriorated significantly (P=0.0001) and the number of patients with abnormal CT scans increased (P=0.001) at longer durations of PTA. The between-group (post hoc Tukey) differences are shown in Table 2.

Psychiatric Data

An examination of the intrusion and avoidant symptomatology on the IES showed that as PTA lengthened, no significant between-group differences were apparent. More specifically, patients showed a similar range of intrusive phenomenology (e.g., dreams and intrusive thoughts) irrespective of duration of PTA. Similarly, no statistically significant differences were found on the avoidant subscale or the GHQ (Table 2). However, when the sample was dichotomized into those with a PTA <1 hour or >1 hour, significant differences were present on the IES, but not on the GHQ. Thus, patients with a PTA <1 hour had more intrusive (t=2.1, df=2,266, P=0.04) and avoidant (t=2.0, df=2,266, P=0.05) symptoms and, as a result, higher total IES scores (t=2.3, df=2,266, P=0.02).

DISCUSSION

We present 282 cases of closed head injury, in which patients were stratified into four groups according to the duration of posttraumatic amnesia. Our data are the first to cover the full spectrum of severity of head injury, ranging from the very mild to the severe. The focus of the study was to assess the degree to which a core feature of posttraumatic stress disorder, namely the reexperiencing (intrusion) symptomatology, varies according to the duration of PTA. Before we examine this question, however, some comment on the data pertaining to the severity of the traumatic brain injury is germane.

Posttraumatic amnesia is considered a sensitive marker of the severity of traumatic brain injury and, as such, a useful predictor of outcome.²² It also correlates significantly with retrograde amnesia, a variable missing from our data set.¹⁸ Although the accuracy of ascertaining PTA retrospectively has been criticized,^23,24 a study designed specifically to compare retrospective and prospective methodologies reported robust correlations between them and concluded that a retrospective assessment was valid.²⁵ Others have endorsed this finding.²⁶ In our study, the concordance of PTA data with two other indices of cerebral involvement, GCS scores and CT scan abnormalities, provides further indirect support for the validity of our PTA assessment.

Despite the evidence that increasing PTA denotes more severe brain damage, patients in all four duration groups endorsed intrusive and avoidant symptoms relating to the trauma. Of note here is our observation that the mechanism of injury (motor vehicle accident versus other, i.e., assault, industrial accident) did not differ between the four PTA groups. Nevertheless, it was noticeable that on a descriptive level, intrusive and avoidant symptomatology diminished when PTA exceeded one hour, and a subsidiary analysis demonstrated that this change was statistically significant. Looking beyond the statistics, however, it is equally noteworthy that symptoms of PTSD were present even when the PTA exceeded 24 hours and, in some cases, one week. In addition, the nature of the intrusive symptomatology did not vary according to duration of PTA. Seven examples of intrusive phenomenology were recorded by the IES and were distributed across all four groups, albeit in varying intensities.

Whereas intrusion and avoidance scores decreased when PTA exceeded 1 hour, the GHQ scores as an index of general psychological distress remained fairly constant. Given that the 28-item GHQ records symptoms of anxiety and depression, the divergence of PTSD-type symptoms and GHQ scores with increasing PTA makes it unlikely that emotional factors alone were responsible for patients endorsing recollections of their trauma and suggests that additional theories must be invoked to explain this aspect of PTSD symptom development. In this regard, our study did not directly address etiology, focusing instead on phenomenology and its relationship to posttraumatic amnesia. However, our data help dispel the notion that for PTSD to develop in the context of PTA, the “amnesia” must represent a subject's psychological repression of the traumatic event. Although the possibility of repression is acknowledged in the DSM-1V criteria for PTSD as “inability to recall an important aspect of the trauma” and is further supported by an association between PTSD and dissociative disorders,²⁷ this explanation breaks down in the face of increasingly severe brain trauma giving rise to more prolonged PTA and accompanied by more frequently observed CT brain abnormalities. The association of unequivocal brain damage with PTSD-type symptoms therefore suggests that repression is applicable only to patients with mild head injury without a loss consciousness or with the briefest alteration in consciousness who go on to develop PTSD reactions without recall for what occurred.

Our results must, however, be interpreted with certain methodological limitations in mind. Although we adopted a consistent approach to the Glasgow Coma Scale ratings by taking only those completed on arrival in the emergency room, the time from injury to arrival varied between subjects and was not always known to us. Nevertheless, our GCS data were confluent with other indices of brain damage such as the percentage of subjects with CT scan abnormalities, indicating that while there may have been individual GCS inaccuracies, for the group the scores provided a valid indicator of the extent of cerebral involvement. A second limitation was the absence of a formal diagnosis of PTSD. Our assessment of phenomenology relied on the Impact of Event Scale, a self-report measure of PTSD intrusive and avoidant symptomatology. The scale correlates highly with a diagnosis of PTSD, but the two are not synonymous.²⁸

In summary, the most notable finding to emerge from our study was that in a sample of head-injured patients assessed within a couple of months of injury and stratified for severity of head injury on the basis of duration of PTA, symptoms of PTSD occurred in all groups. However, when PTA extended beyond one hour, symptoms of reexperiencing the traumatic event (intrusive phenomena) and avoidant behavior were endorsed significantly less often. Because this finding is clinically important, but not linked to an a priori hypothesis on our part, replication is needed. In addition, whether or not these phenomena in the more seriously injured (i.e., PTA >1 week) represent the historical “truth” or an embellishment or confabulation of what occurred, as some have suggested,²⁹ cannot be answered here. Future research focusing on the phenomenology of the intrusive recollections, and establishing the veracity of recall by matching subjective recollection with independent confirmation of what actually occurred, offers one way forward. Meanwhile, as clinicians, we need to recognize that irrespective of the mechanism involved, symptoms of PTSD may occur across the full range of head injury severity and demand our prompt attention.

ACKNOWLEDGMENTS

Dr. Feinstein is supported grants from the Ontario Neurotrauma Program (ONRO-54), the Physicians Services Incorporated (PSI 98–49), and the Medical Research Council of Canada (Grants MT-15001; MT-36535, MT-37535).

TABLE 1. Demographic details and mechanism of injury (N=282)

Enlarge table

TABLE 2. Head injury and posttraumatic symptoms data

Enlarge table