The role of the nutritional screening tools in predicting clinical outcomes among older adults with hemorrhagic cerebrovascular disease

Abstract

Background

Malnutrition is associated with poor prognosis in critically ill elderly patients with hemorrhagic cerebrovascular disease (HCVD), and in this study, the effectiveness of nutritional assessment tools such as Nutritional Risk Screening 2002 (NRS-2002), Controlling Nutritional Status (CONUT), and Geriatric Nutritional Risk Index (GNRI), as well as the severity-of-illness score Acute Physiology and Chronic Health Evaluation II (APACHE-II), was evaluated for predicting prognosis in this population.

Methods

This study was designed as a retrospective cohort study. Patients aged 65 years and older who were admitted to the intensive care unit from the emergency department with a diagnosis of HCVD between January 1, 2018, and June 30, 2024, were included. Patients with incomplete medical records and those with post-traumatic or mass-related intracranial hemorrhage were excluded. Clinical and laboratory findings and outcomes of all patients, were recorded. The NRS-2002 (3 parameters: nutritional status, disease severity, and age), APACHE-II (12 physiological parameters), CONUT (serum albumin, total cholesterol, and lymphocyte count), and GNRI (serum albumin and body mass index) scores were calculated. Statistical analysis was performed using the Chi-square test for categorical variables and the Mann–Whitney U test for continuous variables. Receiver Operating Characteristic (ROC) curve analysis was conducted to evaluate the predictive value of nutritional screening scores on mortality. A p-value of < 0.05 was considered statistically significant.

Results

A total of 113 patients were included, with an overall mortality rate of 57.5%. The NRS-2002 showed weak predictive ability for mortality (AUC = 0.67, p = 0.002). The APACHE-II demonstrated very good predictive performance (AUC = 0.84, p < 0.001). The CONUT score showed excellent discrimination (AUC = 0.92, p < 0.01), while the GNRI also demonstrated very good predictive value (AUC = 0.83, p < 0.001).

Conclusion

The CONUT score may serve as a valuable tool for assessing nutritional risk in patients with HCVD, both in the emergency department and in the intensive care unit.

Introduction

Malnutrition is a common and serious problem, especially in intensive care unit (ICU) patients. Its prevalence is 20–40% in the general hospital population and increases to 38–78% in ICU settings, depending on baseline nutritional status and assessment methods [1]. Malnutrition adversely affects clinical outcomes by delaying wound healing, increasing infection risk, and reducing response to treatment. In critically ill patients, it is an independent risk factor for prolonged ICU stay, recurrent hospitalization, and increased mortality [1, 2]. Therefore, various nutritional screening and assessment tools have been developed to identify malnutrition risk and to predict related complications and mortality [3].

Older adults are at increased risk of malnutrition due to physiological decline and psychosocial factors. Malnutrition is associated with sarcopenia, osteoporosis, geriatric syndromes, and adverse health outcomes. In hospitalized elderly patients, comorbidities and polypharmacy further increase this risk and are associated with higher morbidity and in-hospital mortality [4]. Therefore, accurate nutritional assessment is important for prognostic evaluation in this population, although it is often challenging due to age-related changes, reduced food intake, and multiple comorbidities [2, 5–7].

There is a general consensus among international nutrition and healthcare organizations that nutritional screening is essential for identifying patients who require further nutritional assessment. According to the American Society for Parenteral and Enteral Nutrition (ASPEN) clinical guidelines, although multiple tools are available to assess malnutrition risk, none is considered a gold standard. This underscores the need for practical and reliable assessment methods tailored to specific clinical settings [3, 5, 8]. Studies have emphasized the importance of routinely conducting nutritional screening in stroke patients, while noting that further evidence is needed to determine which assessment tool is most appropriate [9].

Hemorrhagic cerebrovascular disease (HCVD) patients are at increased risk of malnutrition due to impaired consciousness, dysphagia, and increased catabolic stress. This specific population allows assessment of the prognostic impact of malnutrition with fewer confounding factors. In this study, the Nutritional Risk Screening 2002 (NRS-2002), Geriatric Nutritional Risk Index (GNRI), Controlling Nutritional Status (CONUT) score, and the Acute Physiology and Chronic Health Evaluation II (APACHE II) score were applied in geriatric patients with HCVD. All scoring systems used in this study are previously published and validated tools. The APACHE II score is a widely used severity-of-disease classification system for critically ill patients [10]. The NRS-2002, CONUT, and GNRI have been validated in hospitalized and elderly populations [11–13]. These tools were selected based on their clinical applicability, suitability for emergency and intensive care settings, and the availability of required parameters at hospital admission. The aim of this study was to evaluate the prognostic value of these tools in predicting in-hospital mortality and length of hospital stay among geriatric patients presenting to the emergency department (ED).

Method

Study design

This retrospective cohort study included patients aged ≥ 65 years who were admitted from the ED to the neurology and neurosurgery ICU with a diagnosis of HCVD between January 1, 2018, and June 30, 2024. Ethics approval for this study was obtained from the Clinical Research Ethics Committee of the University of Health Sciences, Adana City Training and Research Hospital (Date: 10/10/2024; Meeting No: 6; Decision No: 183). Owing to the retrospective nature of the study and the use of fully anonymized patient data, the requirement for written informed consent to participate was waived by the same Institutional Review Board. The study was conducted in accordance with the principles of the Declaration of Helsinki and the Good Clinical Practice guidelines. The study population was defined as patients with acute intracerebral hemorrhage confirmed by non-contrast computed tomography (CT) in the ED. All patients underwent standard clinical, neurological, and laboratory evaluations at ED admission, which were used to determine eligibility and collect baseline data. Following initial stabilization and diagnostic confirmation, patients were transferred to the intensive care unit. During ED stay, patients were generally kept nil per os, and enteral nutrition was not initiated. The primary outcome of the study was in-hospital mortality, and secondary outcomes included length of hospital stay and other clinical parameters.

Participant

Patients aged ≥ 65 years who were admitted to the ICU from the ED with a diagnosis of HCVD were included in the study. The study focused on elderly patients because advanced age is associated with increased vulnerability to malnutrition and poorer outcomes in critically ill patients. Patients younger than 65 years, those with unavailable or incomplete medical records, or those with insufficient clinical or laboratory data were excluded. In addition, patients with HCVD who were not admitted to the ICU—such as those managed in general wards or transferred to other institutions—were excluded. Patients with post-traumatic intracranial hemorrhage, intracranial masses, or intralesional hemorrhage were also excluded. For the purposes of this study, HCVD was defined as acute intracerebral hemorrhage confirmed by CT and interpreted by an experienced radiologist. All eligible patients during the study period were consecutively included. The patient selection process is summarized in a study flowchart (Fig. 1).

Fig. 1.

Flowchart of the study

Data collection

The standard data collection protocol included demographic characteristics, vital signs, and laboratory parameters obtained from the patients’ medical records. Complete blood count (CBC) parameters were analyzed using an automated hematology analyzer (XN-1000, Sysmex Corporation, Kobe, Japan; 2014), which operates based on flow cytometry with semiconductor laser technology and hydrodynamic focusing, providing a five-part leukocyte differential. Whole blood samples collected in EDTA tubes were used for analysis. Calibration and internal quality control procedures were performed in accordance with the manufacturer’s instructions.Biochemical parameters were measured using a fully automated clinical chemistry analyzer (AU5811, Beckman Coulter Inc., Brea, CA, USA; 2016), which operates on photometric measurement principles and ion-selective electrode technology. Serum samples were used for biochemical analyses. Calibration and internal quality control procedures were conducted according to the manufacturer’s guidelines.

Length of hospital stay and discharge status were recorded. In-hospital mortality was identified through the hospital’s electronic medical record system and review of patient reports.

Body mass index (BMI) was calculated using weight and height data obtained from medical records. When direct measurements were not feasible, these values were estimated based on patient- or family-reported information, in accordance with standard hospital protocols. For calculation of the NRS-2002 score, body weight loss within the previous three months, reduced dietary intake, and the presence of chronic disease were obtained through a detailed history from the patient or their relatives Patients with an NRS-2002 score ≥ 3 were considered at risk of malnutrition [11]. The APACHE-II scoring system utilized after age, twelve physiological measurements, and other parameters based on medical history were recorded [10]. The CONUT score was calculated based on serum albumin concentration, lymphocyte count, and total cholesterol levels measured at the time of initial presentation. CONUT scores were categorized as follows: 0–1, normal nutritional status; 2–4, mild risk; 5–8, moderate risk; and 9–12, severe risk of malnutrition [14]. The GNRI scores were calculated using serum albumin levels measured at admission and the body mass index (BMI) [12]. GNRI values were classified as > 98 (no risk), 92–98 (low risk), 82–92 (moderate risk), and < 82 (high risk of malnutrition). Each parameter was assigned a value based on specific laboratory and numerical ranges and recorded accordingly.

All indices and laboratory measurements were based on data obtained at the time of admission to the emergency department.

Statistical analysis

Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS), version 25.0. Categorical variables were summarized as frequencies and percentages, while continuous variables were expressed as mean ± standard deviation, median, minimum, and maximum values. The chi-square test was used for comparisons of categorical variables. The Shapiro–Wilk test was applied to assess the normality of continuous variables. Normally distributed variables were analyzed using the independent-samples t-test, whereas non-normally distributed variables were analyzed using the Mann–Whitney U test. Spearman’s rho correlation analysis was used to evaluate associations between continuous variables.

A multivariable logistic regression analysis was performed to assess the association between nutritional indices and in-hospital mortality. Mortality was defined as the dependent variable, and GNRI, NRS-2002, CONUT, and APACHE-II scores were included as independent variables. To control for potential confounding factors, age and sex, as well as comorbidities and laboratory parameters considered to be associated with mortality, were included in the model. Results were presented as odds ratios (OR) with 95% confidence intervals.

Receiver Operating Characteristic (ROC) curve analysis was conducted to evaluate the discriminatory performance of nutritional screening scores (GNRI, CONUT, and NRS-2002) and the APACHE-II score in predicting mortality. Optimal cutoff values were determined, and corresponding sensitivity and specificity values were calculated.

Survival analysis was performed using the Kaplan–Meier method, and group comparisons were conducted using the log-rank test. A Cox proportional hazards model was not applied. A p-value of < 0.05 was considered statistically significant for all analyses.

Results

A total of 113 patients were included in the study, of whom 48 (42.5%) were female and 65 (57.5%) were male. The mean age was 76.2 ± 7.6 years. During the treatment and follow-up period, in-hospital mortality was observed in 65 patients (57.5%). Hypertension was the most common comorbidity (82 patients, 72.6%), whereas malignancy and chronic kidney disease were the least frequent (7 patients, 6.2%). Approximately two-thirds of the patients required mechanical ventilatory support, and one-quarter underwent external ventricular drainage placement. The mean total hospital length of stay was 22.9 ± 34.6 days, and the mean ICU length of stay was 19.8 ± 34.1 days (Table 1).

Table 1.

Descriptive characteristics of the study group

Characteristic (n = 113)	n	%
Gender
Female	48	42.5
Male	65	57.5
Mortality
Alive	48	42.5
Deceased	65	57.5
Hypertension
Yes	82	72.6
No	31	27.4
Diabetes Mellitus
Yes	34	30.1
No	79	69.9
Coronary Artery Diseases
Yes	29	25.7
No	84	74.3
Malignancy
Yes	7	6.2
No	106	93.8
Chronical Kidney Disease
Yes	7	6.2
No	106	93.8
History of Stroke
Yes	15	13.3
No	98	86.7
Extraventricular Drainage
Performed	27	23.9
Not Performed	86	76.1
Mechanical ventilation support
Performed	74	65.5
Not Performed	39	34.4
Mean Age (years)	76.2 (± 7.6 )
Mean ICU length of stay (days)	19.8 (± 34.1 )
Median ICU length of stay (days) (min–max)	9 (1– 306)
Mean hospital length of stay (days)	22.9 (± 34.6 )
Median hospital length of stay (days) (min–max)	13.0 (1.0–306.0)

ICU Intensive care unit, min Minimum, max Maximum

Mortality was significantly higher in patients with hypertension (p = 0.027), in those who underwent extraventricular drainage (p = 0.004), and in those who were mechanical ventilatory support (p < 0.001) compared to patients without these conditions (Table 2). No significant associations were observed between mortality and gender, diabetes mellitus, coronary artery disease, malignancy, chronic kidney disease, history of stroke, age, or ICU length of stay (p > 0.05 for all; Table 2). The mean total hospital length of stay was significantly lower in deceased patients compared to survivors (p < 0.001; Table 2).

Table 2.

Impact of various characteristics and parameters on mortality

Characteristic (n = 92)		Survived (n = 48)		Deceased (n = 65)		p* value
		n	%	n	%
Gender	Female	18	37.5	30	62.5	0.358
	Male	30	46.2	35	53.8
Hypertension	Yes	40	48.8	42	51.2	0.027
	No	8	25.8	33	74.2
Diabetes mellitus	Yes	13	38.2	21	61.8	0.549
	No	35	44.3	44	55.7
Coronary artery disease	Yes	11	37.9	18	62.1	0.566
	No	37	44	47	56
Malignancy	Yes	1	14.3	6	85.7	0.235**
	No	47	44.3	59	55.7
Chronic kidney disease	Yes	4	57.1	3	42.9	0.456**
	No	44	41.5	62	58.5
Stroke	Yes	5	33.3	10	66,7	0.442
	No	43	43.9	55	56.1
Extraventricular drainage	Performed	5	18.5	22	81.5	0.004
	Not Performed	43	50,0	43	50
Mechanical ventilation support	Performed	9	12.2	65	87.8	< 0.001
	Not Performed	39	100	0	0

Characteristic (n = 113)		Survived (n = 48)		Deceased (n = 65)		p*** value
		Mean (± Sd)	Median (min-max)	Mean (± Sd)	Median (min-max)
Age (years)		75.7 (± 7.7)	75 ( 65–95)	76.5 (± 7.6)	75 ( 65.0–97.0)	0.606
ICU Length of Stay (days)		24.9 (± 47.5)	10.5 ( 2-306)	16.0 (± 18.4)	9.0 ( 1–91)	0.378****
Total Hospital Length of Stay (days)		32.2 (± 47.3)	20 ( 4-306)	16 (± 18.4)	9 ( 1–91)	< 0.001****

ICU Intensive Care Unit, min Minimum, max Maximum, SD Standard deviation

*Chi-square test was used

**Fisher’s exact test was used

***t-test was used

****Mann-Whitney U test was used

Analysis of laboratory parameters showed that deceased patients had significantly higher levels of glucose, urea, creatinine, alanine aminotransferase, aspartate aminotransferase, sodium, neutrophil count, procalcitonin, and C-reactive protein, and significantly lower levels of calcium, hemoglobin, platelet count, lymphocyte count, albumin, and total protein compared to survivors (p < 0.05 for all; Table 3).

Table 3.

Relationship of vital signs and various laboratory parameters with mortality

Vital Signs and Laboratory Parameters (n = 113)	Survived (n = 48) Mean (± SS)	Deceased (n = 65) Mean (± SS)	p* value
Body Temperature	36.5 ± 0.3	36.5 ± 0.3	0.641
Heart Rate (beats/min)	83.1 ± 14.3	80.9 ± 16.1	0.438
Respiratory Rate (breaths/min)	18.6 ± 2.3	18.9 ± 2.4	0.601
Systolic Blood Pressure (mmHg)	166.0 ± 29.8	174.5 ± 39.7	0.195
Diastolic Blood Pressure (mmHg)	92.0 ± 16.5	95.0 ± 20.8	0.381
Body Mass Index (kg/m²)	27.0 ± 4.8	27.5 ± 4.7	0.526
Oxygen Saturation (%)	97,3 ± 2,4	96.8 ± 3.3	0.934**
Glucose (mg/dL)	163.9 ± 76.3	208.8 ± 115.6	0.031**
Urea (mg/dL)	46.3 ± 24.9	116.8 ± 66.7	< 0.001**
Creatinine (mg/dL)	1.2 ± 1.0	2.0 ± 1.4	< 0.001**
ALT (U/L)	23.1 ± 38.1	104.1 ± 283.8	< 0.001**
AST (U/L)	30.6 ± 30.0	189.2 ± 721.8	< 0.001**
Sodium (mmol/L)	137.6 ± 4.2	147.0 ± 10.8	< 0.001
Potassium (mmol/L)	4.3 ± 0.5	4.3 ± 1.0	0.799
Calcium (mg/dL)	9.1 ± 0.7	8.5 ± 2.2	0.021
CK-MB (ng/mL)	3.3 ± 3.4	3.5 ± 4.2	0.984**
High-Sensitivity Troponin (ng/L)	27.5 ± 31	78.7 ± 229.2	0.776**
Hemoglobin (g/dL)	12.4 ± 2	11.5 ± 2.7	0.041
Hematocrit (%)	36.8 ± 5.8	35.0 ± 8.3	0.180
Platelet (x10³/µL)	213.1 ± 80	159.1 ± 103.8	0.002
Monocyte (count/µL)	572.9 ± 412.9	535.4 ± 348.4	0.724**
Neutrophil (count/µL)	8797.9 ± 4484.6	13255.4 ± 7912.2	0.003**
Lymphocyte (count/µL)	1127.1 ± 630.7	853.8 ± 764.1	0.003**
Procalcitonin (ng/mL)	0.4 ± 1	6.0 ± 8.7	< 0.001**
CRP (mg/L)	21.8 ± 37.2	155.9 ± 126.8	< 0.001**
Albumin (g/L)	37.8 ± 5.3	25.5 ± 6.5	< 0.001
Total Protein (g/L)	66.3 ± 9.2	53.8 ± 8.4	< 0.001
Cholesterol (mg/dL)	189.1 ± 37.9	188.7 ± 50.1	0.959

dl Deciliter, g Gram, mg Milligram, mmHg Millimeters of mercury, L Liter, mEq Milliequivalent, mL Millilitermm, mm Millimeter, mmol Millimoles, min Minute, ng Nanogram, U Unit, µL Microliter

*t-test was used. *Mann-Whitney U test was used

For nutritional risk scores, GNRI was significantly lower in deceased patients (p < 0.001), whereas NRS-2002, APACHE-II, and CONUT scores were significantly higher (p = 0.002, p < 0.001, and p < 0.001, respectively; Table 4). In a logistic regression model including sex, age, GNRI, NRS-2002, APACHE-II, and CONUT, each one-unit increase in APACHE-II increased mortality risk by 1.19 times (95% CI: 1.08–1.32), and each one-unit increase in CONUT increased the risk by 2.96 times (95% CI: 1.73–5.04) (p < 0.001 for both; Table 5).

Table 4.

Association of nutritional risk scores with mortality

Nutritional Risk Score (n = 113)	Survived (n = 48) Mean (± SS)	Deceased (n = 65) Mean (± SS)	p* value
GNRI	107.4 ± 11.9	90.1 ± 13.8	< 0.001
NRS2002	2.9 ± 1.7	3.7 ± 1.1	0.002
APACHE-II	21.4 ± 9.5	34.5 ± 8.6	< 0.001
CONUT	2.9 ± 1.9	7.2 ± 2.2	< 0.001

APACHE II Acute Physiology and Chronic Health Evaluation Score, CONUT Controlling Nutritional Status, GNRI Geriatric Nutritional Risk Index, NRS-2002 Nutritional Risk Score

*t test was used

Table 5.

Effect of nutritional risk scores on mortality

Characteristic (n = 113)	OR	p	%95 CI
Gender
Male	1.00
Female	1.08	0.918	0.26–4.47
Age	1.04	0.502	0.93–1.16
GNRI	1.03	0.442	0.96–1.09
NRS2002	0.86	0.571	0.50–1.46
APACHE-II	1.19	< 0.001	1.08–1.32
CONUT	2.96	< 0.001	1.73–5.04

Constant: B= -15.013 p = 0.043 Exp(B) < 0.001 Nagelkerke R Square = 0.78

APACHE II Acute Physiology and Chronic Health Evaluation Score, CONUT Controlling Nutritional Status, GNRI Geriatric Nutritional Risk Index, NRS-2002 Nutritional Risk Score, OR Odds Ratio

ROC curve analysis demonstrated that the CONUT score had excellent predictive ability for mortality (AUC = 0.92, 95% CI: 0.87–0.96), while GNRI (AUC = 0.83, 95% CI: 0.76–0.91) and APACHE-II (AUC = 0.84, 95% CI: 0.77–0.91) showed very good predictive power. The NRS-2002 score showed weaker predictive ability (AUC = 0.67, 95% CI: 0.56–0.78; Fig. 2).

Fig. 2.

Receiver operating characteristic (ROC) curves for nutritional risk scores

Kaplan–Meier analysis revealed a mean survival time of 69.2 ± 16.3 days (95% CI: 37.1–101.2), with an overall survival rate of 42.5% (Fig. 3). No significant differences in survival were observed with respect to sex, diabetes mellitus, hypertension, coronary artery disease, malignancy, chronic kidney disease, or history of stroke (p > 0.05 for all).

Fig. 3.

Cumulative survival outcomes of the study cohort

Discussion

Malnutrition is associated with increased morbidity and mortality, and ASPEN emphasizes the importance of early nutritional risk assessment in critically ill patients [15–18]. In line with these recommendations, objective nutritional risk indices, such as the CONUT score and the GNRI, are increasingly used for early prognostic evaluation across different clinical settings [13, 19]. In this context, our study evaluated the prognostic performance of commonly used nutritional risk scores for predicting mortality in critically ill patients with HCVD. Our findings demonstrated that the CONUT score exhibited excellent discriminative performance for mortality prediction, followed by the GNRI and APACHE-II scores with very good performance, whereas the NRS-2002 score showed limited predictive ability. These results indicate that the CONUT score may provide greater prognostic utility than other nutritional assessment tools in this specific patient population.

In critically ill elderly patients with multi-organ dysfunction, prognostic evaluation is commonly based on severity-of-illness scores such as APACHE-II or the Sequential Organ Failure Assessment (SOFA) [20]. However, these scores do not directly account for baseline nutritional status or physiological reserve, which are particularly important determinants of outcomes in geriatric populations. In the present study, APACHE-II demonstrated very good performance in predicting mortality, confirming its established prognostic value in the intensive care setting. Our findings also showed that nutritional risk scores, particularly GNRI and CONUT, provided substantial prognostic information. However, it should be noted that serum albumin, which is used in the GNRI calculation, may decrease during critical illness due to inflammation rather than malnutrition. Therefore, careful interpretation is required when using GNRI in this context.Taken together, these results suggest that combining traditional severity scores such as APACHE-II with nutritional risk assessments may offer a more comprehensive prognostic evaluation in elderly patients with HCVD.

There are a limited number of studies comparing scoring systems as predictors of mortality in HCVD patients [9]. Zhu et al. conducted a study involving 349 patients with acute hemorrhagic stroke and found that the CONUT score successfully and independently predicted poor three-month outcomes. However, it did not show clear superiority over the NRS-2002 score. They defined a high CONUT score threshold of ≥ 2 for malnutrition risk, and an mRS score of ≥ 3 was used as the outcome parameter [21]. Xin Huang et al. reported that a higher CONUT score was associated with a 61% increased risk of all-cause mortality in patients with decompensated heart failure [22]. Similarly, Yuan et al. found a significant association between CONUT score and the risk of hemorrhagic transformation in patients with acute ischemic stroke [23]. Hu et al. also demonstrated a significant relationship between malnutrition and poor clinical outcomes for all three scores: CONUT, GNRI, and NRS-2002 [17].

Our study highlights the clinical utility of the CONUT score, which demonstrated superior performance compared with other nutritional risk indices in predicting in-hospital mortality among critically ill elderly patients with HCVD. In our study, the CONUT score was analyzed as a continuous variable to capture the full spectrum of nutritional risk and to avoid potential information loss associated with categorical cut-offs. This approach allowed a more precise assessment of the relationship between nutritional status and mortality in the geriatric ICU population.

CONUT reflects both nutritional and immunological status through multiple objective biochemical parameters, including serum albumin, total cholesterol, and lymphocyte count. In contrast, GNRI primarily relies on serum albumin and body weight measurements, which may be less sensitive to acute metabolic and inflammatory changes in critically ill patients. The strong association between poor nutritional status and increased mortality observed in our cohort is particularly relevant in HCVD patients who experience hypermetabolic and hypercatabolic states that accelerate nutritional deterioration [24]. Elevated inflammatory markers and decreased lymphocyte counts in non-survivors further support the role of systemic inflammation in adverse outcomes. By incorporating lymphocyte count, the CONUT score captures an important aspect of the immunoinflammatory response, which may partly explain its superior prognostic performance compared with other nutritional risk indices. Taken together, these findings suggest that the CONUT score provides a more comprehensive prognostic evaluation than scores that do not include immunological parameters in critically ill elderly patients.

Limitations

As a retrospective cohort study, causality cannot be established, and findings may be affected by unmeasured confounders and missing data. The study was conducted at a single center, and patients with incomplete records were excluded, limiting generalizability. Nutritional risk scores were assessed only at admission, which did not allow evaluation of dynamic changes over time. The patient population included multiple factors that acted as potential confounders, such as advanced age, edentulism, facial muscle weakness, impaired tongue mobility, aphasia, and other stroke-related disabilities. Additionally, the use of anti-lipidemic medications may have influenced cholesterol levels and nutritional scores, while lymphocyte counts could have been affected by critical illness. The study period overlapped with the COVID-19 pandemic, and patient COVID-19 status was not included. Individual anthropometric measures and indicators of sarcopenia, such as low muscle mass, were unavailable, preventing a more comprehensive nutritional assessment.

Selection bias should also be considered, as only critically ill HCVD patients admitted to the ICU were included, which may have strengthened the observed associations and limits generalizability to less severe cases.

Conclusion

Our study provided a comparative evaluation of three nutritional scoring systems with respect to their ability to predict mortality in HCVD patients. The results indicated that the CONUT score exhibited superior predictive performance for mortality compared with the other scores, suggesting that it could serve as a simple and practical assessment tool in clinical practice.

Abbreviations

APACHE II

Acute physiology and chronic health evaluation score

ASPEN

American society for parenteral and enteral nutrition

BMI

Body mass index

CAD

Coronary artery disease

CKD

Chronic kidney disease

CONUT

Controlling nutritional status

CT

Computed tomography

DM

Diabetes mellitus

ED

Emergency department

GNRI

Geriatric nutritional risk index

HCVD

Hemorrhagic cerebrovascular disease

ICU

Intensive care unit

NRS-20002

Nutritional risk score 2002

ROC

Receiver operating characteristic

Authors’ contributions

Ahmet Burak Urfalioglu: Concept, Methodology, Investigation, Data Curation, Formal Analysis, Writing – Original Draft, Writing – Review & Editing, Visualization, Supervision. Erdem Aksay: Concept, Methodology, Investigation, Data Curation, Writing – Original Draft, Writing – Review & Editing, Visualization, Supervision. Yeliz Simsek: Concept, Methodology, Investigation, Data Curation, Writing – Original Draft, Writing – Review & Editing, Visualization, Supervision. Onder Yesiloglu: Concept, Methodology, Investigation, Formal Analysis, Writing – Original Draft. Mustafa Oguz Tuscan: Methodology, Data Curation, Formal Analysis, Writing – Original Draft, Writing – Review & Editing, Visualization, Supervision. Saliha Dilek: Methodology, Data Curation, Formal Analysis, Writing – Original Draft, Writing – Review & Editing, Visualization, Supervision. Behum Seyda Avci: Methodology, Investigation, Data Curation, Writing – Original Draft, Writing – Review & Editing, Visualization, Supervision. Adnan Kuvvetli: Methodology, Investigation, Data Curation, Writing – Original Draft, Writing – Review & Editing, Visualization, Supervision. Akkan Avci: Concept, Methodology, Investigation, Data Curation, Formal Analysis, Writing – Original Draft, Writing – Review & Editing, Visualization, Supervision.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

Data and materials are reachable from hospital automation information systems.

Declarations

Ethics approval and consent to participate

Ethics approval was obtained from the Clinical Research Ethics Committee of the University of Health Sciences, Adana City Training and Research Hospital (Date:10/10/2024, Meeting Number: 6, Decision Number: 183). This manuscript was carried out in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. Written informed consent waived by the ethics committee.

Consent for publication

Not applicaple.

Competing interests

The authors declare no competing interests.