Oro-esophageal feeding for tracheostomized patients with severe traumatic brain injury: a randomized controlled trial Authors: Dandan Lv, Hongyan Liu, Jing Zhao, Yang Liu, Qianqian Li, Xing’na Zhao, Liugen Wang, Xi Zeng, Yu Ding Published: 2026 ABSTRACT AbstractBackgroundThis study reports the clinical effects of intermittent oro-esophageal tube feeding (IOE) versus nasogastric tube feeding (NGT) on nutritional status, aspiration pneumonia, decannulation, and level of consciousness in tracheostomized patients with severe traumatic brain injury (sTBI).MethodsA randomized controlled trial was conducted between March 2024 and October 2025 in China and included tracheostomized patients with sTBI. Participants were randomized 1:1 to the intervention and control groups for 28-day interventions. IOE or NGT was used for nutritional supports, respectively. The primary outcome was nutritional status, including hemoglobin, albumin, prealbumin and body mass index. The secondary outcomes included aspiration pneumonia, decannulation, and level of consciousness assessed using the Glasgow Coma Scale (GCS). Generalized linear mixed-effects models, generalized estimating equations, and Cox regression were used for data analyze.ResultsA total of 104 participants were included in the analysis. After intervention, significant interaction effects were observed in hemoglobin (β = 5.272, 95% CI: 2.707, 7.837), albumin (β = 3.675, 95% CI: 1.854, 5.496), prealbumin (β = 11.835, 95% CI: 6.623, 17.047), body mass index (β = 1.719, 95% CI: 0.868, 2.569), the GCS (β = 0.981, 95% CI: 0.572, 1.390), and aspiration pneumonia (OR= 0.304, 95% CI: 0.133, 0.693). The Cox model revealed that group significantly influenced the decannulation outcomes [HR (95% CI) =5.556 (3.197, 9.657), p < 0.001].ConclusionsIn tracheostomized patients with sTBI who received routine treatment, IOE is more conducive to decannulation and the improvement in nutritional status, aspiration pneumonia, and level of consciousness than NGT.Clinical trial registrationProspectively registered at ClinicalTrials.gov (NCT06328985, 03/18/2024, clinicaltrials.gov/study/NCT06328985).Keywords: Traumatic brain injury, intermittent oro-esophageal tube feeding, dysphagia, tracheostomy, rehabilitation, NGT, nasogastric tube Introduction Traumatic brain injury is a leading cause of disability and death around the world, affecting millions of individuals each year [1]. Based on severity, it is classified into three levels: mild, moderate, and severe. Severe traumatic brain injury (sTBI) refers to extensive skull fractures, widespread cerebral contusions, brainstem injury or intracranial hematoma, with a coma lasting more than 6 h, or worsening of consciousness within one day after injury followed by another coma lasting more than 6 h [2]. sTBI accounts for approximately 8% of all traumatic brain injury cases and has been associated with a mortality rate over 20% and a disability rate exceeding 50% [3,4]. Furthermore, survivors of sTBI commonly experience impairments in cognitive function, movement, and communication [1], limiting their activities of daily living and posing challenges for rehabilitation [5]. Additionally, a notable proportion of patients with sTBI, around 10%–15%, may enter a prolonged coma or vegetative state, further complicating their reintegration into society and placing burdens on families and society [6]. After sTBI, patients typically experience a series of physiological stress responses due to brain tissue damage. This can cause abnormal metabolism, hormonal changes, and inflammatory reactions, resulting in a chronic hypermetabolic state [1], which can lead to significant protein and nutrient loss in a short period [7]. Additionally, tracheotomy is commonly performed after the injury to improve ventilation function and maintain respiratory health [8]. However, prolonged placement of tracheotomy tube is associated with an increased risk of life-threatening complications, such as malnutrition and pulmonary infections [9]. Therefore, early removal of the tracheostomy tube is crucial for patients with sTBI where conditions permit [2]. Nutrition support for patients with sTBI is divided into enteral and parenteral nutrition, with enteral nutrition primarily including tube feeding and gastrostomy [10,11]. In Eastern Asia, due to the invasiveness of gastrostomy, tube feeding is the preferred option [12]. While these tube feedings can address nutrition issues in short term, their compliance is poor and can affect esophageal function. Additionally, long-term placement of feeding tubes increases the risk of pharyngitis [13]. These negative effects not only affect the effectiveness of tube feeding, but also further hinder the removal of the tracheostomy tube [14]. Recently, IOE, a novel and well-established enteral nutrition mode, has demonstrated its effectiveness in the management of dysphagia [15]. When feeding, the tube is inserted via the mouth and removed promptly after feeding, reducing persistent irritation and compression on the pharynx and esophagus [16]. The tube only reaches the upper esophagus rather than entering the stomach, thus not affecting esophageal function and potentially reducing the risks of reflux, vomiting, and pneumonia [17]. Therefore, IOE has the potential to deliver a relatively large amount of nutrients. Additionally, it does not interfere with the patient’s rehabilitation training, nor does it irritate the nasal passage, thereby potentially facilitating the removal of the tracheostomy tube [18]. Despite these speculations, the clinical effects of IOE among tracheostomized patients with sTBI are still unclear. The current trial, therefore, was carried out to investigate the clinical effects of IOE versus nasogastric tube feeding (NGT) in tracheostomized patients with sTBI. Method Study design This was a randomized controlled study, including tracheostomized patients with sTBI transferred to the inpatient rehabilitation unit from intensive care unit in China between March 2024 and October 2025. Participants were 1:1 randomized to intervention and control groups. The intervention period lasted 28 days for each patient. Based on routine treatments, participants were fed via IOE or NGT, respectively. The study was approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University (Ethic number: 2024-KY-1323), prospectively registered at ClinicalTrials.gov (NCT06328985, 03/18/2024, clinicaltrials.gov/study/NCT06328985), and conducted according to the guidelines in the Declaration of Helsinki. Informed consent was signed and obtained from patients’ family members. Participants Patients were invited according to the following criteria: (1) age over 18 years; (2) sTBI confirmed through imaging materials [19]; (3) The Glasgow Coma Scale (GCS) <8 points [20]; (4) stable vital signs; (5) tracheostomized; (6) enteral nutritional support was required; (7) duration of ICU stay 14–28 days. The exclusion criteria were: (1) inability to complete the therapy or assessment because of personal factors; (2) other intracranial lesions; (3) other conditions that might cause consciousness impairment; (4) contraindications for tube feeding; (5) Serious impairment of hepatic, renal, cardiovascular or metabolic function. Especially, participants would drop out in the middle of the study if (1) ones experienced severe intolerance, (2) ones need to switch the nutrition support mode, (3) there was the severe deterioration in condition, or (4) family members voluntarily withdrew. After obtaining ethics approval and clinical registration, we published the study introduction in the hospital’s ICU and official website. Interested patients obtained further information by contacting the staff to decide on participation. Public or patients’ family members did not participate in the protocol development. Data were collected in the rehabilitation department and medical record system. Intervention Both groups received routine treatments, including pharmacological treatment and rehabilitation therapy [1]. Pharmacologically, participants routinely received neurotrophic drugs, drugs for blood pressure and blood sugar control accordingly. Corresponding antibiotic treatment was conducted according to sputum culture, blood culture, urine culture, and other test results. The rehabilitation therapy included (a) passive exercise therapy to improve muscle contracture, joint function, spasms, and blood circulation, once a day, 60 min per session; (b) acupuncture, once every two days, 30 min per session; and (c) others, such as back patting and position changes conducted by caregivers. Participants from the intervention group were fed via IOE, of which the procedure was consistent with previous studies [16]. During feeding, participants were kept in a semi-recumbent posture and the tube was inserted through the mouth. In cases where patients had trouble opening the mouth, a dental pad was used prior to feeding. There are scale marks every five centimeters on the outer side of the tube. Once the tube was inserted approximately 30 cm from the lips, the placement check was to be done as follows: (1) The tube was gently rotated, and presence of the patient’s cough indicated that the tube was misplaced in the airway. (2) The distal end of the tube was put under water while presence of any emerging bubbles meant that the tube was misplaced in the airway. The tube was promptly removed after each meal. The tube was promptly removed. The intervention group participants underwent scheduled IOE three-five times a day, with up to 500 ml of nutrients for each feeding [21]. Water was also given according to the situation. Participants from the control group were fed via NGT based on the NGT consensus [22]. The feeding tube was fixed on the user’s face and the replacement of tubes was conducted every week. Feeding was conducted every two-to-three hour and each feeding amount was limited to <200ml. We set identical nutritional standards for all participants: The daily calorie intake was set between 25 and 35 kcal/kg of body weight and the proportion for protein intake was 0.8–1.2g/kg [23]. Adjustments were made according to the patient’s condition, and the feeding content was arranged by a professional nutritionist. In daily clinical practice, both NGT and IOE can be conducted without having staff present during the feeding. In this study, all feeding processes were conducted by trained therapists to minimize biases. Assessment Nutritional status: (primary outcomes) The nutritional status of the patients was assessed using hemoglobin (Hb), albumin (Alb), prealbumin (PA) and body mass index (BMI). The assessment was conducted in the morning before the first feeding, on Day 1 and Day 28, respectively. Aspiration pneumonia Occurrence of aspiration pneumonia in patients throughout the entire study period was tracked and the prevalence on Day 1 and Day 28 was compared. The diagnostic criteria were as follows: When a patient presented with persistent low-grade fever, abnormal breath sounds (rales, wheezes, and friction rubs) in the lungs, etc., a computed tomography was then conducted to check if there was a gravity-dependent distribution of shadows. The diagnosis was further confirmed through blood tests and sputum if necessary [24]. Decannulation of tracheostomy tube The duration of tracheostomy condition and decannulation outcomes were recorded for each participant. A successful tracheostomy tube removal was defined using the following criteria: the patient showed stable, regular breath and no significant pulmonary complications during the capping trial, such as blood oxygen declines or CO2 retention [25]. Only the patient who did not require re-tracheostomy within 72 h after decannulation would be recorded as successful decannulation. The patients who required require re-tracheostomy after 72 h after decannulation would be separately recorded. Only the successful decannulation cases were included in the records of duration of tracheostomy tube placement. Because including patients who did not successfully decannulated would make it difficult to present the differences in decannulation timing between the two groups. Level of consciousness The level of consciousness was evaluated using the GCS [20]. The scale consists of three components: eye response, verbal response, and motor response. The total scores could range from 3 to 15. The higher scores indicate clearer consciousness. Sample size The sample size was calculated using PASS 2018. The data were extracted from a previous study [26]. Based on the results, we conducted sample size estimations respectively and ultimately chose the largest anticipated sample size to ensure reliability of this study. In this case, M1 = 174.68, M2 = 183.57, S = 14.87, α = 0.05, (1 − β) = 80%, the initial anticipated sample size was 43 for each group. The dropout rate was set at 20% and there should be at least 42 (43*120%) cases in each group. Randomization After enrollment, patient information was uploaded to the database. Their names were replaced with automatic code. The randomization was conducted using a block generated by computer. The entire randomization process was performed by an independent staff, and the grouping information was strictly confidential. SPSS 21.0 was used for randomization. Blinding and details The research staff were instructed to minimize contact with the participants and maintain strict confidentiality regarding the details of the study. Only the assessors were blinded. A specialized group, of which the members were not involved in the intervention and were not aware of the research design, was responsible for collecting the data. All the assessments were conducted by this group to minimize subjective bias. All intervention-related personnel received strict training before the start of the study to ensure appropriate implementation of interventions. Statistical analysis The statistical analysis was based on the intention-to-treat principle. The Kolmogorov–Smirnov test or Shapiro–Wilk test were used to analyze the distribution of continuous data, where p > 0.05 was considered to indicate a normal distribution. Normally distributed continuous data were presented as mean and standard deviation. Continuous data that did not follow a normal distribution and ordinal data were presented as median and quartiles. Categorical data were presented as counts and percentages. The primary outcomes and level of consciousness were analyzed using generalized linear mixed-effects models, which included treatment arm as a covariate and a random effect for participant ID, and adjusted for the baseline value of the corresponding outcome, sex, age, and Rotterdam CT score. Aspiration pneumonia, being a binary outcome, was analyzed using logistic regression-based generalized estimating equations. The generalized estimating equations inherently include an analysis of baseline status and already considered baseline aspiration pneumonia. Therefore, the covariates included sex, age, and Rotterdam CT score. Effect estimates were reported according to model type. For continuous outcomes, regression coefficients represented between-group differences in change from Day 1 to Day 28 in the original measurement units. For aspiration pneumonia, coefficients from the logistic generalized estimating equations were on the log-odds scale, and odds ratios were calculated by exponentiating the coefficients. The group-by-time interaction term was used to estimate whether the change from Day 1 to Day 28 differed between groups. For continuous outcomes, the interaction coefficient represented the additional change in the intervention group compared with the control group over time. For aspiration pneumonia, the group-by-time interaction term represented the between-group difference in change in pneumonia status over time on the log-odds scale. Tracheostomy tube decannulation was analyzed using a Cox regression model to simultaneously account for the effects of the intervention on both survival time (duration of tracheostomy tube use) and survival outcome (whether decannulation was successfully achieved), adjusting for sex, age, and Rotterdam CT score. The proportional hazards assumption was tested to assess model validity, and the C-index was used to evaluate model fit. Age, sex, and Rotterdam CT score were selected as clinically relevant covariates because they may be associated with traumatic brain injury severity, nutritional recovery, pneumonia risk, consciousness recovery, and decannulation outcomes. The same covariate set was used across models to avoid data-driven covariate selection and to improve precision while preserving the randomized group comparison. No imputation was performed as the number of missing cases was small and dropout reasons were documented. Generalized linear mixed models, linear mixed-effects models, and generalized estimating equations can handle missing values. Mixed-effects models used all available repeated measurements under the missing-at-random assumption. For the generalized estimating equations and Cox models, participants with available baseline and follow-up information were included, and censoring was assumed to be non-informative. Participants’ decannulation status was censored at the time of loss to follow-up or at the end of the study. Sensitivity analyses were conducted to examine the robustness of the results. For nutritional indicators, level of consciousness, and tracheostomy tube decannulation, sex, age, and Rotterdam CT score were removed from the models. For aspiration pneumonia, probit regression-based generalized estimating equations were used. Since seven outcome measures were involved, the Bonferroni method was used to adjust the significance level to 0.008 for the result interpretation. SPSS 21.0 statistical was used for statistical analysis and p < 0.008 was set to determine statistical significance. Result Totally, 104 patients were enrolled in the study and six cases quitted the study midway. Specifically, two intervention group cases and one control group cases quitted the study because the patients were transferred to another hospital voluntarily. Another three control group cases changed their feeding modes during the study. All randomized participants were retained in the intention-to-treat analysis and contributed available data according to their randomized groups. Complete Day-28 data were available for 98 participants. A flowchart was shown in Figure 1. The baseline characteristics of the participants are shown in Table 1. Baseline data after excluding dropouts are shown in Appendix S1. Flowchart. Table 1.Baseline information of participants.VariablesThe intervention group (n = 52)The control group (n = 52)Age (years, x¯ ± s)40.47 ± 12.3143.25 ± 11.26Gender [n (%)]Male40 (76.92)36 (69.23)Female12 (23.08)16 (30.77)Ethnics [n (%)]Han50 (96.15)52 (100.00)Others2 (3.85)0 (0.00)Causes of injury [n (%)]Car accident22 (42.31)16 (30.77)Falling20 (38.46)24 (46.15)Others10 (19.23)12 (23.08)Smoking history [n (%)]Yes41 (78.85)33 (63.46)No11 (21.15)19 (36.54)Alcohol history [n (%)]Yes43 (82.69)38 (73.08)No9 (17.31)14 (26.92)Intracranial injury [n (%)]Acute subdural hematoma with cerebral contusion17 (32.69)15 (28.85)Acute intracerebral hematoma with cerebral contusion and laceration13 (25.00)9 (17.31)Extensive cerebral contusion15 (28.85)17 (32.69)Multiple intracranial hematoma7 (13.46)11 (21.15)Rotterdam CT score [n (%)]I–II14 (26.92)11 (21.15)III–IV21 (40.38)19 (36.54)V–VI17 (32.69)22 (42.31)Neurosurgical procedures [n (%)]Evacuation of hematoma18 (34.62)15 (28.85)Craniotomy17 (32.69)22 (42.31)Craniectomy11 (21.15)16 (30.77)External ventricular drain38 (73.08)34 (65.38)Ventriculoperitoneal shunt4 (7.69)6 (11.54)Hypertension33 (63.46)29 (55.77)Type 2 diabetes18 (34.62)14 (26.92)Hyperlipidemia24 (46.15)21 (30.83)Duration of ICU stay (days, x¯ ± s)19.52 ± 4.8821.01 ± 6.12Note: All data were based on the patient’s ID card and medical records. If not applicable, self-reporting would be used instead. Baseline information of participants. Note: All data were based on the patient’s ID card and medical records. If not applicable, self-reporting would be used instead. Results of the normality test are shown in Appendix S2. Generalized linear mixed models showed significant interaction effects for Hb (β = 5.272, 95% CI: 2.707, 7.837), Alb (β = 3.675, 95% CI: 1.854, 5.496), PA (β = 11.835, 95% CI: 6.623, 17.047), BMI (β = 1.719, 95% CI: 0.868, 2.569), and GCS (β = 0.981, 95% CI: 0.572, 1.390), as shown in Table 2. These significant group-by-time interactions indicated that the improvements from Day 1 to Day 28 were greater in the intervention group than in the control group. The likelihood ratio test based on the chibar2 statistic was significant (p < 0.001), supporting the use of a mixed-effects model. Table 2.Generalized linear mixed models for nutritional status and level of consciousness.OutcomesTimeEffectThe intervention groupThe control group β 95% CI p Alb (g/L, x¯ ± s)Day 1 34.66 ± 7.6335.30 ± 5.77 Day 28 41.27 ± 4.4238.22 ± 3.55 Treatment arm 1.148−0.822, 3.1180.253 Time 2.9271.631, 4.224<0.001*** Interaction 3.6751.854, 5.496<0.001***Hb (g/L, x¯ ± s)Day 1 102.39 ± 17.01100.31 ± 12.45 Day 28 119.10 ± 11.16113.04 ± 10.53 Treatment arm 2.478−2.469, 7.4250.326 Time 12.71910.143, 15.294<0.001*** Interaction 5.2722.707, 7.837<0.001***PA (mg/L, x¯ ± s)Day 1 175.69 ± 33.49170.88 ± 29.16 Day 28 233.31 ± 30.43216.62 ± 26.70 Treatment arm 4.579−6.916, 16.0740.435 Time 45.74642.095, 49.398<0.001*** Interaction 11.8356.623, 17.047<0.001***BMI (kg/m2, x¯ ± s)Day 1 19.31 ± 2.3419.72 ± 1.86 Day 28 20.91 ± 2.3919.60 ± 3.19 Treatment arm 0.119−0.485, 0.7240.700 Time 0.351−0.594, 1.2950.467 Interaction 1.7190.868, 2.569<0.001***GCS [M (Q1, Q3)]Day 1 5.00 (4.00, 5.00)5.00 (5.00, 6.00) Day 28 10.00 (7.00, 10.00)9.00 (8.00, 9.00) Treatment arm 0.374−0.587, 0.8070.090 Time 3.6983.407, 3.989<0.001*** Interaction 0.9810.572, 1.390<0.001***Note: Alb: albumin; Hb: hemoglobin; PA: prealbumin; BMI: body mass index; GCS: Glasgow Coma Scale. *p < 0.05, **p < 0.01, ***p < 0.001. The adjusted covariates included treatment arm, age, sex, Rotterdam CT score, and the baseline value of the corresponding outcome. The random effect was participant ID. Prob ≥ chibar2 = 0.0000 for all models. Generalized linear mixed models for nutritional status and level of consciousness. Note: Alb: albumin; Hb: hemoglobin; PA: prealbumin; BMI: body mass index; GCS: Glasgow Coma Scale. *p < 0.05, **p < 0.01, ***p < 0.001. The adjusted covariates included treatment arm, age, sex, Rotterdam CT score, and the baseline value of the corresponding outcome. The random effect was participant ID. Prob ≥ chibar2 = 0.0000 for all models. There were eight new aspiration pneumonia cases from the control group and three from the intervention group. Once pneumonia was diagnosed, the patient received appropriate treatment. Throughout the entire course of treatment, no cases of infectious pneumonia occurred. After treatment, significant interaction effects were revealed in aspiration pneumonia (group-by-time OR = 0.304, 95% CI: 0.133–0.693), as shown in Table 3. This significant group-by-time interaction indicated a greater reduction in aspiration pneumonia over time in the intervention group than in the control group. During the treatment, there were 43 cases of successful decannulation in the intervention group while this number in the control group was 25. During the study, there was no patients who required re-tracheotomy after 72 h of decannulation. Among those successful decannulation cases, the duration of tracheostomy tube placement was (14.92 ± 3.61) d and (21.27 ± 5.80) d in the intervention and control groups, respectively. The Cox model showed that group significantly influenced the decannulation outcomes [HR (95% CI) =5.556 (3.197, 9.657), p < 0.001], as shown in Table 4. The p-values of all variables in the proportional hazard assumption test were >0.05, indicating that the Cox regression models were applicable, and the analysis results were reliable. The survival curves for tracheostomy decannulation are shown in Figure 2. Table 3.Generalized estimating equations for aspiration pneumonia.OutcomesTimeEffectThe intervention groupThe control group β 95% CIOROR 95% CI p Aspiration pneumonia (n, %)Day 1 40 (76.92)37 (71.15) Day 28 18 (50.00)31 (73.80) Group 0.301−0.580, 1.1821.3510.560, 3.2610.503 Time −0.308−1.133, 0.5180.7350.322, 1.6790.465 Interaction −1.192−2.019, −0.3660.3040.133, 0.6940.005**Note: *p < 0.05, **p < 0.01, ***p < 0.001. The adjusted covariates included treatment arm, age, sex, Rotterdam CT score. Coefficients are presented on the log-odds scale; odds ratios can be obtained by exponentiating the coefficients. Generalized estimating equations for aspiration pneumonia. Note: *p < 0.05, **p < 0.01, ***p < 0.001. The adjusted covariates included treatment arm, age, sex, Rotterdam CT score. Coefficients are presented on the log-odds scale; odds ratios can be obtained by exponentiating the coefficients. Table 4.Cox model of decannulation.ModelItemRegression coefficientStandard error p Hazard ratioHR 95% Confidence IntervalPH assumption χ2PH assumption p-valueIGroup1.2450.256<0.001***3.4742.104, 5.7350.7300.393IIGroup1.7150.282<0.001***5.5563.197, 9.6571.9850.159Sex−0.3690.2760.1820.6910.402, 1.1880.5980.439Age−0.7000.142<0.001***0.4970.376, 0.6562.3840.123Rotterdam CT score−0.7510.208<0.001***0.4720.314, 0.7091.7300.188Note: C-index for model I = 0.791; C-index for model II = 0.867. PH: proportional hazards. The adjusted covariates included treatment arm, age, sex, Rotterdam CT score. ***p < 0.001. Cox model of decannulation. Note: C-index for model I = 0.791; C-index for model II = 0.867. PH: proportional hazards. The adjusted covariates included treatment arm, age, sex, Rotterdam CT score. ***p < 0.001. Survival curves for tracheostomy decannulation. Since no invasive treatment was involved, no serious harm was caused to the patients. However, there were still some adverse events related to tube feeding. The common terminology criteria for adverse events were used to enhance clarity. As some patients had significantly reduced consciousness in the early stages of treatment and were unable to subjectively report certain events, such as reflux, all adverse events were recorded based on the following criteria: (1) Identified by family members or healthcare providers and further confirmed by researchers. (2) Subjectively reported by patients and further diagnosed by researchers. No severe adverse events occurred (grade >2), and all adverse events were summarized in Appendix S3. Sensitivity analysis showed that adjustment for covariates did not alter the statistical significance of the model results, which confirmed the robustness of our findings, as shown in Appendices S4 and S5. Discussion Patients with traumatic brain injury often experience an increased release of adrenal cortex hormones when in a comatose state or unable to eat. This can lead to prolonged stress responses and heightened breakdown of nutrients [27]. The condition significantly impacts nitrogen balance. Moreover, as the body works on repairing traumatic tissues, the energy demand is estimated to be 175% to 200% of the average level [1]. Inadequate or delayed nutritional support can potentially lead to a weekly depletion of approximately 15% of the patient’s energy reserves. Notably, previous research has shown a significant rise in mortality rates when energy reserves drop by more than 30% [28]. A review has concluded that nutraceuticals and supplements yields benefits in TBI management, which could be integrated into individualized enteral nutrition strategies in clinical practice [29]. Although tracheostomy can improve the survival rate of patients, it can result in a sustained reduction in immune function in the short term, exacerbating malnutrition, pneumonia, and increasing the risk of nosocomial infections [30]. Therefore, sufficient nutritional support and timely removal of the tracheostomy tube are of significance in promoting patients’ functional recovery, enhancing immunity, reducing the occurrence of complications, and ultimately lowering mortality rates. In clinical practice, hemoglobin, Alb and PA are commonly used as short-term nutritional indicators, and BMI is often chosen as an important indicator reflecting protein-energy malnutrition [31]. Significant outcome intergroup differences were found in the nutrition status. Although we set identical nutritional standards for patients, to prevent reflux, a relatively small amount of feed must be given in multiple times (totaling around 1,800 mL) for patients using NGT [22]. In contrast, IOE does not affect the closure of the cardia, which can reduce the risk of reflux [32,33]. This allows patients using IOE to have a higher single and daily average intake. According to the guidelines, the maximum daily average intake for IOE can reach up to 2,500 mL [34]. During clinical observation, although most patients could not express their appetite, the patients using IOE significantly accepted higher feeding volumes without adverse events. Therefore, under the condition of consistent absorption ability, IOE is more conducive to providing enough nutrients [32]. Secondly, studies have shown that regular diet can not only improve the absorption and metabolism of nutrients, but also reduce the occurrence of complications such as intestinal flora disorder, gastrointestinal inflammation, gastrointestinal ulcer, and intestinal stress diseases [35,36]. When using NGT, frequent small feeding is conducted with interval around 2 h. This obviously breaks the regularity of the daily diet. IOE was generally consistent with the three mealtimes of normal people, which is beneficial to the absorption of nutrients. Moreover, the patients using NGT often face complications such as reflux and pneumonia, which can increase the consumption of nutrients. These complications are associated with inflammation and can subsequently cause low-grade fever. This affects various digestive enzymes and poses a negative impact in digestion of nutrients [37]. By not affecting lower esophageal function, IOE can avoid these risks while improving the participation of the oropharynx and esophagus, thus beneficial to the absorption of nutrients. Feeding routes should be selected according to disease phase, tolerance, and nutritional targets. NGT remains convenient for short-term enteral feeding, whereas IOE may be considered when intermittent feeding is feasible and persistent tube-related irritation, reflux, or delayed decannulation is a concern. For patients requiring prolonged enteral support or poorly tolerating oral/transnasal tubes, gastrostomy or post-pyloric feeding may be considered; parenteral nutrition may be used temporarily or supplementally when enteral targets cannot be safely achieved. Route selection should be reassessed with dietitian-led adjustment and monitoring of reflux and aspiration. Controlling aspiration pneumonia is beneficial for patients’ prognosis [11]. In this study, IOE showed advantages in reducing aspiration pneumonia. When feeding using NGT, the tube is inserted into the stomach, affecting the contraction function of the lower esophageal sphincter. This increases the risk of reflux [38]. Additionally, NGT requires persistent placement, resulting in continuous irritation to the posterior pharynx and esophageal wall. Reflux material may enter the trachea, thus increasing the risk of aspiration pneumonia [13]. In contrast, IOE does not interfere with esophageal function and the tube is removed after feeding, thus reducing the risk of aspiration pneumonia [15]. After tracheostomy, the air entering the airway is not adequately warmed and humidified. This can cause mucosa dryness and secretion retention, increasing the risk of bacterial infections [39]. Early decannulation can also reduce the risk of granulation tissue formation. The timepoint of tracheostomy is also considered to be associated with the duration of mechanical ventilation, length of hospital stay, and length of ICU stay [40]. Additionally, relevant studies have shown that approximately half of the bacterial species present in oropharyngeal secretions are the same as those causing pulmonary inflammation [30]. Therefore, reducing the duration of tracheostomy tube placement is an important rehabilitation goal for tracheostomized patients with pneumonia. The persistent placement of feeding tube can not only cause prolonged irritation to the nasopharynx but also affect the self-cleaning ability of the oral cavity, mucosal protection, and normal pH value [13]. In addition, NGT occupies one nostril persistently, which affects the patient’s respiratory function. Before decannulation, a tracheostomy tube capping trial is routinely performed, during which the patient needs to gradually adapt to breathing through the nose. This process is affected by NGT, which can lead to decannulation delay or even failure. In contrast, IOE ensures normal ventilation of the nasal and pharyngeal cavities while providing patients with relatively adequate nutrients. This can reduce the impact of persistent tube placement on the oral cavity and nasopharyngeal irritation. Therefore, IOE can not only facilitate the successful decannulation but also shorten the tracheostomy tube duration. Experience from acute traumatic spinal cord injury is relevant to our decannulation findings. In these patients, early tracheostomy has been associated with shorter mechanical ventilation, shorter ICU and hospital stay, and fewer pneumonia and tracheostomy-related complications, although a clear mortality benefit has not been established [40]. Although sTBI is a different condition, both patient groups require careful airway protection, secretion management, respiratory rehabilitation, and decannulation planning. In the acute phase of sTBI, elevated intracranial pressure is often managed with temporary hyperosmolar therapy, such as mannitol or hypertonic saline; recent evidence suggests that these agents have similar ICP-lowering effects, with hypertonic saline showing longer effect duration and shorter ICU stay in some analyses [41]. Our trial was conducted after ICU transfer, so the findings mainly apply to the rehabilitation phase, when nutritional support and airway management may jointly affect pneumonia control and decannulation. IOE appears to better improve patients’ level of consciousness than NGT. The main reason is speculated to be that IOE can provide more nutrients. Therefore, it can better meet the high energy consumption of hypermetabolic state, improve negative nitrogen balance, and enhance immunity. This can facilitate overall patient recovery, manifested as an improvement in the level of consciousness [42]. However, there is still no clear research evidence to reasonably explain how nutritional support improves neurological recovery effects. There is also no direct evidence to prove a connection between nutrition and consciousness levels. Additionally, imaging findings may help guide appropriate feeding methods. Severe esophageal dysmotility, significant gastroesophageal reflux, or structural upper gastrointestinal abnormalities visualized on imaging are relative contraindications to IOE or NGT [21]. In such cases, early placement of a gastrostomy tube is considered appropriate to ensure safe and sustained nutrition support. There are still some limitations in this study. Firstly, follow-up was not successfully conducted, limiting our research to exploring the short-term effects and lacking comparisons of long-term impacts. Secondly, due to the poor level of consciousness in the patients with sTBI, their ability to express themselves was restricted. Lastly, we did not set a percutaneous endoscopic gastrostomy group for comparison. In the future, a long-term trial with multiple groups should be conducted to resolve these limitations. Conclusion In conclusion, for tracheostomized patients with sTBI, IOE presents superior advantages over NGT in the improvement the patient’s nutritional status, prompt decannulation, aspiration pneumonia, and level of consciousness. However, further research should be conducted to explore the effects of IOE versus percutaneous endoscopic gastrostomy among this population. Supplementary Material Appendix.docx Funding Statement This research was funded by the Capital Health Development Research Special Project (Code: 2024-2-5032, Recipient: Yu Ding) and NHC Key Laboratory of Prevention and Treatment of Cerebrovascular Disease (Recipient: Xi Zeng) Ethics statement The study was approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University (Ethic number: 2024-KY-1323), prospectively registered at ClinicalTrials.gov (NCT06328985, 03/18/2024, clinicaltrials.gov/study/NCT06328985), and conducted according to the guidelines in the Declaration of Helsinki. Written informed consent was obtained from patients’ family members prior to the study. Consent for publication We have obtained the consent for publication from each participant. Disclosure statement No potential conflict of interest was reported by the author(s). Data availability statement The data which support this study are not publicly available but can be obtained from the corresponding author on reasonable request. References Associated Data Supplementary Materials Appendix.docx Data Availability Statement The data which support this study are not publicly available but can be obtained from the corresponding author on reasonable request.