Article Text

Download PDFPDF

Post-traumatic amnesia and confusional state: hazards of retrospective assessment
  1. Daniel Friedland1,
  2. Michael Swash2,3,4
  1. 1The Neurorehab Clinic, London, UK
  2. 2The London Independent Hospital, London, UK
  3. 3Barts and The London School of Medicine and Dentistry, QMUL at the Royal London Hospital, London, UK
  4. 4Institute of Neuroscience, University of Lisbon, Lisbon, Portugal
  1. Correspondence to Daniel Friedland, The Neurorehab Clinic, P.P.C.S., 14 Devonshire Place, London UK; danielfriedland{at}


Retrospective assessment of post-traumatic amnesia (PTA) must take into account factors other than traumatic brain injury (TBI) which may impact on memory both at the time of injury and subsequent to the injury. These include analgesics, anaesthesia required for surgery, and the development of acute or post-traumatic stress disorder. This is relevant in clinical and medicolegal settings. Repeated assessments of the post-injury state, involving tests for continuing amnesia, risk promoting recall of events suggested by the examiner, or generating confabulations. The PTA syndrome affects the categorical autobiographical memory, and is accompanied by confusion as an essential component; this should be suspected from the initial or early Glasgow Coma Scale score (13–14/15) if not directly recorded by clinical staff. PTA by itself is only one of several indices of severity of TBI. The nature of the head injury, including observers’ accounts, clinical and neuroimaging data, the possible role of other external injuries, blood loss, acute stress disorder and the potential for hypoxic brain injury, must be taken into account as well as concomitant alcohol or substance abuse, and systemic shock. A plausible mechanism for a TBI must be demonstrable, and other causes of amnesia excluded.


Statistics from


Retrospective assessment of the duration of post-traumatic amnesia (PTA) is generally considered a reasonably robust measure of severity of traumatic brain injury (TBI), especially in medicolegal practice. Indeed, PTA is often taken as the principal index of severity of TBI, without attention to other relevant phenomena. It is classically associated with a confusional state, although this may not be immediately obvious to the casual observer. In the clinical context, assessment often takes place months or years after brain injury, and the assessment is not necessarily reliable or consistent. In this review, we consider retrospective clinical assessment of PTA, especially the manner in which it is tested, its interpretation and its relationship to other indices of TBI in the context of the time elapsed after injury. In particular, we stress that multifaceted comparative quantitative studies of memory in people with TBI in the acute and later stages of recovery are lacking. We discuss reasons for caution in relying on the phenomenon of PTA to the exclusion of other information. We argue that PTA may not be reliable and that it is not linearly graded in relation to severity of brain injury as judged, most relevantly, by outcome. In this review, we focus especially on mild TBI (box 1), since this poses the most difficult problems in assessment.1

Box 1

WHO definition of mild traumatic brain injury (MTBI)1

MTBI is an acute brain injury resulting from mechanical energy to the head from external force. Operational criteria for clinical identification include

  1. One or more of the following: confusion or disorientation, loss of consciousness for 30 min or less, post-traumatic amnesia for <24 h, and/or other transient neurological abnormalities such as focal signs, seizure, and intracranial lesion not requiring surgery

  2. Glasgow Coma Scale score of 13–15 after 30 min post head injury or later on presentation for healthcare. These manifestations of MTBI must not be due to drugs, alcohol, medications, or caused by other injuries or treatment for other injuries (eg, systemic injuries, facial injuries or intubation), or by other problems (eg, psychological trauma, language barrier or coexisting medical conditions) or caused by penetrating craniocerebral injury

The sensitivity of PTA as an index of TBI severity

TBI is defined as an alteration in brain function, or other evidence of brain pathology, caused by an external force.1 ,2 This therefore includes loss of consciousness or a decreased level of consciousness, other neurological and neuropsychological features, and abnormalities on investigation, especially neuroimaging.3 The concept of an acute post-traumatic alteration in brain function includes a confusional state, often revealed by a reduced Glasgow Coma Scale (GCS) score (14/15), with loss of memory for events after the injury (PTA) or, rather less prominently, immediately before the injury (retrograde amnesia). PTA has assumed particular importance in the medicolegal assessment after TBI, since it has become widely accepted as an objective and reliable measure, even when assessed long after the injury.4 However, there are problems in accepting this retrospective approach to PTA, especially in mild or minor TBI.

Validity of PTA assessment

In medicolegal practice involving TBI, PTA is almost invariably assessed retrospectively, often several years after the injury. As background to this assessment, much depends on the quality of the contemporaneous medical records. There is usually some information about the accident itself, and the GCS is almost universally available from assessment at the scene both at the scene and in the accident and emergency department at the hospital; the GCS is a validated, ordinal, multidimensional measure of brain function that is equated to the level of consciousness, but in practice the PTA is only rarely prospectively assessed in this setting. The retrospective analysis of PTA therefore assumes particular importance in assessing the severity of the TBI,5 and other aspects of the post head injury state tend to be neglected. The validity of such a retrospective assessment of PTA, perhaps made several years after the injury, is controversial. King et al6 commented on a significant misclassification rate, suggesting that reliance on PTA alone was hazardous, particularly at the milder end of the TBI spectrum. Nonetheless, it can be particularly important to assess PTA retrospectively in a patient with a previous mild TBI, even when, at the time of the injury, no confusional state was apparent.7 When the person has no recollection of the accident, retrospective assessment of PTA is the only possible objective measure of TBI severity. Recently, there has been increased emphasis on TBI and, in particular, mild TBI in US8 ,9 and UK military personnel, including development of a screening procedure.9 ,10 One of the elements in this screening procedure is the retrospective analysis of PTA.9 ,11 It is therefore also an important question whether retrospective assessment of PTA in civilian and military settings is a valid, sensitive and reliable measure of TBI severity, and whether it is linearly related to severity of brain injury.

Definition of PTA

PTA consists of a disorder of episodic memory for personally experienced events and information. Episodic memory is a form of declarative recall particularly involving working memory. It represents an automatic, subconscious recall of a sequence of memories for multisensory events in an individual's immediate environment. Most studies of PTA, especially the seminal early studies, have been made in relatively severely brain-injured patients. Levin et al12 defined PTA as a period following a TBI with loss of consciousness during which there is confusion, amnesia for ongoing events and often a behavioural disturbance. Russell and Smith13 considered that the end of PTA was most easily defined as the point at which the patient could give a clear, consecutive account of what was happening around them. Ahmed et al14 noted that Russell and Nathan,15 in a seminal description of PTA in 1946, emphasised that to be recognised as out of PTA, patients had to demonstrate normal ‘continuous memory’, meaning the ability to reliably recall events from ongoing autobiographical memory. However, autobiographical memory is not continuous, even in normal individuals, but is episodic, and any account of a remembered event in normal individuals will feature both detailed recollections and gaps.16 ,17 Thus, the reliable day-to-day committal of events to memory, and their retrieval, is a more accurate description of the day-to-day categorical, episodic memory process than an attempt to construe memory as a so-called continuous process. Nakase-Richardson et al18 have conducted wide-ranging studies of PTA and post-traumatic confusion. They defined PTA as the interval from the event of the head injury until the patient is orientated and can form and later recall new memories, over a period of three consecutive days, a definition that follows Russell and Nathan's emphasis on both resolution of confusion and restoration of recall of ongoing events as marking the end of PTA.18–20 However, Russell and Nathan15 used a simple notion of subjective awareness of return of awareness for everyday events. This is a loose paradigm that is often followed by neurologists and neurosurgeons in assessing PTA, especially in retrospective assessment made many months or even years later. In prospective studies after acute head injury, Russell and Nathan15 stressed that recovery of orientation, representing resolution of a post-traumatic confusional state, is a key element in the resolution of PTA.

Lack of consensus on the definition of the end of PTA is generally acknowledged in studies of mild TBI.6 ,12 Wilson et al21 found that patients in PTA were distinct from those with an amnesic syndrome or those with chronic memory impairment due to TBI, a distinction characterised by semantic processing errors, impaired verbal fluency and slowed simple reaction time, with impaired backward digit span in the PTA group. They commented that the term ‘PTA’ was misleading, since categorical recall is not the only psychological abnormality present.22

The prominence of additional neurobehavioural manifestations including confusion, sleep-wake cycle disturbance, motor agitation, affective lability, aggressive behaviour and abnormalities in thought processes20 ,23 has given rise to the concept that the term ‘post-traumatic confusional state’ should replace ‘PTA’, a reversion to the terminology first used by Russell in 1932.15 ,22 However, much of the literature on outcome following TBI, especially that related to retrospective assessment, has focused on length of PTA, assessed by the return of the ill-defined notion of normal continuous memory, and not by resolution of the post-traumatic confusional state. The latter, however, can only be accurately assessed prospectively. Since the disorder of categorical memory and the confusional state are closely associated in the acute stage after TBI, any distinction between the two is difficult to address prospectively or retrospectively.24 Tate et al25 conducted a prospective assessment of people with severe TBI (GCS<8 in 68% of participants). They found that, contrary to the expected sequence of recovery of orientation followed by recovery from amnesia, recognition memory reached criterion before orientation to place (at 19 days after injury) and 5 days sooner than orientation to time, that is, before resolution of the post-traumatic confusional state. In this prospective study, these researchers also noted that there was variability in determining the end of PTA according to the scale used for the assessment. Nonetheless, there was a close correlation between the recovery of orientation and return of memory. They noted that ‘in clinical terms, PTA represents a major disturbance of the sensorium’. In this context, it is perhaps relevant to remember that people admitted to hospital without brain damage often do not know the date or day of the week.

Measurement of PTA

In clinical practice, PTA can be assessed prospectively and retrospectively. In the prospective measurement of PTA, scales such as the Westmead PTA Scale and the Galveston Orientation and Amnesia Test (GOAT) are often recommended. The Westmead PTA Scale consists of 12 items relevant to orientation, recall and recognition of new information. The end of PTA is taken as the conclusion of the first of three consecutive days with a score of 12/12.6 ,10 ,26 ,27

Retrospective analysis of PTA is less secure. It involves asking the individual to recount their first memory following injury. The assessor deems recovery from PTA to have occurred when he or she is satisfied that normal ‘continuous memory’ (better expressed as normal categorical recall of memories) is being described. Although essentially a subjective judgement, this is the basis for the Rivermead PTA Protocol.19 Ruff et al7 made recommendations for retrospective assessment of PTA in the context of categorising a mild TBI. One of their key points was that the clinician must specifically ask the patient to describe their personal recall of events and not what they have subsequently learnt.7 However, in practice, recognition of learnt material presented as a memory for events occurring after brain injury is essentially unverifiable, both by the observer and the patient. A further potential hazard arises when the clinician assessing PTA is one of a sequence of assessors, instructed by different parties, as is common in the legal context. The individual may unwittingly present learnt material rather than genuine memories. Any later examiner is particularly vulnerable to this error. This is also relevant in the clinical setting during the early phase of recovery after a mild or moderate TBI when the patient may be prospectively assessed by a sequence of different medical and other health professionals, each asking similar, protocol-driven questions about remembered events. This may encourage the individual to present learnt responses, as though forming a part of ongoing experience and, as in the retrospective context, may be difficult to recognise both by the examiners and by the patient. The variability of prospective and retrospective assessment of PTA often carried out many years after the injury is uncertain, although the validity of retrospective PTA is generally accepted unchallenged by the Courts. The experience of Tate et al24 ,25 suggests that retrospective PTA determination should be subject to cautious interpretation. For example, a patient in PTA may be capable of finding their way in a familiar environment, implying that ecological memory is available and effective.

PTA and severity of TBI—definition

The working definition of PTA has changed over the years. From the 1960s to the early 1990s, a PTA of <1 h was considered a mild TBI, a PTA of 1–24 h was viewed as a moderate TBI, and a PTA longer than 24 h was viewed as a severe TBI. This definition has been modified, so that a PTA of less than a day has become accepted as a mild TBI, a PTA of 1–7 days is considered a moderate TBI, and a PTA of >7 days is viewed as a severe TBI.21 A very brief PTA, duration not strictly defined, but less than an hour, often <30 min, for example in sports injuries, is sometimes now termed ‘minor TBI’. Currently, there is general acceptance that a PTA<24 h is characteristic of a mild TBI, whereas a PTA>24 h indicates at least a moderate TBI.2 ,3 ,26 ,27 Nakase-Richardson et al18 have since extended this classification system using the Mississippi Post Traumatic Amnesia Classification System, suggesting that a PTA of 0–14 days is viewed as a moderate TBI rather than a severe TBI. A severe TBI would be an injury following which the PTA was over 28 days. It is important to note that this change was based on the long-term outcome after TBI as judged by return to work. For example, 67% of individuals with a PTA of up to 14 days were back at work at 1 year, suggesting that the older classification of severe TBI was not appropriate. The distinction between a mild TBI and a moderate or severe TBI is important given the difference in outcome following such injuries, but this distinction is essentially a matter of convenient clinical definition, since it is self-evident that there must be a continuum in the severity of brain injury following head trauma, with a number of additional factors contributing to disability, including focal brain injuries, injuries to other parts of the body and limbs, and psychological factors. The true natural history of an uncomplicated mild TBI ranges from recovery within minutes to months, or even years, after injury.28–30 The outcome following moderate or severe TBI is variable and difficult to predict, often leading to long-term cognitive and neurobehavioural difficulties.28–30 Since retrospective analysis of PTA may be necessary in order to define mild TBI,31 ,32 agreement on terminology is essential.7 However, these observations suggest that it is unlikely that PTA is linearly correlated with outcome, but this relationship has not been explored. Current definitions of mild TBI use several categories of information that are not necessarily closely related to outcome, for example, GCS at first measurement, which itself may be delayed for some time after the accident, duration of PTA, duration of ‘loss of consciousness’, itself defined loosely, and associated injuries, especially cranial injuries. The validity of PTA as the sole measure of severity of TBI has never been objectively tested.

Confounding issues

There are major potential pitfalls in the retrospective analysis of PTA. For example, Kemp et al33 studied 63 patients with orthopaedic injuries who had not suffered a TBI. They were followed up about 10 weeks after injury using the Rivermead PTA Protocol; 38% reported PTA-like phenomena. A third of these individuals reported amnesia of 24 h or more, thus placing them in the category of moderate TBI, despite their never having suffered a brain injury.33 The use of analgesics, and more specifically opiate analgesics, can affect cognition and memory in the absence of a TBI. This is a particular problem in the prospective and retrospective assessment of PTA since opiate analgesia is often given at the scene of a major accident, and during subsequent hospital triage.

McCarter et al,34 in an important study of this issue, tested 17 non-head injured, non-traumatic orthopaedic patients given opiate analgesics in hospital, using the Westmead PTA Scale. Only 20% of participants tested over 4 days reached the criterion of three 12/12 scores defined as indicating the end of PTA.19 ,35 This result indicates that opiate analgesia can induce amnesia resembling PTA in individuals who have not sustained a TBI, potentially confounding assessment of possible PTA in many head-injured patients. Anaesthesia for emergency surgical procedures is frequently accompanied by a period of opiate analgesia, further compromising the reliability of any prospective assessment of PTA following TBI complicated by other injuries. This needs to be borne in mind in the medicolegal setting when relying on the prospective or retrospective measurement of PTA.

Islands of memory

In the retrospective analysis of PTA, individuals may describe isolated ‘islands of memory’ within the period of PTA. These islands of memory occur before continuity of memory is restored.13 ,14 Russell and Nathan15 pointed out in 1946 that this can lead to an underestimation of the period of PTA. Even if the recent definition of PTA with emphasis on the patient being orientated and able to lay down memories that can be recalled at a later stage is followed, identification of these ‘islands of memory’ still remains a problem. Islands of memory are thought to be related to the experience and recall of an event associated with a peak of arousal, or emotional experience, in the context of the TBI,36 ,37 a phenomenon suggesting that learning and recall of events, at least of categorical events, is possible during the period of PTA. Furthermore, islands of memory are not specific to TBI. This phenomenon can have various other causes, including opiate analgesia.34 The assumption that since a patient presents islands of memory this necessarily is indicative of PTA due to TBI is therefore incorrect.

Organic, psychogenic amnesias and malingering

The clinical distinction between organic and psychogenic amnesia; that is, PTA versus psychogenic dissociation occurring in the context of post-traumatic stress disorder (PTSD) or acute stress disorder (ASD), can be difficult, for example, in an assessment made some time after the injury.37 ,38 Thus, it cannot be assumed that since an individual cannot remember the details of their injury this necessarily indicates PTA, and therefore TBI, as it could also result from a psychological dissociation. It has been argued that individuals with TBI, who have suffered amnesia, are not capable of suffering from PTSD. However, the recognised contemporary position is that an individual who has suffered a TBI may develop PTSD, perhaps especially so in the case of mild TBI.37 ,38 This is highlighted by recent research pointing to the coexistence of PTSD and mild TBI, both in military personnel and in civilian populations. In a meta-analysis, Carlson et al39 found that the frequency of PTSD in individuals with a history of mild TBI ranged remarkably widely, from 0% to 89%, the majority of investigators reporting values between 10% and 40%. Thus, the converse assumption, that is, that the presence of features of PTSD does not necessarily mean that the individual did not suffer a TBI, is false. Research into malingering has focused on symptom exaggeration and the failure of symptom validity tests in neuropsychological assessment. What has not been considered in this research is the malingering of PTA, that is, either making up a period of PTA or exaggerating the length of PTA.

The TBI narrative

Recognition of the duration of PTA is dependent on the patient's narrative, as interpreted by the observer. There are very few studies considering the similarities and differences in the narratives between individuals who have sustained a TBI and those with a trauma-related psychological response to motor vehicle accidents. Jones et al38 studied the trauma narratives of 131 road traffic accident survivors with TBI and ASD, focusing on mild TBI, as conventionally defined. Participants were assessed at 1 week, 6 weeks and 3 months after TBI. At 1 and 6 weeks, the narratives of individuals with ASD/PTSD lacked coherence. At 3 months, their narratives were still repetitive. TBI was associated with confusion, defined in this retrospective study as uncertain memory of the event, as a separate characteristic at all three time points. This is a unique and relevant study. Nonetheless, a more precise definition of the operational criteria for identifying confusion in PTA versus amnesia in PTSD/ASD would be helpful, especially in retrospective assessment of PTA. Although difficult, this would be of benefit in clinical and medicolegal settings. The problem can be likened to symptom validity assessment in neuropsychological practice.40

New approaches to grading mild TBI

Concern regarding the consequences of TBI in US military personnel led the USA Department of Defense in 2008 to mandate a screening programme designed to identify deployment-related mild TBIs and associated residual symptoms. Iverson et al11 reviewed this postdeployment health assessment, and delineated the specific steps at which false-positive and false-negative screening results might emerge in determining whether or not an individual had sustained a mild TBI. In this screening process, the goal was to determine whether the person had experienced signs or symptoms of TBI, including PTA. However, as the authors acknowledge, a combatant who is psychologically traumatised, physically injured or both may experience a brief period of amnesia as a psychogenic phenomenon. In the case of physical injury, analgesics may also have been used,11 ,38 a potential confounding variable in retrospectively determining PTA. It is therefore important to recognise that although TBI may be associated with a period of amnesia, this amnesia may not necessarily be directly attributed to the brain injury. It should be noted that the screening process used in the US study has not been validated in terms of the retrospective analysis of PTA.

Arguably, therefore, the diagnostic scenario in the medicolegal, civilian and military settings when a patient has sustained a TBI, especially as assessed retrospectively, presents challenges. PTA lasting for 2–3 days, indicative of a moderate TBI, is frequently not determinable prospectively or retrospectively with reliability because of the use of analgesia or anaesthesia required for associated orthopaedic or other surgery. While this review has focused on false-positive factors, this would be a scenario in which there is also a false-negative risk, that is, falsely diagnosing a patient as not having suffered a TBI when no reliable estimate of the length of PTA is possible.

Validity of PTA analysis

Is retrospective analysis of PTA valid? McMillan et al31 compared prospective and retrospective measurement. There was a high correlation between the two measurements. However, the mean duration of PTA in these patients was 34 days, indicating that the patients studied were at the more severe end of the spectrum of TBI. The validity of retrospective PTA analysis in the mild or moderate range of TBI, especially in a period when many other factors may be concurrently influencing the situation, especially analgesia, remains uncertain. In their study of the reliability of the Rivermead PTA Protocol, King et al6 found that, although overall correlation was good, the test was less reliable when PTA was <24 h.

Ashla et al32 retrospectively assessed PTA in TBI using the Rivermead PTA Protocol, in addition to cognitive testing. They found that the duration of retrospectively measured PTA correlated with cognitive impairments when assessed up to 5 years after TBI. After 5 years, however, the retrospectively assessed PTA no longer correlated with cognitive deficits. This study was based on a small sample, consisting of 23 patients with recent TBI and 23 with remote TBI. In addition, there is no reference as to whether any study participants were seeking compensation, and symptom validity testing was not conducted. The latter is particularly important in assessing individuals seeking compensation.40 ,41 These patients were all in the severe TBI range. When considering the validity of retrospective analysis of PTA, it would be helpful to compare the results in a mild TBI group with those in a moderate TBI group to test whether there was a significant difference in cognitive performance. If there was such a difference, this would add weight to the validity of the retrospective analysis of PTA using the Rivermead PTA Protocol.

Overall, there is a paucity of studies regarding the validity of retrograde analysis of PTA in the assessment of TBI. Sample numbers have been small. Studies have tended to focus on the relationship between the length of PTA, without clear definition of its end point, and outcome.41–43 The duration of PTA, measured quite simply in the clinical context, is viewed as one of the most reliable indicators of outcome following TBI.41 It has been recommended in determining the need for rehabilitation management.41 ,43 Analysis of recall, whether defined purely in terms of memory or taking into account other aspects of the period of PTA, especially without consideration of other indicators of TBI severity, limits the validity of the assessment process.28 ,29 Tate et al found that amnesia resolved before disorientation in 94% of 31 severely injured subjects, and that disorientation resolved first for person, then for place, and then for time.23

Mild TBI and PTA

What can be done to improve assessment and understanding of mild TBI? The duration of PTA is an essential but empirical operational criterion, which should be documented as occurring in association with confusion, and a history of alteration or loss of consciousness of up to 24 h duration. The value of amnesia alone without the other characteristic neuropsychological features of the PTA syndrome (box 2) immediately after injury is less clear. The ending of PTA should be assessed with circumspection, since it is a gradual process, not an abrupt phenomenon.23–25 In addition, the WHO definition of mild TBI stresses the importance of recognising non-traumatic factors as possible causes of altered consciousness and PTA.1 These non-traumatic factors include associated physical injuries, psychological factors including an ASD, a need for intubation and a language barrier (box 2). Assessment of the severity of TBI in the medicolegal setting can be contentious, especially if there have been multiple separate assessments by different experts, and it is therefore important to apply the WHO recommendations. In the overwhelming majority of cases, symptoms following mild TBI remit within 2– 3 months of injury.1 ,28 ,29 In a minority, symptoms may be more prolonged. Although the determinants of disability in such cases may appear to consist of personal and social factors that are unrelated directly to the brain injury itself, MRI and functional MRI (fMRI) studies point towards acute and subacute changes in neural activation during this recovery phase, predominantly affecting attentional networks in the right hemisphere.44 ,45 Nonetheless, litigation, chronic pain, disrupted sleep, and preinjury and postinjury psychiatric factors have been consistently identified as poor prognostic factors in recovery.28

Box 2

Features of the post-traumatic confusional state in mild, or more severe, traumatic brain injury.21

  • Confusion (Glasgow Coma Scale 13–14/15)

  • Episodic autobiographical amnesia for ongoing events after the injury

  • Retrograde amnesia (usually very brief, if present)

  • Impaired accuracy of comprehension

  • Impaired verbal fluency

  • Delayed logical memory

  • Slowed simple reaction time

  • Impaired backward digit span

  • Agitation and restlessness; or uncharacteristically quiet behaviour

  • Repeatedly asking questions. A tendency to wander aimlessly

Imaging and PTA

What is the role of MRI clinical assessment after mild TBI? Diffusion tensor imaging (DTI) in the acute phase after mild TBI has revealed prefrontal and frontal white matter changes, consistent with diffuse axonal injury, as shown by initial increased fractional anisotropy (FA), suggesting gliosis and neurofilamentous compaction, and restricted mean diffusivity.44 ,45 In late phase images, FA is decreased, a feature considered supportive of the hypothesis that it represents damage to axons in white matter.45 However, the possible correlation of specific neuropsychological test profiles with these changes, and their relationship to the nature of the head injury and the symptoms described, remains uncertain. A crucial question in mild TBI is whether such DTI abnormalities imply long-term neuropsychological impairments and correlation of the outcome in patients with frontal cortical abnormalities on DTI with PTA less than or more than 24 h. The conventionally defined cut-off at a PTA of 24 h may be valid if clearly correlated with functional outcome. In all such correlative studies, a precise definition of the clinical features of the post-TBI state, as well as the duration of PTA, and the associated delirium, will be essential.

For example, fMRI studies in a group of 12 concussed American high school athletes, matched against uninjured teammates, showed decreased brain activation patterns in a working memory task, associated with postconcussive symptoms and impaired cognitive performance, with decreased reaction time.46 Abnormalities found 13 h after injury had resolved at retest 7 weeks after injury. The fMRI studies showed decreased activation of right hemisphere attentional networks in the concussed patients relative to controls in the initial studies, with compensatory increases in activation of this network at 7 weeks. In a second fMRI study,47 abnormalities in neural response inhibition were found in regions associated with inhibitory control and with the default mode network. The authors of these studies stress that PTSD was not a factor in their results. Future fMRI studies in mild TBI may help to better define post-traumatic delirium with impaired anterograde memory, leading to a more objective assessment procedure than the subjective interview currently commonly used for detection of PTA, essentially based on retrospective recollection, but susceptible to learnt responses in repeated testing.


It should always be remembered that the episodic autobiographical memory system that is tested regarding subjective recovery of memory for daily events occurring after TBI is only one aspect of the neuropsychological phenomenology (box 2). Episodic autobiographical memory itself depends on encoding and consolidation in the limbic system and prefrontal cortex, storage in the cerebral cortex, mainly in association areas and the limbic region, and retrieval, a function of the right frontotemporal and limbic regions. Procedural memory, memory priming, perceptual memory and semantic memory are strikingly uninvolved. In a critical review of the semiology and brain localisation of these different types of memory, Markowitsch and Staniloiu48 commented that the clinical term PTA was not clearly defined; ‘sometimes it describes the phase after the resolution of delirium that is characterised mainly by memory impairment’ and ‘sometimes the term overlaps with delirium, encompassing a period from injury until recovery of full consciousness and memory’.

The current vague definition of PTA in retrospective assessment arises from its historical context, as a relatively simple method for detecting neuropsychological abnormality in the immediate post-TBI period, with persistent features allowing assessment many months, or even years later.13 ,15 ,21 ,22 Unlike the assessment protocol recommended for prospective evaluation immediately after TBI, retrospective assessment, long after injury, is not robust and is open to unascertainable bias from practice effects,49 especially when isolated retrospective memories are interpreted as occurring within the assumed context of PTA, rather than as representing emotionally laden memories, for example, of a bedside visit by a family member, and thus as a normal episodic memory, occurring at a time when there was no other observed neuropsychological abnormality, for example, confusion (delirium). Emergence from the state of post-traumatic delirium and amnesia is therefore not sudden and abrupt, but gradual, and may not have an easily defined end point. Recall by the patient of learnt material, purporting to be recalled from within the period of PTA, whether consciously learnt or not, is an ever-present hazard in retrospective interpretation of the duration of PTA, especially when the assessment is repeated on several occasions by different observers in the medicolegal context. In addition, confabulated responses may develop.

There are important clinical and, especially, medicolegal reasons for accurately determining the severity of a TBI, particularly at the milder end of the spectrum, and all available evidence should be taken into account.2 For example, misclassification of a patient who has suffered a mild TBI as having a moderate or severe TBI may lead a clinician to suggest that ongoing cognitive symptoms enduring for longer than 3 months are to be expected. This expectation may provoke ‘expectation as aetiology’,50 so that expectation of cognitive symptoms becomes a self-fulfilling prophecy. The converse, failure to identify evidence for TBI, may lead to misclassification of an injured patient as exaggerating symptoms or as suffering from stress-related symptoms. Psychogenic symptoms are often considered additional to the effects of the TBI itself, but the possibility of an organic cause for these symptoms associated with subfrontal white matter disruption requires further study. By itself, PTA—the simple interview to establish return to normality—is a rough and insensitive instrument for determining these factors that is susceptible to unascertainable observer and respondent bias.49 McNally51 memorably stated in 1950 that ‘memories are not videotapes of our experiences’, meaning that they are not immutable records, but are encoded and recalled in relation to external events and internal emotional experiences. Furthermore, PTA is amnesia for current elapsing time after brain trauma; it is not retrieval failure.


View Abstract


  • Competing interests None declared.

  • Provenance and peer review Not commissioned; internally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.