Abstract
The early enteral nutrition feeding intolerance (FI) risk prediction model and corresponding scoring table developed in this study showed good predictive performance and could serve as a useful reference for the clinical assessment of FI risk in neurocritical patients.
Introduction
For patients in a Neurological Intensive Care Unit (NCU), early enteral nutrition (tube feeding) is a clinical priority to support recovery and immune function. However, these patients often suffer from Feeding Intolerance (FI)—a condition where the digestive system fails to process nutrients effectively, leading to complications like vomiting, diarrhea, or gastric reflux. In neurocritical patients, the "brain-gut axis" is often disrupted due to impaired consciousness or vagal inhibition, leading to FI incidence rates as high as 83%.
FI is not merely a nuisance; it is linked to prolonged mechanical ventilation, increased infection rates, and higher mortality. While general scoring systems exist, they often lack the specificity needed for neurological patients. This paper describes the development of a specialized, easy-to-use predictive model tailored for the NCU to identify high-risk patients before complications arise.
Methods and Study Population
The authors conducted a prospective cohort study involving 440 eligible patients admitted to the NCU at a Chinese tertiary hospital between April and December 2022. The study population was divided into a derivation cohort (n=300) to identify risk factors and weights, and a validation cohort (n=140) to verify the model's predictive power. The incidence of FI in the derivation cohort was notably high at 71%.
Variable Selection and Modeling
Researchers initially screened 27 potential risk factors. Using univariate analysis followed by binary logistic regression, they identified seven independent predictors of FI. To make the model practical for clinical use, the complex logistic regression equation was converted into a "Simplified Risk Assessment Scale." Each risk factor was assigned a point value (0, 1, or 2) based on its statistical weight, resulting in a total scale of 13 points.
Results
The model identified seven key variables that significantly increase FI risk:
- Age (≥60 years): Correlated with decreased gastrointestinal motility.
- GCS Score: Lower scores indicate severe consciousness impairment and gut dysfunction.
- APACHE II Score (≥20): Higher overall disease severity.
- Mechanical Ventilation: Can interfere with gastric emptying.
- Feeding Route: Nasogastric tubes carried higher risk than nasoenteric tubes.
- Hyperglycemia (>11.1 mmol/L): High blood sugar slows gastric emptying.
- Hypoalbuminemia (<30 g/L): Low protein levels hinder absorption.
Model Performance
The original model achieved an Area Under the Curve (AUC) of 0.941 in the derivation cohort and 0.924 in the validation cohort, indicating excellent discriminatory power. In the validation phase, the model demonstrated a 96.0% sensitivity. The simplified 13-point scale determined that a score of 5 or higher indicates a "high-risk" patient, with an overall clinical accuracy of 84.3%.
Discussion
The study reinforces that post-pyloric feeding (nasoenteric) significantly reduces FI risk compared to traditional stomach feeding by bypassing gastric motility issues common in brain-injured patients. Interestingly, metabolic markers like serum albumin and blood glucose were as predictive as neurological scores, suggesting that FI is a systemic failure rather than a purely neurological one.
Limitations and Future Work
The study is limited by its single-center design and short-term focus. While internal validation was strong, future research should involve multi-center external validation to ensure the model's generalizability across diverse clinical environments and different ethnic groups.
Summary updated