Acute behavioral changes as a diagnostic factor of intracranial injuries among the elderly population with mild traumatic brain injury - retrospective cross-sectional study

Abstract

Purpose

Mild traumatic brain injury (mTBI) is one of the most common trauma-related diagnoses treated in emergency departments, especially among the geriatric population. Higher age alone is often an indication for a computed tomography (CT) scan, even when, approximately 90% of these scans do not reveal intracranial injuries. Incorporation of new diagnostic parameters into indication schemes for CT scans could improve the efficiency and reduce unnecessary imaging. The primary outcome of this study was to evaluate the association of acute behavioral changes among elderly patients treated for mTBI with the prevalence of intracranial injuries diagnosed by CT scans.

Methods

A retrospective cross-sectional study was conducted at Louis Pasteur University Hospital in Košice. All patients aged 65 and older who presented during the period of 12 months with suspected mTBI and underwent CT imaging were included in the study. Electronic health records were used as a data source.

Results

A total of 586 patients were included in the study. Acute behavioral changes were observed among 60 (10.2%) patients. Intracranial injury was diagnosed in 35 patients (6.0%). There was a statistically significant association between acute behavioral changes and the presence of intracranial injuries (p < 0.05), with those exhibiting behavioral changes having higher odds of injury (OR: 6.51; 3.01–13.7; p < 0.001).

Conclusion

Elderly patients with mTBI who present with acute behavioral changes are more likely to have intracranial injuries detected by CT scans. Incorporating these symptoms into indication schemes for head CT scans may improve strategies aimed at more effective and judicious use of imaging.

Trial registration

Clinical trial number: Not applicable, retrospectively registered.

Background

Mild traumatic brain injury (mTBI) is one of the most frequent injury-related diagnoses among the geriatric population, driven by the aging population and a higher incidence of falls among the elderly. It is one of the leading causes of morbidity and mortality among older adults [1,2,3]. Patients in this cohort are unique due to their physiology, medical conditions, frailty, and polypharmacy. The physiological changes associated with aging, including brain atrophy and increased fragility of cerebral blood vessels, complicate the diagnosis of mTBI in this group [2, 4]. Additionally, psychiatric symptoms such as confusion, agitation, irritability and aggression, which are often linked to qualitative disturbances in consciousness, may further complicate the clinical picture in these patients. These disturbances can serve as important indicators of acute intracranial injuries, especially when classical neurological symptoms like altered Glasgow Coma Scale (GCS) scores are absent [5,6,7].

Computed tomography (CT) is a standard diagnostic tool for assessing mTBI, primarily due to its effectiveness in detecting acute intracranial injuries, which are the main cause of mortality [8]. Although it is an essential diagnostic tool for identifying serious intracranial injuries, its overuse poses concerns, including cumulative exposure to ionizing radiation, cost implications, and the potential for unnecessary healthcare interventions [9, 10]. In the elderly population with mTBI, almost 90% of CT scans are negative concerning the presence of intracranial injuries [3, 11, 12]. Clinical decision rules, such as the Canadian CT Head Rule or the New Orleans Criteria guide indications for CT scans. However, these guidelines are only partially applicable among older adults, because higher age alone necessitates CT scans in several decision rules [8, 13]. Moreover, elderly patients often present with symptoms and comorbidities that complicate the clinical picture. Given these challenges, there is a growing need to refine further the diagnostic strategies for mTBI among older adults, including the possible utilization of new clinical symptoms, laboratory markers, or the enhancement of clinical criteria that would reduce the reliance on CT scans without compromising the timely detection of clinically significant injuries.

Agitation, irritability, impulsiveness, aggressive, or antisocial behavior are frequently observed after traumatic brain injury [14, 15]. These psychiatric manifestations may occur even in the absence of overt neurological symptoms, making their recognition crucial for timely diagnosis. These disturbances can result from cortical or subcortical injuries and are linked to neurochemical changes, including dysregulation of serotonin and glutamate, which can lead to post-injury psychiatric disorders such as aggression or post-traumatic stress disorder [16]. Several studies assessed their development among patients with a history of TBI, weeks or months after the injury [14, 17]. Some studies have highlighted the role of post-injury irritability and cognitive symptoms as important factors for incomplete recovery in elderly populations [18,19,20]. However, the timing of their onset, prevalence during the acute phase after injury, or their association with intracranial injuries detected by CT scans in the geriatric population remain unknown.

The primary aim of this study was to assess the presence of acute behavioral changes as a predictive factor of intracranial injuries detected by CT among geriatric patients treated for mTBI in the emergency department.

Methods

Study design

We conducted a single-center, retrospective, descriptive cross-sectional study of geriatric patients aged 65 years and older treated for suspected mild traumatic brain injury in the emergency department of the Louis Pasteur University Hospital in Košice. The study covered a period of 12 months (July 2023 - June 2024). Patients were identified through the hospital’s electronic health records system. Records of all patients, who were 65 years and older, and treated in the trauma emergency department were screened for possible primary or concomitant traumatic brain injury. On-duty physicians (trauma surgery, orthopedics, or emergency medicine specialties) conducted and documented patient assessments. Additional data points were extracted from patients’ electronic health records. Data analysis was performed in July and August 2024. The primary outcome of this study was to evaluate the association of acute behavioral changes among elderly patients treated at the emergency department due to mTBI with the prevalence of intracranial injuries diagnosed by CT scans. This study has been written according to the “Strengthening The Reporting of OBservational Studies in Epidemiology” (STROBE) guidelines.

Inclusion and exclusion criteria

Electronic records of all patients aged 65 years and older, who were treated in the emergency department for injury-related complaints in the period of July 2023 - June 2024, were screened for suspected or confirmed mild traumatic brain injury up to 48 h before their presentation. The patients had to obtain initial CT scan imaging in the ED. The decision to perform a head CT scan was made by the treating physician. We excluded all patients with severely incomplete or missing data (missing description of patient’s behavior, missing multiple secondary outcomes). We also excluded patients with pre-existing cognitive deficits and chronic behavioral changes, who were initially triaged as mTBI patients but whose chronic GCS was not reliably documented and the presenting GCS did not meet mTBI criteria (GCS 13–15).

Methodology

The primary outcome of the study was an assessment of acute behavioral changes as documented by the treating physician and their association with intracranial lesions found on a head CT scan among patients 65 years and older treated in the ED for mild TBI. The type of behavior was assessed according to the description of patients’ behavior in the written electronic records and was categorized into one of the following options: no acute behavioral changes; confusion; agitation or irritability or severe restlessness; verbal aggression; and physical aggression. Confusion was defined based on the patient’s Glasgow Coma Scale (GCS) verbal response score of 4, or if the medical notes indicated uncertainty on the patient’s part regarding their surroundings, actions, or expectations, without any notable impairment in physical presentation or gestures. Agitation, irritability, or severe restlessness were recorded when patients visibly exhibited these behaviors, accompanied by physical changes such as increased movement or altered gestures. These behaviors were considered in conjunction with mental confusion, reflecting a more physically demonstrative response. Verbal aggression was characterized by any documented verbal threats made towards medical staff, which were expressed verbally but not accompanied by physical actions or attempts. Physical aggression was defined as any documented attempt or actual physical attack directed toward medical personnel.

Patients had to obtain their CT scan in the emergency department. All CT scans were evaluated and interpreted immediately after the procedure by a radiologist through a written electronic report. Intracranial lesions were defined as acute epidural, subdural, or subarachnoid hemorrhages, and acute intraparenchymal hemorrhage or brain contusions. In cases of more than one lesion present, a combination of more than two specific findings was reported as a distinct group. Among secondary outcomes, several parameters were collected and analyzed. Basic demographic data contained age, gender, place of injury, cause and mechanism of injury, usage of blood thinners, GCS score, and symptomatology after injury. The type of injury and cutaneous impact location were recorded according to the written description. Any additional injuries and patient outcomes were documented and reported.

Data analysis

Descriptive statistical analysis was performed for all collected variables. Categorical variables are presented as absolute counts and percentages. Continuous variables are presented as means (standard deviations - SD) or medians (Interquartile range - IQR). The differences in central tendencies of categorical variables were analyzed using Chi-square tests and Fisher exact tests.

Two logistic regression models were fitted for the analysis, with the presence of intracranial injury as the dependent variable and the presence of behavioral change as the main independent variable. The first model was adjusted for the covariates: sex and age. The second model was adjusted for sex, age, and alcohol intoxication. Small sample sizes of several behavioral categories were observed and the decision was then taken to collapse specific variables of acute behavioral changes into a binary variable (patients with and without acute behavioral changes). Given the small proportion of missing data (four cases related to alcohol intoxication as a cause of injury), a complete case analysis approach was adopted for the analysis, without applying any data imputation techniques. Both models were evaluated through Akaike Information Criterion (AIC) comparison and likelihood ratio testing. The model quality was evaluated through a pseudoR2 test, and the model’s output was interpreted by odds ratio, 95% confidence intervals, and p-values. Data analysis was performed with Rstudio v.2024.09.0 + 375.

Results

The inclusion criteria initially identified 603 patients as candidates for potential enrollment in the study. Following a thorough review of the preliminary data, 17 patients were excluded based on the exclusion criteria. A total of 586 patients were included in the final analysis (Fig. 1).

Fig. 1

Flow diagram

As the primary outcome, acute behavioral changes were detected in 60 cases (10.2%), no changes were observed among 526 patients (89.8%). Confusion was present in 48 cases (8.2%). Agitation, irritability, and severe restlessness were documented among 5 patients (0.9%), verbal aggression in 4 cases (0.7%), and physical aggression in 3 cases (0.5%). In total, 35 patients (6.0%) were diagnosed with intracranial bleeding. Subdural hemorrhage was the most prevalent finding (13 patients; 2.2%) followed by subarachnoidal hemorrhage (11 patients; 1.9%). Multiple types of intracranial hemorrhages occurred among 8 patients (1.4%). Intraparenchymal hemorrhage or brain contusion was present in 3 cases (0.5%). mTBI was recorded as an isolated injury in 359 cases (61.8%). Acute behavioral changes among the elderly population with mild traumatic brain injury were associated with the presence of intracranial injuries detected by CT scans (p < 0.05). Among secondary outcomes, the median age of the study population was 79.00 years (IQR 73.00–85.00), from whom 232 (39.6%) were male. Shapiro-Wilk and Kolmogorov-Smirnov tests revealed a non-normal age distribution within the study population. Most accidents happened at home (241 patients; 41.4%) or outside (214 patients; 36.5%). The most frequent mechanism of injury was ground-level fall or injury during ground-level fall (468 patients; 79.9%). The cause of injury was mostly attributed to mechanic reasons (342 patients; 58.4%), followed by alcohol intoxication (89 patients; 15.2%) (Tables 1 and 2).

Table 1 Summary of patient demographics and characteristics (N = 586)

Table 2 Summary of patient clinical findings (N = 586)

On presentation at the emergency department, 499 patients (85.2%) had a GCS score of 15. The majority of patients did not report any subjective symptoms after injury (338 patients; 57.7%). The most prevalent signs and symptoms after injury among the rest of patients were amnesia (124 patients; 21.2%), followed by diffuse headache (32 patients; 5.5%), and a combination of more than 2 complaints (31 patients; 5.3%). Unequal pupils were recorded in one case (0.2%). The usage of blood thinners was recorded among 273 patients (46.6%). Antiplatelet treatment was the most prevalent option (159 patients; 27.1%), followed by DOACs (83 patients; 14.2%) and LMWH (17 patients; 2.9%). No visible signs of injury were observed in 115 cases (19.6%), abrasion / wound not requiring suture was recorded in 157 patients (26.8%), haematoma in 139 patients (23.7%), and wound requiring suture in 175 cases (29.9%). Cutaneous impact location in a facial location was recorded in 89 cases (15.2%), in a frontal location in 163 cases (27.8%) and in scalp location in 219 cases (37.4%). No visible cutaneous impact location was recorded among 115 patients (19.6%). The most prevalent additional finding was peripheral wound or contusion (98 patients; 16.7%), followed by complicated facial bone fracture(s) (39 patients; 6.7%), upper limb fractures (33 patients; 5.6%), lower limb fractures (15 patients; 2.6%), and skull fractures (14 patients; 2.4%). Outpatient treatment was the preferred management for 439 patients (74.9%), 142 patients (24.2%) were admitted to the hospital for any reason. Additionally, neurosurgical treatment during hospital stay was required in 4 cases (0.7%), and one patient died in the emergency department (0.2%), the cause of death was not the traumatic brain injury (Tables 1 and 2). The results of additional analyses are presented in Table 3.

Table 3 Factors associated with intracranial injuries

Two logistic regression models were fitted for the analysis, with the presence of intracranial injury as the dependent variable and the presence of behavioral change as the main independent variable. The first model was adjusted for the covariates: sex and age. The second model was adjusted for sex, age, and alcohol intoxication (to isolate the effect of alcohol intoxication). Small sample sizes of several behavioral categories were observed and the decision was then taken to collapse specific variables of acute behavioral changes into a binary variable (patients with and without acute behavioral changes). Alcohol intoxication was included in the second model, recoded from five categories into a binary one, specifically focusing on whether alcohol intoxication was associated with intracranial injuries, based on evidence in the literature (Table 4).

Table 4 Clinical findings stratified according to the presence of intracranial injury for the purpose of logistical regression models (N = 582)

In the model without alcohol intoxication, adjusted for sex and age, patients with behavioral change upon admission have 6.98 times higher odds of having CT-confirmed intracranial injury compared to those without (OR 6.98; 95% CI: 3.26–14.6; p < 0.00). In the model with alcohol intoxication (other variables held constant) OR was computed with a slightly lower variance (OR: 6.51; 95% CI:3.01–13.7; p < 0.001), the effect has remained significant. Intoxication with alcohol was not found to be significantly correlated with the odds of having intracranial injury in this model (OR: 1.88; 95% CI: 0.76–4.46; p = 0.2) (Table 5). Multiple testing was done to evaluate the quality of the models, the results of the analyses are in the Table 5.

Table 5 Logistic regression models

Discussion

This study aimed to explore the relationship between acute behavioral changes among elderly patients with mild TBI recorded during emergency department management and the presence of intracranial injuries detected by CT scans. The prevalence of intracranial injuries in this cohort was consistent with similar studies (35 patients; 6.0%) [3, 11, 12, 21]. The high prevalence of subdural hemorrhages (2.2%) and subarachnoid hemorrhages (1.9%) is consistent with previous studies that highlight these types of injuries as common findings in elderly patients with mTBI [3, 11]. The prevalence of combined intracranial injuries with two or more types of intracranial bleeding (8 patients; 1.4%) is relatively high as well. Combined injuries are sometimes not reported and the “less severe” type of bleeding is collapsed into another category, which is perceived as “more severe”, which can alter the overall data [2]. Epidemiology and outcomes of such combined injuries are therefore not assessed thoroughly. The results indicate that acute behavioral changes are relatively rare, with 89.8% of patients showing no such symptoms. Confusion was the most common behavioral change, occurring in 8.2% of cases, followed by agitation, verbal aggression, and physical aggression in fewer than 1% of patients. Acute behavioral changes were shown to be significantly associated with the odds of intracranial hemorrhage in both logistic regression models (adjusted for sex, and age as confounding factors) (OR: 6.98; 95% CI: 3.26–14.6; p < 0.001; respectively with alcohol intoxication as an additional confounding factor: OR: 6.51; 95% CI: 3.01–13.7; p < 0.001).

Evaluation of elderly patients with mTBI is challenging due to their unique physiology and specific considerations, including usage of medications, or the impact of TBI itself. Several studies have suggested that GCS score may differ in its interpretation when applied to elderly population (over 65 years old) compared to younger patients. This difference can limit its reliability in accurately reflecting the severity of TBI, particularly during the early stages of patient assessment [5,6,7]. Acute behavioral changes may further complicate thorough assessment of patients’ neurological status, level of cognitive functions and overall their impairment of consciousness. Even in the absence of more overt “classical” symptoms of intracranial injuries, a thorough examination of patients’ psychological status and behavioral alterations may serve as an important indicator in this population. Several theories have been proposed to explain the development of long-term psychiatric problems following mTBI, including violent behavior or post-traumatic stress disorder, including disruptions in neuromodulators’ intake [14, 16]. Their specific dynamics, onset of presentation and their early association with intracranial injuries detected by CT scans are unknown. In this context, confusion, agitation, or aggression may not only reflect post-traumatic disorientation but also indicate more severe intracranial injuries requiring prompt intervention. Moreover, these symptoms might reveal an increased psychiatric vulnerability in elderly patients, where underlying cognitive decline or pre-existing mental health conditions could exacerbate the behavioral impact of mTBI [14].

One notable aspect of this study is the relatively low rate of intracranial hemorrhages compared to the overall number of patients undergoing CT scans, even among patients using blood thinners. This supports the growing concern about the overuse of CT imaging in elderly patients with mTBI, where nearly 90% of scans do not reveal any significant intracranial pathology [3, 11, 12, 22]. Current most widely used clinical decision rules such as the Canadian CT Head Rule (CCHR) and the New Orleans Criteria have limited applicability in older adults since age alone often necessitates a CT scan, without any other factors [8, 13]. The gaps in knowledge regarding the management of elderly patients further include the identification of screening tools in elderly patients with significant confounders, including criteria for brain imaging, the role of biomarkers, or specific CT/MRI rules (or criteria for patients presenting > 24 h. from injury) [23].

In addition to the primary outcomes, the secondary outcomes highlight several risk factors that may contribute to the increased vulnerability of elderly patients to mTBI. Falls were the most common mechanism of injury, which aligns with previous literature suggesting that age-related frailty and balance issues resulting in frequent falls are key contributors to mTBI in this population [2, 4]. Cutaneous impact location was associated with the presence of intracranial injuries, in particular among patients who had an injury in the scalp area (54% of patients with intracranial injuries had an impact in this location). This finding is consistent with several studies documenting this association [22, 24]. The use of blood thinners, particularly antiplatelet agents and direct oral anticoagulants (DOACs), was also prevalent in this cohort, but with no association with the presence of intracranial injuries. Usage of blood thinners as a risk factor for intracranial injuries is a frequently debated question with several studies focusing on this problem. The results are often uncertain, with several studies showing slightly higher risk among patients with dual antiplatelet therapy or warfarin, in contrast with DOACs, which seem to have better safety profiles [25,26,27].

Limitations

The study had several limitations. The retrospective nature of this study posed numerous challenges. The reliance on electronic health records for behavioral assessments introduces potential bias, as these symptoms may have been underreported or inconsistently documented. While the sample size is significant, multiple categories within collected variables made certain statistical analyses less accurate. Increasing the number of observations in future studies may alleviate this and plausibly provide a more accurate representation of the effect size of the exposure variables towards the outcome. Behavioral and psychiatric symptoms, such as confusion, agitation, or aggression, are often subject to underreporting, particularly in the elderly who may be unable to articulate their symptoms effectively. Additionally, the presence of cognitive impairments or pre-existing psychiatric disorders may obscure the true nature of the behavioral changes observed during ED visits, requiring more comprehensive psychiatric assessments in future studies. The same applies to potential confounding factors of acute behavioral changes, which are often underreported or not recognized during the ED visit, such as chronic impairment of cognitive functions, acute intoxications, or influence of other medical and social factors. Future prospective studies with standardized behavioral assessments are necessary to validate these findings and further clarify the role of behavioral symptoms in predicting intracranial injuries in geriatric patients with mTBI.

Generalisability

The findings of this study contribute important insights into the relationship between acute behavioral changes and intracranial injuries in elderly patients with mTBI in the emergency department setting. While the study’s sample size and specific population provide relevant data, its generalisability may be limited to settings with similar demographics, healthcare resources, and clinical practices. The relatively low rate of intracranial hemorrhage observed, even among patients using blood thinners, might not be directly applicable to populations with different anticoagulant usage patterns or access to imaging technology. Alcohol intoxication as a cause of injury and a confounder of patients’ presentation is another setting-specific factor, which may be limited to this cohort. Additionally, the retrospective study design, relying on electronic health records, may limit the consistent assessment of behavioral symptoms, possibly affecting the transferability of results to contexts where behavioral assessments follow standardized protocols.

Interpretation and conclusions

This study contributes to a deeper understanding of the role that acute behavioral symptoms, such as confusion, agitation, and aggression, play in the diagnostic process of traumatic brain injury in the geriatric population. The findings suggest that incorporating behavioral changes into clinical indication protocols for CT scans could lead to more effective and targeted use of CT imaging. Beyond their diagnostic value for intracranial injuries, these behavioral symptoms may also signal underlying psychiatric vulnerabilities in elderly patients, which, if unaddressed, could result in long-term psychiatric complications such as post-traumatic stress disorder, aggression, or cognitive decline. Future prospective studies are needed to validate these results and explore whether refined diagnostic criteria including psychiatric assessments, can optimize patient outcomes while minimizing unnecessary radiation exposure and resource use.