Association between nutritional risk scores and timing of endotracheal intubation in COVID-19-associated acute respiratory distress syndrome: a single-center cohort study in South Korea
Article information
Abstract
Background
The optimal timing of endotracheal intubation in patients with coronavirus disease 2019 (COVID-19)-associated acute respiratory distress syndrome (ARDS) remains uncertain, and delayed intubation is associated with worse outcomes. Nutritional status, known to affect respiratory function and immune response, may help identify patients at risk of rapid deterioration. This study aimed to evaluate whether nutritional risk scores can predict early intubation in COVID-19-associated ARDS.
Methods
We retrospectively analyzed 247 patients with COVID-19-associated ARDS admitted to a tertiary hospital intensive care unit. Nutritional status at admission was assessed using the modified Nutrition Risk in the Critically Ill (mNUTRIC) score and the Prognostic Nutritional Index (PNI). Early intubation was defined as occurring within 24 hours of hospital admission. Receiver operating characteristic curves and multivariate logistic regression were used to evaluate predictive performance
Results
Of 247 patients, 193 (78.1%) required mechanical ventilation, and 133 (68.9%) underwent early intubation. The mNUTRIC score showed moderate discriminatory performance (area under the curve [AUC], 0.705), while PNI performed poorly (AUC, 0.401). In a multivariate analysis adjusted for illness severity, only Acute Physiology and Chronic Health Evaluation II (odds ratio [OR], 1.206; P<0.001) and Sequential Organ Failure Assessment scores (OR, 1.270; P=0.028) were independent predictors of early intubation. The mNUTRIC score was not independently associated (P>0.05), suggesting its value is derived from component severity.
Conclusions
The predictive power of the mNUTRIC score for early intubation in COVID-19 ARDS was primarily driven by its embedded illness severity components. Nevertheless, the score demonstrated practical utility as a single, composite marker for rapid, holistic evaluation of patient risk.
INTRODUCTION
Acute respiratory distress syndrome (ARDS) frequently complicates the clinical course of patients with coronavirus disease 2019 (COVID‑19), often leading to the need for mechanical ventilation. The timing of endotracheal intubation remains challenging because both premature and delayed interventions can impact outcomes. Specifically, delayed intubation beyond 24 hours after intensive care unit (ICU) admission has been linked to higher mortality, underscoring the need for timely risk assessment [1-3]. However, a persistent challenge lies in identifying objective predictors of respiratory deterioration to guide the timing of intubation.
Malnutrition has long been recognized as a factor that worsens the clinical trajectory of critical illness by impairing immune function, diminishing respiratory muscle strength, and prolonging recovery. Timely and accurate evaluation of nutritional status may offer crucial insights to help identify patients who are more vulnerable to respiratory decline and could benefit from earlier airway intervention. Objective evaluation of nutritional risk may help clinicians identify patients who are more likely to require early airway management, supporting timely and appropriate decision-making. Although several nutritional assessment tools have been developed and validated to estimate clinical risk and predict outcomes in ICUs [4-7], their utility in guiding the timing of endotracheal intubation, especially in the context of coronavirus disease 2019 (COVID-19)-associated ARDS, remains insufficiently explored.
To address this evidence gap, we evaluated whether objective nutritional assessment at ICU admission can inform the timing of endotracheal intubation in patients with COVID-19-associated ARDS. Specifically, we assessed two widely used nutritional scoring systems, the modified Nutrition Risk in the Critically Ill (mNUTRIC) score and the Prognostic Nutritional Index (PNI), to determine their associations with early intubation, defined as occurring within 24 hours of hospital admission [4,5]. By retrospectively analyzing a well-characterized cohort of critically ill patients, this study aimed to explore the potential utility of nutritional risk assessment in guiding early airway intervention during the acute phase of severe viral respiratory illness.
MATERIALS AND METHODS
The study population was drawn from a prospectively enrolled cohort of patients with pneumonia-related ARDS that has been maintained since 2013 under the approval of the Institutional Review Board of Pusan National University Hospital (No. 1212-009-010). For the current analysis focusing on patients with COVID-19-associated ARDS, additional approval was obtained from the same IRB (No. 2504-010-150). The requirement for informed consent was waived because of the retrospective and observational nature of this study.
Study Design and Patient Selection
This observational cohort study was conducted at a university-affiliated tertiary hospital with a capacity of 1,100 beds. The study population was drawn from a prospectively enrolled cohort of patients with pneumonia-related ARDS that was established in 2013. Among this cohort, patients diagnosed with COVID-19-induced ARDS were retrospectively identified and included in the analysis. Data were collected between March 2020 and August 2024, including the COVID-19 pandemic.
In the early stages of the pandemic, critically ill patients with COVID-19 were managed in the National Designated Isolated ICU, a specialized unit consisting of 18 beds, which operated from December 30, 2020, to May 30, 2022. This ICU was temporarily implemented within a regional medical center to provide focused care, including cardiovascular monitoring and continuous respiratory support. After this unit ceased operation, care for patients with COVID-19-associated ARDS was transitioned to the hospital’s designated respiratory ICU, in accordance with guidelines of the Busan Civil Facilitation Division.
The study included adult patients (aged ≥18 years) who were confirmed to have COVID-19 by reverse transcription polymerase chain reaction and were diagnosed with ARDS based on the Berlin definition [8]. Eligible cases were identified among patients admitted to the National Designated Isolated ICU or the designated respiratory ICU during the study period. The primary outcome was the incidence of early endotracheal intubation, defined as intubation performed within 24 hours of hospital admission. Secondary outcomes were ICU length of stay, hospital length of stay, 28-day mortality, and in-hospital mortality.
All enrolled patients received care based on lung-protective ventilation strategies [9], alongside standard pharmacologic treatments with dexamethasone and/or remdesivir, as per prevailing therapeutic protocols [10,11]. Mechanical ventilation was supported by basic ICU rehabilitation measures, such as early mobilization and respiratory therapy, when appropriate [12].
Data Collection
Demographic information (age, sex, and body mass index), clinical outcomes (length of stay, 28-day mortality, and in-hospital mortality), and underlying comorbidities were prospectively collected for a cohort of patients with pneumonia-related ARDS. On the day of hospital admission, disease severity was assessed using the Acute Physiology and Chronic Health Evaluation (APACHE) II score [13], and organ dysfunction was evaluated via the Sequential Organ Failure Assessment (SOFA) score [14]. The Charlson Comorbidity Index was used to quantify pre-existing comorbidities [15].
In addition, a range of laboratory parameters, including procalcitonin, C-reactive protein, lactic acid, albumin, pro–B-type natriuretic peptide (pro-BNP), and the ratio of partial pressure of arterial oxygen to fraction of inspired oxygen (PaO2/FiO2), were recorded. The lactate-to-albumin ratio was calculated by dividing the serum lactate concentration (mmol/L) by the serum albumin concentration (g/dl), as previously described [16,17]. The number of patients who received renal replacement therapy, prone positioning, or extracorporeal membrane oxygenation (ECMO) or who underwent tracheostomy during hospitalization was also documented. Renal replacement therapy was defined as any form of hemodialysis initiated on the day of mechanical ventilation or within 72 hours thereafter.
For the identified cohort of patients with COVID-19-associated ARDS, the mNUTRIC score and PNI were retrospectively calculated using clinical and laboratory data obtained on the day of hospital admission [4,5]. This cohort included both patients who underwent endotracheal intubation and those who did not. Non-intubated patients were retrospectively enrolled based on the 2023 global definition of ARDS, which allows inclusion of patients with non-invasive respiratory support who meet standardized diagnostic criteria [18]. In terms of management, prone positioning was based on the criteria outlined in the Prone Positioning in Severe Acute Respiratory Distress Syndrome trial [19,20], and ECMO was initiated in patients who met eligibility criteria from the Extracorporeal Membrane Oxygenation for Severe Acute Respiratory Distress Syndrome trial [21].
The primary outcome of this study was the occurrence of early endotracheal intubation, defined as intubation performed within 24 hours of hospital admission. While many studies use ICU admission as a reference point, the hospital admission standard was chosen for a critical clinical reason. During the COVID-19 pandemic, many patients deteriorated rapidly and required intubation in the emergency department or general wards before an ICU bed became available. Using hospital admission as the zero-time allowed a more comprehensive and accurate capture of the entire acute illness trajectory from the patient's first contact with the healthcare system, avoiding potential misclassification of pre-ICU intubations. The time from admission to intubation was measured for each patient. Patients who underwent intubation within 24 hours were classified as the early intubation group, while those who underwent intubation after 24 hours comprised the late intubation group [3].
Statistical Methods
Continuous variables were summarized as medians with interquartile ranges, and categorical variables were presented as frequencies with percentages. Differences between groups were assessed using the Mann-Whitney U-test for continuous variables and the chi-square or Fisher’s exact test for categorical variables, as appropriate. To identify independent predictors of early endotracheal intubation, a multivariate logistic regression analysis was performed using a fixed model that included variables of clinical importance and those with a P-value <0.1 in the univariate analysis. This was performed to explicitly assess the independent contribution of the mNUTRIC score after adjusting for key confounders such as illness severity. Results were reported as adjusted odds ratios with 95% CIs. Model calibration was assessed using the Hosmer-Lemeshow goodness-of-fit test.
The discriminative performance of the mNUTRIC score and PNI for predicting early intubation was evaluated by receiver operating characteristic (ROC) curve analysis. The area under the curve (AUC), optimal cutoff values (determined by Youden’s index), sensitivity, specificity, and predictive values were calculated [22]. To assess the linear relationship between nutritional scores and the binary outcome of early intubation, point-biserial correlation coefficients were calculated for both the mNUTRIC score and PNI [23]. Violin plots overlaid with swarm plots were constructed to display the distribution and individual values of the mNUTRIC score and PNI according to intubation timing. These visualizations were generated using the Seaborn library in Python to enhance the interpretability of score differences between groups. A two-sided P-value <0.05 was considered statistically significant. All statistical analyses were performed using R software version 4.3.1 (R Foundation for Statistical Computing).
RESULTS
Patient Characteristics
A total of 247 adult patients with COVID‑19‑associated ARDS were included, 193 (78.1%) of whom required mechanical ventilation. Among these 193 patients, 133 were intubated within 24 hours of hospital admission, and 86 of them survived for more than 28 days (Figure 1). Patients in the intubated group were older, had a lower body mass index, and had higher APACHE II and SOFA scores at ICU admission than those in the non-intubated group. Comorbid conditions such as diabetes and chronic neurologic diseases were more common and the Charlson Comorbidity Index was higher in the intubated group than in the non-intubated group. Nutritionally, patients in the intubated group had higher mNUTRIC scores and lower PNIs than those in the non-intubated group. The 28‑day and in‑hospital mortality rates in the intubated group were 32.6% and 40.4%, respectively (Table 1).
Flowchart. MV: mechanical ventilation.
Comparison of intubated and non-intubated patients
Comparison of the Early and Late Intubation Groups
Among patients who required mechanical ventilation, 133 (68.9%) were intubated within 24 hours of hospital admission and were classified as the early intubation group, while the remaining 60 (31.1%) constituted the late intubation group. Patients in the early intubation group were significantly older and exhibited greater illness severity at ICU admission, as reflected by higher APACHE II and SOFA scores, than those in the late intubation group. Despite this, patients in the early intubation group had significantly shorter ICU and hospital lengths of stay than those in the late intubation group.
The mNUTRIC score was higher and the PNI was lower in the early intubation group than in the late intubation group. The PaO2/FiO2 ratio at admission was lower and the levels of C-reactive protein, procalcitonin, and pro-BNP and the lactic acid-to-albumin ratio were higher in the early intubation group than in the late intubation group. However, the proportions of patients who received prone positioning, ECMO, and renal replacement therapy did not significantly differ between the groups. There was no significant difference in 28-day or in-hospital mortality between the groups (Table 2).
Comparison of clinical characteristics between patients in the early and late intubation groups
Discriminatory Performance of Nutritional Scores for Early Intubation
To evaluate the ability of nutritional assessment scores to distinguish early intubation status, ROC curve analyses were performed. The mNUTRIC score had an AUC of 0.705, with a cutoff value of 4 based on Youden’s index. To further explore its clinical applicability, we additionally evaluated its performance at the previously proposed high-risk cutoff value of 5 [24-26]. At this threshold, the mNUTRIC score demonstrated high specificity (0.867; 95% CI, 0.758–0.931) but relatively low sensitivity (0.406; 95% CI, 0.326–0.491). By contrast, the PNI demonstrated limited discriminatory ability for early intubation, with an AUC of 0.401, indicating it has poor diagnostic performance in this clinical context (Table 3).
Predictive performance of the mNUTRIC score and PNI for early intubation
Violin and box plots were used to compare the distributions of mNUTRIC scores and PNIs between the early and late intubation groups. The mNUTRIC score was significantly higher in the early intubation group than in the late intubation group, as shown in violin (Figure 2A) and box (Figure 2B) plots. These plots revealed a broader distribution and higher median mNUTRIC score in the early intubation group than in the late intubation group. By contrast, the PNI was slightly lower in the early intubation group than in the late intubation group, as visualized in violin (Figure 2C) and box (Figure 2D) plots. Point-biserial correlation analysis confirmed a moderate positive association between the mNUTRIC score and early intubation (r=0.334, P<0.001), while the PNI showed a weaker inverse correlation (r=–0.149, P=0.038).
Distributions of modified Nutrition Risk in the Critically Ill (mNUTRIC) score and Prognostic Nutritional Index (PNI) according to intubation timing. (A) Violin plot of mNUTRIC scores, showing a higher distribution in the early intubation group than in the late intubation group. (B) Box plot of mNUTRIC scores comparing the early and late intubation groups. (C) Violin plot of PNIs, showing a slightly lower distribution in the early intubation group than in the late intubation group. (D) Box plot of PNIs comparing the early and late intubation groups. The point-biserial correlation coefficient (r) and corresponding P-value are provided for each score.
Independent Factors Associated with Early Intubation
To clarify the independent predictors of early endotracheal intubation, we performed a multivariate logistic regression analysis. All variables that showed potential significance in the univariate analysis (P<0.1) were included in the comprehensive model to specifically assess the independent contribution of the mNUTRIC score after adjusting for key confounders.
The results of this analysis are presented in Table 4. In the final adjusted model, the higher APACHE II and SOFA scores and presence of pulmonary comorbidity remained significant independent predictors of early intubation. Crucially, after adjusting for these powerful severity and comorbidity factors, the mNUTRIC score was no longer a significant predictor. Similarly, other variables such as age, PNI, and C-reactive protein level also did not show an independent association. This finding strongly suggests that the association of the mNUTRIC score with early intubation, as observed in the univariate analysis, was primarily explained by the significant influence of the severity of illness components embedded within the score.
Multivariate logistic regression analysis of factors associated with early endotracheal intubation
DISCUSSION
This study aimed to evaluate the role of nutritional assessment in predicting early intubation for patients with COVID-19-associated ARDS. Our initial analysis showed a strong association between a high mNUTRIC score and need for early intubation. However, a more rigorous multivariate analysis revealed a crucial finding: this association was predominantly driven by the powerful effect of acute illness severity, a core component of the mNUTRIC score, rather than by nutritional status as an independent factor. This finding underscores that physiological stress in the hyperacute phase of COVID-19 ARDS was the most dominant determinant for rapid clinical deterioration and the need for advanced airway management.
Although the mNUTRIC score did not emerge as an independent predictor after adjusting for its core components, its clinical utility should not be dismissed. In a high-pressure critical care environment, the primary value of the mNUTRIC score lies in its function as a practical, all-in-one risk stratification tool. It efficiently synthesizes several critical prognostic variables—age, comorbidity, inflammation, and severity (APACHE II and SOFA score)—into a single, easily accessible number. For frontline clinicians, this score can facilitate a rapid, holistic assessment of a patient's overall risk, serving as a useful adjunct for comprehensive clinical judgment.
When evaluating each score’s individual discriminatory performance using ROC analysis, only the mNUTRIC score demonstrated clinically useful potential, whereas the PNI showed limited value. This discrepancy may be attributable to the design of each tool. The mNUTRIC score incorporates markers of both illness severity and nutritional risk and is more suitable for critically ill patients with complex physiological stress, such as those with ARDS. The PNI, which is primarily based on albumin and lymphocyte count, may not adequately capture the multifactorial nature of respiratory deterioration in patients with COVID-19-related ARDS. These results indicate that composite tools like the mNUTRIC score, which account for both acute physiological stress and nutritional status, may have superior clinical utility in guiding early intervention decisions in critically ill patients.
In addition to the findings regarding intubated patients, we observed that patients who did not require endotracheal intubation had lower mNUTRIC scores and higher PNIs. This further supports the hypothesis that better baseline nutritional status may be associated with reduced respiratory deterioration and avoidance of invasive ventilation. These differences support the role of nutritional assessment in predicting not only the likelihood of early intubation among ventilated patients but also the capacity to maintain respiratory stability without invasive ventilation in better-nourished individuals.
While a cutoff value of 5 was previously proposed to define high nutritional risk using the mNUTRIC score [24-27], our findings indicate that this threshold may offer limited sensitivity for predicting early intubation in patients with COVID-19-associated ARDS. Although the cutoff demonstrated high specificity, its low sensitivity may reduce its utility as a screening tool for early intervention. A cutoff value of 4, which was identified through Youden’s index in our ROC analysis, provided a more balanced diagnostic performance. Nevertheless, the overall discriminative power of the mNUTRIC score was moderate, with an AUC of 0.705. This suggests that, while nutritional risk is an important clinical factor, it may be insufficient on its own to reliably guide intubation timing. These results underscore the need to establish context-specific thresholds and develop comprehensive predictive models that integrate nutritional, inflammatory, and respiratory parameters to better reflect the complex pathophysiology of acute respiratory failure.
Our findings suggest that nutritional assessment scores may be useful indicators to determine the timing of endotracheal intubation in patients with COVID-19-associated ARDS. While this study focused on two commonly used indices, the mNUTRIC score and PNI, other nutritional scoring systems and biomarkers, such as the Controlling Nutritional Status score [6] and the serum prealbumin level [7], may warrant evaluation of their predictive utility in this context. However, the nutritional data used in our analysis were obtained retrospectively from electronic medical records, and the completeness and consistency of the dataset were inherently limited. Therefore, prospective studies are needed to validate our findings and explore the potential role of additional nutritional assessment tools in guiding early respiratory intervention.
This study has several limitations. First, its retrospective, single-center design limits the generalizability of the findings to other patient populations and healthcare settings. Second, although nutritional assessment was performed using established tools, the data were retrospectively collected from electronic medical records, and the accuracy and completeness of the score components could not be ensured. Third, the decision of whether to perform endotracheal intubation and its timing were determined by individual clinicians rather than using a standardized protocol, introducing potential variability that may have influenced outcomes.
In conclusion, this study found that acute illness severity was the primary determinant for early endotracheal intubation in patients with COVID-19-associated ARDS. The predictive ability of the mNUTRIC score was predominantly explained by its embedded severity components and not by nutritional status as an independent factor. Nevertheless, the mNUTRIC score served as a practical and useful tool for a rapid, integrated assessment of overall patient risk at bedside. These findings emphasize that clinical decisions regarding timely airway management in this patient population should be guided primarily by objective markers of physiological severity.
KEY MESSAGES
▪ In patients with coronavirus disease 2019 (COVID-19)-associated acute respiratory distress syndrome, illness severity was the primary predictor of early endotracheal intubation.
▪ The predictive ability of the modified Nutrition Risk in the Critically Ill (mNUTRIC) score was mainly driven by the severity of its illness components, rather than nutritional status as an independent factor.
▪ The mNUTRIC score served as a practical, all-in-one tool for a rapid overview of patient risk by integrating key prognostic factors into a single score.
Notes
CONFLICT OF INTEREST
Kwangha Lee is an editorial board member of the journal but was not involved in the peer review selection, evaluation, or decision process of this article. No other potential conflict of interest relevant to this article were reported.
FUNDING
This work was supported by a 2-Year Research Grant from Pusan National University (Grant No. 202216520003).
ACKNOWLEDGMENTS
We thank Hee Jin Hong, RN and Young Nam Kim, RN, for their assistance in reviewing the electronic medical records.
AUTHOR CONTRIBUTIONS
Conceptualization: HJ, KL. Methodology: WY, HJ. Formal analysis: WY, HJ. Data curation: WY, HJ. Visualization: HJ, WY, KL. Project administration: HJ, WY, KL. Funding acquisition: KL. Writing - original draft: HJ, KL. Writing - review and editing: HJ, KL. All authors read and agreed to the published version of the manuscript.