Association of the Endothelial Activation and Stress Index (EASIX) with short-term mortality in critically ill patients with congestive heart failure: A retrospective cohort study from the MIMIC database

Abstract

The role of endothelial injury in worsening congestive heart failure (CHF) remains unquantified. This study evaluates the Endothelial Activation and Stress Index (EASIX) for predicting short-term mortality in patients with critical CHF. This was a retrospective cohort study using the Medical Information Mart for Intensive Care-IV (2008–2022). Adults with CHF admitted to the intensive care unit were stratified by ln(EASIX) quartiles. The primary endpoint was 30-day all-cause mortality. Multivariable Cox regression, restricted cubic splines, and subgroup analyses were performed. Among 4556 patients (median age 72.1 years, 42.8% female), the highest EASIX quartile (Q4) had a 44.2% 30-day mortality rate versus 19.2% in Q1 (adjusted hazard ratio [HR] = 1.6, 95% confidence interval [CI]: 1.27–2.02, P < .001). A nonlinear association was observed (nonlinearity P = .008) with an inflection point at ln(EASIX) = 0.05. Beyond this threshold, each unit increase in ln(EASIX) conferred a 14.2% higher mortality risk (HR = 1.142, 95% CI: 1.062–1.227). Ln(EASIX) remained predictive after full adjustment for severity scores and treatments, with the highest quartile (Q4) exhibiting a 60% increased mortality risk (adjusted HR = 1.60, 95% CI: 1.27–2.02). EASIX is a robust predictor of short-term mortality in patients with critical CHF, particularly valuable in nonsepsis populations. Its simple calculation (lactate dehydrogenase/creatinine/platelets) that refines risk stratification beyond conventional severity scores.

1. Introduction

Despite a decline in overall cardiovascular disease mortality, the prevalence and burden of congestive heart failure (CHF) continue to rise.^[1] Heart failure constitutes a primary global health challenge, with its prevalence projected to increase by nearly 50% over the next 15 years.^[2] Despite available effective therapies, the 5-year mortality rate remains as high as 50%.^[2] Accounting for 1% to 2% of all hospital admissions,^[3] heart failure imposes a substantial clinical burden. Hospitalizations due to acute decompensation are strongly associated with reduced life expectancy and impaired quality of life.^[4] Severe complications may occur in CHF, markedly elevating mortality risk in critically ill patients.^[1]

The pathophysiology of CHF involves complex interactions, including endothelial dysfunction. In CHF, endothelial injury promotes a pro-inflammatory state, which impairs systemic endothelial function. This dysfunction is exacerbated by comorbid renal impairment, which reduces the clearance of inflammatory cytokines, thereby amplifying the inflammatory response.^[5] Biomarkers of endothelial activation and dysfunction have been linked to poor outcomes in conditions such as cancer^[6,7] and coronary artery disease.^[8] However, the prognostic utility of an integrated endothelial biomarker in CHF has not been established.

The Endothelial Activation and Stress Index (EASIX) – a novel metric derived from lactate dehydrogenase (LDH), creatinine (Cr), and platelet (PLT) counts – has recently been utilized to assess endothelial integrity. This composite biomarker reflects endothelial damage, coagulation abnormalities, and metabolic dysregulation, potentially offering a superior prognostic assessment compared with single biomarkers. EASIX has shown promise in predicting outcomes, including cardiovascular mortality.^[9]

Therefore, this study aims to investigate the association between EASIX and mortality in critically ill patients with CHF admitted to the intensive care unit (ICU) and to identify prognostic predictors. We hypothesize that EASIX is independently associated with increased mortality risk in this patient population after adjusting for potential confounders. The findings may inform the development of targeted clinical interventions.

2. Materials and methods

2.1. Database

This study utilized the Medical Information Mart for Intensive Care-IV database (MIMIC-IV v3.1), a publicly available repository containing clinical records of >90,000 adult ICU patients.^[10] The database encompasses comprehensive clinical variables, including survival status, vital signs, laboratory parameters, diagnoses, and therapeutic interventions. Given its high-quality data and additional advantages, MIMIC-IV has been increasingly adopted as a research platform.^[11,12] Researchers Xiao Dong (Certification ID: 63109170) and Yingxiu Huang (Certification ID: 56513391) were granted access after completing the required training modules. The Institutional Review Board of Beth Israel Deaconess Medical Center approved this study as it involved the secondary analysis of deidentified data, waiving the need for patient consent and additional ethical review. This cohort study adheres to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.^[13]

2.2. Study population

We included patients diagnosed with CHF during hospitalization from 2008 to 2022. The diagnostic criteria were based on the International Classification of Diseases (ICD), including ICD-9 and ICD-10, with the corresponding codes: 4280, I5032, I5033, I5022, I5023, I5030, I5021, I5020, I5031, I5043, I5042, 39891, I5041, and I5040. Only the first ICU admission record was retained. Finally, the analysis was restricted to individuals with LDH, Cr, and PLT count measurements within 24 hours of ICU admission. For patients with multiple measurements during this period, the worst values were recorded (defined as peak values for LDH/Cr and the nadir for PLT based on clinical significance). The flowchart is detailed in Figure 1.

Figure 1.

Flowchart of the study. ICU = intensive care unit, MIMIC IV = Medical Information Mart for Intensive Care IV.

2.3. Exposure variable

The EASIX was calculated using the following formula proposed by Luft^[14]:

$\mathrm{\backslash newcommand}\mathrm{\backslash e}\mathrm{e}{\displaystyle \mathrm{E}\mathrm{A}\mathrm{S}\mathrm{I}\mathrm{X}=\left(\mathrm{l}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{d}\mathrm{e}\mathrm{h}\mathrm{y}\mathrm{d}\mathrm{r}\mathrm{o}\mathrm{g}\mathrm{e}\mathrm{n}\mathrm{a}\mathrm{s}\mathrm{e}[\mathrm{L}\mathrm{D}\mathrm{H},\mathrm{U}/\mathrm{L}]\times \mathrm{c}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{e}\left[\mathrm{m}\mathrm{g}/\mathrm{d}\mathrm{L}\right]\right)/\mathrm{p}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}(\times {10}^9/\mathrm{L}).}$

Due to a right-skewed distribution, the EASIX values were natural log-transformed to better meet the assumptions of parametric statistical tests. This log-transformed variable (ln[EASIX]) was used in all subsequent analyses.

2.4. Outcomes

The primary outcome was 30-day all-cause mortality following ICU admission. The secondary outcome was 90-day all-cause mortality.

2.5. Covariates

Data were extracted from the MIMIC-IV database using Structured Query Language (SQL) and stored in a PostgreSQL (University of California, Berkeley) database system. Core variables were defined according to established prognostic studies of EASIX. Extracted data included the following: Demographics: age, sex, and ethnicity. Vital signs: heart rate, mean arterial pressure, respiratory rate, oxygen saturation, and temperature. Comorbidities (coded by ICD): myocardial infarction, chronic pulmonary disease, diabetes with complications, chronic kidney disease, cerebrovascular disease, malignancy, AIDS, and sepsis. Laboratory parameters (worst values within 24 hours of ICU admission): white blood cell count, hemoglobin, anion gap, glucose, blood urea nitrogen, Cr, calcium, sodium, potassium, and lactate. Disease severity scores: Charlson Comorbidity Index, Acute Physiology and Chronic Health Evaluation III (APACHE III), and Sequential Organ Failure Assessment (SOFA). Treatments: beta-blockers, vasopressors, mechanical circulatory support, and dialysis. Sepsis-3 refers to organ dysfunction that endangers life due to a disordered host response to infection.^[15]

2.6. Statistical analysis

Observations with missing values were excluded before statistical analysis. Patients were stratified by quartiles (Q1–Q4) of ln(EASIX). Categorical variables are presented as frequencies (percentages). Normally distributed continuous variables are expressed as mean ± standard deviation, while non-normally distributed variables are reported as median (interquartile range). Intergroup comparisons utilized the following: chi-square test for categorical variables, one-way ANOVA for normally distributed continuous variables, Kruskal–Wallis H test for non-normally distributed continuous variables.

To investigate the association between EASIX and mortality, univariate and multivariate logistic regression models were sequentially constructed. In accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines, an unadjusted model (model 1) was initially developed, followed by stepwise covariate adjustment until the final model (model 4). Natural cubic spline regression of model 4 was employed to evaluate the independent dose-response relationship between EASIX and 30-day mortality. A 2-piece linear regression model with smooth curve fitting identified potential thresholds in the EASIX-mortality association, with inflection points validated by likelihood ratio tests and bootstrap resampling (1000 iterations). Stratified subgroup analyses assessed potential effect modification by age (<65 vs ≥65 years), sex, sepsis, malignancy, chronic kidney disease, and diabetes with complications. Kaplan–Meier curves with log-rank tests compared survival probabilities across EASIX quartiles.

All analyses were performed using R (version 4.4.2; R Foundation for Statistical Computing, Vienna, Austria) and Free Statistics software (version 2.1; FreeClinical Medical Technology Co., Ltd., Beijing, China). Statistical significance was considered when the 2-sided P value was <.05.

3. Results

3.1. Baseline characteristics

Table 1 summarizes baseline characteristics stratified by ln(EASIX) quartiles. The overall cohort had a median age of 72.1 years, and 42.8% were female. Both 30- and 90-day mortality rates were significantly higher in the highest EASIX quartile (Q4) compared with the lowest (Q1; 30-day: 44.2% vs 29.7%, P < .001; 90-day: 47.1% vs 32.8%, P < .001). Relative to lower quartiles, patients in Q4 had a higher proportion of males and a greater prevalence of myocardial infarction, diabetes with complications, and chronic kidney disease. Physiological derangements in Q4 included elevated heart rate, white blood cell count, anion gap, lactate, Cr, and blood urea nitrogen (all P < .05). Higher disease severity scores (SOFA and APACHE III) were observed in Q4, whereas mean arterial pressure was significantly lower.

Table 1.

Baseline characteristics and outcomes of the study population.

Category/variable	Total (n = 4556)	Q1 (n = 1139)	Q2 (n = 1139)	Q3 (n = 1139)	Q4 (n = 1139)	P value
Demographics
Age, yr	72.1 ± 13.9	71.6 ± 14.0	73.6 ± 14.0	72.8 ± 14.1	70.2 ± 13.4	<.001
Sex, n (%)						<.001
Female	1949 (42.8)	632 (55.5)	514 (45.1)	438 (38.5)	365 (32.0)
Race (White), n (%)	2924 (64.2)	782 (68.7)	768 (67.4)	721 (63.3)	653 (57.3)	<.001
Vital signs
Heart rate, bpm	86.3 ± 17.6	87.2 ± 17.3	85.0 ± 17.6	85.4 ± 16.9	87.6 ± 18.2	<.001
MAP, mm Hg	77.3 ± 11.2	78.6 ± 11.5	77.4 ± 10.7	77.0 ± 10.8	76.2 ± 11.5	<.001
Respiratory rate, /min	20.5 ± 3.9	20.5 ± 3.9	20.2 ± 3.6	20.4 ± 3.9	21.0 ± 4.2	<.001
Temperature, °C	36.8 ± 0.6	36.8 ± 0.4	36.8 ± 0.5	36.8 ± 0.6	36.7 ± 0.7	.001
SpO2, %	96.4 ± 2.5	96.3 ± 2.0	96.4 ± 2.0	96.5 ± 2.3	96.3 ± 3.3	.141
Laboratory results
WBC, ×103/μL,	12.9 (9.0–18.2)	12.1 (8.6–16.4)	12.1 (8.8–17.1)	13.1 (9.1–18.2)	14.1 (9.7–20.7)	<.001
Hemoglobin, g/dL	9.9 ± 2.4	10.3 ± 2.2	10.1 ± 2.3	9.8 ± 2.4	9.3 ± 2.4	<.001
Anion gap, mEq/L	17.6 ± 5.3	15.2 ± 3.5	16.4 ± 4.3	17.7 ± 4.4	21.2 ± 6.3	<.001
BUN	34.0 (21.0–53.0)	21.0 (15.0–29.0)	30.0 (20.0–43.0)	41.0 (27.0–61.0)	52.0 (36.0–78.0)	<.001
Calcium	8.1 ± 0.9	8.3 ± 0.8	8.2 ± 0.8	8.1 ± 0.9	7.9 ± 0.9	<.001
Creatinine	1.5 (1.0–2.5)	1.0 (0.8–1.2)	1.3 (1.0–1.8)	1.9 (1.4–2.8)	2.9 (1.9–4.7)	<.001
Sodium	136.0 ± 5.7	136.1 ± 5.6	136.6 ± 5.5	136.0 ± 5.5	135.2 ± 6.0	<.001
Potassium	4.8 ± 1.0	4.5 ± 0.8	4.7 ± 0.9	4.9 ± 0.9	5.2 ± 1.1	<.001
Lactate	2.2 (1.4–4.0)	1.6 (1.2–2.4)	2.0 (1.4–3.2)	2.3 (1.5–3.9)	3.4 (1.8–7.4)	<.001
Comorbidities, n (%)
Myocardial infarction	1645 (36.1)	300 (26.3)	394 (34.6)	442 (38.8)	509 (44.7)	<.001
Chronic pulmonary disease	1552 (34.1)	438 (38.5)	407 (35.7)	378 (33.2)	329 (28.9)	<.001
Renal disease	1881 (41.3)	193 (16.9)	433 (38.0)	599 (52.6)	656 (57.6)	<.001
Cerebrovascular disease	590 (12.9)	171 (15)	148 (13)	142 (12.5)	129 (11.3)	.066
Diabetes with chronic complications	933 (20.5)	132 (11.6)	198 (17.4)	284 (24.9)	319 (28)	<.001
Malignancy	574 (12.6)	143 (12.6)	113 (9.9)	148 (13)	170 (14.9)	.004
AIDS, n (%)	22 (0.5)	5 (0.4)	4 (0.4)	5 (0.4)	8 (0.7)	.649
Sepsis-3	2808 (61.6)	547 (48)	656 (57.6)	761 (66.8)	844 (74.1)	<.001
Scores
Charlson Index	7.7 ± 2.7	6.9 ± 2.5	7.4 ± 2.5	8.1 ± 2.8	8.2 ± 2.7	<.001
APACHE III	58.1 ± 25.1	46.3 ± 19.6	51.8 ± 20.8	60.9 ± 24.0	73.4 ± 26.8	<.001
SOFA	5.7 ± 3.6	3.2 ± 2.4	4.6 ± 2.6	6.3 ± 3.0	8.6 ± 3.6	<.001
Outcomes
30-d mortality	1354 (29.7)	219 (19.2)	280 (24.6)	351 (30.8)	504 (44.2)	<.001
90-d mortality	1493 (32.8)	241 (21.2)	322 (28.3)	393 (34.5)	537 (47.1)	<.001

AIDS = acquired immunodeficiency syndrome, APACHE III = Acute Physiology and Chronic Health Evaluation III, BUN = blood urea nitrogen, MAP = mean arterial pressure, SOFA = Sequential Organ Failure Assessment, SpO₂ = percutaneous oxygen saturation, WBC = white blood cell count.

3.2. The relationship of EASIX with mortality

Table S1, Supplemental Digital Content, https://links.lww.com/MD/R510 presents the univariate Cox regression analysis of risk factors for mortality. After adjusting for potential confounders, sequential models evaluated the independent impact of ln(EASIX) on 30-day mortality in patients with CHF (Table 2). The univariate model (model 1) demonstrated a significantly elevated mortality risk in the highest ln(EASIX) quartile (Q4) versus the lowest (Q1; hazard ratio [HR] = 3.17, 95% confidence interval [CI] 2.65–3.78, P < .001). After adjusting for demographics and vital signs (model 2), the highest ln(EASIX) quartile (Q4) remained independently associated with mortality (HR = 3.10, 95% CI: 2.58–3.72, P < .001). In the fully adjusted model (model 4, including disease severity scores and treatments), this association persisted (HR = 1.6, 95% CI: 1.27–2.02, P < .001). A consistent pattern was observed for 90-day mortality, with each unit increase in ln(EASIX) conferring a 15% higher risk (HR = 1.15, 95% CI: 1.09–1.21, P < .001; model 4). Kaplan–Meier curves (Fig. 2) confirmed a dose-dependent increase in cumulative mortality with higher ln(EASIX) levels (log-rank P < .001). We extended the analysis to 365-day all-cause mortality, which confirmed the consistency of the results, as detailed in Table S2, Supplemental Digital Content, https://links.lww.com/MD/R510.

Table 2.

Risk of 30- and 90-day mortality according to ln(EASIX).

Variable	Model 1		Model 2		Model 3		Model 4
	HR (95% CI)	P	HR (95% CI)	P	HR (95% CI)	P	HR (95% CI)	P
30-d mortality
ln(EASIX)	1.34 (1.29–1.39)	<.001	1.36 (1.30–1.41)	<.001	1.23 (1.17–1.30)	<.001	1.21 (1.05–1.18)	<.001
ln(EASIX) quartiles
Q1 (reference)	1.00 (Ref)	–	1.00 (Ref)	–	1.00 (Ref)	–	1.00 (Ref)	–
Q2	1.36 (1.11–1.66)	.002	1.29 (1.05–1.57)	.014	1.16 (0.94–1.42)	.164	1.08 (0.88–1.33)	.461
Q3	1.85 (1.53–2.24)	<.001	1.78 (1.47–2.15)	<.001	1.37 (1.12–1.69)	.002	117 (0.95–1.44)	.149
Q4	3.17 (2.65–3.78)	<.001	3.10 (2.58–3.72)	<.001	2.07 (1.67–2.58)	<.001	1.53 (1.21–1.92)	<.001
P for trend		<.001		<.001		<.001		<.001
90-d mortality
ln(EASIX)	1.3 (1.26–1.35)	<.001	1.33 (1.28–1.38)	<.001	1.22 (1.16–1.28)	<.001	1.13 (1.08–1.19)	<.001
ln(EASIX) quartiles
Q1 (reference)	1.00 (Ref)		1.00 (Ref)		1.00 (Ref)		1.00 (Ref)
Q2	1.4 (1.18–1.65)	<.001	1.32 (1.11–1.56)	.001	1.43 (1.2–1.7)	<.001	1.17 (0.98–1.39)	.074
Q3	1.79 (1.53–2.1)	<.001	1.74 (1.48–2.05)	<.001	2.1 (1.74–2.53)	<.001	1.24 (1.04–1.49)	.016
Q4	2.83 (2.43–3.29)	<.001	2.83 (2.42–3.31)	<.001	1.27 (1.2–1.35)	<.001	1.61 (1.32–1.96)	<.001
P for trend	1.41 (1.34–1.48)	<.001	1.42 (1.35–1.49)	<.001	1.24 (1.04–1.47)	.015	1.16 (1.09–1.24)	<.001

Model 1: unadjusted.

Model 2: adjusted for age, sex, race, heart rate, mean blood pressure (MBP), temperature, respiratory rate (RR), and SpO₂.

Model 3: further adjusted for laboratory parameters (WBC, hemoglobin, glucose, anion gap, BUN, calcium, creatinine, sodium, potassium) and comorbidities (Charlson Comorbidity Index, cerebrovascular disease, diabetes with chronic complications, malignancy, severe liver disease, renal disease, and Sepsis-3).

Model 4: fully adjusted for severity scores (SOFA and APACHE III).

Quartile definition: Q1–Q4 represent increasing quartiles of ln(EASIX), with Q4 indicating the highest-risk group.

Statistical conventions: HRs and 95% CIs rounded to 2 decimal places; P values reported as “<.001” if below .001. Trend analysis was performed using the linear-by-linear association test.

APACHE III = Acute Physiology and Chronic Health Evaluation III, BUN = blood urea nitrogen, CI = confidence interval, HR = hazard ratio, ln(EASIX) = natural log-transformed Endothelial Activation and Stress Index, SOFA = Sequential Organ Failure Assessment, SpO₂ = percutaneous oxygen saturation, WBC = white blood cell count.

Figure 2.

Kaplan–Meier survival curves stratified by ln(EASIX) quartiles in patients with congestive heart failure (CHF) admitted to the intensive care unit. Patients were categorized into 4 groups (Q1–Q4) according to ln(EASIX) quartiles, with Q1 representing the lowest and Q4 the highest values. Survival probability was significantly different among the 4 groups as determined by the log-rank test (P < .0001). The number of patients at risk at each time point (0, 10, 20, and 30 days) is shown in the corresponding table below the figure. ln(EASIX) = natural log-transformed Endothelial Activation and Stress Index.

3.3. Restricted cubic spline and threshold effect analysis

Multivariate Cox proportional hazards regression with smooth curve fitting revealed a nonlinear relationship between ln(EASIX) and 30-day mortality (Fig. 3). A significant nonlinearity was detected (P = .008, Table 3), prompting the use of a 2-piece linear regression model. An inflection point was identified at ln(EASIX) = 0.5 (Fig. 3). To the left of the threshold, the HR was 1.114 (95% CI: 0.981–1.591, P = .553), indicating no significant association. To the right of the threshold, each unit increase in ln(EASIX) significantly elevated mortality risk (HR = 1.142, 95% CI: 1.062–1.227, P < .001). This translates to a 14.2% increased mortality risk per unit ln(EASIX) increment above 0.5.

Figure 3.

Dose-response relationship between ln(EASIX) and 30-day mortality in critically ill patients with congestive heart failure. Adjusted for age, sex, race, heart rate (HR), mean blood pressure (MBP), temperature, respiratory rate (RR), SpO₂, WBC, hemoglobin, glucose, anion gap, BUN, calcium, creatinine, sodium, potassium, Charlson Comorbidity Index, cerebrovascular disease, diabetes with chronic complications, malignancy, severe liver disease, renal disease, Sepsis-3, SOFA, and APACHE III. Only 99.5% of the data is shown. APACHE III = Acute Physiology and Chronic Health Evaluation III, BUN = blood urea nitrogen, ln(EASIX) = natural log-transformed Endothelial Activation and Stress Index, SOFA = Sequential Organ Failure Assessment, SpO₂ = percutaneous oxygen saturation, WBC = white blood cell count.

Table 3.

Threshold effect analysis of ln(EASIX) on 30-day mortality.

Threshold of ln(EASIX)	HR	95% CI	P value
<0.05	1.114	0.78–1.5919	.553
≥0.05	1.142	1.062–1.227	<.001
Nonlinear test			.008

CI = confidence interval, HR = hazard ratio, ln(EASIX) = natural log-transformed Endothelial Activation and Stress Index.

3.4. Subgroup analysis

Prespecified subgroup analyses stratified by age (<65 or ≥65 years), sex, sepsis, malignancy, chronic kidney disease, and diabetes with complications were presented in Figure 4. Significant associations between elevated EASIX and 30-day mortality were observed in elderly patients (≥65 years) and those with comorbidities (diabetes, malignancy, and chronic kidney disease; all P < .05), with no significant interaction effects (P-interaction > .05). The interaction analysis revealed that sepsis had an interaction with ln(EASIX) in predicting mortality (P < .05).

Figure 4.

Forest plot of subgroup analyses for the association between ln(EASIX) and 30-day mortality in critically ill patients with congestive heart failure (CHF). This figure presents hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between ln(EASIX) and 30-day mortality across various patient subgroups. The HR (95% CI) represents the risk increase per unit increase in ln(EASIX). A common reference line is set at HR = 1.0. ln(EASIX) = natural log-transformed Endothelial Activation and Stress Index.

4. Discussion

This study provides the first comprehensive evidence of an independent positive association between ln(EASIX) and short-term mortality in critically ill CHF patients. After multivariable adjustment, each unit increase in ln(EASIX) conferred a 14.2% and 13% higher risk of 30- and 90-day all-cause mortality, respectively. The consistent directionality across clinical subgroups and the identified nonlinear dose-response relationship (with accelerated risk above ln(EASIX) > 0.5) reinforce EASIX as a robust prognostic biomarker. These findings position EASIX as a novel risk-stratification tool for CHF patients in critical care settings.

Our findings further substantiate the pervasive role of endothelial dysfunction as a central driver in diverse disease states. The EASIX, which integrates LDH, Cr, and PLT count, provides a practical surrogate marker for this systemic pathological condition. Its prognostic value has been validated across multiple clinical domains. In cardiovascular diseases, Sang et al reported that elevated EASIX predicted 30-day mortality in acute myocardial infarction patients^[16] (adjusted HR = 1.70, 95% CI: 1.17–2.46, P = .005). In critically ill patients with atrial fibrillation, EASIX predicts mortality with accuracy comparable to the SOFA score (AUC: 0.79 vs 0.77) and significantly outperforms CHA₂DS₂-VASc (AUC: 0.62).^[17] In hypertensive populations, EASIX independently predicts all-cause mortality (adjusted HR = 1.82, 95% CI: 1.35–2.45).^[9] In infectious diseases, elevated EASIX levels were associated with increased 28-day mortality risk in ICU patients with sepsis (HR = 1.10; 95% CI: 1.07–1.13; P < .001).^[18] Similarly, a multicenter cohort study demonstrated significantly reduced survival in COVID-19 patients with high baseline EASIX scores compared with those with lower values^[19] (43% lower survival probability; adjusted HR = 2.33, 95% CI: 1.98–2.74; P < .001), further supporting EASIX as a pan-infectious prognostic biomarker. Furthermore, in hematologic malignancies, including lymphoma,^[20] acute myeloid leukemia,^[21] and multiple myeloma,^[22] elevated EASIX levels exhibit a strong positive association with disease-specific mortality. Our study crucially addresses this gap by demonstrating, for the first time, the independent prognostic value of EASIX in the specific cohort of critically ill CHF patients.

The association between elevated EASIX and mortality in CHF is primarily mediated by endothelial dysfunction. First, EASIX directly quantifies endothelial activation and damage through its components: LDH, Cr, and PLT counts. As a cornerstone of cardiovascular pathology, endothelial dysfunction promotes increased vascular permeability, a hypercoagulable state, and amplified systemic inflammation.^[9,16,23] Supporting this, elevated EASIX predicts mortality in coronary artery disease (adjusted HR = 1.63–4.65),^[23] suggesting that endothelial injury may exacerbate acute heart failure decompensation by accelerating atherosclerotic progression and plaque destabilization. Second, systemic inflammation and oxidative stress are key contributors. An increase in LDH indicates tissue ischemia or cell lysis, acting as a biomarker for inflammation and oxidative stress,^[24,25] and it can independently predict sepsis mortality.^[26] Thrombocytopenia is associated with consumptive coagulopathy,^[27,28] creating a positive feedback loop of “endothelium-platelet-inflammation.” Third, microcirculatory dysfunction and hypoperfusion play a critical role. Elevated Cr reflects glomerular endothelial damage. Endothelium-dependent microcirculatory failure results in a decreased glomerular filtration rate. Disruption of podocyte tight junctions reduces the effective filtration area. Fluid retention increases cardiac preload, thereby exacerbating heart failure.^[29] Finally, synergistic multi-organ endothelial injury is a key factor. EASIX incorporates hepatic (LDH), renal (Cr), and hematologic (PLT) parameters. Studies on cardiac surgery confirm that a high preoperative EASIX predicts postoperative mortality (OR: 3.12, 95% CI: 2.18–4.46), showing that the collapse of the “endothelium-organaxis” can trigger multiple organ dysfunction syndrome.^[30]

This study establishes that ascending ln(EASIX) quartiles are strongly associated with reduced survival in CHF patients, with the highest quartile (Q4) exhibiting a 38.8% 30-day mortality rate – confirming EASIX as an independent prognostic biomarker for early high-risk identification. Notably, EASIX retained significant predictive capacity after adjusting for SOFA and APACHE III scores (adjusted HR = 1.53, 95% CI: 1.32–1.77), underscoring endothelial activation as a risk factor beyond conventional severity metrics. Clinically, EASIX offers practical advantages through its simple calculation using 3 routinely measured parameters: LDH, Cr, and PLT count. Biologically, these markers reflect endothelial pathophysiology linked to inflammation, oxidative stress, and microthrombosis, positioning EASIX as a direct indicator of endothelial dysfunction in critical CHF. As depicted in Figure 2 (Kaplan–Meier curves), patients with elevated ln(EASIX) showed rapid survival decline within 10 days, signaling impending acute organ failure or septic shock that warrants prioritized intensive monitoring and multiorgan support. Subgroup analysis revealed a significant interaction effect in sepsis patients (P-interaction < .001). A key and novel insight from our analysis is the modifying effect of sepsis.

This study elucidates the clinical significance of ln(EASIX) as a novel predictor of short-term mortality risk in patients with CHF. By providing granular insights into risk stratification, ln(EASIX) enables clinicians to rapidly identify high-risk patients for intensified therapeutic interventions. Critically, we demonstrate that conventional severity scores (e.g., SOFA and APACHE III) may fail to capture endothelial-specific injury, whereas EASIX serves as an independent prognostic factor that addresses this limitation. However, several limitations of our single-center retrospective study must be acknowledged. First, similar to all retrospective studies, there may be residual confounding factors despite multivariable adjustment. Second, the measured values of LDH, Cr, and PLT counts were only obtained within the first 24 hours after admission. This restriction prevents us from assessing the dynamic changes in EASIX levels following ICU admission and may influence the accuracy of the results. Therefore, future prospective studies should investigate the relationship between dynamic EASIX trajectories and clinical outcomes. Third, we observed a correlation between EASIX levels and patient prognosis; however, causality remains to be determined. Fourth, the MIMIC-IV database systematically lacks patient-reported functional outcomes and quality of life assessments, which is a common constraint of such large public critical care repositories. Despite these limitations, our study utilized multiple analysis strategies, such as evaluating nonlinear relationships and conducting stratified subgroup analyses with different thresholds, to verify the robustness of this correlation. These findings highlight the need for confirmatory studies to explore potential causal mechanisms.

5. Conclusions

The study identified a significant positive correlation between elevated ln(EASIX) levels and the risks of 30- and 90-day mortality in critically ill patients with CHF. These findings suggest that ln(EASIX) may serve as a valuable prognostic marker in clinical practice. Early identification of high-risk patients using ln(EASIX) could facilitate optimized treatment approaches and reduce mortality associated patients with CHF admitted to the ICU.

Acknowledgments

We thank Dr Liu Jie (People’s Liberation Army of China General Hospital, Beijing, China) for helping in this revision.

Author contributions

Methodology: Yingxiu Huang.

Investigation: Ming Hu.

Project administration: Ming Hu.

Supervision: Peng Zhen, Xinzhen Li.

Writing – original draft: Xiao Dong.

Writing – review & editing: Xinzhen Li.