Long-lasting neurobehavioral disorders after traumatic brain injury: potential predictors and functional outcomes

Abstract

BACKGROUND

Traumatic brain injury (TBI) often leads to neurobehavioral disorders (NBDs) that hinder functional recovery. Although demographic (gender, age, years of education) and clinical factors (post-traumatic amnesia duration [PTA], Glasgow Coma Scale severity) have been studied as potential predictors of NBDs, the impact of the levels of cognitive functioning (LCF) and traumatic axonal injury (TAI) has received less attention.

AIM

This study investigates the relationship between the variables and the onset of NBDs following TBI. It also examines the correlation between NBDs and patients’ functional outcome and community participation, as measured by the Glasgow Outcome Scale Extended (GOSE) and the Community Integration Questionnaire (CIQ).

DESIGN

Observational, longitudinal study.

SETTING

Inpatient rehabilitation setting.

POPULATION

The study cohort comprised 54 TBI patients (12 females, 42 males; mean age 46.1 years).

METHODS

Patients underwent comprehensive neuropsychological, neurobehavioral, and psychological assessments at 12 months. Clinical variables were collected during the acute/subacute phase, and functional outcomes were measured in the chronic phase (GOSE and CIQ).

RESULTS

The most frequent NBDs observed by caregivers included anger, difficulty controlling temper, impulsivity, and irritability. The findings highlight years of education, PTA duration, LCF score at rehabilitation admission (LCFa) and TAI as the key drivers of long-lasting NBDs (R²≈0.4-0.5). There was a significant moderate negative correlation between NBDs and GOSE (r=-0.67, P<0.001) as well as CIQ (r=-0.71, P<0.001).

CONCLUSIONS

The study highlights that lower education levels, prolonged PTA duration, lower LCFa, and presence of TAI are linked to a higher likelihood of developing persistent NBDs, which negatively impact functional outcomes and community participation.

CLINICAL REHABILITATION IMPACT

Regular monitoring and early intervention for patients with these risk factors – lower education, prolonged PTA, lower LCFa and TAI – could help mitigate the long-term effects of NBDs, improving rehabilitation outcomes through timely and targeted therapeutic approaches.

Traumatic brain injury (TBI) poses a significant global public health and socioeconomic challenge, being a primary cause of death and disability among young adults. Its incidence is increasing, particularly in low- to middle-income countries due to rising motor vehicle usage. In high-income countries, TBI is predominantly caused by road accidents, falls, violence, and sports activities.¹

Cognitive and neurobehavioral disorders (NBDs) are common long-term consequences of TBI. Cognitive issues encompass reduced processing speed, attention difficulties, and challenges in memory and executive functions.² Common NBDs include impulsivity, irritability, verbal aggression, socially inappropriate behavior, self-centeredness, lack of awareness, emotional lability, low frustration tolerance, and diminished drive.^{3, 4}

Over the past two decades, the literature has explored various predictors for long-term outcomes of TBI, initially focusing on demographic variables such as age, sex, and education level^5-8 and clinical measures of brain injury severity, such as post-traumatic amnesia (PTA) duration⁹ and Glasgow Coma Scale (GCS).¹⁰

The PTA – typically measured in hours or days – reflects a clinical condition of generalized cognitive disturbance characterized by disorientation, inability to store and retrieve new memories (anterograde amnesia), recall events preceding the brain injury (retrograde amnesia), and sometimes agitation and delusions. Its duration includes the period of coma and is calculated from the time of injury until the recovery of the ability to acquire and retrieve new memories. It is considered a complex condition, affecting not only orientation and memory, but also working memory, processing speed, and attention.

Similar to the GCS, PTA duration serves as an index of TBI severity and as a behavioral indicator of altered states of consciousness.^{10, 11} While GCS is primarily associated with predicting acute-phase mortality, PTA emerges as a more robust predictor of long-term outcomes.^{9, 12, 13}

Despite its clinical value, early cognitive assessment still lacks rapid and reliable bedside tools. The Levels of Cognitive Functioning (LCF)¹⁴ scale provides an alternative, applicable even in patients with low responsiveness. Unlike the GCS and PTA, which assess consciousness and confusional posttraumatic state, respectively, the LCF combines both, capturing early responsiveness and confusion (levels 1-6) and cognitive recovery at higher levels. Importantly, LCF scores at rehabilitation admission (LCFa) are linked to global and functional outcomes, aiding in rehabilitation planning and communication with families and caregivers.^15-17

Beyond clinical measures, advanced imaging techniques such as diffusion tensor imaging have begun to better identify risk factors for persistent deficits after TBI. Structural brain damage, both focal and diffuse, significantly influences long-term prognosis,^{18, 19} often co-occurring in moderate-severe cases.

Among the long-term outcomes of the TBI, the NBDs play a crucial role in the life of patients. Increasing evidence points to a negative association between NBDs, functional outcomes^{4, 20, 21} and community participation.²² Moreover, the NBDs can even lead to significant stress on the patient’s relatives.³

Several studies have explored predictors of NBDs, identifying younger age and lower education level at the time of injury as risk factors for their persistence.^{8, 23} Findings on gender differences remain inconsistent.²⁴

Among commonly used indicators of injury severity, only PTA – not the GCS – has consistently been associated with the development of long-lasting NBDs.^{9, 25, 26}

Beyond clinical severity, structural brain damage – especially focal lesions in the frontal and temporal lobes – has been linked to NBDs, while the impact of diffuse axonal injury remains less explored. A recent systematic review by Bryant and colleagues²⁷ highlighted the limited evidence supporting diffuse injuries such as TAI- traumatic axonal injury (a modern term for Diffuse Axonal Injury (DAI), defined as scattered axonal damage in white matter) as predictors of NBDs. TAI prevalence can reach up to 70% in moderate to severe TBI^{28, 29} and significantly disrupts neural communication, potentially linking it to cognitive disorders post-TBI.³⁰ Despite its prevalence, TAI’s association with NBDs remains underexplored in the literature.²⁷

In this context, additional tools that provide clinically relevant insights into early cognitive-behavioral functioning may help bridge this gap. While LCFa has been linked to rehabilitation outcomes, its role in relation to long-term NBDs remains to be clarified. Considered alongside established measures like PTA, it may contribute to a more comprehensive understanding of the factors underlying persistent NBDs.

The primary purpose of this study was to explore the role of different predictors, including LCFa and TAI, in the development of long-lasting NBDs (using The Head Injury Behavior Scale, HIBS as the outcome measure³¹ at 12 months post-injury) in a single-center, homogeneous cohort of chronic TBI patients.

A secondary objective was to explore the impact of long-lasting NBDs on the functional outcome (assessed using Glasgow Outcome Scale Extended, GOSE) and on community participation (measured by the Community Integration Questionnaire, CIQ) at 12 months post- injury.

Materials and methods

Study design, setting and participants

The study included comprehensive NBDs assessments and neuropsychological evaluations as part of a prospective monocentric observational study involving 225 TBI patients of all ages, admitted to the Neurosurgical ICU of Bergamo Hospital from January 1^st, 2017, to January 31^st, 2020. Follow-up assessment at 6 months (magnetic resonance imaging [MRI]) and 12 months (neuropsychological, neurobehavioral, and psychological evaluations and functional outcome measures) post-injury were conducted at the same hospital departments. Included patients were considered for NBDs assessment if they had: evidence of acute PTA or positive neuroimaging, a chronic recovery phase (12 months post-injury), and age ≥18 years (pediatric patients were excluded due to developmental differences and the use of age-specific assessment tools), and an Italian-speaking caregiver. Exclusion criteria: pre-traumatic developmental, neurological, or psychiatric disorders, and history of alcohol or drug abuse. TAI diagnosis was based on initial Computed Tomography (CT) scan and confirmed by a 1.5 Tesla MRI. Follow-up MRI was performed 6 months post-injury in the referring neuroradiological department. The protocol included T1- (longitudinal relaxation time) and T2- (transverse relaxation time) weighted sequences, T2-weighted gradient echo, fluid-attenuated inversion recovery (FLAIR) sequences, and diffusion-weighted imaging (DWI) for a comprehensive assessment of structural and functional brain changes following TBI. The data were analyzed by a neuroradiologist and a neurosurgeon both blinded to the clinical outcomes. The Standards for Reporting of Diagnostic Accuracy Studies Statement and Strengthening the Reporting of Observational Studies in Epidemiology were followed (https://www.strobe-statement.org).

The study was approved by the Hospital Ethics Committee on December 10, 2016 (register number: 291/16) and conducted according to the Declaration of Helsinki guidelines. Study ID: NCT03810222.

Variables and data collection methods

Clinical measures related to the injury were collected in the acute or sub-acute phase: sociodemographic factors (gender, age, years of education), injury-specific factors (GCS, PTA, LCFa) and the occurrence of TAI (initially detected during the acute phase and later confirmed by MRI).

All the patients underwent a neuropsychological, neurobehavioral, and psychological assessment, including functional outcome measures in the chronic phase (i.e., 12 months after the TBI).

Clinical measures

Post traumatic amnesia

PTA duration was assessed using the Galveston Orientation Amnesia Test (GOAT),³² a 10-item scale measuring orientation to person, place, time, and memory for events around the TBI. PTA resolution is defined by a GOAT score of 75 or higher. PTA duration was recorded as a continuous variable (days).

Glasgow Coma Scale

GCS assesses consciousness and TBI severity using three criteria: eye-opening (max 4 points), verbal response (max 5 points), and motor response (max 6 points). Scores range from 3 to 15, classifying TBI as mild (15-13), moderate (12-9), or severe (8-3).¹⁰

Level of cognitive functioning

The LCF scale is a measure of post-coma cognitive functioning, used at rehabilitation admission (LCFa) in line with the Rehabilitation Assessment Protocol of the Italian Society of Physical and Rehabilitation Medicine (SIMFER,³³ www.sinferweb.net). It supports decisions on treatment eligibility, guides personalized rehabilitation planning, and monitors recovery progress both during the rehabilitation process and at discharge (LCF at discharge). The scale evaluates patient responses in several domains, including stimulus responsiveness, orientation, memory, attention, communication, and behavior. The scale consists of eight levels, where 8 is the best performance.

Neuropsychological and psychological tests

A battery of standardized tests assessed cognitive domains commonly impaired in TBI: memory (Rey Auditory-Verbal Learning Test (RAVLT) and Digit Span Backward)³⁴ and attention/executive functions (Trail Making Test A and B – TMT A-B),³⁵ the Symbol Digit Modalities Test, oral version (SDMT),³⁶ the Letter Semantic Fluency Tests,³⁷ and the Wisconsin Card Sorting Test (WCST).³⁸ All the neuropsychological tests used as normative data the equivalent scores³⁹ except for the SDMT which used a cut-off value (34.2).

The Hospital Anxiety and Depression Scale (HADS) assessed psychological symptoms, with scores indicating normal to severe depression (HADS-D) and anxiety (HADS-A).⁴⁰

The HADS measures symptoms by means of two scales, each with seven items scored on a 4-point Likert scale (0-3). Final scores are classified as normal (0-7), mild (8-10), moderate (11-15), or severe (16-21).

Functional outcome measures

Glasgow Outcome Scale Extended

GOSE provides an ordinal classification of disability into eight categories, from death (score 1) to Upper Good Recovery (score 8), via a structured interview.⁹

Community Integration Questionnaire

CIQ assesses home integration (score 0-10), social integration (score 0-12), and productive activities (score 0-7), with total scores ranging from 0 to 29. Higher scores indicate greater integration and fewer restrictions.²²

Neurobehavioral assessment

The Head Injury Behavior Scale (HIBS) is one of the most widely used scales to assess behavioral disorders. The scale is a 20-item questionnaire completed by caregivers. Scores range from 0 to 20, indicating the severity of NBDs.³¹

Statistical analysis

Descriptive statistics were used, with mean and standard deviation (SD) or median and interquartile range (IQR) for continuous variables, and frequencies and percentages for categorical variables. Normality was checked using the Shapiro-Wilk test. Chi-square and Mann-Whitney U Tests were used for group comparisons.

Spearman’s correlation coefficients were used to explore the relationships between variables and multicollinearity (r>0.7) to inform model construction,

An explorative linear regression (robust estimation) was used, with variables entered in sequential blocks of three, based on logical and temporal ordering: 1) sociodemographic variables (gender, age, years of education) were entered as the first block, establishing a baseline; 2) injury severity during hospitalization (GCS, PTA); 3 and 4) LCFa and TAI occurrence (1=Y, 0=N). Variables were retained based on statistical significance. The original sample size of 225 patients was determined based on the expected effect sizes for functional outcomes in TBI patients. A sensitivity analysis confirmed the adequacy of the sample for detecting significant effects of NBDs in the regression models. To minimize observer bias, MRI data were analyzed by a neuroradiologist and a neurosurgeon, both blinded to the clinical outcomes. Patients who were lost to follow-up were excluded from the final analysis, and the potential impacts of this exclusion were discussed. All statistical analyses were conducted using R software.

Data availability

The data associated with the paper are not publicly available but are available from the corresponding author on reasonable request.

Results

Patient cohorts

Overall cohort. During the study period, 225 consecutive patients admitted to the Neurosurgical ICU were included.

Of them, 201 were adult patients (aged ≥18 years), 58 of whom died during hospital stay (corresponding to an in-hospital mortality rate of 25.7%) and a further 10 died within 12 months, resulting in an overall mortality rate of 30.2%. Five patients were excluded due to the occurrence of severe disorders of consciousness (vegetative state or minimally conscious state). These results were representative of a population of moderate-severe TBI, in line with consistent previously published series (versus a 12-month mortality of 26.8% (N.=92/343), P=0.858),⁴¹ and series published by the authors (12-month mortality of 30% (N.=58/193), P=1).⁴²

NBDs cohort. Of the 128 TBI patients surviving at 12 months with GOSE>2, two were excluded for linguistic barriers and two had a psychiatric disorder. In 70 (56.4%) eligible patients the NBDs assessments were not performed for the following reasons: refusal, return to domicile (different geographic area), and different rehabilitation programs (including long-term hospitalization).

Then, the sample for the analysis included 54 (43.5%) patients with TBI: 42 male, 12 female, a mean age of 46.1±16.5 years (range: 18-79) and a median of 11 years of education (IQR: 8-13). Patient characteristics are presented in Table I. The 6-month MRI was performed after an average of 234 days (SD: 110, range: 11-462 days). In 79% of patients, the evaluation was performed centrally at the referring neuroradiology department; for the remaining patients, whose MRIs were performed at other institutions, only the images were reviewed.

Table I. —Patient characteristics (N.=54).

Variables	Value
Age (years)	46.1±16.5
Sex (F/M)
Female	12 (22.2%)
Male	42 (77.8%)
Education (years), median (IQR)	11.0 (8.0-13.0)
GCS, median (IQR)	6.5 (3.0-10.8)
Injury type
Motor vehicle accident	25 (46.3%)
Motor vehicle accident (pedestrian)	6 (11.1%)
Fall	20 (37.0%)
Assault	2 (3.7%)
Sports-related head injury	1 (1.9%)
PTA duration (days), median (IQR)	25.0 (10.0-41.8)
TAI (occurrence)	28 (51.9%)
LCFa scale, median (IQR)	5.0 (5.0-7.0)

Number and relative percentage (in brackets). Mean±SD for normally distributed variables; median with interquartile range (IQR) for non-normal variables. GCS: Glasgow Coma Scale (total score); PTA: post-traumatic amnesia; TAI: traumatic axonal injury; LCFa: level of cognitive functioning at admission to rehabilitation.

The 12-month NBDs, neuropsychological and psychological evaluations were performed on average at 402±16.4 days (range: 375-437 days).

The main characteristics of patients who underwent NBDs and neuropsychological assessments at 12 months were compared with those who did not (Supplementary Digital Material 1: Supplementary Table I). As expected, patients who were not assessed at follow-up in the outpatient clinic were, on average, older and more severely disabled at 12 months than those who were assessed. However, the severity of the initial injury did not differ between the two groups. These findings confirmed that, although initial injury severity was compared between groups, the excluded group had more severe injuries, which likely impacted their ability to participate in the follow-up assessments.

Clinical measures

About 52% of patients presented a diagnosis of TAI; the remaining 26 patients had the following focal lesions: frontal (N.=6, 23.1%), temporal (N.=2, 7.7%), fronto-temporal (N.=11, 42.3%), other sites (N.=7, 26.9%). Most patients suffered a severe brain injury due to a motor vehicle accident (Table I).

Neuropsychological and psychological profile

Four patients were excluded from the assessment for the following reasons: two patients were still in the stage of PTA, one refused the assessment and one was excluded as non-Italian-speaking. In addition, the RAVLT was not administered to one patient, because he presented with anomic aphasia that could influence the performance. One patient refused to perform WCST and TMT B and one patient was excluded from the SDMT administration because of primary visual difficulties.

The neuropsychological and the psychological profiles of patients (mean score, standard deviation and range of score) are presented in Supplementary Digital Material 2 (Supplementary Table II).³⁹

Patients with chronic severe TBI performed poorly on the long-term verbal memory test (i.e., RAVLT) compared to normative scores.

The median scores for HADS-A and HADS-D were within the normal range, with values of 5 (IQR: 3-7) and 4 (1-7), respectively. Fourteen percent of patients had anxiety scores above the normal range, while 22% of patients had a score above the cut-off on the depression scale.

Functional outcome measures

The median GOSE score was 7 (IQR: 6-8); 1.9% of patients presented a lower severe disability (GOSE: 3), 9.3%, an upper severe disability (GOSE: 4), 7.4% a lower moderate disability (GOSE: 5), 18.5% an upper moderate disability (GOSE: 6); 22.2% a lower good recovery (GOSE 7) and remaining 40.7% had an upper good recovery (GOSE 8).

The median CIQ total score was 19.9 (IQR: 14-26). The median values of the three subscales were: 7 (IQR: 3-9) for Home Integration, 9 (IQR: 7-12) for Social Integration and 5 (IQR: 2-6) for Productive Activity.

Neurobehavioral measures

The median HIBS total score was 4 (IQR: 1.2-10). The most frequent NBDs that the caregivers observed were: anger/difficulty controlling temper (N.=26, 48.1%), impulsivity (N.=25, 46.3%), irritability, anxiety, emotional lability and impatience (N.=22, 40.7%). However, all these behaviors were above the 75^th percentile (Table II).

Table II. —Neurobehavioral disorders captured by HIBS.

Behavior	Caregiver
(1) Anger	26 (48.1%)
(2) Impulsivity	25 (46.3%)
(3) Impatience	22 (40.7%)
(4) Irritable	22 (40.7%)
(5) Anxious	22 (40.7%)
(6) Overly sensitive	22 (40.7%)
(7) Argumentative	21 (38.9%)
(8) Poor insight	17 (31.5%)
(9) Sudden/rapid mood change	17 (31.5%)
(10) Lacks control over (social) behavior	14 (25.9%)
(11) Depressed	14 (25.9%)
(12) Childish	13 (24.1%)
(13) Difficulty in becoming interested in things	13 (24.1%)
(14) Poor decision making	12 (22.2%)
(15) Lacks motivation	10 (18.5%)
(16) Frequent complaining	9 (16.7%)
(17) Irresponsible	8 (14.8%)
(18) Lack of initiative	7 (12.9%)
(19) Overly dependent	5 (9.3%)
(20) Aggression	2 (3.7%)

The table shows the NBDs (frequency and percentage) reported by the caregivers.

Correlational and predictive analyses of outcome measures

PTA and LCFa were strongly inversely correlated (r=-0.89, Figure 1), reflecting an inverse relationship between amnesia duration and cognitive functioning at rehabilitation admission. Both variables were significantly associated with 12-month outcomes: PTA correlated negatively with GOSE and CIQ, while LCFa showed positive correlations (all |r|>0.70, P<0.001). Similarly, PTA was positively associated with HIBS scores (r=0.60), and LCFa negatively (r=-0.64), both P<0.001.

Figure 1.

—Correlation matrix among study variables. Spearman’s rank correlation coefficients (r) are reported for statistically significant associations (P<0.05); non-significant correlations are left blank. Correlation strength is represented by a grayscale gradient ranging from -1 to 1. GCS: Glasgow Coma Scale at admission; PTA: duration of post-traumatic amnesia, in days; TAI: presence of traumatic axonal injury; LCFa: Level of Cognitive Functioning at admission to rehabilitation; HIBS total score: Head Injury Behavior Scale; GOSE: Glasgow Outcome Scale; CIQ: Community Integration Questionnaire.

Exploratory linear regression models were tested using HIBS total score as the dependent variable (Table III and Supplementary Digital Material 3: Supplementary Table III), revealing that years of education, PTA, LCFa and TAI were significantly associated with the outcome measure.

Table III. —Key results from sequential exploratory model building with HIBS as the outcome variable.

Variable	Step 1	Step 2	Step 3	Step 4
Education	-0.46 (0.21)*	-0.38 (0.17)*	-0.39 (0.17)*	NS
GCS	NA	NS	NA	NA
PTA	NA	0.03 (0.01)**	0.02 (0.01)**	NA
LCFa	NA	NA	NA	-1.85 (0.37)***
TAI	NA	NA	2.32 (1.14)*	2.19 (1.02)*

These columns show the models built at each step: Step 1 (Education + Gender + Age); Step 2 (Education + PTA + GCS); Step 3 (Education + PTA + TAI); Step 4 (Education + LCF + TAI). Values are presented as coefficient (standard error). Only significant results are included (*P<0.05, **P<0.01, ***P<0.001, full regression data are available in Supplementary Table III). NA: the variable is not included in that specific step of the model; NS: the variable is not significant in that specific step of the model. GCS: Glasgow Coma Scale at admission; PTA: duration of post-traumatic amnesia, in days; LCFa: Level of Cognitive Functioning at admission to rehabilitation; TAI: presence of traumatic axonal injury.

Specifically, years of education were associated with lower HIBS scores (β≈-0.46 to β≈-0.39, P<0.05). PTA emerged as a reliable positive predictor when included (β ≈ 0.02, P<0.01), showing a moderate effect size with relatively precise estimates. However, due to collinearity with LCFa (Figure 1), the two were not tested together to avoid redundancy. LCFa was significantly associated with HIBS scores (β=-1.85, P<0.001), and TAI also showed a significant effect when included (β≈2.20, P<0.05).

Here is a breakdown of each explorative model: Step 1) Demographic variables (age, education, gender) were included, with education showing a significant association with HIBS. Step 2) Clinical severity variables (PTA, GCS) were added, increasing the model’s variance (R²=0.35). Step 3) GCS was replaced by TAI, improving the model fit (R²=0.39). All variables – education, PTA, and TAI – were significant. Step 4) PTA was replaced by LCFa due to collinearity, explaining the highest variance (R²=0.49), with both LCFa and TAI remaining significant.

The correlation between the HIBS total score and long-term functional status (12-month outcome measures: GOSE and CIQ) was then explored. In both cases, a significant moderate-to-strong negative correlation was observed (r=-0.67 for GOSE and r=-0.71 for CIQ, both P<0.001).

Discussion

In this study, we reported the clinical characteristics and outcomes of a single, homogeneous group of 54 patients with long-lasting NBDs 12 months after TBI.

Firstly, the present study confirmed that patients with TBI often present long-term NBDs;^{3, 4, 23} among these, externalized disturbances such as anger, difficulty in controlling temper and impulsivity emerged as the most frequent NBDs observed by the caregivers.

Secondly, this study aimed to detect which variables could be associated with the development of long-lasting NBDs and to explore the relationship between these disturbances and the functional outcome measured 12 months post-TBI.

According to on the literature, we considered as predictors of NBDs demographic variables (i.e., age, gender and years of education)^{8, 23} and clinical measures of injury severity (i.e., PTA duration,^{9, 12, 13} GCS severity, LCFa).^15-17 In addition, we explored the role of brain damage, particularly the TAI, as a predictor that has been scarcely studied.²⁷

The findings highlight education, PTA duration, LCFa, and TAI as potential key drivers of persistent NBDs. However, the results should be interpreted with caution, taking into account both the sample size and the estimates obtained.

Education appeared to play a protective role, with each additional year of schooling modestly associated with lower HIBS scores. This interpretation aligns with the literature.

Pre-injury education level is a well-documented predictor of different outcomes, such as the quality of life after TBI,⁷ the global functional outcome measured by the GOSE⁴³ and employment status 12-month post-TBI.⁴⁴ To our knowledge, very few studies have provided a relationship with the occurrence of NBDs.^{8, 20}

Based on the literature, the PTA is the best predictor of functional and cognitive outcomes.²⁵ In agreement with our results, some studies provided evidence that PTA is also a predictor of long-term NBDs.^{25, 26}

The GCS did not emerge as a predictor of behavioral disturbances, consistent with literature linking it more closely to short-term outcomes such as mortality.⁹

While LCFa remains a practical tool for early rehabilitation planning, its association with long-lasting NBDs – although promising – requires further confirmation. Its predictive value may be constrained by the high proportion of patients clustered within levels 4 to 6, which limits its granularity compared to more continuous and nuanced measures such as PTA duration. Nevertheless, these findings align with previous research suggesting that early cognitive assessment can help predict long-term functional outcomes.^{11, 45, 46}

The superior power of LCFa and PTA compared to GCS supports the assertion that comprehensive cognitive assessments better capture the neural networks underlying both cognitive and behavioral recovery. Moreover, the synergistic effect observed when combining PTA or LCFa with TAI presence suggests that behavioral outcomes depend on both initial cognitive status and structural damage patterns.

Despite being representative of the pathophysiology of TBI and despite increasing evidence that TAI is related to long-term outcomes,²⁹ it is still a scarcely studied predictor of NBDs.²⁷ TAI significantly interferes with neural communication and, as such, it disrupts the neural networks that link brain structure to function.³⁰ Given these considerations, it would be worth investigating the relationship between TAI and NBDs. To our knowledge, there are very few studies concerning this relationship and they have mixed results.

Wallesch et al.,⁴⁷ compared seven patients with TAI (mainly severe) with a group of 54 non-TAI patients in the subacute phase on the NBDs and on a scale related to disturbances of the frontal lobe functioning. They found a difference in both measures that disappeared after Bonferroni correction analysis.

Fork et al.,⁴⁸ compared the occurrence of NBDs among patients in the subacute phase (5-29 days) with healthy controls. They studied a small sample of 11 patients with TAI of mixed severity, 11 patients with frontal contusions and 17 healthy controls. A significant difference emerged between patients and healthy controls, but not among patients. All the differences disappeared at the 5-8 months follow-up.

Finnanger et al.,²⁰ explored the NBDs of 47 chronic patients with a TAI. The presence of TAI significantly predicted the occurrence of internalizing symptoms; this difference was reduced after adjustment for age and education.

The results of the present study confirm a predictive role of TAI on the occurrence of NBDs and for the first time, to our knowledge, we also extended this evidence also to the externalizing behaviors (e.g., anger/difficulty controlling temper, impulsivity, irritability, impatience).

Finally, as already suggested by the literature,^{4, 49} we confirmed that a higher frequency of NBDs is related to a greater impairment of the global functional outcome (GOSE) and the community re-integration (CIQ), assessed at 12-month post-TBI.

Limitations of the study

The present study has several limitations that should be considered when interpreting the findings and planning future research. First, the sample size was relatively small: over a 3-year period, only 54 patients completed the follow-up program. Although the statistical models demonstrated adequate power (94.32%), the limited cohort size constrains the generalizability and stability of the results. To ensure clinical relevance, the analysis focused only on a restricted set of variables to avoid model overspecification given the sample size – years of education, GCS, PTA, LCFa, and TAI – striking a balance between well-established indicators and emerging, underexplored variables. While a larger cohort would enhance reliability, long-term follow-up in TBI research remains challenging due to high mortality and disability rates, as echoed in prior studies. This limitation reflects well-documented challenges in TBI research, where high mortality and injury severity often hinder long-term follow-up and data completeness.^{1, 41} Despite efforts to comprehensively assess neuropsychological, behavioral, and psychological factors, the exclusion of older and more functionally impaired patients – who had similar initial injury severity – may have introduced selection bias and limits the generalizability of our findings. Furthermore, the data are not broadly generalizable, as older patients may have been excluded. Second, the diagnosis of TAI was based on the clinical presentation of the patient and on the use of conventional CT and/or MRI imaging. However, these methods may underestimate TAI.¹⁹ Diffusion tensor imaging (DTI) provides a more sensitive measure of the microstructural changes to white matter present in TAI.³⁰ Despite the clear evidence of the utility of DTI, this method is rarely used in clinical practice and is mainly adopted in the research setting.

Conclusions

In the context of TBI, it is essential to identify patients with lower education, prolonged duration of PTA, lower LCFa, and occurrence of TAI, during the acute (ICU stay) and sub-acute (Neurosurgical and Rehabilitation wards) phases, as these individuals may be at greater risk of developing long-lasting NBDs that can impact not only their own well-being but also that of their families. Paying attention to these patients would also allow an early implementation of an integrated rehabilitation approach (i.e., pharmacological and non-pharmacological interventions)⁵⁰ for their management.

Supplementary Digital Material 1

Supplementary Table I

Comparison between patients who were examined and patients who were not assessed by NBDs and neuropsychological evaluations.

Supplementary Digital Material 2

Supplementary Table II

Mean scores, standard deviations and range of scores obtained at the tests administered to the TBI patients.

Supplementary Digital Material 3

Supplementary Table III

Results from sequential exploratory model building with HIBS as the outcome variable.