Nutritional Risk Assessment and Its Prognostic Impact in Critically Ill Patients with Neurological Diseases

Summary

Here, we present a standardized protocol for nutritional risk assessment, nutritional support evaluation, and prognostic prediction in critically ill patients with nervous system diseases.

Abstract

The research aims to apply and evaluate a standardized protocol for nutritional risk assessment and nutritional support management, and to assess its predictive value in individuals with severe neurological disorders who are at risk of malnutrition. A retrospective analysis was performed on 930 suitable patients hospitalized in the Neurology Intensive Care Unit (NICU) at Xuanwu Hospital, Capital Medical University, between January 2012 and December 2019. At discharge, patients were categorized based on modified Rankin Scale (mRS) results (excellent prognosis: mRS 0-2, n = 600; bad prognosis: mRS 3-6, n = 330) and survival status (survivor: n = 869; deceased: n = 61). The protocol included baseline characteristics and biochemical evaluations, admission Nutritional Risk Screening 2002 (NRS 2002) scores, identifying the nutritional support modality, estimation of typical energy intake throughout the first 7 days, and level of disease evaluation using standardized assessment instruments such as the Glasgow Coma Scale (GCS), Acute Physiology and Chronic Health Evaluation II (APACHE II), and National Institutes of Health Stroke Scale (NIHSS).

Univariate studies, multivariate logistic regression estimation, and receiver operating characteristic (ROC) curve assessment were used to assess predictive significance. The standardized protocol enabled the systematic identification of nutritional risk, stratification of clinical severity, and early guidance for planning nutritional interventions. Key risk indicators, including elevated NRS 2002, APACHE II, and NIHSS scores, advanced age, and inadequate early energy intake, were associated with unfavorable outcomes. In contrast, higher GCS and adequate vitamin B12 levels demonstrated protective tendencies. ROC analysis confirmed that NRS 2002 provided the most effective predictive utility for both functional outcome and survival status. The findings demonstrate that structured nutritional risk assessment and standardized support implementation offer practical value for early prognostic judgment and Nutritional decision-making for severely ill individuals suffering from neurological diseases.

Introduction

Neurocritically ill patients are victims of a severe primary or secondary neurological illness that is often accompanied by dysfunction of systemic organs that necessitate continuous hemodynamic observation, mechanical ventilation, and a multifaceted therapeutic care in the Intensive Care Unit (ICU)^1,2. The patients, with impaired consciousness, dysphagia, mechanical ventilation, and poor spontaneous intake, often cannot engage in adequate nutritional intake, which results in a high rate of nutritional deficiency. Together with hypermetabolic and hypercatabolic stress reactions and physiological deterioration with age, these factors significantly enhance the risk of malnutrition^3,4. It has been seen that malnutrition is a factor that induces considerable protracted hospitalization, sluggishness in neurological functional recuperation, vulnerability to complications, and mortality in neurological ICUs⁵. Recent epidemiological evidence shows that malnutrition levels up to 79% exist among neurocritical patients, so there is a need to identify nutritional risks early and accurately.

Although nutritional assessment in general ICU populations has been studied, research specifically focusing on nutritional risk and its prognostic value in neurological ICUs has not been adequately explored⁶. The majority of the literature is devoted to individual types of diseases, like ischemic stroke, and often, nutritional evaluation is done at the time of admission and not on a dynamic basis on nutritional risk or standardized nutritional support planning⁷. Traditional evaluation tools, such as the Subjective Global Assessment (SGA) and the Malnutrition Universal Screening Tool (MUST), have biases and lack accuracy in constantly evolving neurocritical conditions⁸. The Nutritional Risk Screening 2002 (NRS-2002), on the other hand, integrates the level of decline in eating habits with the severity of disease, making it an additional suitable tool in the setting of seriously ill neurological conditions due to its organized scoring system and supporting function in medical decisions. Integrating the NRS-2002 with neurological severity scales, including the Glasgow Coma Scale (GCS), the National Institutes of Health Stroke Scale (NIHSS), and the Acute Physiology and Chronic Health Evaluation II (APACHE II), allows for a comprehensive evaluation of neurological health, overall health, and prediction⁹.

Previous research has investigated multifunctional feeding approaches to improve outcomes in individuals with severe neurological illnesses¹⁰, with increased tolerance and biomarkers resulting from coordinated interventions. However, the larger sample size, heterogeneity of conditions, and lack of long-term follow-ups have limited the ability to generalize the clinical results. Studies comparing the efficacy of structured nutritional screening in the ICU indicated that the use of validated instruments, such as MUST, yields better mortality forecasts and clinical performance¹¹, albeit with different patient characteristics, historical designs, and inconsistencies in intervention strategies. The practice of medical nutrition therapy in neuro-ICUs demonstrates significant variation in adherence to protocols, assessment strategies, and methods of monitoring¹². Even though the use of structured SOPs enhances consistency, little standardization, the use of insufficient calorimetry, and non-homogeneous implementation diminishes comparability and reliability of outcomes.

The Prognostic Nutritional Index has been shown to be a highly predictive measure of stroke outcomes in hospitalized patients, thereby enhancing mortality estimation and clinical decision-making¹³. Subtype variations, retrospective design limitations, and insufficient external verification, however, limit further application in a wide variety of neurological populations. There are significant relationships between better protein-related biomarkers and functional recovery with early Enteral Nutrition (EN) in acute ischemic stroke¹⁴. However, dependence on retrospective data, short-term observation, and incomplete nutritional inputs makes the causal conclusion and subsequent clinical extrapolation difficult.

Though long-term hospitalization, the poor neurological outcome, and high mortality rate in neurocritical care unit patients are linked to malnutrition, there is also a lack of data on the predictive value of organized dietary risk assessments in neurological ICUs. The existing literature primarily focuses on individual disease categories and conducts nutritional screenings at the point of admission, rather than dynamically monitoring them. Additionally, the subjective assessment measures, including SGA and MUST, lack sensitivity in a rapidly changing neurocritical setting¹⁰. There is minimal evidence to support the use of NRS-2002 in combination with neurological severity scales, including GCS, NIHSS, and APACHE II, to predict outcomes¹¹. Moreover, generalizability is limited by variations in nutritional therapy practices, heterogeneity of small samples, and short longitudinal follow-up¹³. That is why the prognostic evaluation of nutritional risk in neurocritical groups should be thoroughly done. Therefore, the purpose of this research was to test the predictive value of a combined nutritional risk evaluation regimen including NRS-2002, neurological severity scales (GCS, NIHSS, and APACHE II), and early energy intake monitoring to identify high-risk neurocritical patients and guide timely nutritional intervention.

Protocol

This protocol was carried out in accordance with the Declaration of Helsinki¹⁵ and approved by Xuanwu Hospital's Ethics Committee at Capital Medical University. All patients, as well as their legal guardians, provided informed permission.

1. Objects and methods

1. Define inclusion and exclusion criteria.
   1. Set inclusion criteria as follows: age ≥ 14 years, an established diagnostic of serious neurological disorders, a GCS score ≤ 12, or an NIHSS score ≥ 11.
   2. Set exclusion criteria: NCU treatment duration < 7 days, patients with advanced malignancies, patients with non-acute diseases, and those with a history of gastrectomy or bowel resection.
      NOTE: Although Parenteral Nutrition (PN) was available, patients with a history of gastrectomy or bowel resection were excluded because their altered gastrointestinal physiology could lead to inaccurate nutritional risk screening (e.g., NRS 2002) and potentially bias the study outcomes.
2. Screen and group patients.
   NOTE: The initial collection comprised 1,227 patients with serious neurological illnesses who were hospitalized between January 2012 and December 2019. Among them, 297 cases were excluded, including 254 with NCU treatment duration of less than 7 days, 23 with advanced malignancies, 18 with non-acute diseases, and 2 with a history of bowel resection. Finally, 930 patients were incorporated in the evaluation.
   1. Divide patients into two groups depending on their modified Rankin Scale (mRS) scores: good prognosis (mRS 0-2; n = 600) and bad prognosis (mRS 3-6; n = 330). Consider a discharged mRS score of 3-6 to be a poor outlook and scores of 0-2 to be excellent.

2. Clinical data collection

1. Collect demographic and basic clinical data.
   1. Record the patient's age, gender, disease category, comorbidities (coronary heart disease, hypertension, diabetes), drinking or smoking information, and length of hospitalization in the NCU.
   2. Document nutritional support mode: PN, EN, or combined EN + PN.
   3. Calculate average energy intake in the first 7 days: define it as the total energy intake over 7 days divided by the patient's actual body weight (kg), with the unit expressed as kcal/kg.
      NOTE: We used the 6^th edition (2018) of the China Food Composition Table to obtain energy values (kcal/100 g) of each food, added the total energy intake for 7 days, and divided it by the patient's actual body weight to get the average energy intake in the first 7 days.
2. Perform nutritional risk and gastrointestinal function assessment.
   1. Conduct NRS 2002¹⁶ within 24 h of admission: score disease severity (0-3), nutritional impairment (0-3), and add 1 point for age ≥70 years (total score 0-7; ≥3 indicates nutritional risk).
      NOTE: The NRS 2002 score was chosen because it is an unbiased screening measure recommended by ESPEN for patients in hospitals, allows for the rapid detection of patients at risk of malnutrition, and has already been thoroughly validated in ICU environments. Compared with other tools (e.g., CONUT, mNUTRIC), NRS 2002 incorporates disease severity and recent nutritional decline, making it more suitable for neurocritical patients in the early admission stage.
   2. Evaluate swallowing function using the Water Swallow Test (WST) and gastrointestinal function using Acute Gastrointestinal Injury (AGI) grading at admission. Perform the WST within 24 h of admission using the standardized five-grade scoring system. Instruct the patient to drink 30 mL of water and observe for coughing, choking, or dysphagia; higher scores indicate impaired swallowing safety. Assess gastrointestinal function using AGI grading (grades I-IV), based on tolerance to enteral nutrition, abdominal distension, vomiting, and bowel motility.
   3. Monitor gastric residual volume during hospitalization via gastric tube aspiration and bedside ultrasound to determine the appropriate nutritional support mode. Assess gastric residual volume through routine gastric tube aspiration, where gastric contents are withdrawn via a nasogastric tube. In cases where aspiration results are uncertain, perform bedside ultrasound as a non-invasive adjunct method to estimate gastric volume and tolerance to enteral feeding. Monitor gastric residual volume every 4-6 h via nasogastric tube aspiration; values > 250 mL indicate feeding intolerance and require adjustment of the EN strategy.
3. Assess disease severity using standardized scales.
   1. Administer the NIHSS¹⁷: evaluate 11 items (consciousness, gaze, motor function, etc.) with an overall score from 0 to 42 (greater scores imply more serious neurological dysfunction).
   2. Administer the GCS¹⁸: assess eye-opening, verbal response, and motor response (non-hemiplegic side) with an overall value of 3 to 15 (fewer values imply more serious awareness disruption).
   3. Administer the APACHE II¹⁹: score 12 physiological factors (the lowest results within 24 h of NCU admissions), age, and chronic medical history for an overall score of 0-71 (greater scores imply more serious disease).
4. Collect and test laboratory indicators.
   1. Obtain 5 mL of fasting venous blood the day after hospitalization.
      CAUTION: Use disposable vacuum blood collection tubes to minimize hemolysis. Wear gloves during sample handling to prevent biological contamination.
   2. Centrifuge the blood sample at 3,000 × g for 10 min at 4 °C to separate serum. Check that the serum layer appears clear and well-separated from the cellular phase; if hemolysis or turbidity is observed, repeat sampling.
   3. Detect serum indicators: protein markers (hemoglobin [HB], albumin [ALB], free triiodothyronine [FT3]), prealbumin [PA]), infection markers (C-reactive protein [CRP], liver/kidney function markers (total bilirubin [TBIL], coagulation markers (international normalized ratio [INR], thyroid function markers (thyroid-stimulating hormone [TSH], creatinine [Cr]), procalcitonin [PCT]), D-dimer [D-D], and vitamin B12).
   4. Dispose of all used blood sample tubes, gloves, and related materials following institutional biohazard waste management protocols to prevent contamination and ensure laboratory safety.

3. Data quality control and statistical analysis

1. Data quality control
   1. Have two trained neurologists independently extract data from electronic medical records.
   2. Cross-verify extracted data; resolve inconsistencies through review.
      NOTE: The review was done by a senior neurology professor.
   3. Handle missing data: impute missing values with the mean for indicators with <5% missing rate; exclude indicators with ≥5% missing rate.
2. Perform statistical analysis.
   1. Open SPSS 26.0 software, create a new dataset, and input data following the structure: 'patient ID - baseline data - assessment indicators - outcome indicators'.
   2. Conduct descriptive statistics:
      1. Evaluate distribution-normal continuous numbers (such as FT3) as mean ± standard deviation (x±s) using a samples-independent t-test.
      2. Expressing irregularly distributed continuous numbers (e.g., APACHE II score) as the median (25th percentile, 75th percentile) [M (P25, P75)]. and use the Kruskal-Wallis H test.
      3. Express categorical variables (e.g., gender) as count (percentage) [n (%)] and analyze with χ² test.
   3. Perform logistic regression analysis:
      1. Set dependent variables: poor prognosis (1 = poor, 2 = good) or death (1 = death, 2 = survival).
      2. In a univariate analysis, consider factors having a P-value < 0.05 as independent factors (continuous variables entered directly; categorical variables coded as dummy variables, e.g., nutritional support mode: PN = 1, EN + PN = 2, EN = 3 with EN as reference).
      3. Navigate to Analyze | Regression | Binary Logistic, add variables to the equation, select Enter method, and output β, standard error (SE), Wald χ², P-value, and 95% confidence interval (95% CI).
      4. To account for confounding factors, all covariates having P < 0.05 in univariate analysis were incorporated in the multivariate logistic regression models.
   4. Perform ROC curve analysis:
      1. Navigate to Analyze | ROC Curve, set the indicator to test (e.g., NRS 2002 score) and the outcome as the state variable (state value = 1).
      2. Check Display ROC curve, Compute area under the curve, and SE and confidence interval, then run the analysis to obtain AUC, optimal cut-off value, sensitivity, and specificity.
         NOTE: Dichotomize average energy intake in the first 7 days using the optimal cut-off value from ROC analysis for logistic regression; all analyses are two-tailed, and P < 0.05 indicates statistical importance.

Results

Comparison of general data between the two groups of data
Table 1 shows that patients with serious neurological illnesses have an adverse outcome due to variables including age, APACHE II score, GCS score, NIHSS score, dietary intake techniques, vitamin B12 levels, and albumin levels (P < 0.05).

Logistic regression analysis
The poor prognosis was used as the dependent variable (poor prognosis = 1, good prognosis = 2). Factors with P...

Discussion

Previous research has studied the relationship between dietary habits and prognosis in general ICU populations; very limited studies have specifically focused on critically ill neurological patients. The present study is novel in that it integrates nutritional risk screening, energy intake analysis, and neurological severity scales into a single prognostic model. This multidisciplinary approach bridges the existing gap and provides a clinically applicable protocol for early identification and targeted intervention in pat...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Automated Biochemical Analyzer	Guilin Youlit Medical Electronics Co., Ltd.	Gui 's equipment is approved for 20172220142	Used for detecting serum protein, liver/kidney function, and infection markers
Bedside Ultrasound Machine	Shenzhen Mindray Biomedical Electronics Co., Ltd.	Guangdong machinery injection 20152061288	Equipped with a 3.5–5 MHz abdominal probe; used for monitoring gastric residual volume
Centrifuge	Eppendorf	5810R	Maximum speed 15,000 × g; used for separating serum from venous blood
Disposable Gloves	Kimberly-Clark	KC500	Nitrile, powder-free; worn during blood sample handling
Gastric Tube	ConvaTec	14Fr	Silicone material, sterile; used for EN delivery and gastric residual volume aspiration
Nutritional Risk Screening 2002 (NRS 2002) Scale	European Society of Parenteral and Enteral Nutrition (ESPEN)	N/A	Standardized scale for nutritional risk assessment; scores 0–7 points
Statistical Analysis Software	IBM	SPSS 26.0	Used for descriptive statistics, Logistic regression, and ROC curve analysis
Vacuum Blood Collection Tubes	Shanghai Kehua Laboratory Medical Products Co., Ltd.	Shanghai Machinery Registration 20142410117	used for collecting venous blood samples
Water Swallow Test (WST) Checklist	Xuanwu Hospital Neurology Department	N/A	Customized checklist for evaluating swallowing function in neurocritical patients