An Exploratory Assessment on Differing Supplemental Feeding Methods and Associated Pediatric Neurorehabilitation Outcomes Authors: Jennifer Wu, Lea C. Mason, Patricia A. Orme, Stephanie E. Cohen, Brian D. Wishart, Bridgid M. O’Brien, Lynette Holmes, Katherine Barrett, A. Kaytee Welsh Published: 2026 ABSTRACT ObjectiveTo identify functional differences of continued nasogastric tube (NG) compared with new gastrostomy (GT) tube for enteral feeding support in pediatric neurorehabilitation patients admitted to inpatient rehabilitation (IPR).DesignRetrospective cohort study.SettingPediatric IPR facility.ParticipantsA total of 74 children and adolescents (N=74) (47 boys/27 girls, age mean ± SD=10.7±5.1y) with new neurologic impairments because of acquired brain injury and admitted to pediatric IPR with either NG or new GT tube for supplemental feeding.InterventionsStandard of care IPR treatment.Main Outcome MeasuresThe primary outcome measure was the change in International Dysphagia Diet Standardisation Initiative Functional Diet Scale (IDDSI-FDS) score at IPR admission and discharge. Secondary outcome measures included the cognitive components of the Functional Independence Measure in Children (WeeFIM) and body mass index.ResultsAt admission, rates of continued NG (51%) and new GT (49%) were similar. IPR treatment resulted in statistically significant improvements in eating function (IDDSI-FDS score median [interquartile range], 0 [0-2.25] to 8 [4.75-8] points; χ2=44.9; P<.001) and cognitive function (WeeFIM Cognition subtotal score mean ± SD=12.2±6.5 to 19.4±7.9 points, t=9.25, P<.001). For nil per os patients, admission with GT was associated with a higher discharge IDDSI-FDS discharge score (GT=7 [0-8] vs NG=5 [0.75-8]; χ2=11.1; P=.09) and a larger effect size in score change across all WeeFIM Cognition components. In contrast, NG was associated with lower body mass index at admission and lesser weight gain across IPR treatment.ConclusionsFeeding tube type at IPR admission may be correlated with functional outcome at discharge home in patients with new neurologic impairments because of acquired brain injury. Acquired brain injury is a leading cause of disability in children and adolescents, and nutrition is a critical component to their recovery.1,2 There is substantial variability in defining best practices pertaining to pediatric feeding tubes,3, 4, 5 particularly as it relates to prolonged nasogastric (NG) versus early gastrostomy (GT) enteral tube feeding. Both NG and GT are associated with their respective advantages and disadvantages. Although NG placement can be performed in many clinical settings, tube insertion is associated with high psychosocial distress,6 frequent dislodgement causes inadequate feeding,7 and repeated reinsertions can lead to eating difficulty.8 Conversely, GT offers increased stability but is more invasive to place and may be associated with postoperative complications, including skin irritation, wound infection, and leakage.9,10 There are wide-ranging functional implications associated with feeding tube type in pediatric brain injury. Although studies generally support GT feeding in conditions in which prolonged eating impairment is expected, comparisons between NG and GT have primarily focused on duration of treatment and anthropometric indicators of nutrition and growth. Inpatient rehabilitation (IPR) represents an ideal setting to compare a range of functional effects with respect to NG versus GT supplemental feeding. In general, pediatric patients with new neurologic impairments because of acquired brain injury demonstrate robust functional gains across IPR.11, 12, 13, 14 Furthermore, a significant proportion of IPR admissions require supplemental feeding support by either NG or GT.15 This study aimed to compare patients admitted with NG versus GT at IPR admission with respect to the: (1) eating function demonstrated by a standardized measure of functional diet; (2) cognitive function, including communication, expression, social interaction, problem solving, and memory; and (3) anthropomorphic markers of nutrition. We hypothesized that admission to IPR with GT would be associated with better functional outcomes in eating, cognition, and nutrition at discharge. Methods This retrospective cohort study included patients aged 3 to 21 years who were admitted to pediatric IPR at Spaulding Rehabilitation Hospital between November 1, 2018, and May 1, 2024, with new neurologic impairments because of an acquired brain injury. Potential subjects were identified through medical chart review of all admitted patients during the study period of interest. In total, 86 patients admitted to IPR were considered for study inclusion and had: (1) a new diagnosis of acquired brain injury; (2) either an NG or new GT at admission; and (3) required supplemental enteral feeding for basal nutrition and hydration. Ten patients were identified who had either an NG (n=9) or new GT (n=1) at IPR admission for medication administration, but were otherwise tolerating a regular texture, thin liquid diet and thus were outside of the scope of the current study. Two patients were excluded because of an interval IPR admission at a different facility after acute hospital discharge. The group also included 3 patients who had NG-to-GT conversion during IPR treatment. For these patients, only the initial IPR admission with NG was included in analyses. The final study cohort included 74 unique individuals. Patient data, including demographics, disease etiology, diet orders, functional measurements of cognition, and anthropometric metrics, were obtained from the medical records. Clinical variables of interest included patient sex, age, admitting rehabilitation diagnosis, height, weight, and feeding tube status. Eating function was derived from diet orders which were classified according to the International Dysphagia Diet Standardisation Initiative (IDDSI) framework.16 To support statistical analysis, the IDDSI level for solid and liquids was then converted to an IDDSI Functional Diet Scale (IDDSI-FDS) score,17 which is an ordinal scale ranging from 0 to 8, where nil per os (NPO) is ranked “0” and regular solid and thin liquid oral diet is ranked “8.” Height and weight were converted to body mass index (BMI) and age-corrected BMI percentile for each patient using pediatric growth charts published by the Centers for Disease Control.18 Cognitive function was assessed using the cognitive subdomain of the Functional Independence Measure in Children (WeeFIM), which is commonly used to track functional changes in the pediatric IPR setting.11,19 The cognitive components of the WeeFIM include 5 items (Comprehension, Expression, Social Interaction, Problem Solving, and Memory) with each item scored on an ordinal scale ranging from 1 (Total Assistance) to 7 (Complete Independence).20 WeeFIM Cognitive components assessments were scored by trained pediatric speech therapists and recorded as part of standard clinical care operations on our pediatric IPR unit at admission and discharge.11,21 To account for age-expected differences in cognitive performance, WeeFIM scores were normalized to the age-expected maximum score for each item such that the maximum WeeFIM cognitive score was 35 across all patients.22 Statistics Descriptive statistics were used to characterize the patients. Statistical analyses were performed with MATLAB R2015a statistics toolboxa (Mathworks). Nonparametric methods were used for ordinal measures, including IDDSI-FDS and WeeFIM Cognitive component scores, whereas parametric methods were used for continuous measures, including WeeFIM Component subtotal score and BMI. Within-subject changes in function across IPR used Friedman’s test for nonparametric measures, paired t test for parametric measures, and Kendall w was used to compare effect sizes across groups. To assess functional differences for patients admitted to IPR with an NG compared with a GT, ordinal logistic regression was used for ordinal measures, while analysis of variance was used for continuous measures. All analyses were 2-tailed with statistical significance set at α=0.05. Study procedures were approved by the Mass General Brigham Institutional Review Board. Because of the retrospective study design, the study was exempted from written and verbal consent procedures. Results The group of 74 patients included 47 boys/27 girls with a mean ± SD age 10.7±5.1 years. Admitting diagnoses for IPR included traumatic brain injury (n=28), brain cancer (n=23), childhood stroke (n=15), and nontraumatic acquired brain injury (n=8). Diagnoses of nontraumatic acquired brain injury included autoimmune encephalitis/encephalopathy (n=3), hypoxic-ischemic brain injury (n=3), and infectious encephalitis (n=2). Acute hospital duration preceding IPR admission averaged 36.7±28.6 days, and the group subsequently completed an average of 43.2±29.1 days of IPR treatment. At admission, rates of NG (n=34, 46%) and GT (n=40, 54%) were similar (χ2=0.05; P>.4). Patients admitted with NG were younger compared with those admitted with GT (9.0±4.6 vs 12.2±5.0y; t=2.86; P=.006). For patients who were NPO at admission (IDDSI-FDS score=0), more patients had a GT (n=36) compared with an NG (n=16; χ2=7.89; P=.005). For all patients, IPR treatment resulted in statistically significant improvement in IDDSI-FDS score (fig 1A; median [interquartile range], 0 [0-2.25] to 8 [4.75-8] points; χ2=44.9; P<.001). At admission, patients with NG had a higher IDDSI-FDS score (NG=1 [0-5] vs GT=0 [0-0]), but this difference was not seen in discharge IDDSI-FDS scores (fig 1B). Among patients who were NPO at admission (n=52), GT was associated with higher IDDSI-FDS at discharge (fig 2; GT=7 [0-8] vs NG=5 [0.75-8]; χ2=11.1; P=.09).Fig 1International Dysphagia Diet Standardisation Initiative Functional Diet Scale (IDDSI-FDS) score at inpatient rehabilitation (IPR) admission and discharge for (A) all subjects and (B) comparing NG and GT subjects. Boxplots demonstrate group median (red line), interquartile range (gray box), 2.5 SDs (black whiskers), and outliers (red plus).Fig 1 dummy alt textFig 2International Dysphagia Diet Standardisation Initiative Functional Diet Scale (IDDSI-FDS) score at inpatient rehabilitation discharge for patients admitted NPO with NG versus GT. Boxplots demonstrate group median (red line), interquartile range (gray box), and 2.5 SDs (black whiskers); there were no outliers in these groups.Fig 2 dummy alt text International Dysphagia Diet Standardisation Initiative Functional Diet Scale (IDDSI-FDS) score at inpatient rehabilitation (IPR) admission and discharge for (A) all subjects and (B) comparing NG and GT subjects. Boxplots demonstrate group median (red line), interquartile range (gray box), 2.5 SDs (black whiskers), and outliers (red plus). International Dysphagia Diet Standardisation Initiative Functional Diet Scale (IDDSI-FDS) score at inpatient rehabilitation discharge for patients admitted NPO with NG versus GT. Boxplots demonstrate group median (red line), interquartile range (gray box), and 2.5 SDs (black whiskers); there were no outliers in these groups. The group showed strong gains in cognition across IPR treatment with WeeFIM Cognition component subtotal scores improving from mean ± SD=12.2±6.5 points at admission to 19.4±7.9 points at discharge (t=11.5; P<.001). Across the group, there was not a significant effect of tube type on WeeFIM Cognition subtotal score at admission or discharge (P>.8). However, in NPO patients, GT was associated with larger effect sizes in all WeeFIM Cognition components score changes across IPR treatment (table 1).Table 1Comparison of nasogastric tube (NG) versus gastrostomy tube (GT) groups for Functional Independence Measure in Children (WeeFIM) Cognition subtotal and component scores for patients who are admitted nil per os (NPO) at inpatient rehabilitation admission.Table 1 dummy alt textNGGTZ (GT-NG)PWeeFIM Cognition Subtotal score Admit9.7±1.616.5±2.11.01>.3 Discharge11.7±1.119.4±1.41.11>.2 t (discharge-admit)4.87.8 P.002<.001Comprehension Admit2.6 (1.2-3.4)3 (1-4)0.3>.8 Discharge3 (2.5-5.8)4.6 (3-6)1.3>.1 χ2 (discharge-admit)3.520.0 P>.06<.001 W0.220.56Expression Admit1 (1-2)1 (1-3)0.6>.5 Discharge3 (2-4)4 (2-5)1.1>.2 χ2 (discharge-admit)8.8220.8 P.03<.001 W0.550.58Social Admit1.6 (1-2.7)2 (1-3)1.0>.3 Discharge3.8 (2.2-5.9)5 (2.9-6)0.4>.60 χ2 (discharge-admit)5.919.4 P>.01<.001 W0.370.54Problem solving Admit1.2 (1-1.9)2 (1-2.8)0.7>.5 Discharge2.7 (1.2-2.7)4 (1.4-4.5)0.9>.3 χ2 (discharge-admit)5.717.4 P>.01<.001 W0.360.48Memory Admit1.3 (1.0-2)2 (1-3)0.7>.4 Discharge3 (1.4-4.5)3 (2-4)0.7>.4 χ2 (discharge-admit)6.014.3 P>.01.002 W0.380.40NOTE. The WeeFIM Cognition subtotal ranges 5-35 points, with each component (Comprehension, Expression, Social, Problem Solving, and Memory) scoring 1-7 points for dependent to independent level of function.W represents the Wilcoxon test statistic for nonparametric group testing. Table 1Comparison of nasogastric tube (NG) versus gastrostomy tube (GT) groups for Functional Independence Measure in Children (WeeFIM) Cognition subtotal and component scores for patients who are admitted nil per os (NPO) at inpatient rehabilitation admission.Table 1 dummy alt textNGGTZ (GT-NG)PWeeFIM Cognition Subtotal score Admit9.7±1.616.5±2.11.01>.3 Discharge11.7±1.119.4±1.41.11>.2 t (discharge-admit)4.87.8 P.002<.001Comprehension Admit2.6 (1.2-3.4)3 (1-4)0.3>.8 Discharge3 (2.5-5.8)4.6 (3-6)1.3>.1 χ2 (discharge-admit)3.520.0 P>.06<.001 W0.220.56Expression Admit1 (1-2)1 (1-3)0.6>.5 Discharge3 (2-4)4 (2-5)1.1>.2 χ2 (discharge-admit)8.8220.8 P.03<.001 W0.550.58Social Admit1.6 (1-2.7)2 (1-3)1.0>.3 Discharge3.8 (2.2-5.9)5 (2.9-6)0.4>.60 χ2 (discharge-admit)5.919.4 P>.01<.001 W0.370.54Problem solving Admit1.2 (1-1.9)2 (1-2.8)0.7>.5 Discharge2.7 (1.2-2.7)4 (1.4-4.5)0.9>.3 χ2 (discharge-admit)5.717.4 P>.01<.001 W0.360.48Memory Admit1.3 (1.0-2)2 (1-3)0.7>.4 Discharge3 (1.4-4.5)3 (2-4)0.7>.4 χ2 (discharge-admit)6.014.3 P>.01.002 W0.380.40NOTE. The WeeFIM Cognition subtotal ranges 5-35 points, with each component (Comprehension, Expression, Social, Problem Solving, and Memory) scoring 1-7 points for dependent to independent level of function.W represents the Wilcoxon test statistic for nonparametric group testing. Comparison of nasogastric tube (NG) versus gastrostomy tube (GT) groups for Functional Independence Measure in Children (WeeFIM) Cognition subtotal and component scores for patients who are admitted nil per os (NPO) at inpatient rehabilitation admission. NOTE. The WeeFIM Cognition subtotal ranges 5-35 points, with each component (Comprehension, Expression, Social, Problem Solving, and Memory) scoring 1-7 points for dependent to independent level of function. W represents the Wilcoxon test statistic for nonparametric group testing. At IPR admission, 16 patients (19%) had a BMI lower than the fifth percentile for age-expected values and were underweight. Among underweight patients, most were admitted with GT (GT=11/16; NG=5/16; χ2=2.3; P=.1). NG status at admission was associated with lower BMI (NG=13.2 [1.3] vs GT=14.8 [1.7]; t=2.0; P=.07) and lesser weight gain across IPR (NG =2.9[2.8] vs GT=5.1[4.3] kg; t=1.23; P=.2). NG was also associated with more frequent plain radiographs to check tube positioning (NG=1.42 [2.43], range, 0-10; GT=0.2 [0.4], range, 0-1; χ2=3.6; P=.004). Discussion Tube feeding dependence in the setting of pediatric acquired brain injury has wide-ranging functional implications. Although we were restricted by the presence of previously placed GT and NG tubes, the current study found that GT placement preceding IPR admission was associated with larger functional gains in oral feeding and cognitive function. Conversely, NG feeding in the IPR setting was associated with higher rates of low BMI and decreased weight gain. Together, the current results represent the first study to concurrently examine functional effects of early GT placement across feeding, cognition, and nutritional outcomes. This study also represents the first evaluation of the effects of early GT placement versus continued NG feeding in the pediatric IPR setting. Eating is a complex oral motor-cognitive function that is frequently impaired in the setting of new brain injury.23,24 For patients with brain injury and significant functional impairments, IPR is a setting in which early neurorecovery has compounding consequences for improved long-term outcomes.25 Therefore, the transition from acute care hospital to IPR represents a critical checkpoint for assessing the benefits and risks of GT placement versus prolonged NG nutritional support. Notably, the current study found that the acute hospital duration preceding IPR admission was 37.4 days, a timeframe that aligns with practice recommendations in the United Kingdom, which recommend prompt GT in patients requiring NG support longer than 4 weeks.3 Previous studies also report an association between prolonged NG exposure and increased rates of food refusal,8 which is also reflected in the current study in patients admitted with NG demonstrating slower gains in eating function. In the pediatric brain injury population, the sensory irritations and psychosensory deprivations associated with an NG’s presence in the nares and oropharynx have widespread effects on eating function.26 After brain injury, cortical-visual impairment can limit the patient’s ability to register food in front of them; disordered sensory perception causes difficulty processing food/fluid in the mouth; low arousal slows automatic swallow reflexes; and inattention can make patients highly distractible, leading to a slow rate of intake.27 External lines and tubes can also contribute to increased agitation and distractibility.28 In addition, the physical presence of an NG has been demonstrated to slow multiple elements of chewing and swallow function even among normal, control subjects.29 Therefore, in pediatric patients with new neurologic impairment and associated sensory processing disorders, the effects of feeding tube type on eating function are multifactorial and would ideally be addressed preceding transition to the IPR setting. For individuals with severe dysphagia, the neurologic impairments that limit eating also affect multiple areas of cognition. First, oral motor impairments which impede chewing and swallowing also impact speech articulation and expression.30 Sensory dysfunctions after pediatric brain injury disrupt the sensory processing of eating and impair the child’s ability to process their surroundings.31 In addition, high rates of concurrent neuro-irritability and impairments affecting self-regulation32,33 result in difficulty attending to the mechanics of eating but also have broad impacts on a child’s ability to engage in social interactions, problem solve, and integrate new memories.34 With respect to the current study results, patients who were feeding tube dependent and admitted to IPR with GT achieved larger functional gains across multiple cognitive domains compared with those admitted with NG requirement. These differences in cognitive outcomes in relation to tube type likely reflect compounding effects of the NG within the nares and oropharynx on sensory dysfunction, neuro-irritability, and impairments in self-regulation. Specifically, the presence of the NG tube impedes oral motor control,29 is associated with heightened sensory dysfunction,26 and pain and discomfort from the NG35 worsens irritability and self-dysregulation in the neurologically impaired child. In the IPR setting, these combined effects of prolonged NG dependence can lead to lost therapeutic minutes because of NG dislodgement interrupting therapy schedules, increased NG discomfort worsening neuro-irritability and limiting therapeutic engagement, and repeated NG replacements decreasing tolerance of oral stimulatory activities. Furthermore, increased activity demands in the IPR setting likely increase the risk of NG dislodgement and associated consequences of replacement. Compared with the general pediatrics population in the United States, in which an estimated 3%-7% of children and adolescents are underweight,36 a significant proportion of patients in the current study were underweight (19% with BMI <5th percentile for age). In the current group, the high percentage of underweight patients reflects both the prolonged acute hospitalization (mean=37.4d) and the high medical complexity of pediatric patients with acquired brain injury admitted to IPR. Significant weight loss because of increased metabolic demand is commonly observed in severe acquired brain injuries of childhood, with multiple studies demonstrating early and robust nutrition support improving overall outcomes.37,38 Previous studies have demonstrated that GT placement is associated with significant height and weight catch-up growth,39 and GT status in the current group was associated with increased weight gain during IPR. Furthermore, compared with NG feeding, GT feeding is linked to improved long-term health outcomes,40 reflecting the expansive consequences of malnutrition across all organ systems41, 42, 43 and highlighting the broad effects of nutritional inadequacies for pediatric patients with brain injury who are at high risk of nutritional vulnerability. Study limitations The current study is limited by its retrospective design, relatively narrow inclusion criteria, and nonrandom assignment of GT and NG tubes at IPR admission. As a result, the degree to which the current findings extend to pediatric rehabilitation patients without neurologic impairment or other diagnoses without brain injury is limited. Furthermore, institutional differences related to indications for GT placement, availability of pediatric GT placement services, and utilization of IPR programs also affect the interpretation of the current results. In the current study cohort, feeding tube type at IPR admission was not randomly assigned, instead reflecting clinical decision-making completed in the acute care setting. As such, the current findings are susceptible to confounding by indication, whereby children selected for GT may have differed systematically from those discharged with continued NG feeding in ways that also influence rehabilitation outcomes. Although we compared groups using regression-based approaches, confounding by unmeasured factors, such as granular indices of injury severity, airway status, preinjury developmental function, swallow study findings, or center-specific GT placement thresholds, may have biased estimates of the association between tube type and outcomes. Thus, the presently observed relationships between GT and NG status and functional gains should be interpreted as associated rather than causal, and highlight the need for prospective studies with standardized criteria to better isolate the independent effects of feeding tube type on neurorehabilitation outcomes. Regardless, brain injury remains a leading indication for admission to pediatric IPR. Conclusions As advancements in acute care medical and surgical interventions improve survival rates in children and adolescents with significant neurologic injury, clinical considerations that affect functional outcomes, such as early GT placement, may have systemic effects.44 Although the current study did not specifically evaluate psychosocial effects of feeding tube type, previous studies have generally found that the impact of GT placement is positive,45,46 with studies citing reduction in feeding times, increased ease of drug administration, and reduced concern about their child’s nutritional status as specific GT benefits. Rates of satisfaction after GT are generally very high with many parents reporting they wish the intervention had been completed earlier.45 Furthermore, considering the sensory dysfunctions and self-regulatory disorders associated with acquired pediatric brain injury, future studies should examine the functional consequences of feeding tube type on specific subgroups of patients with underlying sensory dysregulation: those who are developmentally immature in sensory integration47 (children<6y old), those who are neurodivergent,48 and those with baseline inattention or hyperactivity disorders.49 Supplier a.MATLAB R2015; Mathworks. a.MATLAB R2015; Mathworks. MATLAB R2015; Mathworks. Disclosure The investigators have no financial or nonfinancial disclosures to make in relation to this project.