Comparative prognostic performance of nutritional assessment tools in critically ill surgical patients after emergency surgery for sepsis or septic shock: a single-center retrospective study in South Korea

Sung Bak An; Dong Woo Shin

doi:10.4266/acc.001170

Articles

Page Path: HOME > Acute Crit Care > Volume 41(2); 2026 > Article

Original Article Nutrition Comparative prognostic performance of nutritional assessment tools in critically ill surgical patients after emergency surgery for sepsis or septic shock: a single-center retrospective study in South Korea: Acute and Critical Care 2026;41(2):399-410.
DOI: https://doi.org/10.4266/acc.001170
Published online: May 28, 2026

Sung Bak An

, Dong Woo Shin

Department of Surgery, Hallym University Dongtan Sacred Heart Hospital, Hwaseong, Korea

Corresponding author: Dong Woo Shin Department of Surgery, Hallym University Dongtan Sacred Heart Hospital, 7 Keunjaebong-gil, Hwaseong 18450, Korea Tel: +82-31-8086-2430 Fax: +82-31-8086-3459 Email: shin519@hallym.or.kr

• Received: January 31, 2026 • Revised: April 17, 2026 • Accepted: April 19, 2026

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

546 Views
14 Download

prev next

Full Article

Download PDF

Abstract
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
KEY MESSAGES
NOTES
REFERENCES

Abstract

Background
Malnutrition is prevalent in critically ill patients. Although various nutritional assessment tools are used in intensive care units (ICUs), their comparative prognostic performance in critically ill surgical patients requiring postoperative ICU care after emergency surgery for sepsis or septic shock remains uncertain. We compared the performance of four tools in predicting ICU mortality.
Methods
We retrospectively analyzed 218 adult critically ill surgical patients requiring postoperative ICU care after emergency surgery for sepsis or septic shock (January 2015–July 2025). Nutritional status was assessed within 24 hours of ICU admission using the Global Leadership Initiative on Malnutrition (GLIM), modified Nutrition Risk in the Critically Ill (mNUTRIC), Nutritional Risk Screening 2002 (NRS-2002), and Malnutrition Universal Screening Tool (MUST). The primary outcome was ICU mortality. Discriminative performance was evaluated using the area under the receiver operating characteristic curve and incremental value over a baseline model (age, sex, Acute Physiology and Chronic Health Evaluation [APACHE] II score, and Sequential Organ Failure Assessment [SOFA] score).
Results
The mNUTRIC score showed the highest crude discrimination for ICU mortality, followed by MUST and GLIM, whereas NRS-2002 demonstrated limited discriminatory ability. When added to the baseline clinical model, MUST and GLIM improved mortality prediction, unlike mNUTRIC. Higher risk categories in MUST and GLIM were associated with increased ICU mortality.
Conclusions
Prognostic performance varies among nutritional assessment tools. MUST and GLIM provide prognostic information beyond conventional severity scores, whereas the prognostic value of mNUTRIC is largely driven by disease severity, supporting their use for risk stratification in critically ill surgical patients.
Key Words: critical illness; malnutrition; nutritional assessment; septic shock; surgical intensive care units

INTRODUCTION

Malnutrition is common but often underdiagnosed in critically ill patients and is strongly associated with adverse clinical outcomes, including increased mortality, infectious complications, prolonged mechanical ventilation, and extended intensive care unit (ICU) and hospital stay [1-3]. Several nutritional assessment and screening tools have been developed and are widely used in both general and critical care settings. Commonly used tools include the Subjective Global Assessment, Nutritional Risk Screening 2002 (NRS-2002), Malnutrition Universal Screening Tool (MUST), and Nutrition Risk in the Critically ill (NUTRIC) score and its modified version (mNUTRIC) [4-7]. Each tool reflects a distinct conceptual framework, ranging from clinician judgment-based global assessment to systematic screening algorithms and ICU-specific risk scores, with predictive performance varying by patient population.

In 2018, the Global Leadership Initiative on Malnutrition (GLIM) proposed an international consensus-based diagnostic framework integrating phenotypic criteria (involuntary weight loss, low body mass index (BMI), and reduced skeletal muscle mass) with etiological criteria (reduced food intake or inflammation) [8]. Since then, GLIM has been widely validated across diverse populations, including hospitalized medical patients, older populations, and various disease-specific cohorts, as well as its prognostic value for outcomes, such as prolonged length of stay, readmission, and short-term mortality [9-11]. More recently, GLIM has been applied to critically ill populations, including neurocritical care and ICU cohorts, to explore its predictive capacity for mortality and other clinically relevant outcomes [10,11].

However, despite extensive research in different settings, studies evaluating GLIM or directly comparing multiple nutritional assessment tools in critically ill surgical patients requiring postoperative ICU care after emergency general surgery for sepsis or septic shock are lacking. Critically ill surgical patients requiring postoperative ICU care after emergency general surgery for septic shock represent a unique and highly vulnerable group characterized by profound systemic inflammation, severe catabolic stress, hemodynamic instability, rapid loss of muscle mass, and functional reserve. These pathophysiological features may modify the prognostic performance of existing nutritional tools, leaving uncertainty about which instrument best discriminates mortality risk in this population.

Therefore, this study aimed to compare several widely used nutritional assessment tools, including GLIM, NRS-2002, MUST, and mNUTRIC, in critically ill surgical patients requiring postoperative ICU care after emergency surgery for sepsis or septic shock. Specifically, we aimed to determine the nutritional assessment tool with the strongest association with ICU mortality in this critically ill surgical cohort. Clarifying the relative prognostic values of these tools may help clinicians select the most appropriate instrument for early risk stratification and guide timely targeted nutritional interventions in critically ill surgical patients after emergency surgery for sepsis or septic shock.

MATERIALS AND METHODS

Study Design and Ethical Approval

This retrospective observational cohort study was conducted at a secondary hospital in South Korea. The Institutional Review Board of Hallym University Dongtan Sacred Heart Hospital approved the study protocol (No. HDT 2025-07-014-001) and waived informed consent because only de-identified electronic medical record (EMR) data were analyzed. The study adhered to the ethical principles of the Declaration of Helsinki and followed Good Clinical Practice standards. Sex reporting complied with the current guidelines. Biological sex (male or female), as documented in the EMR, was used for sex-stratified analyses when physiologically relevant, particularly for skeletal muscle mass. Gender identity information was unavailable. Reproducibility was ensured through prespecified variable definitions, standardized protocols for nutritional assessment tools, and complete documentation of analytic procedures.

Study Population

The study included adult critically ill surgical patients who were admitted to the surgical ICU (SICU) between January 1, 2015, and July 31, 2025, immediately after emergency or urgent surgery for source control of an infectious focus. The study specifically focused on patients who met the diagnostic criteria for sepsis or septic shock and required postoperative critical care. To minimize missed cases and ensure comprehensive identification of postoperative septic shock, we additionally screened all critically ill surgical patients in which norepinephrine infusion was initiated upon ICU arrival, reflecting hemodynamic instability consistent with Sepsis-3 definitions [12]. Only adult patients who met the criteria for this study were included. Patients were classified as having sepsis or septic shock according to the Sepsis-3 criteria, and the proportion of patients with septic shock was described in the baseline characteristics.

Data Collection and Case Identification

Eligible encounters were identified by reviewing the EMR. Perioperative anesthesia records, operative notes, and ICU medication administration logs were examined to confirm vasopressor use upon SICU admission. Laboratory values, vital signs, anthropometric data, and severity indices (Acute Physiology and Chronic Health Evaluation [APACHE] II, Sequential Organ Failure Assessment [SOFA]) were collected within the first 24 hours of admission. Each patient was assigned a unique study identification code to ensure anonymization. Data extraction followed a prespecified protocol to ensure uniformity and reproducibility.

Inclusion and Exclusion Criteria

Inclusion criteria were: (1) age ≥18 years; (2) diagnosis of sepsis or septic shock according to the Sepsis-3 criteria (suspected or confirmed infection plus an increase in SOFA score ≥2); (3) postoperative SICU admission following operative source control for an infectious focus; (4) availability of comprehensive clinical and laboratory data within 24 hours of SICU admission; and (5) sufficient EMR information to apply to all four nutritional tools (GLIM, mNUTRIC, NRS-2002, and MUST).

Exclusion criteria were: (1) ICU length of stay <48 hours; (2) presence of treatment-limiting directives (Physician Orders for Life-Sustaining Treatment, Do Not Resuscitate); (3) surgery for non-infectious indications (trauma, transplantation, cardiovascular procedures, or elective oncological operations); (4) missing essential variables required for nutritional scoring; (5) pregnancy or age <18 years; and (6) repeated ICU admissions (only the first admission recorded).

Nutritional Assessment Tools

Four nutritional assessment tools were retrospectively applied to all eligible patients using data obtained within the first 24 hours of SICU admission.

Global Leadership Initiative on Malnutrition

The GLIM framework requires at least one phenotypic and etiological criterion to diagnose malnutrition. Phenotypic criteria included: (1) unintentional weight loss (≥5% within 6 months or ≥10% within 1 year); (2) low BMI (<20 kg/m² for adults <70 years and <22 kg/m² for adults ≥ 70 years); and (3) reduced skeletal muscle mass. Etiologic criteria included: (1) reduced nutrient intake or absorption and (2) presence of inflammation. Because all patients presented with sepsis or septic shock, the inflammatory criterion was universally satisfied.

Assessment of Skeletal Muscle Mass

Skeletal muscle mass was quantified using abdominal computed tomography (CT) scans obtained before or immediately after surgery. Skeletal muscle area (SMA) at the third lumbar vertebral level (L3) was measured using 3D Slicer software (version 5.8.1; https://www.slicer.org). Semi-automated segmentation techniques were used to delineate the psoas, paraspinal, and abdominal wall muscles, with manual verification by trained reviewers.

Skeletal muscle index (SMI) was calculated as:

SMI=SMA (cm²)/height2 (m²)

Sex-specific cutoffs validated in healthy Korean populations were used [13,14].

Men: mild <46.7 cm²/m²; severe ≤39.8 cm²/m²

Women: mild <33.6 cm²/m²; severe ≤28.5 cm²/m²

These thresholds were incorporated into the GLIM phenotypic assessment.

Modified Nutrition Risk in the Critically Ill

The modified NUTRIC score incorporates: (1) age; (2) APACHE II score; (3) SOFA score; (4) comorbidity count; and (5) pre-ICU length of stay. Scores range from 0 to 9, with ≥5 indicating high nutritional risk.

Nutritional Risk Screening 2002

NRS-2002 evaluates: (1) recent nutritional impairment (BMI, unintentional weight loss, reduced dietary intake); (2) disease severity; and (3) age ≥70 years (additional 1 point). A score ≥3 indicates nutritional risk.

Malnutrition Universal Screening Tool

MUST evaluates: (1) BMI category; (2) unintentional weight loss within the previous 3-6 months; and (3) acute disease effect (expected no intake >5 days). Scores of 0, 1, and ≥2 indicate low, medium, and high nutritional risk, respectively. All nutritional tools were applied independently by trained investigators following a standardized operation manual to ensure consistency and reproducibility.

Definition of Complications

Complications were defined as clinically significant adverse events documented in the EMR during the ICU and hospital stay. These were identified by retrospective chart review of the complication fields and included infectious complications (e.g., pneumonia, wound infection, intra-abdominal abscess), surgical complications (e.g., anastomotic leakage, fistula, postoperative bleeding, wound or stoma dehiscence), and other clinically relevant events such as ischemic or neurologic complications. For analysis, complications were treated as a binary variable indicating the presence of at least one documented complication.

Outcomes

Primary outcome

The primary outcome of this study was ICU mortality, and the study evaluated the ability of four nutritional assessment tools—GLIM, mNUTRIC, NRS-2002, and MUST—to predict ICU mortality in critically ill surgical patients requiring postoperative ICU care after emergency surgery for sepsis or septic shock.

Secondary outcome

The secondary outcome was to assess whether adding each nutritional tool to a baseline clinical model composed of age, sex, APACHE II score, and SOFA score improved ICU mortality prediction compared to the clinical model alone.

Sample Size Considerations

Sample size estimation was based on expected differences in ICU mortality prediction performance among nutritional tools. Previous studies have reported area under the curve (AUC) values of approximately 0.70–0.75 for GLIM and mNUTRIC and 0.60–0.65 for NRS-2002 and MUST [1,7,15,16]. Using the Hanley–McNeil method with a two-sided α of 0.05, 80% power, an estimated mortality rate of 25%, and an anticipated ΔAUC of 0.03, the minimum required sample size for adequate discriminative comparison was 112 patients. To ensure sufficient power for subgroup and multivariate analyses, the study targeted ≥120 patients.

Statistical Analysis

Continuous variables were summarized as means±standard deviations or medians with interquartile ranges, depending on the distribution. Categorical variables were presented as frequencies and percentages. Group comparisons were performed using Student t-test or the Mann-Whitney U-test for continuous variables and the χ² or Fisher’s exact test for categorical variables. To evaluate the predictive performance of each nutritional tool for ICU mortality, receiver operating characteristic (ROC) curves were generated, and the AUC was calculated. AUCs were compared using DeLong’s method for correlated ROC curves.

A baseline clinical model consisting of age, sex, APACHE II score, and SOFA score was constructed. Each nutritional tool was then added separately to this model to evaluate its incremental predictive value as follows: (1) change in AUC (ΔAUC); (2) integrated discrimination improvement (IDI); (3) continuous net reclassification improvement (NRI); and (4) change in Nagelkerke’s R². Logistic regression was used to estimate the odds ratios for ICU mortality in relation to nutritional classification. All statistical analyses were performed using R software version 4.3.0 (R Foundation for Statistical Computing). Statistical significance was set at P<0.05.

As an exploratory analysis, individual nutrition-related components of the evaluated nutritional assessment tools were examined for their association with ICU mortality. Univariable logistic regression analyses were first performed for each nutrition-related component, and variables with P<0.10 were entered into a multivariable logistic regression model adjusted for age and sex. To focus on nutrition-related elements, severity-related variables embedded in some tools, such as APACHE II and SOFA scores, were not included in this component-level analysis. Reduced skeletal muscle mass was defined using sex-specific SMI cutoffs.

RESULTS

Patient Screening and Study Population

During the study period, 653 critically ill surgical patients requiring norepinephrine infusion after surgery and admitted to the SICU were screened. Among them, 42 patients who died within 48 hours of ICU admission and 52 organ donor harvest cases were excluded. In addition, 341 patients admitted after elective surgery were excluded to ensure the cohort represented true patients with postoperative sepsis or septic shock requiring critical care support. Ultimately, 218 patients met the inclusion criteria and were analyzed (Figure 1).

Baseline Characteristics

Baseline demographic and clinical characteristics are summarized in Table 1. Among the 218 included patients, 144 (66.1%) met the criteria for septic shock. The mean age of the patients was 70.9±14.7 years, and approximately half were male (50.5%). The mean BMI was 22.6±4.0 kg/m². Most patients had at least one chronic comorbidity, with hypertension (49.1%) and diabetes mellitus (26.6%) being the most prevalent. Gastrointestinal perforation was the leading surgical diagnosis (63.8%), followed by ischemic bowel disease (17.0%) and localized intra-abdominal infection (17.9%). Segmental bowel resection was the most common surgical procedure (32.1%), followed by anastomosis (27.5), diversion procedures, and primary repair. Disease severity scores at SICU admission indicated substantial physiological derangement: the mean SOFA score was 7.94±3.75, and the mean APACHE II score was 17.04±7.29.

Comparison between Survivors and Non-survivors

Comparison between survivors and non-survivors is summarized in Table 2. Non-survivors were older and had lower BMI, lower CT-derived SMI, lower serum albumin levels, and lower Glasgow Coma Scale scores than survivors. They also had higher lactate levels and greater illness severity, as reflected by higher SOFA and APACHE II scores. In addition, severe GLIM-defined malnutrition and higher mNUTRIC, NRS-2002, and MUST scores were more frequently observed in non-survivors.

Clinical Outcomes According to Nutritional Assessment Tools

Clinical outcomes, including ICU mortality, stratified by nutritional assessment tools are presented in Tables 3-5. Under the GLIM criteria, ICU and hospital length of stay and the need for mechanical ventilation were similar across groups. However, ICU mortality showed a clear and significant stepwise increase with worsening GLIM-defined malnutrition. ICU mortality increased from 4.7% among patients without malnutrition to 19.7% among those with moderate malnutrition and 29.0% among those with severe malnutrition (P<0.001). Although the ventilator days, ventilator-free days, and complication rates tended to worsen with higher GLIM categories, the differences were not significant.

In contrast to GLIM, the mNUTRIC score was strongly associated with multiple adverse clinical outcomes. Patients classified as high-risk (mNUTRIC ≥5) required longer ICU stays (6.5 vs. 11.7 days), a markedly higher need for mechanical ventilation (26.3% vs. 94.2%), and more frequent renal replacement therapy (1.8% vs. 21.2%). The number of ventilator-free days was substantially lower, and the overall complication rate was higher in the high-risk group. Importantly, ICU mortality was more than four times higher in the high mNUTRIC group (7.9% vs. 31.7%, P<0.001).

The MUST also showed strong discriminatory capacity. Patients in the high-risk MUST category (≥2 points) had significantly longer ICU and hospital stays, greater ventilator dependence, fewer ventilator-free days, and substantially higher complication rates and ICU mortality than those in the low-risk category. The medium-risk group showed intermediate outcomes. Conversely, the NRS-2002 tool showed limited discriminatory power in the population of critically ill surgical patients requiring postoperative ICU care after emergency surgery for sepsis or septic shock. Nearly all patients (216 of 218) scored ≥3, which placed them automatically in the “at risk” category. The extremely small size of the low-risk group (n=2) limited meaningful statistical comparisons.

Crude Predictive Performance of Nutritional Tools

Figure 2 presents crude ROC curves for the four nutritional assessment tools used to predict ICU mortality. The mNUTRIC score demonstrated the highest discriminatory performance, with an area under the receiver operating characteristic curve (AUROC) of 0.728, followed by the MUST score (AUROC, 0.720). The GLIM criteria showed moderate discrimination (AUROC, 0.673), whereas NRS-2002 displayed limited predictive ability (AUROC, 0.610), reflecting the restricted variability of this tool in the study cohort.

Logistic Regression Analysis for ICU Mortality

In univariable logistic regression analysis, all four nutritional assessment tools were significantly associated with ICU mortality: GLIM (odds ratio [OR], 2.449; 95% CI,1.501–3.994; P<0.001), mNUTRIC (OR, 1.667; 95% CI, 1.367–2.032; P<0.001), NRS-2002 (OR, 1.525; 95% CI, 1.169–1.990; P=0.002), and MUST (OR, 1.584; 95% CI,1.278–1.965; P<0.001) (Table 6).

After adjustment for age, sex, APACHE II score, and SOFA score, GLIM category remained significantly associated with ICU mortality (adjusted OR, 2.64; 95% CI, 1.47–4.73; P=0.001) and MUST score also retained independent prognostic significance (adjusted OR, 1.53; 95% CI, 1.20–1.96; P<0.001). In contrast, mNUTRIC was no longer significantly associated with ICU mortality after adjustment (adjusted OR, 1.17; 95% CI, 0.79–1.74; P=0.439). NRS-2002 was not estimable in the adjusted model because nearly all patients were classified as being at nutritional risk.

Incremental Prognostic Value beyond the Clinical Model

To determine whether nutritional assessment tools add prognostic value beyond conventional severity indices, a baseline logistic regression model was constructed using age, sex, APACHE II scores, and SOFA scores (Figure 3). The baseline model demonstrated good discrimination, with an AUROC of 0.791 (95% CI, 0.711–0.871). When GLIM was added, the AUROC increased to 0.822, accompanied by meaningful improvements in the IDI and NRI. Similarly, MUST increased the AUROC to 0.831 and produced the largest NRI improvement. In contrast, adding mNUTRIC to the clinical model did not improve discrimination (AUROC, 0.792), and both the IDI and NRI values were negative. Figure 3 summarizes the AUROC values of the baseline and combined models.

Exploratory Component-Level Analysis

To further explore which nutrition-related elements of the evaluated assessment tools were most strongly associated with ICU mortality, we performed an exploratory component-level analysis. In univariable analysis, expected no nutritional intake for >5 days, BMI decrease, and reduced skeletal muscle mass based on sex-specific SMI cutoffs were significantly associated with ICU mortality, whereas weight loss history and decreased feeding before ICU admission were not. In the multivariable model including candidate variables selected from the univariable analysis, expected no nutritional intake for >5 days and reduced skeletal muscle mass remained independently associated with ICU mortality, whereas BMI did not retain significance (Table 7).

DISCUSSION

This study compared the prognostic performance of four commonly used nutritional assessment tools—GLIM, mNUTRIC, NRS-2002, and MUST—in critically ill surgical patients requiring postoperative ICU care after emergency general surgery for sepsis or septic shock. Although all tools demonstrated some association with ICU mortality, their predictive performances and incremental prognostic values differed substantially. Contrary to several prior reports on heterogeneous hospitalized populations, GLIM did not emerge as the single best-performing tool in this critically ill surgical cohort. Therefore, this finding warrants careful interpretation.

The mNUTRIC score demonstrated the strongest crude discrimination of ICU mortality, consistent with previous studies in mixed ICU populations [7,17,18]. This result is largely attributable to the structure of the mNUTRIC, which explicitly identified patients at high-risk of ICU mortality in unadjusted analyses. However, when evaluated in conjunction with a baseline clinical model already including APACHE II and SOFA scores, the addition of mNUTRIC failed to improve discrimination and resulted in negative IDI and NRI values. This finding suggests that the prognostic value of mNUTRIC in this cohort was predominantly driven by overlap with established severity indices rather than by independent nutritional risk, a limitation previously acknowledged in critically ill populations [19]. This interpretation was further supported by the logistic regression analysis. Although mNUTRIC was significantly associated with ICU mortality in univariable analysis, its association was no longer significant after adjustment for age, sex, APACHE II scores, and SOFA scores. This finding suggests that much of its prognostic signal in this cohort was explained by overlapping with established severity variables rather than by independent nutritional information.

In contrast, MUST demonstrated both strong crude discrimination and meaningful incremental prognostic value beyond the clinical model. The MUST is a simple screening tool based on BMI, recent unintentional weight loss, and acute disease effects, thereby reflecting baseline nutritional reserve rather than acute physiological derangement. The consistent associations between high MUST scores and prolonged ICU stay, ventilator dependence, complications, and ICU mortality observed in this study suggest that depleted nutritional reserves play a pivotal role in postoperative resilience to septic insults. Importantly, MUST retained prognostic relevance even after adjustment for severity scores, supporting its role as a complementary rather than redundant risk stratification tool.

GLIM showed a distinct and more nuanced performance. GLIM-defined malnutrition was strongly associated with ICU mortality, particularly in patients classified as having severe malnutrition, but demonstrated a weaker association with length of ICU stay, ventilator use, and other short-term clinical course variables. This discrepancy likely reflects the conceptual framework of GLIM, which was developed to diagnose malnutrition rather than to predict acute ICU outcomes. GLIM integrates phenotypic criteria with etiologic criteria to identify chronic or subacute nutritional vulnerability.

Several factors may explain why GLIM did not outperform MUST or mNUTRIC in this postoperative septic shock cohort. First, GLIM places substantial weight on reduced muscle mass, which was assessed using CT-based SMI in our study. Although the SMI is a robust marker of sarcopenia and long-term prognosis, its impact on early ICU outcomes, such as ventilator duration or length of stay, may be less pronounced than its effect on mortality or long-term functional decline. Second, the inflammatory criterion for the GLIM is almost universally met in patients with septic shock, potentially diminishing its discriminatory contribution in this setting. Consequently, GLIM may be better suited to identifying patients at risk of late mortality, frailty-related death, or impaired recovery, rather than short-term organ support requirements.

These findings align with prior studies showing that GLIM has been increasingly adopted and validated across diverse patient groups and various disease-specific cohorts and has short-term mortality prognostic value for outcomes such as prolonged length of stay, readmission, and short-term mortality [9-11]. In contrast, tools such as MUST, which emphasize preexisting nutritional reserves, may better capture vulnerability relevant to immediate postoperative and septic stress responses.

Taken together, these findings highlight that the prognostic utility of nutritional assessment tools is highly context dependent. In patients with postoperative septic shock, tools reflecting acute severity (mNUTRIC) or baseline nutritional reserve (MUST) demonstrated superior crude discrimination, whereas tools designed to diagnose malnutrition across broader clinical contexts (GLIM) may preferentially predict mortality rather than short-term ICU trajectories. Importantly, MUST and GLIM provided independent prognostic information beyond traditional severity indices, supporting their complementary role in risk stratification frameworks for critically ill surgical patients.

The exploratory component-level analysis further suggested that expected no nutritional intake for more than 5 days and reduced skeletal muscle mass were the nutrition-related factors most consistently associated with ICU mortality. These findings support the clinical relevance of prolonged nutritional deprivation and body composition in postoperative septic surgical patients and may partly explain the prognostic contribution of tools incorporating acute disease effect and phenotypic nutritional reserve. However, the acute disease effect component of MUST should be interpreted with caution in retrospective analyses, because the expectation of no nutritional intake for more than 5 days may be partly subject to clinical judgment and may not always be explicitly documented in the medical record.

This study has several strengths. To the best of our knowledge, this is one of the largest single-cohort analyses focusing specifically on critically ill surgical patients requiring postoperative ICU care after emergency general surgery for sepsis or septic shock. The long study period and systematic identification of patients using vasopressors minimized selection bias, and the inclusion of CT-based SMI measurements allowed objective assessment of muscle mass. In addition, we evaluated both the crude discrimination and incremental prognostic value using complementary metrics, providing a nuanced comparison of nutritional tools.

This study had some limitations. This single-center retrospective study was conducted at a secondary hospital, which may have limited the generalizability of the results. Nutritional interventions during ICU stay were not standardized or analyzed, precluding the assessment of treatment effect modifications. In addition, although CT-based SMI measurements enhance objectivity, they may not be feasible in all clinical settings. Furthermore, patients who died within 48 hours of ICU admission were excluded, which may have introduced selection bias by underrepresenting the most severely ill patients. This exclusion criterion was intended to reduce inclusion of patients whose outcomes were likely driven predominantly by overwhelming acute physiological deterioration rather than nutritional status. Finally, mortality was limited to ICU mortality, and long-term outcomes were not evaluated.

This study compared the prognostic performance of four commonly used nutritional assessment tools—GLIM, mNUTRIC, NRS-2002, and MUST—in critically ill surgical patients requiring postoperative ICU care after emergency general surgery for sepsis or septic shock. Although the mNUTRIC score showed strong crude discrimination for ICU mortality, its predictive performance was largely driven by overlap with established severity indices and did not provide incremental prognostic value beyond age, sex, APACHE II, and SOFA scores. In contrast, MUST score demonstrated the most consistent and robust prognostic performance, providing independent and incremental predictive value beyond conventional severity-based models, although GLIM also contributed additional prognostic information, particularly for mortality risk stratification.

These findings indicate that nutritional tools that capture baseline nutritional reserves and body composition rather than acute physiological severity alone offer clinically meaningful information regarding postoperative septic shock. Incorporating MUST, with complementary use of GLIM, may improve risk stratification and support the development of targeted nutritional strategies in critically ill surgical patients requiring postoperative ICU care after emergency general surgery for sepsis or septic shock.

KEY MESSAGES

▪ In critically ill surgical patients requiring postoperative intensive care unit (ICU) care after emergency general surgery for sepsis or septic shock, commonly used nutritional assessment tools demonstrated substantial differences in prognostic performance for ICU mortality.

▪ Although the modified Nutrition Risk in the Critically Ill score showed strong crude discrimination, its prognostic value was largely driven by an overlap with established severity scores and did not provide incremental predictive value.

▪ In contrast, Malnutrition Universal Screening Tool and Global Leadership Initiative on Malnutrition provided additional prognostic information beyond conventional severity indices, supporting their selective use for nutritional risk stratification in critically ill surgical patients.

NOTES

CONFLICT OF INTEREST

No potential conflict of interest relevant to this article was reported.

FUNDING

None.

ACKNOWLEDGMENTS

None.

AUTHOR CONTRIBUTIONS

Conceptualization: SBA, DWS. Data curation: SBA. Formal analysis: SBA. Methodology: SBA, DWS. Project administration: DWS. Visualization: SBA. Writing - original draft: SBA. Writing - review & editing: DWS. All authors read and agreed to the published version of the manuscript.

Figure 1.

Patient selection flowchart. ICU: intensive care unit.

Figure 2.

Receiver operating characteristic curves for intensive care unit mortality according to the nutritional assessment tools. GLIM, Global Leadership Initiative on Malnutrition; AUC, area under the curve; mNUTRIC, modified Nutrition Risk in the Critically Ill; NRS-2002: Nutritional Risk Screening 2002; MUST, Malnutrition Universal Screening Tool.

Figure 3.

Receiver operating characteristic curves for intensive care unit mortality comparing the discrimination of the baseline clinical model (age, sex, APACHE II, and SOFA) with models incorporating individual nutritional assessment tools. AUC: area under the curve; GLIM: Global Leadership Initiative on Malnutrition; AUC: area under the curve; mNUTRIC: modified Nutrition Risk in the Critically Ill; NRS-2002: Nutritional Risk Screening 2002; MUST: Malnutrition Universal Screening Tool; APACHE: Acute Physiology and Chronic Health Evaluation; SOFA: Sequential Organ Failure Assessment.

Table 1.

Patient demographics and characteristics

Category	Variable	Value
Demographics	Age (yr)	71±15
	Sex, male	110 (50.5)
	BMI (kg/m²)	22.6±4.0
Comorbidity	At least 1 comorbidity	187 (85.8)
	Hypertension	105 (48.2)
	Diabetes mellitus	58 (26.6)
	Chronic lung disease	9 (4.1)
	Chronic kidney disease	12 (5.5)
	Chronic liver disease	14 (6.4)
	Cardiovascular disease	26 (11.9)
	Neurologic disease	26 (11.9)
	Cancer	60 (27.5)
	Others	24 (11.0)
Diagnosis	Perforation	139 (63.8)
	Colorectal	69 (49.6)
	Upper gastrointestinal	33 (23.7)
	Small bowel	31 (22.3)
	Gallbladder	4 (2.9)
	Other/unspecified	2 (1.4)
	Ischemia	37 (17.0)
	Localized infection	39 (17.9)
	Others	3 (1.4)
Operation	Diversion	34 (15.6)
	Primary repair	24 (11.0)
	Segmental resection	70 (32.1)
	Anastomosis	60 (27.5)
	Others	30 (13.8)
Severity of illness	Septic shock	144 (66.1)
	P/F ratio (mm Hg)	334.8±131.3
	GCS score	9.7±4.7
	SOFA score	7.94±3.75
	APACHE II score	17.04±7.29
Key laboratory value	Albumin (g/dl)	2.65±0.56
	CRP (mg/L)	16.8±11.9
	Lactic acid (mmol/L)	3.08±3.38
	WBC (×10³/µl)	13.6±7.7
	Platelet (×10³/µl)	211±119
	Hematocrit (%)	32.1±7.8
Organ support	Ventilator use	128 (58.7)
	Renal replacement therapy	24 (11.0)

Values are presented as mean±standard deviation or number (%).

BMI: body mass index; P/F ratio: PaO₂/FiO₂ ratio; GCS: Glasgow Coma Scale; SOFA: Sequential Organ Failure Assessment; APACHE: Acute Physiology and Chronic Health Evaluation; CRP: C-reactive protein; WBC: white blood cell.

Table 2.

Baseline characteristics of survivors and non-survivors

Variable	Survivor (n=176)	Non-survivor (n=42)	P-value
Age (yr)	70±15	77±11	0.002
Sex, male	90 (51.1)	20 (47.6)	0.733
BMI (kg/m²)	22.8±3.9	20.7±3.6	<0.001
SMI (cm²/m²)	36.3±8.5	29.5±7.2	<0.001
GCS score	10.3±4.6	7.1±3.9	<0.001
P/F ratio (mm Hg)	341.0±130.9	309.0±131.6	0.247
Lactic acid (mmol/L)	2.9±1.8	5.0±3.5	<0.001
Albumin (g/dl)	2.6±0.6	2.2±0.5	<0.001
Septic shock	109 (61.9)	35 (83.3)	0.010
SOFA score	7.4±3.5	10.3±3.7	<0.001
APACHE II score	15.7±6.7	22.6±7.0	<0.001
GLIM severe malnutrition	66 (37.5)	27 (64.3)	0.003
mNUTRIC score	4.1±2.1	6.1±1.9	<0.001
mNUTRIC high risk	71 (40.3)	33 (78.6)	<0.001
NRS-2002 score	4.2±1.2	4.9±1.2	<0.001
MUST score	1.0±1.4	2.2±1.5	<0.001
MUST high risk	61 (34.7)	29 (69.0)	<0.001

Values are presented as mean±standard deviation or number (%).

BMI: body mass index; SMI: skeletal muscle index; GCS: Glasgow Coma Scale; P/F ratio: PaO₂/FiO₂ ratio;

SOFA: Sequential Organ Failure Assessment; APACHE: Acute Physiology and Chronic Health Evaluation;

GLIM: Global Leadership Initiative on Malnutrition; mNUTRIC: modified Nutrition Risk in the Critically Ill; NRS-2002: Nutritional Risk Screening 2002; MUST: Malnutrition Universal Screening Tool.

Table 3.

Clinical course according to nutritional assessment tools: GLIM

Variable	No malnutrition (n=64)	Moderate (n=61)	Severe (n=93)	P-value
ICU LOS (day)	8.3±7.7	9.1±10.3	9.4±8.0	0.271
Hospital LOS (day)	23.5±14.1	23.3±16.3	28.5±21.8	0.302
Ventilator use	39 (60.9)	31 (50.8)	57 (61.3)	0.351
Ventilator day	3.6±6.3	5.1±9.0	4.7±7.3	0.581
30-Day ventilator-free days	26.5±5.8	24.7±8.4	25.7±10.1	0.578
RRT	8 (12.5)	5 (8.2)	11 (11.8)	0.704
Complication	25 (39.1)	28 (45.9)	47 (50.5)	0.366
ICU mortality	3 (4.7)	12 (19.7)	27 (29.0)	<0.001

Values are presented as mean±standard deviation or number (%).

GLIM: Global Leadership Initiative on Malnutrition; ICU: intensive care unit; LOS: length of stay; RRT: renal replacement therapy.

Table 4.

Clinical course according to nutritional assessment tools: mNUTRIC

Variable	Low (n=114)	High (n=104)	P-value
ICU LOS (day)	7±7	12±9	<0.001
Hospital LOS (day)	22.3±16.8	29.2±19.5	0.001
Ventilator use	30 (26.3)	97 (93.3)	<0.001
Ventilator day	2.0±6.0	7.2±8.1	<0.001
30-Day ventilator-free days	28.1±8.5	22.9±7.8	<0.001
RRT	2 (1.8)	22 (21.2)	<0.001
Complication	42 (36.8)	58 (55.8)	0.006
ICU mortality	9 (7.9)	33 (31.7)	<0.001

Values are presented as mean±standard deviation or number (%).

mNUTRIC: modified Nutrition Risk in the Critically Ill; ICU: intensive care unit; LOS: length of stay; RRT: renal replacement therapy.

Table 5.

Clinical course according to nutritional assessment tools: MUST

Variable	Low (n=112)	Medium (n=16)	High (n=90)	P-value
ICU LOS (day)	6.5±5.5	7.8±9.6	12.4±10.3	<0.001
Hospital LOS (day)	19.5±11.7	24.2±18.5	33.3±22.1	<0.001
Ventilator use	57 (50.9)	10 (62.5)	60 (66.7)	0.058
Ventilator days	2.7±4.5	4.3±6.4	6.8±9.8	0.002
30-Day ventilator-free day	27.9±7.5	23.8±8.9	23.2±9.1	<0.001
RRT	10 (8.9)	1 (6.3)	13 (14.4)	0.377
Complication	38 (33.9)	7 (43.8)	55 (61.1)	<0.001
ICU mortality	9 (8.0)	4 (25.0)	29 (32.2)	<0.001

Values are presented as mean±standard deviation or number (%).

MUST: Malnutrition Universal Screening Tool; ICU: intensive care unit; LOS: length of stay; RRT, renal replacement therapy.

Table 6.

Logistic regression analysis of nutritional assessment tools for ICU mortality

Nutritional assessment tool	Univariable OR (95% CI)	P-value	Adjusted OR^a) (95% CI)	P-value
GLIM category (per 1-category increase)	2.449 (1.501–3.994)	<0.001	2.64 (1.47–4.73)	0.001
mNUTRIC score (per 1-point increase)	1.667 (1.367–2.032)	<0.001	1.17 (0.79–1.74)	0.439
NRS-2002 score (per 1-point increase)	1.525 (1.169–1.990)	0.002	Not estimable^b)	0.999
MUST score (per 1-point increase)	1.584 (1.278–1.965)	<0.001	1.53 (1.20–1.96)	<0.001

ICU: intensive care unit; OR: odds ratio; GLIM: Global Leadership Initiative on Malnutrition; mNUTRIC: modified Nutrition Risk in the Critically Ill; NRS-2002: Nutritional Risk Screening 2002; MUST: Malnutrition Universal Screening Tool; APACHE: Acute Physiology and Chronic Health Evaluation; SOFA: Sequential Organ Failure Assessment.

^a)Adjusted for age, sex, APACHE II score, and SOFA score;

^b)NRS-2002 was not estimable in the adjusted model because nearly all patients were classified as being at nutritional risk.

Table 7.

Exploratory univariable and multivariable logistic regression analyses of nutrition-related components for ICU mortality

Component	Univariable OR (95% CI)	P-value	Multivariable OR^a) (95% CI)	P-value
Weight loss history	0.96 (0.26–3.53)	0.949	-	-
Decreased feeding before ICU admission	0.60 (0.30–1.18)	0.137	-	-
Expected no nutritional intake for >5 days	5.68 (2.70–11.95)	<0.001	4.59 (2.08–10.11)	<0.001
BMI decrease (per 1 kg/m²)	1.17 (1.05–1.29)	0.003	1.08 (0.95–1.22)	0.228
Reduced SMI (sex-specific cutoff)	5.30 (1.81–15.52)	0.002	4.14 (1.07–15.98)	0.039

ICU: intensive care unit; OR: odds ratio; BMI: body mass index; SMI: skeletal muscle index.

^a)Multivariable model included candidate nutrition-related variables selected from univariable analysis (P<0.10), with adjustment for age and sex.

REFERENCES

1. Lew CCH, Wong GJY, Cheung KP, Chua AP, Chong MFF, Miller M, et al. Association between malnutrition and 28-day mortality and intensive care length-of-stay in the critically ill: A prospective cohort study. Nutrients 2017;10:10.Article PubMed PMC
2. Eldehily K, Tawfik M, Elkholy M, Sabry S. Impact of malnutrition in critically ill patients on intensive care unit outcomes. Egypt J Intensiv Care Emerg Med 2023;3:57-69.Article
3. Viner Smith E, Lambell K, Tatucu-Babet OA, Ridley E, Chapple LA. Nutrition considerations for patients with persistent critical illness: a narrative review. JPEN J Parenter Enteral Nutr 2024;48:658-66.Article PubMed
4. Detsky AS, McLaughlin JR, Baker JP, Johnston N, Whittaker S, Mendelson RA, et al. What is subjective global assessment of nutritional status? JPEN J Parenter Enteral Nutr 1987;11:8-13.Article PubMed
5. Kondrup J, Rasmussen HH, Hamberg O, Stanga Z; Ad Hoc ESPEN Working Group. Nutritional risk screening (NRS 2002): a new method based on an analysis of controlled clinical trials. Clin Nutr 2003;22:321-36.Article PubMed
6. Aloy Dos Santos T, Luft VC, Souza GC, de Albuquerque Santos Z, Keller Jochims AM, Carnevale de Almeida J, et al. Malnutrition screening tool and malnutrition universal screening tool as a predictors of prolonged hospital stay and hospital mortality: a cohort study. Clin Nutr ESPEN 2023;54:430-5.Article PubMed
7. Heyland DK, Dhaliwal R, Jiang X, Day AG. Identifying critically ill patients who benefit the most from nutrition therapy: the development and initial validation of a novel risk assessment tool. Crit Care 2011;15:R268.Article PubMed PMC PDF
8. Cederholm T, Jensen GL, Correia MI, Gonzalez MC, Fukushima R, Higashiguchi T, et al. GLIM criteria for the diagnosis of malnutrition - a consensus report from the global clinical nutrition community. Clin Nutr 2019;38:1-9.Article PubMed
9. de van der Schueren MAE, Keller H; GLIM Consortium; Cederholm T, Barazzoni R, Compher C, et al. Global leadership initiative on malnutrition (GLIM): guidance on validation of the operational criteria for the diagnosis of protein-energy malnutrition in adults. Clin Nutr 2020;39:2872-80.Article PubMed
10. Jazinaki MS, Safarian M, Arabi SM, Jamali J, Norouzy A. Validation of GLIM criteria for hospital malnutrition diagnosis by comparison of three different anthropometric approaches to evaluate reduced muscle mass: a prospective cohort study. Front Nutr 2024;11:1438158.Article PubMed PMC
11. Liu P, Tian H, Gao L, Zhong T, Wang Y, Chen L, et al. The predictive value of GLIM criteria on clinical outcomes and responses to nutritional support in patients with neurocritical illnesses. Sci Rep 2024;14:15061.Article PubMed PMC PDF
12. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al. The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 2016;315:801-10.Article PubMed PMC
13. Yun JH, Song GJ, Son MW, Lee MS. Global Leadership Initiative on Malnutrition criteria and immunonutritional status predict chemoadherence and survival in stage II/III gastric cancer treated with XELOX chemotherapy. Nutrients 2024;16:3468.Article PubMed PMC
14. Kim EH, Kim KW, Shin Y, Lee J, Ko Y, Kim YJ, et al. Reference data and T-Scores of lumbar skeletal muscle area and its skeletal muscle indices measured by CT scan in a healthy Korean population. J Gerontol A Biol Sci Med Sci 2021;76:265-71.Article PubMed PDF
15. Milanez DSJ, Razzera EL, Lima J, Silva FM. Feasibility and criterion validity of the GLIM criteria in the critically ill: a prospective cohort study. JPEN J Parenter Enteral Nutr 2023;47:754-65.Article PubMed
16. Brascher JM, Peres WA, Padilha PC. Use of the modified "Nutrition Risk in the critically ill" score and its association with the death of critically ill patients. Clin Nutr ESPEN 2020;35:162-6.Article PubMed
17. Rahman A, Hasan RM, Agarwala R, Martin C, Day AG, Heyland DK, et al. Identifying critically-ill patients who will benefit most from nutritional therapy: further validation of the "modified NUTRIC" nutritional risk assessment tool. Clin Nutr 2016;35:158-62.Article PubMed
18. Kim SJ, Cho YJ, Park YS, et al. Comparison of mNUTRIC-S2 and mNUTRIC scores to assess nutritional risk and predict intensive care unit mortality. Acute Crit Care 2022;37:382-90.Article PubMed PMC PDF
19. Compher C, Chittams J, Sammarco T, Higashibeppu N, Higashiguchi T, Heyland DK, et al. Greater nutrient intake is associated with lower mortality in western and eastern critically Ill patients with low BMI: a multicenter, multinational observational study. JPEN J Parenter Enteral Nutr 2019;43:63-9.Article PubMed

Figure & Data

References

Citations

Citations to this article as recorded by

PubReader

Cite

CITE: export Copy Format; Close

Download Citation

Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.

Format:

RIS — For EndNote, ProCite, RefWorks, and most other reference management software
BibTeX — For JabRef, BibDesk, and other BibTeX-specific software

Include:

Citation for the content below
Citation and abstract for the content below

Comparative prognostic performance of nutritional assessment tools in critically ill surgical patients after emergency surgery for sepsis or septic shock: a single-center retrospective study in South Korea

Acute Crit Care. 2026;41(2):399-410. Published online May 28, 2026

DOI: https://doi.org/10.4266/acc.001170

XML Download

Figure

Figure 1. Patient selection flowchart. ICU: intensive care unit.

Figure 2. Receiver operating characteristic curves for intensive care unit mortality according to the nutritional assessment tools. GLIM, Global Leadership Initiative on Malnutrition; AUC, area under the curve; mNUTRIC, modified Nutrition Risk in the Critically Ill; NRS-2002: Nutritional Risk Screening 2002; MUST, Malnutrition Universal Screening Tool.

Figure 3. Receiver operating characteristic curves for intensive care unit mortality comparing the discrimination of the baseline clinical model (age, sex, APACHE II, and SOFA) with models incorporating individual nutritional assessment tools. AUC: area under the curve; GLIM: Global Leadership Initiative on Malnutrition; AUC: area under the curve; mNUTRIC: modified Nutrition Risk in the Critically Ill; NRS-2002: Nutritional Risk Screening 2002; MUST: Malnutrition Universal Screening Tool; APACHE: Acute Physiology and Chronic Health Evaluation; SOFA: Sequential Organ Failure Assessment.

Figure 1.

Figure 2.

Figure 3.

Category	Variable	Value
Demographics	Age (yr)	71±15
	Sex, male	110 (50.5)
	BMI (kg/m²)	22.6±4.0
Comorbidity	At least 1 comorbidity	187 (85.8)
	Hypertension	105 (48.2)
	Diabetes mellitus	58 (26.6)
	Chronic lung disease	9 (4.1)
	Chronic kidney disease	12 (5.5)
	Chronic liver disease	14 (6.4)
	Cardiovascular disease	26 (11.9)
	Neurologic disease	26 (11.9)
	Cancer	60 (27.5)
	Others	24 (11.0)
Diagnosis	Perforation	139 (63.8)
	Colorectal	69 (49.6)
	Upper gastrointestinal	33 (23.7)
	Small bowel	31 (22.3)
	Gallbladder	4 (2.9)
	Other/unspecified	2 (1.4)
	Ischemia	37 (17.0)
	Localized infection	39 (17.9)
	Others	3 (1.4)
Operation	Diversion	34 (15.6)
	Primary repair	24 (11.0)
	Segmental resection	70 (32.1)
	Anastomosis	60 (27.5)
	Others	30 (13.8)
Severity of illness	Septic shock	144 (66.1)
	P/F ratio (mm Hg)	334.8±131.3
	GCS score	9.7±4.7
	SOFA score	7.94±3.75
	APACHE II score	17.04±7.29
Key laboratory value	Albumin (g/dl)	2.65±0.56
	CRP (mg/L)	16.8±11.9
	Lactic acid (mmol/L)	3.08±3.38
	WBC (×10³/µl)	13.6±7.7
	Platelet (×10³/µl)	211±119
	Hematocrit (%)	32.1±7.8
Organ support	Ventilator use	128 (58.7)
	Renal replacement therapy	24 (11.0)

Variable	Survivor (n=176)	Non-survivor (n=42)	P-value
Age (yr)	70±15	77±11	0.002
Sex, male	90 (51.1)	20 (47.6)	0.733
BMI (kg/m²)	22.8±3.9	20.7±3.6	<0.001
SMI (cm²/m²)	36.3±8.5	29.5±7.2	<0.001
GCS score	10.3±4.6	7.1±3.9	<0.001
P/F ratio (mm Hg)	341.0±130.9	309.0±131.6	0.247
Lactic acid (mmol/L)	2.9±1.8	5.0±3.5	<0.001
Albumin (g/dl)	2.6±0.6	2.2±0.5	<0.001
Septic shock	109 (61.9)	35 (83.3)	0.010
SOFA score	7.4±3.5	10.3±3.7	<0.001
APACHE II score	15.7±6.7	22.6±7.0	<0.001
GLIM severe malnutrition	66 (37.5)	27 (64.3)	0.003
mNUTRIC score	4.1±2.1	6.1±1.9	<0.001
mNUTRIC high risk	71 (40.3)	33 (78.6)	<0.001
NRS-2002 score	4.2±1.2	4.9±1.2	<0.001
MUST score	1.0±1.4	2.2±1.5	<0.001
MUST high risk	61 (34.7)	29 (69.0)	<0.001

Variable	No malnutrition (n=64)	Moderate (n=61)	Severe (n=93)	P-value
ICU LOS (day)	8.3±7.7	9.1±10.3	9.4±8.0	0.271
Hospital LOS (day)	23.5±14.1	23.3±16.3	28.5±21.8	0.302
Ventilator use	39 (60.9)	31 (50.8)	57 (61.3)	0.351
Ventilator day	3.6±6.3	5.1±9.0	4.7±7.3	0.581
30-Day ventilator-free days	26.5±5.8	24.7±8.4	25.7±10.1	0.578
RRT	8 (12.5)	5 (8.2)	11 (11.8)	0.704
Complication	25 (39.1)	28 (45.9)	47 (50.5)	0.366
ICU mortality	3 (4.7)	12 (19.7)	27 (29.0)	<0.001

Variable	Low (n=114)	High (n=104)	P-value
ICU LOS (day)	7±7	12±9	<0.001
Hospital LOS (day)	22.3±16.8	29.2±19.5	0.001
Ventilator use	30 (26.3)	97 (93.3)	<0.001
Ventilator day	2.0±6.0	7.2±8.1	<0.001
30-Day ventilator-free days	28.1±8.5	22.9±7.8	<0.001
RRT	2 (1.8)	22 (21.2)	<0.001
Complication	42 (36.8)	58 (55.8)	0.006
ICU mortality	9 (7.9)	33 (31.7)	<0.001

Variable	Low (n=112)	Medium (n=16)	High (n=90)	P-value
ICU LOS (day)	6.5±5.5	7.8±9.6	12.4±10.3	<0.001
Hospital LOS (day)	19.5±11.7	24.2±18.5	33.3±22.1	<0.001
Ventilator use	57 (50.9)	10 (62.5)	60 (66.7)	0.058
Ventilator days	2.7±4.5	4.3±6.4	6.8±9.8	0.002
30-Day ventilator-free day	27.9±7.5	23.8±8.9	23.2±9.1	<0.001
RRT	10 (8.9)	1 (6.3)	13 (14.4)	0.377
Complication	38 (33.9)	7 (43.8)	55 (61.1)	<0.001
ICU mortality	9 (8.0)	4 (25.0)	29 (32.2)	<0.001

Nutritional assessment tool	Univariable OR (95% CI)	P-value	Adjusted OR^a) (95% CI)	P-value
GLIM category (per 1-category increase)	2.449 (1.501–3.994)	<0.001	2.64 (1.47–4.73)	0.001
mNUTRIC score (per 1-point increase)	1.667 (1.367–2.032)	<0.001	1.17 (0.79–1.74)	0.439
NRS-2002 score (per 1-point increase)	1.525 (1.169–1.990)	0.002	Not estimable^b)	0.999
MUST score (per 1-point increase)	1.584 (1.278–1.965)	<0.001	1.53 (1.20–1.96)	<0.001

Component	Univariable OR (95% CI)	P-value	Multivariable OR^a) (95% CI)	P-value
Weight loss history	0.96 (0.26–3.53)	0.949	-	-
Decreased feeding before ICU admission	0.60 (0.30–1.18)	0.137	-	-
Expected no nutritional intake for >5 days	5.68 (2.70–11.95)	<0.001	4.59 (2.08–10.11)	<0.001
BMI decrease (per 1 kg/m²)	1.17 (1.05–1.29)	0.003	1.08 (0.95–1.22)	0.228
Reduced SMI (sex-specific cutoff)	5.30 (1.81–15.52)	0.002	4.14 (1.07–15.98)	0.039

Table 1. Patient demographics and characteristics

Values are presented as mean±standard deviation or number (%).

Table 2. Baseline characteristics of survivors and non-survivors

Values are presented as mean±standard deviation or number (%).

BMI: body mass index; SMI: skeletal muscle index; GCS: Glasgow Coma Scale; P/F ratio: PaO₂/FiO₂ ratio;

SOFA: Sequential Organ Failure Assessment; APACHE: Acute Physiology and Chronic Health Evaluation;

GLIM: Global Leadership Initiative on Malnutrition; mNUTRIC: modified Nutrition Risk in the Critically Ill; NRS-2002: Nutritional Risk Screening 2002; MUST: Malnutrition Universal Screening Tool.

Table 3. Clinical course according to nutritional assessment tools: GLIM

Values are presented as mean±standard deviation or number (%).

GLIM: Global Leadership Initiative on Malnutrition; ICU: intensive care unit; LOS: length of stay; RRT: renal replacement therapy.

Table 4. Clinical course according to nutritional assessment tools: mNUTRIC

Values are presented as mean±standard deviation or number (%).

mNUTRIC: modified Nutrition Risk in the Critically Ill; ICU: intensive care unit; LOS: length of stay; RRT: renal replacement therapy.

Table 5. Clinical course according to nutritional assessment tools: MUST

Values are presented as mean±standard deviation or number (%).

MUST: Malnutrition Universal Screening Tool; ICU: intensive care unit; LOS: length of stay; RRT, renal replacement therapy.

Table 6. Logistic regression analysis of nutritional assessment tools for ICU mortality

Adjusted for age, sex, APACHE II score, and SOFA score;

NRS-2002 was not estimable in the adjusted model because nearly all patients were classified as being at nutritional risk.

Table 7. Exploratory univariable and multivariable logistic regression analyses of nutrition-related components for ICU mortality

ICU: intensive care unit; OR: odds ratio; BMI: body mass index; SMI: skeletal muscle index.

Multivariable model included candidate nutrition-related variables selected from univariable analysis (P<0.10), with adjustment for age and sex.