Authors: Shu-Mei Yang, Ting-Ju Lai, Ya-Chu Hsu, Yu-Lin Lu, Hsing-Yu Chen, Hsiao-Ting Tsai, Sheng-Hao Cheng, Ming-Yen Hsiao, Meng-Ting Lin
Categories: Original Article, Traumatic brain injury, Dysphagia, Recovery of function, Rehabilitation
Source: Annals of Rehabilitation Medicine
Doi: 10.5535/arm.250157
Authors: Shu-Mei Yang, Ting-Ju Lai, Ya-Chu Hsu, Yu-Lin Lu, Hsing-Yu Chen, Hsiao-Ting Tsai, Sheng-Hao Cheng, Ming-Yen Hsiao, Meng-Ting Lin
To identify early clinical predictors associated with dysphagia and delayed swallowing recovery in patients with traumatic brain injury (TBI).
In this retrospective study, we enrolled adult TBI patients admitted to the rehabilitation unit of a tertiary medical center between June 2019 and June 2023. Data on baseline characteristics, neurological status, imaging findings, and rehabilitation-related variables were collected. Swallowing function was assessed using two (1) nasogastric (NG) tube retention and (2) the Functional Oral Intake Scale (FOIS) scores at 1, 4, and 12 weeks post-injury. Regression analyses were conducted to identify predictors associated with dysphagia and swallowing recovery.
A total of 160 patients were included. At 1 week post-injury, longer intensive care unit (ICU) stay, poor initial sitting balance and use of sedative medication in ICU were associated with NG tube retention. At 4 weeks, lower initial Rancho Los Amigos Scale (RLAS) scores, immobility-related complications, longer hospitalization, and temporal lobe hematomas were associated with persistent NG tube dependence. By 12 weeks, older age, delayed ability to follow commands, and poor initial sitting balance remained associated with NG tube retention. FOIS outcomes were also associated with older age, delayed time to follow commands, impaired initial sitting balance, prolonged ICU stay, temporal lobe hematomas, lower initial RLAS scores, immobility-related complications, prolonged endotracheal tube placement and extended hospital stays.
Impaired cognitive status, poor physical function, immobility-related complications, and temporal lobe hematomas were key factors associated with dysphagia and delayed oral intake in individuals with TBI.
Traumatic brain injury (TBI) is a major cause of long-term disability in adults, often resulting in persistent physical, cognitive, behavioral, and emotional impairments [1]. Dysphagia is a common and clinically significant complication of TBI, affecting a substantial proportion of patients in both acute and rehabilitation settings [2]. The reported prevalence of dysphagia among individuals with TBI varies widely, ranging from 25% to 93% in inpatient rehabilitation settings, with approximately 37% of severe TBI cases failing to regain an unrestricted diet even after prolonged rehabilitation [3,4].
Post-TBI dysphagia results from combined neuromuscular dysfunction, cognitive-communication deficits, and behavioral impairments across different swallowing phases [5-8]. The pathophysiology of dysphagia in TBI is distinct from that observed in stroke and other neurological conditions [9]. Unlike cerebrovascular accidents that typically cause focal brain lesions, TBI often results in a combination of focal injury, diffuse axonal injury, and secondary damage due to hypoxia or raised intracranial pressure [10]. Neuroimaging studies have implicated multiple cortical and subcortical structures, such as the opercular-insular region, basal ganglia, corona radiata, thalamus, and internal capsule, in the control of swallowing [11,12]. However, in contrast to stroke, where lesion location is a primary determinant of dysphagia severity, TBI-related dysphagia is influenced by the diffuse and multifactorial nature of the injury, making prediction and management more complex [9].
Previous studies have identified several factors associated with dysphagia severity and recovery following TBI, including lower Glasgow Coma Scale (GCS) scores, prolonged mechanical ventilation, tracheostomy, reduced cognitive function, and neurosurgical interventions [4,8,9,13-15]. However, much of the existing literature includes heterogeneous neurological populations or focuses on dysphagia at a single time point, limiting conclusions specific to TBI and obscuring how swallowing function evolves during recovery.
Despite an increasing amount of research on dysphagia in TBI, significant gaps still exist in our understanding of its disease progress and key predictive factors for post-TBI dysphagia. Many studies on neurogenic dysphagia include heterogeneous populations, incorporating patients with cerebrovascular accidents, brain neoplasms, or progressive neurological diseases, making it difficult to draw definitive conclusions specific to TBI [9,16,17]. Additionally, while previous studies have identified general predictors of dysphagia, they have not consistently examined its progression at different time points post-TBI [8,18].
Given these considerations, the present study aims to identify prognostic factors for dysphagia recovery in patients with TBI at 1-, 4-, and 12-weeks post-injury. By analyzing neurological, imaging, and rehabilitation-related predictors, we seek to improve the understanding of swallowing dysfunction in TBI, facilitate early intervention, and optimize long-term management, thereby informing more precise rehabilitation strategies.
This study was approved by the Institutional Ethics Review Board of National Taiwan University Hospital (IRB No. 202410055RINA). Informed consent was waived by the ethics committee as the study involved retrospective analysis of de-identified data.
This retrospective study was conducted at a university hospital and included patients admitted to the rehabilitation ward between June 1, 2019, and June 30, 2023. This research followed the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines [19]. Patients were included if they had a confirmed diagnosis of TBI by board-certified physical medicine and rehabilitation physicians, were 20 years or older, and had been admitted to the rehabilitation ward for post-TBI care. Exclusion criteria included a history of brain tumors or prior brain surgeries. Patients determined to be in end-of-life status by two specialists and subsequently placed on palliative care were also excluded.
Baseline demographic and clinical data were extracted from medical records, including age, sex, body mass index, employment situation, and living status. Pre-existing medical conditions were documented. Additionally, prior neurological disorders, including ischemic stroke, hemorrhagic stroke, previous TBI, and neurodegenerative diseases were recorded. The mechanism of injury, including motor vehicle accidents, falls, assaults, or suicide attempts, along with the presence and type of additional injuries (e.g., maxillofacial trauma, chest or abdominal injuries, pelvic fractures, polytrauma), was reviewed. Baseline laboratory measurements were obtained upon hospital admission.
To align with the clinical care pathway after TBI, hospitalization was categorized into two phases. The acute phase was defined as the interval from emergency department arrival to admission to the inpatient rehabilitation ward. The post-acute phase was defined as the interval from rehabilitation ward admission to hospital discharge. Total hospitalization duration was calculated as the sum of these two phases. All intervals were obtained from the electronic medical record system.
Neurological status was evaluated using the GCS score upon hospital admission. Additional assessments included coma duration, time to follow commands, RLAS scores at acute admission, and Mini-Mental State Examination (MMSE) scores at acute admission. Time to follow commands was defined as the number of days from injury (or hospital admission) to the first documented ability to consistently follow simple verbal commands, as assessed by the treating rehabilitation team, such as the appropriate response to one-step commands during routine neurological or functional assessments. Data regarding acute management were also extracted, including steroid use in the intensive care unit (ICU), presence of hydrocephalus, placement of external ventricular drains (EVDs), and history of craniectomy or craniotomy. Information on airway management and critical care exposure was additionally collected, including duration of endotracheal tube (ETT) placement, history of tracheostomy during hospitalization, and use of sedative medications in the ICU. Post-traumatic amnesia was obtained from acute-care records and rehabilitation admission notes. Post-traumatic amnesia was defined as a period of impaired memory and disorientation after injury, with resolution determined by the documented return to continuous orientation. When a formal assessment was unavailable, resolution was defined as the first day with physician or nursing documentation of continuous orientation for ≥24 hours. Post-traumatic amnesia was coded as present versus absent at acute admission. Brain computed tomography (CT) scans performed at admission were analyzed for the presence and classification of hematomas, including epidural hematoma (EDH), subdural hematoma (SDH), intracerebral hemorrhage (ICH), subarachnoid hemorrhage (SAH), and intraventricular hemorrhage (IVH). Hematoma location after TBI was categorized as frontal, parietal, temporal, occipital lobes, basal ganglia, brainstem, or cerebellum.
Rehabilitation-related variables included time from injury to the first speech therapy session. Dysphagia therapy was delivered by experienced speech-language pathologists and typically provided 5 days per week, with each session lasting approximately 30–45 minutes Swallowing therapy included a combination of compensatory strategies such as postural adjustments and diet texture modification, direct swallowing exercises, oral–motor and sensory stimulation, and caregiver or patient education, as guided by bedside swallowing evaluations. The selection and progression of interventions were individualized based on the patient’s level of alertness, cognitive status, swallowing safety, and observed clinical performance, rather than following a fixed protocol. The total number of dysphagia therapy sessions each patient received during hospitalization was recorded. Initial sitting ability was graded using a modified Functional Independence Measure (FIM) scale [20]. Only the helper-dependent categories were 5=supervision (subject performs 100%), 4=minimal assistance (≥75%), 3=moderate assistance (≥50%), 2=maximal assistance (≥25%), and 1=total assistance or not testable (<25%). We did not apply the “no-helper” levels (6=modified independence, 7=complete independence). In this study, a score of 5 was classified as normal sitting ability, whereas scores of 1–4 were classified as impaired. Initial sitting balance was assessed utilizing the “sitting-unsupported” item in Berg Balance Scale [21], with a score of 4 indicating normal sitting balance and 0–3 classified as impaired. These assessments were completed within 24–72 hours of rehabilitation admission. Complications of immobility were defined as pneumonia, urinary tract infection, pressure sores, or deep vein thrombosis during hospitalization.
The primary outcome was swallowing function, assessed using the Functional Oral Intake Scale (FOIS) at 1-, 4-, and 12-weeks post-injury. Additionally, nasogastric (NG) tube retention at the same time points was documented. No patients underwent gastrostomy during the study period. FOIS scores and NG tube status at 1 and 4 weeks were recorded during inpatient hospitalization. At 12 weeks, follow-up data were obtained from outpatient visits or, for patients transferred, through standardized telephone interviews conducted by speech-language pathologists or research staff. The FOIS, which uses an ordinal scale, assigns scores of 1–3 to indicate tube-dependent feeding, while scores of 4–7 reflect varying levels of oral intake, either with or without dietary modifications [22].
Descriptive statistics were reported for baseline characteristics, neurological, imaging, and rehabilitation-related variables. The normality of continuous variables was analyzed with Shapiro–Wilk test. The outcome measures included NG tube dependence as a binary variable and FOIS scores as an ordinal variable (levels 1–7). Associations with FOIS were examined using ordinal logistic regression, assuming proportional odds across FOIS categories. The reported odds ratios (ORs) represent cumulative ORs, indicating the odds of being at or below a given FOIS threshold, corresponding to worse swallowing function, for each one-unit increase in the predictor variable. Variables were prespecified based on prior literature and clinical relevance [8,10,23,24]. The univariable ordinal logistic regression/logistic regression screened all candidate predictors against FOIS and NG outcomes at 1, 4, and 12 weeks (screening p<0.05), followed by multivariable analysis using bidirectional stepwise selection on variables that passed screening, with entry and removal criteria both set at p<0.05. The stepwise procedure automatically evaluated each variable's independent contribution while controlling for other variables, retaining only those maintaining significance in the presence of other predictors. Although no variables were forced to remain in the final models, most clinically relevant factors (such as admission RLAS scores and ICU length of stay) reached significance in univariable analysis and were therefore included as candidate variables, indicating concordance between clinical importance and statistical selection. We also performed sensitivity analyses by excluding potential mediators, including ICU length of stay and total hospitalization duration, and re-estimated the multivariable models accordingly. We also performed subgroup analyses comparing outcomes across TBI severity groups defined by the GCS (mild 13–15, moderate 9–12, severe 3–8). Differences in NG tube retention were evaluated with the chi-square test, and differences in FOIS scores with the Kruskal–Wallis test, given the ordinal nature of FOIS score. A p-value<0.05 was considered statistically significant. All statistical analyses were carried out using SAS software (version 9.4, SAS Institute).
The baseline characteristics of the 160 patients with TBI are summarized in Table 1. These patients represent the final eligible cohort after systematic screening and exclusion, as detailed in the study flow diagram (Supplementary Fig. S1). The mean age was 66.04±18.91 years, and 67.50% of the patients were male. Most patients (88.13%) had an independent premorbid functional status, defined as a modified Rankin Scale score of 0–2. None of the patients had NG tube placement or received dysphagia therapy within 1 month before the TBI.
The most prevalent comorbidities were hypertension (56.88%), diabetes mellitus (31.88%), and dyslipidemia (22.5%). Additionally, 20% of patients had a neurodegenerative disease, and 11.88% had a history of prior TBI. Neurological assessments showed a median initial GCS score of 13, with 52.5% of patients classified as having mild TBI, 22.5% as moderate, and 25.0% as severe. Neuroimaging findings revealed that SDH was the most common hemorrhage type (71.25%), followed by SAH (66.88%) and ICH (48.13%); 28.75% of patients had multiple hemorrhage types. The frontal (83.13%), temporal (68.75%), and parietal (46.88%) lobes were the most frequently affected regions.
The mean total hospitalization duration was 53.44±33.32 days, comprising a mean of 21.72±19.39 days from emergency department arrival to rehabilitation ward admission (acute phase) and 29.37±12.79 days from rehabilitation ward admission to discharge (post-acute phase). During the rehabilitation stay, 7 patients (4.38%) required acute care 5 patients (3.1%) to the internal medicine ICU (2 for septic shock and 3 for respiratory failure) and 2 patients (1.3%) to the neurosurgery ICU due to deterioration of neurological status. Among patients receiving dysphagia therapy, the mean time from admission to initiation of therapy was 16.44±16.79 days. The total number of dysphagia therapy sessions during hospitalization was 25.48±12.55 sessions on average.
Swallowing function assessments demonstrated progressive improvement in FOIS scores over time, with a median score of 1 at 1 week, 3 at 4 weeks, and 7 at 12 weeks. The proportion of patients requiring NG tube feeding declined from 70.62% at 1 week to 51.25% at 4 weeks and 20.63% at 12 weeks.
Univariate analysis identified several factors significantly associated with NG tube retention at 1 week, including a lower initial GCS score (OR 0.760, p=0.0007), lower initial RLAS score (OR 0.603, p=0.002), lower initial MMSE score (OR 0.924, p=0.011), and longer ICU stay (OR 1.283, p<0.001). A longer time to follow commands (OR 1.316, p=0.018), the presence of post-traumatic amnesia (OR 2.921, p=0.008), and use of sedative medication in ICU (OR 2.428, p=0.019) were also identified as potential risk factors. Additionally, patients with additional injuries (OR 3.010, p=0.016) and those with temporal lobe hematomas (OR 3.520, p=0.001) had a significantly higher likelihood of NG tube dependence. Rehabilitation-related factors, including reduced initial sitting ability (OR 0.569, p<0.001) and impaired initial sitting balance (OR 0.513, p<0.001), were also associated with NG tube retention (Supplementary Table S1). Multivariate analysis confirmed that longer ICU stay (adjusted OR [aOR] 1.196, p=0.003), impaired initial sitting balance (aOR 0.581, p=0.006), temporal lobe hematoma (aOR 2.857, p=0.043) and use of sedative medications in ICU (aOR 3.068, p=0.032) were independently associated with NG tube retention at 1 week (Table 2).
At 4 weeks (Supplementary Table S2), univariate analysis identified lower hemoglobin levels, lower initial GCS score, lower initial RLAS score, and lower initial MMSE score as factors significantly associated with NG tube retention. Additionally, patients with hydrocephalus, EVD placement, seizures, post-traumatic amnesia, sedative medication use in ICU and complications of immobility were at increased risk of prolonged NG tube dependence. A longer time to follow commands, extended ICU stay and longer hospitalization duration were also associated with persistent NG tube dependence. Furthermore, temporal lobe hematomas and delayed initiation of speech therapy were significant factors. Rehabilitation-associated characteristics, including impaired initial sitting ability and reduced initial sitting balance, were predictive of prolonged NG tube retention. Multivariate analysis further demonstrated that lower initial RLAS score (aOR 0.566, p=0.022), complications of immobility (aOR 2.842, p=0.033), longer hospitalization duration (aOR 1.044, p<0.001), and temporal lobe hematomas (aOR 3.005, p=0.038) were independently associated with NG tube retention at 4 weeks (Table 2).
At 12 weeks (Supplementary Table S3), univariate analysis identified several factors significantly associated with prolonged NG tube retention. These included older age, lower initial RLAS score, lower initial MMSE score, longer time to follow commands, and extended ICU stay. Structural and clinical factors such as the presence of hydrocephalus, EVD placement, prolonged ETT placement, history of tracheostomy during hospitalization, craniectomy, seizures, post-traumatic amnesia, and complications of immobility also showed strong associations with prolonged NG tube dependence. Additionally, longer total hospitalization duration and temporal lobe hematomas remained significant predictors. The cause of injury was a significant predictor, with falls associated with a higher risk of prolonged NG tube dependence. Rehabilitation-related factors, including delayed initiation of speech therapy, impaired initial sitting ability, and poor initial sitting balance, were also associated with poor swallowing outcomes. Multivariate analysis further demonstrated that older age (aOR 1.211, p=0.004), delayed time to follow commands (aOR 1.129, p=0.005), and poor initial sitting balance (aOR 0.175, p=0.005) were independently associated with NG tube retention at 12 weeks (Table 2). In addition, sensitivity analyses excluding potential mediators, including ICU length of stay and total hospitalization duration, yielded consistent directions of association for the main predictors of NG tube retention at 1, 4, and 12 weeks (Supplementary Table S4).
In the univariate analysis for lower FOIS levels at 1 week, several factors were significantly associated with lower FOIS scores, including a lower initial GCS score (OR 0.931, p=0.001), lower initial RLAS score (OR 0.852, p<0.001), lower initial MMSE score (OR 0.970, p=0.009), presence of additional injuries (OR 1.360, p=0.014), and longer ICU stay (OR 1.077, p<0.001). Patients with post-traumatic amnesia (OR 1.357, p=0.004), longer hospitalization duration (OR 1.006, p=0.011), use of sedative medications in ICU (OR 1.273, p=0.017), and temporal lobe hematomas (OR 1.430, p<0.001) had a significantly higher risk of poor oral intake. Rehabilitation-related factors, such as poor initial sitting ability (OR 0.855, p<0.001) and poor initial sitting balance (OR 0.822, p<0.001), were also associated with lower FOIS scores (Supplementary Table S5). In the multivariate analysis, longer ICU stay (aOR 1.079, p=0.001), temporal lobe hematoma (aOR 1.694, p=0.008), poor initial sitting balance (aOR 0.789, p=0.002) and use of sedative medication in ICU (aOR 1.503, p=0.041) were independently associated with lower FOIS levels at 1 week (Table 3).
For FOIS levels at 4 weeks (Supplementary Table S6), univariate analysis identified lower hemoglobin levels, lower initial RLAS score, lower initial MMSE score, longer time to follow commands, longer ICU stay, and longer hospitalization duration as factors associated with lower FOIS scores. Additionally, the presence of hydrocephalus, EVD placement, seizures, prolonged ETT placement, post-traumatic amnesia, and complications of immobility were associated with worse FOIS outcomes. Temporal lobe hematomas and delayed initiation of speech therapy further increased the risk of poor swallowing function. The mechanism of injury demonstrated a significant correlation with FOIS scores, with falls being associated with an increased risk of poor swallowing recovery. Rehabilitation-related factors, including poor initial sitting ability and poor initial sitting balance, were also correlated with lower FOIS scores. Multivariate analysis demonstrated that lower initial RLAS score (aOR 0.854, p=0.004), presence of complications of immobility (aOR 1.265, p=0.031), longer hospitalization (aOR 1.009, p<0.001), and temporal lobe hematoma (aOR 1.324, p=0.017) were independently associated with lower FOIS scores at 4 weeks (Table 3).
For FOIS levels at 12 weeks (Supplementary Table S7), univariate analysis demonstrated that older age, lower initial RLAS score, and lower initial MMSE score were associated with worse swallowing function. Additionally, a longer time to follow commands, extended ICU stay, presence of hydrocephalus, prolonged ETT placement, history of tracheostomy during hospitalization, EVD placement, craniectomy, post-traumatic amnesia, and complications of immobility significantly increased the risk of persistent dysphagia. Temporal lobe hematomas, delayed initiation of speech therapy, poor initial sitting ability, and poor initial sitting balance were also correlated with lower FOIS scores. The cause of injury was also significantly associated with lower FOIS scores, with falls linked to a higher risk of poor swallowing recovery. In the multivariate analysis, older age (aOR 1.069, p=0.004), delayed time to follow commands (aOR 1.116, p=0.001), presence of complications of immobility (aOR 4.067, p=0.042), poor initial sitting balance (aOR 0.322, p=0.002) and prolonged ETT placement (aOR 1.131, p=0.003) were independently associated with lower FOIS scores at 12 weeks (Table 3). Similarly, sensitivity analyses excluding potential mediators, including ICU length of stay and total hospitalization duration, yielded consistent directions of association for the main predictors of FOIS outcomes at 1, 4, and 12 weeks (Supplementary Table S8).
Subgroup analyses stratified by TBI severity revealed significant differences in swallowing outcomes (Supplementary Table S9). For NG tube retention, patients with moderate and severe TBI showed substantially higher dependence than those with mild TBI: at 1 week, rates were 58.75%, 90.91%, and 89.74% in the mild, moderate, and severe groups, respectively (p<0.001); at 4 weeks, the corresponding rates were 41.46%, 63.64%, and 66.67% (p=0.012); and at 12 weeks, 15.58%, 32.26%, and 27.50% (p=0.111). FOIS scores also differed across groups, with a significant group effect at 1 week (p<0.001) but not at 4 weeks (p=0.164) or 12 weeks (p=0.512).
Our study identified several factors associated with dysphagia recovery following TBI. Prolonged NG tube retention was significantly associated with ICU length of stay, impaired initial sitting balance, temporal lobe hematoma and use of sedative medication in ICU at 1 week; lower initial RLAS score, complications of immobility, longer hospitalization duration, and temporal lobe hematomas at 4 weeks; and older age, delayed time to follow commands, and impaired initial sitting balance at 12 weeks. Similarly, lower FOIS scores were independently linked to ICU length of stay, temporal lobe hematomas, impaired initial sitting balance and use of sedative medication in ICU at 1 week; lower initial RLAS score, complications of immobility, longer hospitalization duration, and temporal lobe hematomas at 4 weeks; and older age, delayed time to follow commands, complications of immobility, poor initial sitting balance and prolonged ETT placement at 12 weeks. According to our study, multiple factors are associated with swallowing dysfunction, emphasizing the complex and multifactorial nature of post-TBI dysphagia.
Our findings are consistent with previous studies showing that cognitive impairment is a key factor influencing dysphagia recovery in patients with TBI [8,9]. In our study, lower initial RLAS scores were significantly associated with NG tube placement at 4 weeks and lower FOIS scores at 4 weeks post-TBI. Deficits in cognitive-communication and behavioral regulation can compromise swallowing safety through impaired attention, awareness, and executive control, leading to delayed swallow initiation, poor adherence to safe swallowing strategies, and increased aspiration risk [8,9,25]. Previous research has consistently demonstrated an association between cognitive function and swallowing safety [7,14,18,26]. Terré and Mearin [18] reported a significant correlation between cognitive impairment and swallowing dysfunction in patients with TBI, noting deficits in tongue control, prolonged oral transit time, and an increased risk of penetration, as evaluated separately using RLAS and videofluoroscopy. Similarly, Mackay et al. [14] identified RLAS as an independent predictor of both the initiation and attainment of total oral feeding. A retrospective chart review further showed that lower RLAS scores were common among non-oral feeders, with recovery of oral intake paralleling cognitive improvement over time [2]. However, early TBI-related physiological and procedural factors, such as head and neck trauma, endotracheal intubation, ventilator dependence, and tracheostomy, may have contributed to swallowing dysfunction [13,15]. These factors may account for the absence of a significant association between initial RLAS score and dysphagia at 1 week post-TBI in our study.
Our study found that temporal lobe hematomas were independently associated with NG tube placement at 1 and 4 weeks and FOIS scores at 1and 4 weeks post-TBI. The temporal lobe has been implicated in cognitive, sensory, and integrative processes that may influence swallowing performance [27]. Deficits in attention, awareness, or comprehension may interfere with timely swallow initiation and adherence to swallowing instructions, particularly in the early recovery phase [28,29]. Sensory processing disturbances may further affect bolus control and swallowing responses [9]. In addition, the temporal lobe is anatomically connected with regions involved in swallowing control, including the insula, frontal lobe, and brainstem; disruption of these networks may contribute to delayed or inefficient swallowing [9,30]. However, the present study did not include lesion volumetry, laterality-specific analyses, or instrumented swallowing measures, which limits the ability to determine the specific pathways underlying this association. In patients with TBI, common findings from videofluoroscopic swallow study (VFSS) demonstrated aspiration or penetration, reduced laryngeal elevation, and impaired epiglottic inversion [8]. These patterns suggest that dysphagia in patients with TBI involves both oral-phase impairments and pharyngeal dysfunction. As cognitive function improves, swallowing function may also recover [9], which may explain why temporal lobe hematomas were not significantly associated with dysphagia at 12 weeks post-TBI in our study. Interestingly, the type of hematoma (e.g., SDH, EDH, ICH, SAH, IVH) was not significantly associated with dysphagia outcomes in our study, consistent with findings from a previous study [8]. This contrasts with findings from stroke-related research, where ICH volume and the presence of IVH were key predictors of swallowing dysfunction [31,32]. The absence of such an association in our TBI cohort may be attributed to the diffuse and multifocal nature of brain injury in TBI [9], in contrast to the more localized lesions seen in stroke [32]. Swallowing deficits in TBI are often the result of widespread cortical and subcortical damage rather than the effects of a single focal hematoma [16]. This finding demonstrates the intricate relationship between injury mechanisms, lesion distribution, and dysphagia severity in patients with TBI.
Our study demonstrated a significant association between impaired sitting balance and prolonged post-TBI dysphagia. In patients with TBI, early structured interventions are essential for improving overall functional recovery and facilitating swallowing rehabilitation [33]. Sitting balance evaluation provides crucial insights into trunk stability, which is a key determinant of motor and functional recovery [34]. Beyond postural control, sitting balance may also represent an integrated functional marker encompassing arousal, attention, and motor coordination, all of which are essential for safe and effective swallowing [35]. Adequate trunk stability supports optimal head and neck alignment and facilitates coordinated activation of swallowing-related musculature, potentially contributing to improved airway protection [36]. Furthermore, previous studies have identified FIM scores recorded at admission as reliable predictors of oral feeding recovery throughout the rehabilitation process [37].
These findings emphasize the need for early functional assessments and personalized rehabilitation plans to optimize motor recovery and support dysphagia rehabilitation in individuals with TBI.
Interestingly, GCS score, a commonly used predictor of TBI severity, was not independently associated with dysphagia outcomes in our study. This differs from previous research, where lower GCS scores were linked to prolonged dysphagia [4,14,15]. The discrepancy may be due to differences in study populations, as our cohort included 52.5% patients with mild TBI (GCS 13–15), whereas prior studies predominantly examined patients with moderate-to-severe TBI [4,14,15]. In our healthcare system, patients with mild TBI may still be admitted for inpatient rehabilitation when they present with clinically significant functional impairments, such as dysphagia, cognitive-communication deficits, or balance limitations. Therefore, although classified as mild by GCS, many of these patients had substantial functional limitations that warranted inpatient rehabilitation. This suggests that GCS alone may not fully capture the complexity of swallowing impairment in a heterogeneous TBI population. At the same time, our severity-stratified analyses demonstrated a clear gradient in early swallowing outcomes, with higher NG tube dependence and worse FOIS scores at 1 week in moderate and severe TBI, and attenuation of these differences by 12 weeks. Taken together, these findings indicate that injury severity influences early dysphagia, but its effect is partially mediated or modulated by other clinical factors over the course of recovery. Consistent with this interpretation, prolonged ICU stays, complications of immobility, and extended hospitalization were associated with worse dysphagia outcomes in our multivariable models. These findings suggest that post-TBI dysphagia may be associated not only with the severity of the initial injury but also with ventilation dependence, tracheostomy, surgical interventions, and secondary complications [5,8,15]; however, some of these factors may also arise as consequences of persistent dysphagia or a prolonged recovery course rather than acting as independent predictors. In our cohort, tracheostomy and neurosurgical procedures occurred predominantly among patients with moderate-to-severe TBI, as expected, whereas prolonged dysphagia was less common in mild TBI but could still be observed in selected individuals with associated injuries, cognitive-communication deficits, or medical complications. Therefore, TBI-related dysphagia results from a combination of cognitive, structural, and medical factors rather than being solely determined by the initial injury severity.
In our study, EVD placement, craniectomy, seizures, and the presence of hydrocephalus were significantly associated with poorer swallowing outcomes in univariate analysis but did not remain significant in the multivariate model. This suggests that while these factors may contribute to dysphagia, their effects are likely mediated by other underlying variables, such as the severity of neurological impairment, CT findings, cognitive dysfunction, ventilation status, and associated complications [3,15,18]. Previous studies have highlighted that severe neuroimaging abnormalities requiring surgical intervention, such as midline shift, extensive hemorrhage, or brainstem involvement, are associated with an increased risk of dysphagia in patients with TBI [18]. The need for EVD placement and craniectomy often arises from low GCS scores and increased intracranial pressure [38], both of which reflect severe brain injury rather than serving as independent predictors of prolonged swallowing dysfunction. Similarly, hydrocephalus impairs cognition and motor control, and may disrupt corticospinal and corticobulbar tracts, which play a crucial role in coordinating swallowing reflexes and airway protection [39]. Seizures have also been reported as a risk factor for post-TBI dysphagia, likely due to their association with cortical excitability changes, transient hypoxia, and disruptions in sensorimotor control [40]. Additionally, patients experiencing recurrent seizures often require prolonged sedation, mechanical ventilation, and ICU stays, exacerbating oropharyngeal muscle deconditioning and aspiration risk [41]. While EVD placement, craniectomy, seizures, and hydrocephalus were associated with dysphagia in univariate analysis, their lack of significance in multivariate analysis suggests that their impact on swallowing function is mediated by the overall severity of brain injury and prolonged hospitalization.
Airway management and critical care–related factors were also associated with swallowing outcomes in this study. Prolonged ETT placement, tracheostomy during hospitalization, and sedative medication use in the ICU were associated with NG tube retention as well as poorer FOIS outcomes at specific time points. These factors may contribute to dysphagia through mechanisms such as laryngeal trauma, reduced sensory input, impaired airway protection, and deconditioning related to prolonged critical illness [42,43]. However, given the retrospective design, these variables may also represent markers of greater injury severity or more complicated clinical courses rather than independent causal factors.
Regional differences in enteral feeding practice should be considered when interpreting our findings. In many Western centers and in gastroenterology guidelines, NG tubes are typically used for short-term enteral access, with transition to percutaneous devices (e.g., gastrostomy or jejunostomy) recommended when feeding is expected to exceed approximately four weeks [44]. By contrast, prolonged NG feeding remains more common in many Asian centers, including our facility, due to a combination of anticipated swallowing recovery within the subacute period, temporary contraindications to percutaneous access, institutional protocols, and patient or family preferences [31]. Consistent with this pattern, no patients in our cohort underwent gastrostomy tube placement during the study period, and we therefore used NG tube retention as an indicator of prolonged dysphagia.
Our study has several limitations. First, as a retrospective, single-center study, it is susceptible to selection bias and missing data, which may have influenced our findings. Second, although clinically relevant key covariates were selected based on prior literature, the multivariable analyses incorporated a stepwise selection procedure. Stepwise modeling may be sensitive to sample-specific variation and is associated with risks of multicollinearity, model instability and overfitting; therefore, the identified associations should be interpreted as exploratory and associative rather than causal. Third, the limited number of patients restricts the generalizability of our results to broader TBI populations, and the results are most applicable to similar inpatient rehabilitation settings; generalizability to acute-care, outpatient, or other healthcare systems may be limited. The case mix, including a substantial proportion of mild TBI, may also influence effect estimates. Another important limitation is the inconsistent availability of standardized instrumental swallowing assessments, such as VFSS and fiberoptic endoscopic evaluation of swallowing. As a result, we relied primarily on clinical evaluations, NG tube dependency, and FOIS outcomes, which may underestimate the true prevalence and severity of dysphagia and limit detailed characterization of swallowing physiology. Future research should incorporate prospective, multicenter studies with standardized swallowing assessments to enhance the accuracy, validity and clinical applicability of these findings.
In conclusion, our study identified key factors associated with dysphagia recovery in patients with TBI, including age, ICU length of stay, initial RLAS score, initial sitting balance, complications of immobility, hospitalization duration, and temporal lobe hematomas. Early assessment and targeted rehabilitation are essential to improve swallowing outcomes and guide clinical management.