Mild traumatic brain injury (MTBI) represents 70 - 90% of all treated brain injuries. It is by far most common in teenagers and young adults, and typically caused by falls and motor-vehicle collisions. The estimated population-based incident rate ranges above 600/100000 [1].

MTBI typically induces a range of symptoms such as: headaches, blurred vision, poor concentration, sleep disturbance, depressed mood or irritability. These post-injury effects are referred to as Post-Concussion Syndrome (PCS), a transient condition which is thought to reflect a fully recoverable disturbance of neural function [2–5]. However, long beyond the typical recovery interval of one to three months, at least 15% of persons with a history of MTBI continue to see their GPs because of persistent problems [6–10]. The clinical validity of these sequeale is not undisputed, in part because there is no readily evident physiological damage or deficit that could be made accountable. Consequently, a range of alternative explanations such as involvement in litigation [1] and pre - morbid psychological problems [11] are discussed. In addition, one may question the correctness of the assumption that MTBI is fully recoverable in all cases.

The view that MTBI leads to transient disturbances only is further supported by the absence of structural brain damage in diagnostic MRI images acquired with standard recording sequences[12]. However, the combination of high resolution MRI with specifically tailored scanning protocols provides evidence for microstructural abnormalities in MTBI patients [13–15]. Ultra-structural studies further indicate that MTBI may damage the structure of neurofilaments and cause traumatic axonal injury [16, 17]. Together, these findings support the idea that MTBI can introduce primary structural damage in reaction to the axon misalignment, disconnection and swelling that occurs in response to the physical forces inflicted on the neural tissue [17–19]. It therefore appears plausible to assume that microstructural damage can persist, and that this tissue damage may form the pathophysiological foundation of the persisting sequeale some MTBI patients' experience. This assumption gives rise to the hypotheses that the neural damage affects information processing, and henceforth the prediction that only those MTBI patients who suffer chronic PCS symptoms show neurobehavioral deficits.

This idea is not new and various research groups have studied the role of neurobehavioral performance in PCS. However, the existing data do not provide a coherent picture as to whether cognitive performance in longterm MTBI patients is disturbed or not, and if, to what extent symptoms and performance are related. While some studies report a lack of evidence for sustained postconcussive effects of MTBI [20, 21], others report general cognitive deficits [22–24], as well as impairments in specific information processing domains such as attention [25, 27], working memory [26, 27], and processing speed [28]. Two meta-analyses highlight this empirical controversy and come to the conclusion that the relationship between PCS and cognitive impairment is 'generally weak' [29, 30].

We propose that the inconsistent findings on the relationship of neurobehavioural performance and PCS might, in part, be due to the fact that most experiments investigated long-term MTBI consequences by studying a cohort of MTBI participants without taking the status of PCS specifically into account. However, it is clear that only a subgroup of MTBI patients continue to report PCS symptoms, and hence a separate assessment of MTBI patients with and without sustained PCS should provide a further insight. The incoherent picture emerging from the literature represents a serious problem for both suffers and health care providers. For example, little diagnostic advice is provided for clinicians regarding the classification and prognosis of PCS, and ICD-10 suggests guidelines for research purposes only. Furthermore, the chronic effects of MTBI compromise quality of life and general well being [30], and given the high incident rate of MTBI, particularly amongst young adults, the socio-economic consequences are drastic [2]. Henceforth there is a need to characterise and understand chronic MTBI effects. In this context, the present experiment studied cognitive performance in a cohort of MTBI patients at least 12 month post-incident, whereby those with and without PCS were studied separately. To control for potential confounding factors, all participants were in employment or studying at University and were not involved in any litigation.

The Rivermead Post-Concussion Symptoms Questionnaire (RPQ) [31] was used to categorise MTBI patients as those with PCS (PCS⁺) and those without PCS (PCS^-). Thereby the categorisation criterion was based on ICD-10 guidelines. The RPQ contains a list of symptoms commonly associated with PCS and participants are asked to rate the severity of each item on a scale from 0 - 4. Day-to-day cognitive function/dysfunction was further measured with the Cognitive Failures Questionnaire (CFQ) [32]. The rationale for employing the CFQ as an additional diagnostic tool was based on findings, which suggest that disruptions of daily activities represent a significant correlate of the PCS symptomatic [33–35]. To establish the relationship between subjectively experienced deficits and objectively measurable indices, we further employed a series of neurobehavioral measures. Based on the assumption that frontal lobe functions play a crucial role in modulating information processing and cognition, as well as on previous evidence suggesting deficits in these domains [2, 25], the Test of Attentional Performance (TAP) [36] and the Cambridge Neuropsychological Test Automated Battery expedio, (CANTABe) [37] were used. The principle study design was based on the rationale that only a sub-group of MTBI patients would have contained sustained PCS, presumably because of persistent microstructural damage. We therefore predict that MTBI patients with PCS would have lower performance scores on the objective measures than those without PCS and the control group. We further hypothesize that PCS severity, indexed by the RPQ sum score and the CFQ, would correlate with the performance indices reaction time and error rate.

The present experiment was designed to further our understanding of chronic effects following mild head injuries, which remains a largely controversial but equally important public health issue. More specifically, we aimed to characterize sustained PCS and its neurobehavioural concomitants within a group of long term MTBI sufferers. Thereby, a dissociation of PCS symptomatic and MTBI was introduced through group allocation methods. By and large, previous research has failed to produce a clear picture on sustained PCS, mainly because the majority of these experiments did not distinguish between MTBI participants with and without PCS. To this end, the present experiment studied a group of participants who had experienced a one-off MTBI at least twelve months prior to testing, and a matched control group, for PCS. Indexed by the RPQ cut-off criterion, 11 persons (29%) in the MTBI cohort were found to suffer from sustained PCS (PCS⁺), while the remaining 27 persons (71%) showed similar RPQ scores to healthy control participants. In addition, cognitive failures in everyday situations were reported more frequently in the PCS⁺ group as compared to participants in the PCS^- and healthy control groups. These results have two main implications. First, they highlight that mild head injury leads to a chronically elevated level of PCS symptoms in some persons with MTBI, while in others the experienced symptom level is no different from persons who did not have an MTBI. Second, they indicate that neurobehavioral deficits are related to high levels of PCS symptoms but not the MTBI per se. Most importantly, these findings were obtained by using a cut-off criterion in the RPQ for the initial group allocation, but the RPQ sum score, i.e. the complete RPQ symptom profile, for all subsequent analysis.

Our study further revealed interesting results on the role of loss of consciousness in sustained PCS and/or its severity. According to ICD - 10, a 'history of head trauma with loss of consciousness preceding the onset of symptoms by a period of up to four weeks' represents one of the diagnostic criteria (criterion 'B'). However, in our experiment 4 out of 11 participants in the PCS⁺ group (36%) had not experienced loss of consciousness but nevertheless satisfied all other PCS criteria. At the same time, 14 out of 27 participants (52 %) in the PCS^- group had lost consciousness during the incident but did not develop chronic PCS. This is consistent with other studies [41, 42], where PCS was found in the absence of loss of consciousness. Most notably in this context, Umile et al. [15] recently reported that participants with PCS performed poorer in neuropsychological tests and showed structural abnormalities in high-resolution MR scans after MTBI without loss of consciousness. Together, these and our findings provide initial evidence that a critical re-evaluation of the loss of consciousness criterion in ICD-10 may be considered. Of course one may argue that our findings on loss of consciousness are of questionable reliability, due to the circumstances under which the majority of incidents occur. However, the patient themselves typically remain the only source of information, henceforth the subjectively perceived loss of consciousness - whether it was truly experienced or not - is probably of limited diagnostic value.

A further aim of the present study was concerned with characterizing potential neurobehavioral correlates of sustained PCS. Based on the hypothesis that mild brain injuries may involve minimal structural damage, we presumed that sustained PCS should be reflected in diminished performance during cognitive tasks. To this end, a correlational analysis of test performance and symptom severity, indexed by the RPQ scores, revealed positive correlations with error rates for 50% of the neurobehavioral subtests in the PCS⁺ group. Thus, our data clearly suggests that PCS severity is associated with objectively measurable performance deficits in some cognitive tasks. These results highlight the fact that only those MTBI participants who suffer from PCS show deficits in cognitive processing, and provide further evidence that the level of PCS severity and not the experience of MTBI per se is the critical factor. The findings obtained by the correlational analysis have further importance with regards to the often criticized 'unspecific everyday' nature of the RPQ symptom checklist. Studies have shown non-clinical groups to report symptoms similar to those with PCS (e.g. [39]), which led to claims that the RPQ is an unreliable measure for PCS diagnosis (e.g. [43, 44]). However, the absence of a systematic relationship between RPQ and cognitive performance indices in the PCS^-/control groups, but highly correlated scores in the PCS⁺group found in the present study, identify the RPQ as a valid diagnostic measure despite the unspecific nature of checklist items [4].

The assessment of neurocognitive function revealed significantly poorer response accuracy in the PCS⁺ group than in PCS^- and controls, but no systematic effects for reaction time. This leaves three possible explanations. One could either conclude that persisting PCS is unlikely to affect a person's speed of response to stimuli, that RT non-result is due to low test power and thus an effect of sample size, or that participants in the PCS⁺ group may have greater motivation to perform well in the task. The latter explanation is strongly supported by Potter et al. [24], who suggested that some people with MTBI have heightened anxiety in novel testing situations, because of fears that their performance is affected by their condition. As a result, these participants may be in a state of higher arousal which impacts on their response execution time. For the present study, Potter's argument gains further credibility when error rates are taken into the equation. Thus, participants in the PCS⁺ group showed higher error rates than participants in both other groups, while reaction times were unaffected. Peloso [44] further argues that the simultaneous management of response accuracy and response speed becomes less efficient when the cognitive load exceeds a certain threshold level, and that this level may be lower following minor brain damage. This idea is well compatible with our finding that the highest error rates were found for the subtests with greatest cognitive demand. Furthermore, cognitive and emotional problems are known to persists longer after the incident than somatic symptoms [45]. In line with this observation, the RPQ symptom profile shows that the experienced symptoms in the PCS⁺ group relate mostly to cognitive processing, with poor concentration and irritability receiving the highest scores.

A continuing controversy in the discussion of chronic PCS concerns other contributing factors such as malingering [11, 46–49]. These concerns are partly driven by the lack of clear evidence for objectively measurable correlates of PCS. We cannot totally rule out the possibility of malingering or other contributing factors to our findings. However, this influence is probably of negligable magnitude for several reasons. First and foremost, participants involved in litigation of any kind were excluded from the study. Secondly, our volunteers were in employment or studying, and generally in good health and spirits. Finally, the severity of PCS symptoms systematically varied with the level of performance across a range cognitive tasks, and such a data distribution across a group is unlikely to be explained by malingering or exaggerated responses in the RPQ.

The research hypotheses of the present paper were based on the assumption that mild head injury can cause permanent microstructural damage and disturbances of neural function. As a result, information processing may become less efficient or more effortful and lead to deficits in neurobehavioral performance. Of course the question regarding the neural origin and mechanisms underlying sustained PCS are not addressed in the present study, however the data is in line with these ideas. Only a subgroup of MTBI participants show PCS symptoms, cognitive failures in everyday situations and diminished neurobehavioural performance in standardized test batteries. This neuroscience-based interpretation of our data might be challenged by the view that sustained PCS is driven by psychological factors rather than structural consequences of brain damage [45, 53, 54]. For example, Van Zomeren [55] and Wong [56] argued that PCS related symptoms may be manifested to compensate for 'behavioural faults' which patients attribute to the head injury. While this theory could explain why common complaints reported in the PCS⁺ group are more psychological in origin than somatic, it falls short of a plausible explanation for the strong correlation of objectively measurable performance indices and symptom severity.

In conclusion, the present study demonstrates a link between chronic PCS and neurobehavioral performance with a study design that controlled for the influence of MTBI per se. Our results are in line with some of the literature on long-term consequence of mild head injuries, and particularly highlight the role neuropsychological concomitants. Mild head-injuries are not always to be as mild as the name would suggest, and long-term consequence may very well have a neural underpinning.