Globally, mild traumatic brain injury (mTBI) makes up 85% of traumatic brain injuries, equating to approximately 42 million people being injured [1, 2]. In Australia, an estimated 99 in 100,000 people experience a traumatic brain injury [3], while global incidence rates are as high as 369 in 100,000 people [4]. Causes of mTBI include motor vehicle accidents, cycling accidents, falls, sports injuries and assaults [5, 6]. Symptoms of mTBI can be clustered as cognitive, vestibular, ocular, anxiety/mood, headache/migraines and fatigue [7]. On average, symptoms last two weeks in adults; however, 10–20% develop persistent post-concussion symptoms (PPCS) [8, 9]. Although cognitive function is often rapidly regained after mTBI, the literature suggests that in adults with mTBI, 50% have persistent cognitive dysfunction three months after injury [10]. Furthermore, deficits in episodic, immediate, and delayed memory, as well as working memory and executive function were reported at 12 and 24 months post-mTBI [11, 12]. While it has been suggested that the cognitive symptoms of mTBI can be longstanding, research investigating the longer term cognitive effects of mTBI is limited [10]. Additionally, a hit to the head may result in sub-concussion [13], where the individual does not display the symptoms of concussion [14]. The cumulative effect of multiple sub-concussions is just as detrimental and may lead to functional and microstructural brain changes [14, 15], chronic traumatic encephalopathy [16], dementia-like brain changes such as beta-amyloid and tau accumulation [17, 18]. The relationship between sub-concussion-type injuries and the subsequent pathological and behavioural outcomes is not yet well understood [19, 20, 21, 22]. While several biopsychosocial factors impact the outcome following mTBI [23], recent research has highlighted that gastrointestinal health is an important factor in injury response regulation [24]. Conversely, a neurological injury such as mTBI may lead to dysbiosis of the gut. Although gastrointestinal health is impacted by a myriad of factors [24, 25], a factor commonly at play is medication [26, 27, 28, 29]. Regardless of the causative factors, persistent symptoms may impact return to premorbid activities and quality of life [30], and hence understanding post-mTBI symptoms was an important focus of this study.

A widely used self-report measure of symptoms following mTBI is the Post-Concussion Symptom Scale (PCSS) [31]. Although it is a reliable and valid measure of PPCS [32, 33], the vast and heterogeneous nature of PPCS makes diagnostic specificity challenging. In fact, PPCS-like symptoms are also seen following trauma without brain injury [34], as well as in healthy individuals [35, 36]. In a general population sample of 11,759 individuals, 45.1% met criteria for PPCS [37]. Hence, understanding the underlying neurobiological characteristics that distinguish high and low PCSS raters may provide further clarification on this phenomenon. It has been suggested that mTBI does not have a single pathophysiological mechanism but rather a complex interaction of multiple processes leading to both structural and functional changes resulting from both primary and secondary mechanisms of injury [23, 38]. Consequently, it has been described as a disease process rather than a single event. Whilst the primary damage occurs at the point of injury, secondary damage may evolve over days, months or sometimes longer [39]. Moreover, altered brain activation has been demonstrated to persist beyond the resolution of clinical symptoms following TBI in athletes, suggesting that being asymptomatic is not a complete indicator of recovery [40, 41, 42]. mTBI has also been referred to as “a disorder of brain connectivity” [43], indicating that one potential contributor to post-concussion symptoms is altered brain network activity. In fact, disrupted brain network function has been associated with symptom severity and cognitive function post-mTBI [44].

The purpose of this study was to gain a better understanding of the neurobiological characteristics of post-concussion symptoms by exploring the association between PCSS scores, brain network connectivity and cognition in a non-clinical sample. The first hypothesis that network dysregulation would be higher in high PCSS scorers compared to low PCSS scorers was not supported. In fact, network dysregulation (FPN and DMN) was significantly higher in low PCSS scorers; however, only the FPN difference remained significant after adjusting for multiple comparisons. The second hypothesis that high PCSS scorers would have more cognitive dysfunction was also not supported. These findings are not consistent with previous reports that disrupted brain network function is associated with post-mTBI symptom severity and cognitive function [44].

While the finding that low PCSS scorers had more network dysregulation was not anticipated, certain characteristics of this group may offer some explanation for the findings. The low PCSS group had more recent concussions overall, which may have caused them to be more ‘interested’ in participating in the study and therefore be more engaged with the process. As reflected in the literature [40, 41, 94], it is possible that individuals will continue to experience network dysfunction after clinical symptom recovery. This appears to be supported by our finding that network dysfunction was present even though clinical symptom ratings were low in the low PCSS group. It is also supportive of the conceptual understanding of sub-concussive injury described in Dioso, Cerillo [13], where functional changes exist in the absence of clinical symptoms. In fact, these findings may provide preliminary support for the mechanisms of sub-concussive functional pathology. Moreover, the high PCSS group had a higher rate of medical conditions, and their PCSS score may have been inflated by their chronic conditions (e.g., mental health challenges), reflecting the fact that it is possible that PCSS symptoms are not specific to concussion and that they might overlap with other conditions, or in fact healthy individuals as reported in the literature [34, 35, 36]. While there was no statistically significant difference in the number of participants using prescribed medications between the high and low PCSS groups in this study, the high PCSS group may also have used different types of medication, potentially confounding the results by altering brain network functioning. Considering that medication and chronic conditions are also known to effect gastrointestinal health, which in turn impacts neurological health, exploring the relationship between gastrointestinal health and post-concussion symptom ratings may provide further explanation for the variation between the low and high scorers.

Regarding the lack of significant difference in cognitive dysfunction across the high and low PCSS scorers, it is possible that since the participants were all university students, their attention and processing speed may have been actively trained through their education. Hence, while they may have post-concussion symptoms, they are less likely to display cognitive dysfunction due to higher cognitive reserve. Additionally, this was not a clinical sample, and of those with a reported concussion, most were historical injuries (average time since injury; high group 61.9 months, low group 7.6 months). As indicated in the literature [8], 80–90% of individuals with mTBI would not be expected to have persistent symptoms (e.g., cognitive deficits) at this time point post-mTBI.

The third post-hoc hypothesis was supported, finding that network dysregulation (DMN, SN, FPN) was greater in participants with a more recent mTBI. While this was a non-clinical sample, and participants were not followed up from their injury baseline, the finding that those who had experienced a concussion had significantly less network dysregulation with a greater time since injury, supports the expectation that with time, the functional neurophysiological impacts of concussion are likely to resolve. This was supportive of previous research demonstrating normalisation of network connectivity as time since injury increased when measured at six months post-injury [44]. Although this does not provide a timepoint when normalisation of network function occurs, or what percentage of individuals with mTBI might be expected to have a delayed neurophysiological recovery, and the factors influencing that recovery, it warrants further investigation to inform expectations for neurophysiological recovery from mTBI. Especially since in the present study, the subset of individuals who had experienced concussion was obtained from both high and low PCSS scoring groups, suggesting that symptom score alone was not sufficiently indicative of concussion or recovery status. In fact, it supports previous literature suggesting that neurophysiological dysfunction may be present beyond the point of clinical recovery [40].

While the findings are intriguing, the study was limited by a small sample size with limited power to detect significant differences, especially considering the very large number of variables measured and the highly varied time since injury. The large standard deviation for time since last concussion makes the conclusion challenging. With a larger sample, future research would benefit from separating the groups into acute and chronic mTBI or analysing data from participants with a narrower timeline post mTBI. Additionally, university students may be engaged in a higher level of cognitive activity than the general population. Considering the functional networks measured are closely aligned with cognitive functioning, participants in this study may have had a higher level of compensation than the general population, which may have skewed the results resulting in less cognitive dysfunction. Additionally, the neuropsychological measures used may have been too ‘simplistic’ to capture mild deficits in individuals with higher cognitive performance. Selecting a more comprehensive neuropsychological battery which is catered to higher functioning individuals, particularly in the domain of executive functioning, may have provided a better understanding of participants’ cognition. Selection bias was also a limitation since volunteers may have been particularly invested in the study due to an interest in the topic of research. Importantly, this study was not conducted in a clinical population, and concussion history was determined by self-report. Additionally, this study had a majority female sample (68%), and analyses to determine the impact of covariates such as sex and mechanism of injury were not conducted due to the limited sample size. Moreover, confounding variables such as medication and medical history were not controlled for in the statistical analyses conducted. Although more thorough inclusion criteria such as recruiting confirmed mTBI cases based on neuroimaging or a biological biomarker, would be beneficial, they were beyond the scope of this study. Future research into concussion samples would benefit from a clearly differentiated concussion sample. Furthermore, participant recruitment for this study was significantly impacted by the COVID-19 pandemic. It would be ideal to recruit many more participants so that stratification of the sample by mechanism of injury can be used in statistical analysis, ensuring specificity in our understanding of mTBI. While categorising the participants by injury mechanism would be ideal, it was considered that further dividing a subset sample of 12 participants for statistical analysis would reduce the power of the analyses relating to this subset of participants. Participant recruitment is challenging in this field of research and particularly in a sparsely populated location such as Western Australia, with the additional restrictions of a pandemic. The authors believe that appropriate caution was used in the selection of statistical analyses, interpretation of findings and transparency of the study limitations.

Future research may improve the generalisability of the findings by recruiting a non-university sample to reduce potential cohort effects relating to undergraduate students. Having a larger sample would allow for further stratification of time since injury, mechanism of injury, sex and PCSS symptom scores which would enable a broader understanding of the relationships suggested in this study. Controlling for confounding variables such as medication intake, medical history and gastrointestinal health would also be important in future studies. Considering that participants with a history of concussion were several months to years post-injury on average, conducting the study in a clinical sample during various stages of recovery (acute, subacute, chronic), as well as exploring repeated concussion and sub-concussion would further improve the understanding of brain network and cognitive function following mTBI. Using concussion diagnostic criteria or at least medically diagnosed concussion inclusion criteria would be more reliable than self-report in future studies. Lastly, analysing the relationships between post-concussion symptoms and other conditions would further enhance our understanding of the specificity of the PCSS, particularly as it relates to the large proportion of the normal population who meet the criteria for PPCS.

In conclusion, the present study highlighted that in a sample of undergraduate students, preliminary findings suggest that brain network dysregulation appears to be greater in the early post-injury phase compared to later. Additionally, the PCSS does not relate to network connectivity and cognition. This suggests that measuring post-concussion symptoms using the PCSS does not necessarily inform us about alterations in underlying neural mechanisms. That is, higher PCSS scores do not necessarily indicate more dysfunction in network connectivity or cognition. In fact, the finding that the low PCSS group had more network dysfunction provides a possible explanation for the mechanisms of sub-concussive injury. Moreover, the high PCSS group had a higher number of medical and mental health conditions, suggesting that PCSS scores may be influenced by other health-related conditions and are not specifically measuring concussion-related symptoms. Additionally, non-specific symptoms like fatigue, dizziness and sleep quality may be influenced by medication or general daily activities that are unrelated to the concussion injury. The PCSS involves rating of non-specific symptoms; it does not consider confounding factors such as medication intake or other possible reasons for the symptoms, e.g., hormonal fluctuations or activity levels and bases the post-concussion symptom score on a single timepoint rather than the entire time-period since the concussion injury. This means that the PCSS score is likely confounded by a multitude of variables that are clouding the assessment of concussion recovery, especially in the non-acute phase. While the PCSS was developed for the athlete population, it seems that administering it to the general population may bring more complexities to post-concussion recovery that are not captured by this scale. Considering its widespread use, the measurement of post-concussion symptoms in the general population requires further investigation, particularly the underlying PCSS constructs in a non-sporting population. Furthermore, further investigation in clinical samples is warranted.

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

FB, CP, MF and JB designed the research study. FB performed the research, supervised by CP, MF and JB. SM developed extensive code for data extraction and processing. FB and SH analyzed the data in consultation with CP, MF and JB. FB, CP, MF, JB wrote the manuscript. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

Ethics approval was provided by the University of Western Australia (UWA)’s Human Research Ethics Committee (2021/ET000033). Participants provided written consent.

We thank psychology honours students Emmelie and Rachel who assisted with data collection for this study. Francesca Buhagiar is supported by an Australian Government Research and Training Scholarship.

We thank the Perron Institute for Neurological and Translational Science for its support for this research through the award of a Perron Internal Grant.

The authors declare no conflict of interest.

See Tables 9,10.